Loading...
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. */
102extern int suid_dumpable;
103#ifdef CONFIG_COREDUMP
104extern int core_uses_pid;
105extern char core_pattern[];
106extern unsigned int core_pipe_limit;
107#endif
108extern int pid_max;
109extern int pid_max_min, pid_max_max;
110extern int percpu_pagelist_fraction;
111extern int latencytop_enabled;
112extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
113#ifndef CONFIG_MMU
114extern int sysctl_nr_trim_pages;
115#endif
116
117/* Constants used for minimum and maximum */
118#ifdef CONFIG_LOCKUP_DETECTOR
119static int sixty = 60;
120#endif
121
122static int __maybe_unused neg_one = -1;
123
124static int zero;
125static int __maybe_unused one = 1;
126static int __maybe_unused two = 2;
127static int __maybe_unused four = 4;
128static unsigned long one_ul = 1;
129static int one_hundred = 100;
130static int one_thousand = 1000;
131#ifdef CONFIG_PRINTK
132static int ten_thousand = 10000;
133#endif
134#ifdef CONFIG_PERF_EVENTS
135static int six_hundred_forty_kb = 640 * 1024;
136#endif
137
138/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
139static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
140
141/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
142static int maxolduid = 65535;
143static int minolduid;
144
145static int ngroups_max = NGROUPS_MAX;
146static 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
150static 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__
160extern int pwrsw_enabled;
161#endif
162
163#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
164extern int unaligned_enabled;
165#endif
166
167#ifdef CONFIG_IA64
168extern int unaligned_dump_stack;
169#endif
170
171#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
172extern 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 */
196enum sysctl_writes_mode {
197 SYSCTL_WRITES_LEGACY = -1,
198 SYSCTL_WRITES_WARN = 0,
199 SYSCTL_WRITES_STRICT = 1,
200};
201
202static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
203
204static int proc_do_cad_pid(struct ctl_table *table, int write,
205 void __user *buffer, size_t *lenp, loff_t *ppos);
206static 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
211static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
212 void __user *buffer, size_t *lenp, loff_t *ppos);
213#endif
214
215static 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
218static int proc_dostring_coredump(struct ctl_table *table, int write,
219 void __user *buffer, size_t *lenp, loff_t *ppos);
220#endif
221static int proc_dopipe_max_size(struct ctl_table *table, int write,
222 void __user *buffer, size_t *lenp, loff_t *ppos);
223
224#ifdef CONFIG_MAGIC_SYSRQ
225/* Note: sysrq code uses it's own private copy */
226static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
227
228static int sysrq_sysctl_handler(struct ctl_table *table, int write,
229 void __user *buffer, size_t *lenp,
230 loff_t *ppos)
231{
232 int error;
233
234 error = proc_dointvec(table, write, buffer, lenp, ppos);
235 if (error)
236 return error;
237
238 if (write)
239 sysrq_toggle_support(__sysrq_enabled);
240
241 return 0;
242}
243
244#endif
245
246static struct ctl_table kern_table[];
247static struct ctl_table vm_table[];
248static struct ctl_table fs_table[];
249static struct ctl_table debug_table[];
250static struct ctl_table dev_table[];
251extern struct ctl_table random_table[];
252#ifdef CONFIG_EPOLL
253extern struct ctl_table epoll_table[];
254#endif
255
256#ifdef CONFIG_FW_LOADER_USER_HELPER
257extern struct ctl_table firmware_config_table[];
258#endif
259
260#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
261int sysctl_legacy_va_layout;
262#endif
263
264/* The default sysctl tables: */
265
266static struct ctl_table sysctl_base_table[] = {
267 {
268 .procname = "kernel",
269 .mode = 0555,
270 .child = kern_table,
271 },
272 {
273 .procname = "vm",
274 .mode = 0555,
275 .child = vm_table,
276 },
277 {
278 .procname = "fs",
279 .mode = 0555,
280 .child = fs_table,
281 },
282 {
283 .procname = "debug",
284 .mode = 0555,
285 .child = debug_table,
286 },
287 {
288 .procname = "dev",
289 .mode = 0555,
290 .child = dev_table,
291 },
292 { }
293};
294
295#ifdef CONFIG_SCHED_DEBUG
296static int min_sched_granularity_ns = 100000; /* 100 usecs */
297static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
298static int min_wakeup_granularity_ns; /* 0 usecs */
299static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
300#ifdef CONFIG_SMP
301static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
302static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
303#endif /* CONFIG_SMP */
304#endif /* CONFIG_SCHED_DEBUG */
305
306#ifdef CONFIG_COMPACTION
307static int min_extfrag_threshold;
308static int max_extfrag_threshold = 1000;
309#endif
310
311static struct ctl_table kern_table[] = {
312 {
313 .procname = "sched_child_runs_first",
314 .data = &sysctl_sched_child_runs_first,
315 .maxlen = sizeof(unsigned int),
316 .mode = 0644,
317 .proc_handler = proc_dointvec,
318 },
319#ifdef CONFIG_SCHED_DEBUG
320 {
321 .procname = "sched_min_granularity_ns",
322 .data = &sysctl_sched_min_granularity,
323 .maxlen = sizeof(unsigned int),
324 .mode = 0644,
325 .proc_handler = sched_proc_update_handler,
326 .extra1 = &min_sched_granularity_ns,
327 .extra2 = &max_sched_granularity_ns,
328 },
329 {
330 .procname = "sched_latency_ns",
331 .data = &sysctl_sched_latency,
332 .maxlen = sizeof(unsigned int),
333 .mode = 0644,
334 .proc_handler = sched_proc_update_handler,
335 .extra1 = &min_sched_granularity_ns,
336 .extra2 = &max_sched_granularity_ns,
337 },
338 {
339 .procname = "sched_wakeup_granularity_ns",
340 .data = &sysctl_sched_wakeup_granularity,
341 .maxlen = sizeof(unsigned int),
342 .mode = 0644,
343 .proc_handler = sched_proc_update_handler,
344 .extra1 = &min_wakeup_granularity_ns,
345 .extra2 = &max_wakeup_granularity_ns,
346 },
347#ifdef CONFIG_SMP
348 {
349 .procname = "sched_tunable_scaling",
350 .data = &sysctl_sched_tunable_scaling,
351 .maxlen = sizeof(enum sched_tunable_scaling),
352 .mode = 0644,
353 .proc_handler = sched_proc_update_handler,
354 .extra1 = &min_sched_tunable_scaling,
355 .extra2 = &max_sched_tunable_scaling,
356 },
357 {
358 .procname = "sched_migration_cost_ns",
359 .data = &sysctl_sched_migration_cost,
360 .maxlen = sizeof(unsigned int),
361 .mode = 0644,
362 .proc_handler = proc_dointvec,
363 },
364 {
365 .procname = "sched_nr_migrate",
366 .data = &sysctl_sched_nr_migrate,
367 .maxlen = sizeof(unsigned int),
368 .mode = 0644,
369 .proc_handler = proc_dointvec,
370 },
371 {
372 .procname = "sched_time_avg_ms",
373 .data = &sysctl_sched_time_avg,
374 .maxlen = sizeof(unsigned int),
375 .mode = 0644,
376 .proc_handler = proc_dointvec_minmax,
377 .extra1 = &one,
378 },
379#ifdef CONFIG_SCHEDSTATS
380 {
381 .procname = "sched_schedstats",
382 .data = NULL,
383 .maxlen = sizeof(unsigned int),
384 .mode = 0644,
385 .proc_handler = sysctl_schedstats,
386 .extra1 = &zero,
387 .extra2 = &one,
388 },
389#endif /* CONFIG_SCHEDSTATS */
390#endif /* CONFIG_SMP */
391#ifdef CONFIG_NUMA_BALANCING
392 {
393 .procname = "numa_balancing_scan_delay_ms",
394 .data = &sysctl_numa_balancing_scan_delay,
395 .maxlen = sizeof(unsigned int),
396 .mode = 0644,
397 .proc_handler = proc_dointvec,
398 },
399 {
400 .procname = "numa_balancing_scan_period_min_ms",
401 .data = &sysctl_numa_balancing_scan_period_min,
402 .maxlen = sizeof(unsigned int),
403 .mode = 0644,
404 .proc_handler = proc_dointvec,
405 },
406 {
407 .procname = "numa_balancing_scan_period_max_ms",
408 .data = &sysctl_numa_balancing_scan_period_max,
409 .maxlen = sizeof(unsigned int),
410 .mode = 0644,
411 .proc_handler = proc_dointvec,
412 },
413 {
414 .procname = "numa_balancing_scan_size_mb",
415 .data = &sysctl_numa_balancing_scan_size,
416 .maxlen = sizeof(unsigned int),
417 .mode = 0644,
418 .proc_handler = proc_dointvec_minmax,
419 .extra1 = &one,
420 },
421 {
422 .procname = "numa_balancing",
423 .data = NULL, /* filled in by handler */
424 .maxlen = sizeof(unsigned int),
425 .mode = 0644,
426 .proc_handler = sysctl_numa_balancing,
427 .extra1 = &zero,
428 .extra2 = &one,
429 },
430#endif /* CONFIG_NUMA_BALANCING */
431#endif /* CONFIG_SCHED_DEBUG */
432 {
433 .procname = "sched_rt_period_us",
434 .data = &sysctl_sched_rt_period,
435 .maxlen = sizeof(unsigned int),
436 .mode = 0644,
437 .proc_handler = sched_rt_handler,
438 },
439 {
440 .procname = "sched_rt_runtime_us",
441 .data = &sysctl_sched_rt_runtime,
442 .maxlen = sizeof(int),
443 .mode = 0644,
444 .proc_handler = sched_rt_handler,
445 },
446 {
447 .procname = "sched_rr_timeslice_ms",
448 .data = &sysctl_sched_rr_timeslice,
449 .maxlen = sizeof(int),
450 .mode = 0644,
451 .proc_handler = sched_rr_handler,
452 },
453#ifdef CONFIG_SCHED_AUTOGROUP
454 {
455 .procname = "sched_autogroup_enabled",
456 .data = &sysctl_sched_autogroup_enabled,
457 .maxlen = sizeof(unsigned int),
458 .mode = 0644,
459 .proc_handler = proc_dointvec_minmax,
460 .extra1 = &zero,
461 .extra2 = &one,
462 },
463#endif
464#ifdef CONFIG_CFS_BANDWIDTH
465 {
466 .procname = "sched_cfs_bandwidth_slice_us",
467 .data = &sysctl_sched_cfs_bandwidth_slice,
468 .maxlen = sizeof(unsigned int),
469 .mode = 0644,
470 .proc_handler = proc_dointvec_minmax,
471 .extra1 = &one,
472 },
473#endif
474#ifdef CONFIG_PROVE_LOCKING
475 {
476 .procname = "prove_locking",
477 .data = &prove_locking,
478 .maxlen = sizeof(int),
479 .mode = 0644,
480 .proc_handler = proc_dointvec,
481 },
482#endif
483#ifdef CONFIG_LOCK_STAT
484 {
485 .procname = "lock_stat",
486 .data = &lock_stat,
487 .maxlen = sizeof(int),
488 .mode = 0644,
489 .proc_handler = proc_dointvec,
490 },
491#endif
492 {
493 .procname = "panic",
494 .data = &panic_timeout,
495 .maxlen = sizeof(int),
496 .mode = 0644,
497 .proc_handler = proc_dointvec,
498 },
499#ifdef CONFIG_COREDUMP
500 {
501 .procname = "core_uses_pid",
502 .data = &core_uses_pid,
503 .maxlen = sizeof(int),
504 .mode = 0644,
505 .proc_handler = proc_dointvec,
506 },
507 {
508 .procname = "core_pattern",
509 .data = core_pattern,
510 .maxlen = CORENAME_MAX_SIZE,
511 .mode = 0644,
512 .proc_handler = proc_dostring_coredump,
513 },
514 {
515 .procname = "core_pipe_limit",
516 .data = &core_pipe_limit,
517 .maxlen = sizeof(unsigned int),
518 .mode = 0644,
519 .proc_handler = proc_dointvec,
520 },
521#endif
522#ifdef CONFIG_PROC_SYSCTL
523 {
524 .procname = "tainted",
525 .maxlen = sizeof(long),
526 .mode = 0644,
527 .proc_handler = proc_taint,
528 },
529 {
530 .procname = "sysctl_writes_strict",
531 .data = &sysctl_writes_strict,
532 .maxlen = sizeof(int),
533 .mode = 0644,
534 .proc_handler = proc_dointvec_minmax,
535 .extra1 = &neg_one,
536 .extra2 = &one,
537 },
538#endif
539#ifdef CONFIG_LATENCYTOP
540 {
541 .procname = "latencytop",
542 .data = &latencytop_enabled,
543 .maxlen = sizeof(int),
544 .mode = 0644,
545 .proc_handler = sysctl_latencytop,
546 },
547#endif
548#ifdef CONFIG_BLK_DEV_INITRD
549 {
550 .procname = "real-root-dev",
551 .data = &real_root_dev,
552 .maxlen = sizeof(int),
553 .mode = 0644,
554 .proc_handler = proc_dointvec,
555 },
556#endif
557 {
558 .procname = "print-fatal-signals",
559 .data = &print_fatal_signals,
560 .maxlen = sizeof(int),
561 .mode = 0644,
562 .proc_handler = proc_dointvec,
563 },
564#ifdef CONFIG_SPARC
565 {
566 .procname = "reboot-cmd",
567 .data = reboot_command,
568 .maxlen = 256,
569 .mode = 0644,
570 .proc_handler = proc_dostring,
571 },
572 {
573 .procname = "stop-a",
574 .data = &stop_a_enabled,
575 .maxlen = sizeof (int),
576 .mode = 0644,
577 .proc_handler = proc_dointvec,
578 },
579 {
580 .procname = "scons-poweroff",
581 .data = &scons_pwroff,
582 .maxlen = sizeof (int),
583 .mode = 0644,
584 .proc_handler = proc_dointvec,
585 },
586#endif
587#ifdef CONFIG_SPARC64
588 {
589 .procname = "tsb-ratio",
590 .data = &sysctl_tsb_ratio,
591 .maxlen = sizeof (int),
592 .mode = 0644,
593 .proc_handler = proc_dointvec,
594 },
595#endif
596#ifdef __hppa__
597 {
598 .procname = "soft-power",
599 .data = &pwrsw_enabled,
600 .maxlen = sizeof (int),
601 .mode = 0644,
602 .proc_handler = proc_dointvec,
603 },
604#endif
605#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
606 {
607 .procname = "unaligned-trap",
608 .data = &unaligned_enabled,
609 .maxlen = sizeof (int),
610 .mode = 0644,
611 .proc_handler = proc_dointvec,
612 },
613#endif
614 {
615 .procname = "ctrl-alt-del",
616 .data = &C_A_D,
617 .maxlen = sizeof(int),
618 .mode = 0644,
619 .proc_handler = proc_dointvec,
620 },
621#ifdef CONFIG_FUNCTION_TRACER
622 {
623 .procname = "ftrace_enabled",
624 .data = &ftrace_enabled,
625 .maxlen = sizeof(int),
626 .mode = 0644,
627 .proc_handler = ftrace_enable_sysctl,
628 },
629#endif
630#ifdef CONFIG_STACK_TRACER
631 {
632 .procname = "stack_tracer_enabled",
633 .data = &stack_tracer_enabled,
634 .maxlen = sizeof(int),
635 .mode = 0644,
636 .proc_handler = stack_trace_sysctl,
637 },
638#endif
639#ifdef CONFIG_TRACING
640 {
641 .procname = "ftrace_dump_on_oops",
642 .data = &ftrace_dump_on_oops,
643 .maxlen = sizeof(int),
644 .mode = 0644,
645 .proc_handler = proc_dointvec,
646 },
647 {
648 .procname = "traceoff_on_warning",
649 .data = &__disable_trace_on_warning,
650 .maxlen = sizeof(__disable_trace_on_warning),
651 .mode = 0644,
652 .proc_handler = proc_dointvec,
653 },
654 {
655 .procname = "tracepoint_printk",
656 .data = &tracepoint_printk,
657 .maxlen = sizeof(tracepoint_printk),
658 .mode = 0644,
659 .proc_handler = tracepoint_printk_sysctl,
660 },
661#endif
662#ifdef CONFIG_KEXEC_CORE
663 {
664 .procname = "kexec_load_disabled",
665 .data = &kexec_load_disabled,
666 .maxlen = sizeof(int),
667 .mode = 0644,
668 /* only handle a transition from default "0" to "1" */
669 .proc_handler = proc_dointvec_minmax,
670 .extra1 = &one,
671 .extra2 = &one,
672 },
673#endif
674#ifdef CONFIG_MODULES
675 {
676 .procname = "modprobe",
677 .data = &modprobe_path,
678 .maxlen = KMOD_PATH_LEN,
679 .mode = 0644,
680 .proc_handler = proc_dostring,
681 },
682 {
683 .procname = "modules_disabled",
684 .data = &modules_disabled,
685 .maxlen = sizeof(int),
686 .mode = 0644,
687 /* only handle a transition from default "0" to "1" */
688 .proc_handler = proc_dointvec_minmax,
689 .extra1 = &one,
690 .extra2 = &one,
691 },
692#endif
693#ifdef CONFIG_UEVENT_HELPER
694 {
695 .procname = "hotplug",
696 .data = &uevent_helper,
697 .maxlen = UEVENT_HELPER_PATH_LEN,
698 .mode = 0644,
699 .proc_handler = proc_dostring,
700 },
701#endif
702#ifdef CONFIG_CHR_DEV_SG
703 {
704 .procname = "sg-big-buff",
705 .data = &sg_big_buff,
706 .maxlen = sizeof (int),
707 .mode = 0444,
708 .proc_handler = proc_dointvec,
709 },
710#endif
711#ifdef CONFIG_BSD_PROCESS_ACCT
712 {
713 .procname = "acct",
714 .data = &acct_parm,
715 .maxlen = 3*sizeof(int),
716 .mode = 0644,
717 .proc_handler = proc_dointvec,
718 },
719#endif
720#ifdef CONFIG_MAGIC_SYSRQ
721 {
722 .procname = "sysrq",
723 .data = &__sysrq_enabled,
724 .maxlen = sizeof (int),
725 .mode = 0644,
726 .proc_handler = sysrq_sysctl_handler,
727 },
728#endif
729#ifdef CONFIG_PROC_SYSCTL
730 {
731 .procname = "cad_pid",
732 .data = NULL,
733 .maxlen = sizeof (int),
734 .mode = 0600,
735 .proc_handler = proc_do_cad_pid,
736 },
737#endif
738 {
739 .procname = "threads-max",
740 .data = NULL,
741 .maxlen = sizeof(int),
742 .mode = 0644,
743 .proc_handler = sysctl_max_threads,
744 },
745 {
746 .procname = "random",
747 .mode = 0555,
748 .child = random_table,
749 },
750 {
751 .procname = "usermodehelper",
752 .mode = 0555,
753 .child = usermodehelper_table,
754 },
755#ifdef CONFIG_FW_LOADER_USER_HELPER
756 {
757 .procname = "firmware_config",
758 .mode = 0555,
759 .child = firmware_config_table,
760 },
761#endif
762 {
763 .procname = "overflowuid",
764 .data = &overflowuid,
765 .maxlen = sizeof(int),
766 .mode = 0644,
767 .proc_handler = proc_dointvec_minmax,
768 .extra1 = &minolduid,
769 .extra2 = &maxolduid,
770 },
771 {
772 .procname = "overflowgid",
773 .data = &overflowgid,
774 .maxlen = sizeof(int),
775 .mode = 0644,
776 .proc_handler = proc_dointvec_minmax,
777 .extra1 = &minolduid,
778 .extra2 = &maxolduid,
779 },
780#ifdef CONFIG_S390
781#ifdef CONFIG_MATHEMU
782 {
783 .procname = "ieee_emulation_warnings",
784 .data = &sysctl_ieee_emulation_warnings,
785 .maxlen = sizeof(int),
786 .mode = 0644,
787 .proc_handler = proc_dointvec,
788 },
789#endif
790 {
791 .procname = "userprocess_debug",
792 .data = &show_unhandled_signals,
793 .maxlen = sizeof(int),
794 .mode = 0644,
795 .proc_handler = proc_dointvec,
796 },
797#endif
798 {
799 .procname = "pid_max",
800 .data = &pid_max,
801 .maxlen = sizeof (int),
802 .mode = 0644,
803 .proc_handler = proc_dointvec_minmax,
804 .extra1 = &pid_max_min,
805 .extra2 = &pid_max_max,
806 },
807 {
808 .procname = "panic_on_oops",
809 .data = &panic_on_oops,
810 .maxlen = sizeof(int),
811 .mode = 0644,
812 .proc_handler = proc_dointvec,
813 },
814#if defined CONFIG_PRINTK
815 {
816 .procname = "printk",
817 .data = &console_loglevel,
818 .maxlen = 4*sizeof(int),
819 .mode = 0644,
820 .proc_handler = proc_dointvec,
821 },
822 {
823 .procname = "printk_ratelimit",
824 .data = &printk_ratelimit_state.interval,
825 .maxlen = sizeof(int),
826 .mode = 0644,
827 .proc_handler = proc_dointvec_jiffies,
828 },
829 {
830 .procname = "printk_ratelimit_burst",
831 .data = &printk_ratelimit_state.burst,
832 .maxlen = sizeof(int),
833 .mode = 0644,
834 .proc_handler = proc_dointvec,
835 },
836 {
837 .procname = "printk_delay",
838 .data = &printk_delay_msec,
839 .maxlen = sizeof(int),
840 .mode = 0644,
841 .proc_handler = proc_dointvec_minmax,
842 .extra1 = &zero,
843 .extra2 = &ten_thousand,
844 },
845 {
846 .procname = "printk_devkmsg",
847 .data = devkmsg_log_str,
848 .maxlen = DEVKMSG_STR_MAX_SIZE,
849 .mode = 0644,
850 .proc_handler = devkmsg_sysctl_set_loglvl,
851 },
852 {
853 .procname = "dmesg_restrict",
854 .data = &dmesg_restrict,
855 .maxlen = sizeof(int),
856 .mode = 0644,
857 .proc_handler = proc_dointvec_minmax_sysadmin,
858 .extra1 = &zero,
859 .extra2 = &one,
860 },
861 {
862 .procname = "kptr_restrict",
863 .data = &kptr_restrict,
864 .maxlen = sizeof(int),
865 .mode = 0644,
866 .proc_handler = proc_dointvec_minmax_sysadmin,
867 .extra1 = &zero,
868 .extra2 = &two,
869 },
870#endif
871 {
872 .procname = "ngroups_max",
873 .data = &ngroups_max,
874 .maxlen = sizeof (int),
875 .mode = 0444,
876 .proc_handler = proc_dointvec,
877 },
878 {
879 .procname = "cap_last_cap",
880 .data = (void *)&cap_last_cap,
881 .maxlen = sizeof(int),
882 .mode = 0444,
883 .proc_handler = proc_dointvec,
884 },
885#if defined(CONFIG_LOCKUP_DETECTOR)
886 {
887 .procname = "watchdog",
888 .data = &watchdog_user_enabled,
889 .maxlen = sizeof(int),
890 .mode = 0644,
891 .proc_handler = proc_watchdog,
892 .extra1 = &zero,
893 .extra2 = &one,
894 },
895 {
896 .procname = "watchdog_thresh",
897 .data = &watchdog_thresh,
898 .maxlen = sizeof(int),
899 .mode = 0644,
900 .proc_handler = proc_watchdog_thresh,
901 .extra1 = &zero,
902 .extra2 = &sixty,
903 },
904 {
905 .procname = "nmi_watchdog",
906 .data = &nmi_watchdog_user_enabled,
907 .maxlen = sizeof(int),
908 .mode = NMI_WATCHDOG_SYSCTL_PERM,
909 .proc_handler = proc_nmi_watchdog,
910 .extra1 = &zero,
911 .extra2 = &one,
912 },
913 {
914 .procname = "watchdog_cpumask",
915 .data = &watchdog_cpumask_bits,
916 .maxlen = NR_CPUS,
917 .mode = 0644,
918 .proc_handler = proc_watchdog_cpumask,
919 },
920#ifdef CONFIG_SOFTLOCKUP_DETECTOR
921 {
922 .procname = "soft_watchdog",
923 .data = &soft_watchdog_user_enabled,
924 .maxlen = sizeof(int),
925 .mode = 0644,
926 .proc_handler = proc_soft_watchdog,
927 .extra1 = &zero,
928 .extra2 = &one,
929 },
930 {
931 .procname = "softlockup_panic",
932 .data = &softlockup_panic,
933 .maxlen = sizeof(int),
934 .mode = 0644,
935 .proc_handler = proc_dointvec_minmax,
936 .extra1 = &zero,
937 .extra2 = &one,
938 },
939#ifdef CONFIG_SMP
940 {
941 .procname = "softlockup_all_cpu_backtrace",
942 .data = &sysctl_softlockup_all_cpu_backtrace,
943 .maxlen = sizeof(int),
944 .mode = 0644,
945 .proc_handler = proc_dointvec_minmax,
946 .extra1 = &zero,
947 .extra2 = &one,
948 },
949#endif /* CONFIG_SMP */
950#endif
951#ifdef CONFIG_HARDLOCKUP_DETECTOR
952 {
953 .procname = "hardlockup_panic",
954 .data = &hardlockup_panic,
955 .maxlen = sizeof(int),
956 .mode = 0644,
957 .proc_handler = proc_dointvec_minmax,
958 .extra1 = &zero,
959 .extra2 = &one,
960 },
961#ifdef CONFIG_SMP
962 {
963 .procname = "hardlockup_all_cpu_backtrace",
964 .data = &sysctl_hardlockup_all_cpu_backtrace,
965 .maxlen = sizeof(int),
966 .mode = 0644,
967 .proc_handler = proc_dointvec_minmax,
968 .extra1 = &zero,
969 .extra2 = &one,
970 },
971#endif /* CONFIG_SMP */
972#endif
973#endif
974
975#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
976 {
977 .procname = "unknown_nmi_panic",
978 .data = &unknown_nmi_panic,
979 .maxlen = sizeof (int),
980 .mode = 0644,
981 .proc_handler = proc_dointvec,
982 },
983#endif
984#if defined(CONFIG_X86)
985 {
986 .procname = "panic_on_unrecovered_nmi",
987 .data = &panic_on_unrecovered_nmi,
988 .maxlen = sizeof(int),
989 .mode = 0644,
990 .proc_handler = proc_dointvec,
991 },
992 {
993 .procname = "panic_on_io_nmi",
994 .data = &panic_on_io_nmi,
995 .maxlen = sizeof(int),
996 .mode = 0644,
997 .proc_handler = proc_dointvec,
998 },
999#ifdef CONFIG_DEBUG_STACKOVERFLOW
1000 {
1001 .procname = "panic_on_stackoverflow",
1002 .data = &sysctl_panic_on_stackoverflow,
1003 .maxlen = sizeof(int),
1004 .mode = 0644,
1005 .proc_handler = proc_dointvec,
1006 },
1007#endif
1008 {
1009 .procname = "bootloader_type",
1010 .data = &bootloader_type,
1011 .maxlen = sizeof (int),
1012 .mode = 0444,
1013 .proc_handler = proc_dointvec,
1014 },
1015 {
1016 .procname = "bootloader_version",
1017 .data = &bootloader_version,
1018 .maxlen = sizeof (int),
1019 .mode = 0444,
1020 .proc_handler = proc_dointvec,
1021 },
1022 {
1023 .procname = "io_delay_type",
1024 .data = &io_delay_type,
1025 .maxlen = sizeof(int),
1026 .mode = 0644,
1027 .proc_handler = proc_dointvec,
1028 },
1029#endif
1030#if defined(CONFIG_MMU)
1031 {
1032 .procname = "randomize_va_space",
1033 .data = &randomize_va_space,
1034 .maxlen = sizeof(int),
1035 .mode = 0644,
1036 .proc_handler = proc_dointvec,
1037 },
1038#endif
1039#if defined(CONFIG_S390) && defined(CONFIG_SMP)
1040 {
1041 .procname = "spin_retry",
1042 .data = &spin_retry,
1043 .maxlen = sizeof (int),
1044 .mode = 0644,
1045 .proc_handler = proc_dointvec,
1046 },
1047#endif
1048#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1049 {
1050 .procname = "acpi_video_flags",
1051 .data = &acpi_realmode_flags,
1052 .maxlen = sizeof (unsigned long),
1053 .mode = 0644,
1054 .proc_handler = proc_doulongvec_minmax,
1055 },
1056#endif
1057#ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1058 {
1059 .procname = "ignore-unaligned-usertrap",
1060 .data = &no_unaligned_warning,
1061 .maxlen = sizeof (int),
1062 .mode = 0644,
1063 .proc_handler = proc_dointvec,
1064 },
1065#endif
1066#ifdef CONFIG_IA64
1067 {
1068 .procname = "unaligned-dump-stack",
1069 .data = &unaligned_dump_stack,
1070 .maxlen = sizeof (int),
1071 .mode = 0644,
1072 .proc_handler = proc_dointvec,
1073 },
1074#endif
1075#ifdef CONFIG_DETECT_HUNG_TASK
1076 {
1077 .procname = "hung_task_panic",
1078 .data = &sysctl_hung_task_panic,
1079 .maxlen = sizeof(int),
1080 .mode = 0644,
1081 .proc_handler = proc_dointvec_minmax,
1082 .extra1 = &zero,
1083 .extra2 = &one,
1084 },
1085 {
1086 .procname = "hung_task_check_count",
1087 .data = &sysctl_hung_task_check_count,
1088 .maxlen = sizeof(int),
1089 .mode = 0644,
1090 .proc_handler = proc_dointvec_minmax,
1091 .extra1 = &zero,
1092 },
1093 {
1094 .procname = "hung_task_timeout_secs",
1095 .data = &sysctl_hung_task_timeout_secs,
1096 .maxlen = sizeof(unsigned long),
1097 .mode = 0644,
1098 .proc_handler = proc_dohung_task_timeout_secs,
1099 .extra2 = &hung_task_timeout_max,
1100 },
1101 {
1102 .procname = "hung_task_warnings",
1103 .data = &sysctl_hung_task_warnings,
1104 .maxlen = sizeof(int),
1105 .mode = 0644,
1106 .proc_handler = proc_dointvec_minmax,
1107 .extra1 = &neg_one,
1108 },
1109#endif
1110#ifdef CONFIG_RT_MUTEXES
1111 {
1112 .procname = "max_lock_depth",
1113 .data = &max_lock_depth,
1114 .maxlen = sizeof(int),
1115 .mode = 0644,
1116 .proc_handler = proc_dointvec,
1117 },
1118#endif
1119 {
1120 .procname = "poweroff_cmd",
1121 .data = &poweroff_cmd,
1122 .maxlen = POWEROFF_CMD_PATH_LEN,
1123 .mode = 0644,
1124 .proc_handler = proc_dostring,
1125 },
1126#ifdef CONFIG_KEYS
1127 {
1128 .procname = "keys",
1129 .mode = 0555,
1130 .child = key_sysctls,
1131 },
1132#endif
1133#ifdef CONFIG_PERF_EVENTS
1134 /*
1135 * User-space scripts rely on the existence of this file
1136 * as a feature check for perf_events being enabled.
1137 *
1138 * So it's an ABI, do not remove!
1139 */
1140 {
1141 .procname = "perf_event_paranoid",
1142 .data = &sysctl_perf_event_paranoid,
1143 .maxlen = sizeof(sysctl_perf_event_paranoid),
1144 .mode = 0644,
1145 .proc_handler = proc_dointvec,
1146 },
1147 {
1148 .procname = "perf_event_mlock_kb",
1149 .data = &sysctl_perf_event_mlock,
1150 .maxlen = sizeof(sysctl_perf_event_mlock),
1151 .mode = 0644,
1152 .proc_handler = proc_dointvec,
1153 },
1154 {
1155 .procname = "perf_event_max_sample_rate",
1156 .data = &sysctl_perf_event_sample_rate,
1157 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1158 .mode = 0644,
1159 .proc_handler = perf_proc_update_handler,
1160 .extra1 = &one,
1161 },
1162 {
1163 .procname = "perf_cpu_time_max_percent",
1164 .data = &sysctl_perf_cpu_time_max_percent,
1165 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1166 .mode = 0644,
1167 .proc_handler = perf_cpu_time_max_percent_handler,
1168 .extra1 = &zero,
1169 .extra2 = &one_hundred,
1170 },
1171 {
1172 .procname = "perf_event_max_stack",
1173 .data = &sysctl_perf_event_max_stack,
1174 .maxlen = sizeof(sysctl_perf_event_max_stack),
1175 .mode = 0644,
1176 .proc_handler = perf_event_max_stack_handler,
1177 .extra1 = &zero,
1178 .extra2 = &six_hundred_forty_kb,
1179 },
1180 {
1181 .procname = "perf_event_max_contexts_per_stack",
1182 .data = &sysctl_perf_event_max_contexts_per_stack,
1183 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
1184 .mode = 0644,
1185 .proc_handler = perf_event_max_stack_handler,
1186 .extra1 = &zero,
1187 .extra2 = &one_thousand,
1188 },
1189#endif
1190 {
1191 .procname = "panic_on_warn",
1192 .data = &panic_on_warn,
1193 .maxlen = sizeof(int),
1194 .mode = 0644,
1195 .proc_handler = proc_dointvec_minmax,
1196 .extra1 = &zero,
1197 .extra2 = &one,
1198 },
1199#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1200 {
1201 .procname = "timer_migration",
1202 .data = &sysctl_timer_migration,
1203 .maxlen = sizeof(unsigned int),
1204 .mode = 0644,
1205 .proc_handler = timer_migration_handler,
1206 .extra1 = &zero,
1207 .extra2 = &one,
1208 },
1209#endif
1210#ifdef CONFIG_BPF_SYSCALL
1211 {
1212 .procname = "unprivileged_bpf_disabled",
1213 .data = &sysctl_unprivileged_bpf_disabled,
1214 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
1215 .mode = 0644,
1216 /* only handle a transition from default "0" to "1" */
1217 .proc_handler = proc_dointvec_minmax,
1218 .extra1 = &one,
1219 .extra2 = &one,
1220 },
1221#endif
1222#if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1223 {
1224 .procname = "panic_on_rcu_stall",
1225 .data = &sysctl_panic_on_rcu_stall,
1226 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
1227 .mode = 0644,
1228 .proc_handler = proc_dointvec_minmax,
1229 .extra1 = &zero,
1230 .extra2 = &one,
1231 },
1232#endif
1233 { }
1234};
1235
1236static struct ctl_table vm_table[] = {
1237 {
1238 .procname = "overcommit_memory",
1239 .data = &sysctl_overcommit_memory,
1240 .maxlen = sizeof(sysctl_overcommit_memory),
1241 .mode = 0644,
1242 .proc_handler = proc_dointvec_minmax,
1243 .extra1 = &zero,
1244 .extra2 = &two,
1245 },
1246 {
1247 .procname = "panic_on_oom",
1248 .data = &sysctl_panic_on_oom,
1249 .maxlen = sizeof(sysctl_panic_on_oom),
1250 .mode = 0644,
1251 .proc_handler = proc_dointvec_minmax,
1252 .extra1 = &zero,
1253 .extra2 = &two,
1254 },
1255 {
1256 .procname = "oom_kill_allocating_task",
1257 .data = &sysctl_oom_kill_allocating_task,
1258 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1259 .mode = 0644,
1260 .proc_handler = proc_dointvec,
1261 },
1262 {
1263 .procname = "oom_dump_tasks",
1264 .data = &sysctl_oom_dump_tasks,
1265 .maxlen = sizeof(sysctl_oom_dump_tasks),
1266 .mode = 0644,
1267 .proc_handler = proc_dointvec,
1268 },
1269 {
1270 .procname = "overcommit_ratio",
1271 .data = &sysctl_overcommit_ratio,
1272 .maxlen = sizeof(sysctl_overcommit_ratio),
1273 .mode = 0644,
1274 .proc_handler = overcommit_ratio_handler,
1275 },
1276 {
1277 .procname = "overcommit_kbytes",
1278 .data = &sysctl_overcommit_kbytes,
1279 .maxlen = sizeof(sysctl_overcommit_kbytes),
1280 .mode = 0644,
1281 .proc_handler = overcommit_kbytes_handler,
1282 },
1283 {
1284 .procname = "page-cluster",
1285 .data = &page_cluster,
1286 .maxlen = sizeof(int),
1287 .mode = 0644,
1288 .proc_handler = proc_dointvec_minmax,
1289 .extra1 = &zero,
1290 },
1291 {
1292 .procname = "dirty_background_ratio",
1293 .data = &dirty_background_ratio,
1294 .maxlen = sizeof(dirty_background_ratio),
1295 .mode = 0644,
1296 .proc_handler = dirty_background_ratio_handler,
1297 .extra1 = &zero,
1298 .extra2 = &one_hundred,
1299 },
1300 {
1301 .procname = "dirty_background_bytes",
1302 .data = &dirty_background_bytes,
1303 .maxlen = sizeof(dirty_background_bytes),
1304 .mode = 0644,
1305 .proc_handler = dirty_background_bytes_handler,
1306 .extra1 = &one_ul,
1307 },
1308 {
1309 .procname = "dirty_ratio",
1310 .data = &vm_dirty_ratio,
1311 .maxlen = sizeof(vm_dirty_ratio),
1312 .mode = 0644,
1313 .proc_handler = dirty_ratio_handler,
1314 .extra1 = &zero,
1315 .extra2 = &one_hundred,
1316 },
1317 {
1318 .procname = "dirty_bytes",
1319 .data = &vm_dirty_bytes,
1320 .maxlen = sizeof(vm_dirty_bytes),
1321 .mode = 0644,
1322 .proc_handler = dirty_bytes_handler,
1323 .extra1 = &dirty_bytes_min,
1324 },
1325 {
1326 .procname = "dirty_writeback_centisecs",
1327 .data = &dirty_writeback_interval,
1328 .maxlen = sizeof(dirty_writeback_interval),
1329 .mode = 0644,
1330 .proc_handler = dirty_writeback_centisecs_handler,
1331 },
1332 {
1333 .procname = "dirty_expire_centisecs",
1334 .data = &dirty_expire_interval,
1335 .maxlen = sizeof(dirty_expire_interval),
1336 .mode = 0644,
1337 .proc_handler = proc_dointvec_minmax,
1338 .extra1 = &zero,
1339 },
1340 {
1341 .procname = "dirtytime_expire_seconds",
1342 .data = &dirtytime_expire_interval,
1343 .maxlen = sizeof(dirtytime_expire_interval),
1344 .mode = 0644,
1345 .proc_handler = dirtytime_interval_handler,
1346 .extra1 = &zero,
1347 },
1348 {
1349 .procname = "swappiness",
1350 .data = &vm_swappiness,
1351 .maxlen = sizeof(vm_swappiness),
1352 .mode = 0644,
1353 .proc_handler = proc_dointvec_minmax,
1354 .extra1 = &zero,
1355 .extra2 = &one_hundred,
1356 },
1357#ifdef CONFIG_HUGETLB_PAGE
1358 {
1359 .procname = "nr_hugepages",
1360 .data = NULL,
1361 .maxlen = sizeof(unsigned long),
1362 .mode = 0644,
1363 .proc_handler = hugetlb_sysctl_handler,
1364 },
1365#ifdef CONFIG_NUMA
1366 {
1367 .procname = "nr_hugepages_mempolicy",
1368 .data = NULL,
1369 .maxlen = sizeof(unsigned long),
1370 .mode = 0644,
1371 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1372 },
1373 {
1374 .procname = "numa_stat",
1375 .data = &sysctl_vm_numa_stat,
1376 .maxlen = sizeof(int),
1377 .mode = 0644,
1378 .proc_handler = sysctl_vm_numa_stat_handler,
1379 .extra1 = &zero,
1380 .extra2 = &one,
1381 },
1382#endif
1383 {
1384 .procname = "hugetlb_shm_group",
1385 .data = &sysctl_hugetlb_shm_group,
1386 .maxlen = sizeof(gid_t),
1387 .mode = 0644,
1388 .proc_handler = proc_dointvec,
1389 },
1390 {
1391 .procname = "nr_overcommit_hugepages",
1392 .data = NULL,
1393 .maxlen = sizeof(unsigned long),
1394 .mode = 0644,
1395 .proc_handler = hugetlb_overcommit_handler,
1396 },
1397#endif
1398 {
1399 .procname = "lowmem_reserve_ratio",
1400 .data = &sysctl_lowmem_reserve_ratio,
1401 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1402 .mode = 0644,
1403 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1404 },
1405 {
1406 .procname = "drop_caches",
1407 .data = &sysctl_drop_caches,
1408 .maxlen = sizeof(int),
1409 .mode = 0644,
1410 .proc_handler = drop_caches_sysctl_handler,
1411 .extra1 = &one,
1412 .extra2 = &four,
1413 },
1414#ifdef CONFIG_COMPACTION
1415 {
1416 .procname = "compact_memory",
1417 .data = &sysctl_compact_memory,
1418 .maxlen = sizeof(int),
1419 .mode = 0200,
1420 .proc_handler = sysctl_compaction_handler,
1421 },
1422 {
1423 .procname = "extfrag_threshold",
1424 .data = &sysctl_extfrag_threshold,
1425 .maxlen = sizeof(int),
1426 .mode = 0644,
1427 .proc_handler = sysctl_extfrag_handler,
1428 .extra1 = &min_extfrag_threshold,
1429 .extra2 = &max_extfrag_threshold,
1430 },
1431 {
1432 .procname = "compact_unevictable_allowed",
1433 .data = &sysctl_compact_unevictable_allowed,
1434 .maxlen = sizeof(int),
1435 .mode = 0644,
1436 .proc_handler = proc_dointvec,
1437 .extra1 = &zero,
1438 .extra2 = &one,
1439 },
1440
1441#endif /* CONFIG_COMPACTION */
1442 {
1443 .procname = "min_free_kbytes",
1444 .data = &min_free_kbytes,
1445 .maxlen = sizeof(min_free_kbytes),
1446 .mode = 0644,
1447 .proc_handler = min_free_kbytes_sysctl_handler,
1448 .extra1 = &zero,
1449 },
1450 {
1451 .procname = "watermark_scale_factor",
1452 .data = &watermark_scale_factor,
1453 .maxlen = sizeof(watermark_scale_factor),
1454 .mode = 0644,
1455 .proc_handler = watermark_scale_factor_sysctl_handler,
1456 .extra1 = &one,
1457 .extra2 = &one_thousand,
1458 },
1459 {
1460 .procname = "percpu_pagelist_fraction",
1461 .data = &percpu_pagelist_fraction,
1462 .maxlen = sizeof(percpu_pagelist_fraction),
1463 .mode = 0644,
1464 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1465 .extra1 = &zero,
1466 },
1467#ifdef CONFIG_MMU
1468 {
1469 .procname = "max_map_count",
1470 .data = &sysctl_max_map_count,
1471 .maxlen = sizeof(sysctl_max_map_count),
1472 .mode = 0644,
1473 .proc_handler = proc_dointvec_minmax,
1474 .extra1 = &zero,
1475 },
1476#else
1477 {
1478 .procname = "nr_trim_pages",
1479 .data = &sysctl_nr_trim_pages,
1480 .maxlen = sizeof(sysctl_nr_trim_pages),
1481 .mode = 0644,
1482 .proc_handler = proc_dointvec_minmax,
1483 .extra1 = &zero,
1484 },
1485#endif
1486 {
1487 .procname = "laptop_mode",
1488 .data = &laptop_mode,
1489 .maxlen = sizeof(laptop_mode),
1490 .mode = 0644,
1491 .proc_handler = proc_dointvec_jiffies,
1492 },
1493 {
1494 .procname = "block_dump",
1495 .data = &block_dump,
1496 .maxlen = sizeof(block_dump),
1497 .mode = 0644,
1498 .proc_handler = proc_dointvec,
1499 .extra1 = &zero,
1500 },
1501 {
1502 .procname = "vfs_cache_pressure",
1503 .data = &sysctl_vfs_cache_pressure,
1504 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1505 .mode = 0644,
1506 .proc_handler = proc_dointvec,
1507 .extra1 = &zero,
1508 },
1509#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1510 {
1511 .procname = "legacy_va_layout",
1512 .data = &sysctl_legacy_va_layout,
1513 .maxlen = sizeof(sysctl_legacy_va_layout),
1514 .mode = 0644,
1515 .proc_handler = proc_dointvec,
1516 .extra1 = &zero,
1517 },
1518#endif
1519#ifdef CONFIG_NUMA
1520 {
1521 .procname = "zone_reclaim_mode",
1522 .data = &node_reclaim_mode,
1523 .maxlen = sizeof(node_reclaim_mode),
1524 .mode = 0644,
1525 .proc_handler = proc_dointvec,
1526 .extra1 = &zero,
1527 },
1528 {
1529 .procname = "min_unmapped_ratio",
1530 .data = &sysctl_min_unmapped_ratio,
1531 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1532 .mode = 0644,
1533 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1534 .extra1 = &zero,
1535 .extra2 = &one_hundred,
1536 },
1537 {
1538 .procname = "min_slab_ratio",
1539 .data = &sysctl_min_slab_ratio,
1540 .maxlen = sizeof(sysctl_min_slab_ratio),
1541 .mode = 0644,
1542 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1543 .extra1 = &zero,
1544 .extra2 = &one_hundred,
1545 },
1546#endif
1547#ifdef CONFIG_SMP
1548 {
1549 .procname = "stat_interval",
1550 .data = &sysctl_stat_interval,
1551 .maxlen = sizeof(sysctl_stat_interval),
1552 .mode = 0644,
1553 .proc_handler = proc_dointvec_jiffies,
1554 },
1555 {
1556 .procname = "stat_refresh",
1557 .data = NULL,
1558 .maxlen = 0,
1559 .mode = 0600,
1560 .proc_handler = vmstat_refresh,
1561 },
1562#endif
1563#ifdef CONFIG_MMU
1564 {
1565 .procname = "mmap_min_addr",
1566 .data = &dac_mmap_min_addr,
1567 .maxlen = sizeof(unsigned long),
1568 .mode = 0644,
1569 .proc_handler = mmap_min_addr_handler,
1570 },
1571#endif
1572#ifdef CONFIG_NUMA
1573 {
1574 .procname = "numa_zonelist_order",
1575 .data = &numa_zonelist_order,
1576 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1577 .mode = 0644,
1578 .proc_handler = numa_zonelist_order_handler,
1579 },
1580#endif
1581#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1582 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1583 {
1584 .procname = "vdso_enabled",
1585#ifdef CONFIG_X86_32
1586 .data = &vdso32_enabled,
1587 .maxlen = sizeof(vdso32_enabled),
1588#else
1589 .data = &vdso_enabled,
1590 .maxlen = sizeof(vdso_enabled),
1591#endif
1592 .mode = 0644,
1593 .proc_handler = proc_dointvec,
1594 .extra1 = &zero,
1595 },
1596#endif
1597#ifdef CONFIG_HIGHMEM
1598 {
1599 .procname = "highmem_is_dirtyable",
1600 .data = &vm_highmem_is_dirtyable,
1601 .maxlen = sizeof(vm_highmem_is_dirtyable),
1602 .mode = 0644,
1603 .proc_handler = proc_dointvec_minmax,
1604 .extra1 = &zero,
1605 .extra2 = &one,
1606 },
1607#endif
1608#ifdef CONFIG_MEMORY_FAILURE
1609 {
1610 .procname = "memory_failure_early_kill",
1611 .data = &sysctl_memory_failure_early_kill,
1612 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1613 .mode = 0644,
1614 .proc_handler = proc_dointvec_minmax,
1615 .extra1 = &zero,
1616 .extra2 = &one,
1617 },
1618 {
1619 .procname = "memory_failure_recovery",
1620 .data = &sysctl_memory_failure_recovery,
1621 .maxlen = sizeof(sysctl_memory_failure_recovery),
1622 .mode = 0644,
1623 .proc_handler = proc_dointvec_minmax,
1624 .extra1 = &zero,
1625 .extra2 = &one,
1626 },
1627#endif
1628 {
1629 .procname = "user_reserve_kbytes",
1630 .data = &sysctl_user_reserve_kbytes,
1631 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1632 .mode = 0644,
1633 .proc_handler = proc_doulongvec_minmax,
1634 },
1635 {
1636 .procname = "admin_reserve_kbytes",
1637 .data = &sysctl_admin_reserve_kbytes,
1638 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1639 .mode = 0644,
1640 .proc_handler = proc_doulongvec_minmax,
1641 },
1642#ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1643 {
1644 .procname = "mmap_rnd_bits",
1645 .data = &mmap_rnd_bits,
1646 .maxlen = sizeof(mmap_rnd_bits),
1647 .mode = 0600,
1648 .proc_handler = proc_dointvec_minmax,
1649 .extra1 = (void *)&mmap_rnd_bits_min,
1650 .extra2 = (void *)&mmap_rnd_bits_max,
1651 },
1652#endif
1653#ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1654 {
1655 .procname = "mmap_rnd_compat_bits",
1656 .data = &mmap_rnd_compat_bits,
1657 .maxlen = sizeof(mmap_rnd_compat_bits),
1658 .mode = 0600,
1659 .proc_handler = proc_dointvec_minmax,
1660 .extra1 = (void *)&mmap_rnd_compat_bits_min,
1661 .extra2 = (void *)&mmap_rnd_compat_bits_max,
1662 },
1663#endif
1664 { }
1665};
1666
1667static struct ctl_table fs_table[] = {
1668 {
1669 .procname = "inode-nr",
1670 .data = &inodes_stat,
1671 .maxlen = 2*sizeof(long),
1672 .mode = 0444,
1673 .proc_handler = proc_nr_inodes,
1674 },
1675 {
1676 .procname = "inode-state",
1677 .data = &inodes_stat,
1678 .maxlen = 7*sizeof(long),
1679 .mode = 0444,
1680 .proc_handler = proc_nr_inodes,
1681 },
1682 {
1683 .procname = "file-nr",
1684 .data = &files_stat,
1685 .maxlen = sizeof(files_stat),
1686 .mode = 0444,
1687 .proc_handler = proc_nr_files,
1688 },
1689 {
1690 .procname = "file-max",
1691 .data = &files_stat.max_files,
1692 .maxlen = sizeof(files_stat.max_files),
1693 .mode = 0644,
1694 .proc_handler = proc_doulongvec_minmax,
1695 },
1696 {
1697 .procname = "nr_open",
1698 .data = &sysctl_nr_open,
1699 .maxlen = sizeof(unsigned int),
1700 .mode = 0644,
1701 .proc_handler = proc_dointvec_minmax,
1702 .extra1 = &sysctl_nr_open_min,
1703 .extra2 = &sysctl_nr_open_max,
1704 },
1705 {
1706 .procname = "dentry-state",
1707 .data = &dentry_stat,
1708 .maxlen = 6*sizeof(long),
1709 .mode = 0444,
1710 .proc_handler = proc_nr_dentry,
1711 },
1712 {
1713 .procname = "overflowuid",
1714 .data = &fs_overflowuid,
1715 .maxlen = sizeof(int),
1716 .mode = 0644,
1717 .proc_handler = proc_dointvec_minmax,
1718 .extra1 = &minolduid,
1719 .extra2 = &maxolduid,
1720 },
1721 {
1722 .procname = "overflowgid",
1723 .data = &fs_overflowgid,
1724 .maxlen = sizeof(int),
1725 .mode = 0644,
1726 .proc_handler = proc_dointvec_minmax,
1727 .extra1 = &minolduid,
1728 .extra2 = &maxolduid,
1729 },
1730#ifdef CONFIG_FILE_LOCKING
1731 {
1732 .procname = "leases-enable",
1733 .data = &leases_enable,
1734 .maxlen = sizeof(int),
1735 .mode = 0644,
1736 .proc_handler = proc_dointvec,
1737 },
1738#endif
1739#ifdef CONFIG_DNOTIFY
1740 {
1741 .procname = "dir-notify-enable",
1742 .data = &dir_notify_enable,
1743 .maxlen = sizeof(int),
1744 .mode = 0644,
1745 .proc_handler = proc_dointvec,
1746 },
1747#endif
1748#ifdef CONFIG_MMU
1749#ifdef CONFIG_FILE_LOCKING
1750 {
1751 .procname = "lease-break-time",
1752 .data = &lease_break_time,
1753 .maxlen = sizeof(int),
1754 .mode = 0644,
1755 .proc_handler = proc_dointvec,
1756 },
1757#endif
1758#ifdef CONFIG_AIO
1759 {
1760 .procname = "aio-nr",
1761 .data = &aio_nr,
1762 .maxlen = sizeof(aio_nr),
1763 .mode = 0444,
1764 .proc_handler = proc_doulongvec_minmax,
1765 },
1766 {
1767 .procname = "aio-max-nr",
1768 .data = &aio_max_nr,
1769 .maxlen = sizeof(aio_max_nr),
1770 .mode = 0644,
1771 .proc_handler = proc_doulongvec_minmax,
1772 },
1773#endif /* CONFIG_AIO */
1774#ifdef CONFIG_INOTIFY_USER
1775 {
1776 .procname = "inotify",
1777 .mode = 0555,
1778 .child = inotify_table,
1779 },
1780#endif
1781#ifdef CONFIG_EPOLL
1782 {
1783 .procname = "epoll",
1784 .mode = 0555,
1785 .child = epoll_table,
1786 },
1787#endif
1788#endif
1789 {
1790 .procname = "protected_symlinks",
1791 .data = &sysctl_protected_symlinks,
1792 .maxlen = sizeof(int),
1793 .mode = 0600,
1794 .proc_handler = proc_dointvec_minmax,
1795 .extra1 = &zero,
1796 .extra2 = &one,
1797 },
1798 {
1799 .procname = "protected_hardlinks",
1800 .data = &sysctl_protected_hardlinks,
1801 .maxlen = sizeof(int),
1802 .mode = 0600,
1803 .proc_handler = proc_dointvec_minmax,
1804 .extra1 = &zero,
1805 .extra2 = &one,
1806 },
1807 {
1808 .procname = "suid_dumpable",
1809 .data = &suid_dumpable,
1810 .maxlen = sizeof(int),
1811 .mode = 0644,
1812 .proc_handler = proc_dointvec_minmax_coredump,
1813 .extra1 = &zero,
1814 .extra2 = &two,
1815 },
1816#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1817 {
1818 .procname = "binfmt_misc",
1819 .mode = 0555,
1820 .child = sysctl_mount_point,
1821 },
1822#endif
1823 {
1824 .procname = "pipe-max-size",
1825 .data = &pipe_max_size,
1826 .maxlen = sizeof(pipe_max_size),
1827 .mode = 0644,
1828 .proc_handler = proc_dopipe_max_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
1855static 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
1879static struct ctl_table dev_table[] = {
1880 { }
1881};
1882
1883int __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
1900static 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
1967static 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 */
1984static 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 */
2018int 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
2028static 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
2037static 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 */
2064static 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 */
2123static 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
2141static 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
2153static 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
2180static 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
2195static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2196
2197static 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;
2274out:
2275 *ppos += *lenp;
2276 return err;
2277}
2278
2279static 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
2289static 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
2338out_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() */
2346bail_early:
2347 *ppos += *lenp;
2348 return err;
2349}
2350
2351static 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
2375out:
2376 *lenp -= left;
2377 *ppos += *lenp;
2378
2379 return err;
2380}
2381
2382static 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
2418static 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 */
2442int 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 */
2461int 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 */
2472static 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
2504static 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/**
2515 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
2516 * @min: pointer to minimum allowable value
2517 * @max: pointer to maximum allowable value
2518 *
2519 * The do_proc_dointvec_minmax_conv_param structure provides the
2520 * minimum and maximum values for doing range checking for those sysctl
2521 * parameters that use the proc_dointvec_minmax() handler.
2522 */
2523struct do_proc_dointvec_minmax_conv_param {
2524 int *min;
2525 int *max;
2526};
2527
2528static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2529 int *valp,
2530 int write, void *data)
2531{
2532 struct do_proc_dointvec_minmax_conv_param *param = data;
2533 if (write) {
2534 int val = *negp ? -*lvalp : *lvalp;
2535 if ((param->min && *param->min > val) ||
2536 (param->max && *param->max < val))
2537 return -EINVAL;
2538 *valp = val;
2539 } else {
2540 int val = *valp;
2541 if (val < 0) {
2542 *negp = true;
2543 *lvalp = -(unsigned long)val;
2544 } else {
2545 *negp = false;
2546 *lvalp = (unsigned long)val;
2547 }
2548 }
2549 return 0;
2550}
2551
2552/**
2553 * proc_dointvec_minmax - read a vector of integers with min/max values
2554 * @table: the sysctl table
2555 * @write: %TRUE if this is a write to the sysctl file
2556 * @buffer: the user buffer
2557 * @lenp: the size of the user buffer
2558 * @ppos: file position
2559 *
2560 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2561 * values from/to the user buffer, treated as an ASCII string.
2562 *
2563 * This routine will ensure the values are within the range specified by
2564 * table->extra1 (min) and table->extra2 (max).
2565 *
2566 * Returns 0 on success or -EINVAL on write when the range check fails.
2567 */
2568int proc_dointvec_minmax(struct ctl_table *table, int write,
2569 void __user *buffer, size_t *lenp, loff_t *ppos)
2570{
2571 struct do_proc_dointvec_minmax_conv_param param = {
2572 .min = (int *) table->extra1,
2573 .max = (int *) table->extra2,
2574 };
2575 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2576 do_proc_dointvec_minmax_conv, ¶m);
2577}
2578
2579/**
2580 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
2581 * @min: pointer to minimum allowable value
2582 * @max: pointer to maximum allowable value
2583 *
2584 * The do_proc_douintvec_minmax_conv_param structure provides the
2585 * minimum and maximum values for doing range checking for those sysctl
2586 * parameters that use the proc_douintvec_minmax() handler.
2587 */
2588struct do_proc_douintvec_minmax_conv_param {
2589 unsigned int *min;
2590 unsigned int *max;
2591};
2592
2593static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2594 unsigned int *valp,
2595 int write, void *data)
2596{
2597 struct do_proc_douintvec_minmax_conv_param *param = data;
2598
2599 if (write) {
2600 unsigned int val = *lvalp;
2601
2602 if (*lvalp > UINT_MAX)
2603 return -EINVAL;
2604
2605 if ((param->min && *param->min > val) ||
2606 (param->max && *param->max < val))
2607 return -ERANGE;
2608
2609 *valp = val;
2610 } else {
2611 unsigned int val = *valp;
2612 *lvalp = (unsigned long) val;
2613 }
2614
2615 return 0;
2616}
2617
2618/**
2619 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2620 * @table: the sysctl table
2621 * @write: %TRUE if this is a write to the sysctl file
2622 * @buffer: the user buffer
2623 * @lenp: the size of the user buffer
2624 * @ppos: file position
2625 *
2626 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2627 * values from/to the user buffer, treated as an ASCII string. Negative
2628 * strings are not allowed.
2629 *
2630 * This routine will ensure the values are within the range specified by
2631 * table->extra1 (min) and table->extra2 (max). There is a final sanity
2632 * check for UINT_MAX to avoid having to support wrap around uses from
2633 * userspace.
2634 *
2635 * Returns 0 on success or -ERANGE on write when the range check fails.
2636 */
2637int proc_douintvec_minmax(struct ctl_table *table, int write,
2638 void __user *buffer, size_t *lenp, loff_t *ppos)
2639{
2640 struct do_proc_douintvec_minmax_conv_param param = {
2641 .min = (unsigned int *) table->extra1,
2642 .max = (unsigned int *) table->extra2,
2643 };
2644 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2645 do_proc_douintvec_minmax_conv, ¶m);
2646}
2647
2648static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
2649 unsigned int *valp,
2650 int write, void *data)
2651{
2652 if (write) {
2653 unsigned int val;
2654
2655 val = round_pipe_size(*lvalp);
2656 if (val == 0)
2657 return -EINVAL;
2658
2659 *valp = val;
2660 } else {
2661 unsigned int val = *valp;
2662 *lvalp = (unsigned long) val;
2663 }
2664
2665 return 0;
2666}
2667
2668static int proc_dopipe_max_size(struct ctl_table *table, int write,
2669 void __user *buffer, size_t *lenp, loff_t *ppos)
2670{
2671 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2672 do_proc_dopipe_max_size_conv, NULL);
2673}
2674
2675static void validate_coredump_safety(void)
2676{
2677#ifdef CONFIG_COREDUMP
2678 if (suid_dumpable == SUID_DUMP_ROOT &&
2679 core_pattern[0] != '/' && core_pattern[0] != '|') {
2680 printk(KERN_WARNING
2681"Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2682"Pipe handler or fully qualified core dump path required.\n"
2683"Set kernel.core_pattern before fs.suid_dumpable.\n"
2684 );
2685 }
2686#endif
2687}
2688
2689static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2690 void __user *buffer, size_t *lenp, loff_t *ppos)
2691{
2692 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2693 if (!error)
2694 validate_coredump_safety();
2695 return error;
2696}
2697
2698#ifdef CONFIG_COREDUMP
2699static int proc_dostring_coredump(struct ctl_table *table, int write,
2700 void __user *buffer, size_t *lenp, loff_t *ppos)
2701{
2702 int error = proc_dostring(table, write, buffer, lenp, ppos);
2703 if (!error)
2704 validate_coredump_safety();
2705 return error;
2706}
2707#endif
2708
2709static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2710 void __user *buffer,
2711 size_t *lenp, loff_t *ppos,
2712 unsigned long convmul,
2713 unsigned long convdiv)
2714{
2715 unsigned long *i, *min, *max;
2716 int vleft, first = 1, err = 0;
2717 size_t left;
2718 char *kbuf = NULL, *p;
2719
2720 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2721 *lenp = 0;
2722 return 0;
2723 }
2724
2725 i = (unsigned long *) data;
2726 min = (unsigned long *) table->extra1;
2727 max = (unsigned long *) table->extra2;
2728 vleft = table->maxlen / sizeof(unsigned long);
2729 left = *lenp;
2730
2731 if (write) {
2732 if (proc_first_pos_non_zero_ignore(ppos, table))
2733 goto out;
2734
2735 if (left > PAGE_SIZE - 1)
2736 left = PAGE_SIZE - 1;
2737 p = kbuf = memdup_user_nul(buffer, left);
2738 if (IS_ERR(kbuf))
2739 return PTR_ERR(kbuf);
2740 }
2741
2742 for (; left && vleft--; i++, first = 0) {
2743 unsigned long val;
2744
2745 if (write) {
2746 bool neg;
2747
2748 left -= proc_skip_spaces(&p);
2749
2750 err = proc_get_long(&p, &left, &val, &neg,
2751 proc_wspace_sep,
2752 sizeof(proc_wspace_sep), NULL);
2753 if (err)
2754 break;
2755 if (neg)
2756 continue;
2757 val = convmul * val / convdiv;
2758 if ((min && val < *min) || (max && val > *max))
2759 continue;
2760 *i = val;
2761 } else {
2762 val = convdiv * (*i) / convmul;
2763 if (!first) {
2764 err = proc_put_char(&buffer, &left, '\t');
2765 if (err)
2766 break;
2767 }
2768 err = proc_put_long(&buffer, &left, val, false);
2769 if (err)
2770 break;
2771 }
2772 }
2773
2774 if (!write && !first && left && !err)
2775 err = proc_put_char(&buffer, &left, '\n');
2776 if (write && !err)
2777 left -= proc_skip_spaces(&p);
2778 if (write) {
2779 kfree(kbuf);
2780 if (first)
2781 return err ? : -EINVAL;
2782 }
2783 *lenp -= left;
2784out:
2785 *ppos += *lenp;
2786 return err;
2787}
2788
2789static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2790 void __user *buffer,
2791 size_t *lenp, loff_t *ppos,
2792 unsigned long convmul,
2793 unsigned long convdiv)
2794{
2795 return __do_proc_doulongvec_minmax(table->data, table, write,
2796 buffer, lenp, ppos, convmul, convdiv);
2797}
2798
2799/**
2800 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2801 * @table: the sysctl table
2802 * @write: %TRUE if this is a write to the sysctl file
2803 * @buffer: the user buffer
2804 * @lenp: the size of the user buffer
2805 * @ppos: file position
2806 *
2807 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2808 * values from/to the user buffer, treated as an ASCII string.
2809 *
2810 * This routine will ensure the values are within the range specified by
2811 * table->extra1 (min) and table->extra2 (max).
2812 *
2813 * Returns 0 on success.
2814 */
2815int proc_doulongvec_minmax(struct ctl_table *table, int write,
2816 void __user *buffer, size_t *lenp, loff_t *ppos)
2817{
2818 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2819}
2820
2821/**
2822 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2823 * @table: the sysctl table
2824 * @write: %TRUE if this is a write to the sysctl file
2825 * @buffer: the user buffer
2826 * @lenp: the size of the user buffer
2827 * @ppos: file position
2828 *
2829 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2830 * values from/to the user buffer, treated as an ASCII string. The values
2831 * are treated as milliseconds, and converted to jiffies when they are stored.
2832 *
2833 * This routine will ensure the values are within the range specified by
2834 * table->extra1 (min) and table->extra2 (max).
2835 *
2836 * Returns 0 on success.
2837 */
2838int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2839 void __user *buffer,
2840 size_t *lenp, loff_t *ppos)
2841{
2842 return do_proc_doulongvec_minmax(table, write, buffer,
2843 lenp, ppos, HZ, 1000l);
2844}
2845
2846
2847static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2848 int *valp,
2849 int write, void *data)
2850{
2851 if (write) {
2852 if (*lvalp > INT_MAX / HZ)
2853 return 1;
2854 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2855 } else {
2856 int val = *valp;
2857 unsigned long lval;
2858 if (val < 0) {
2859 *negp = true;
2860 lval = -(unsigned long)val;
2861 } else {
2862 *negp = false;
2863 lval = (unsigned long)val;
2864 }
2865 *lvalp = lval / HZ;
2866 }
2867 return 0;
2868}
2869
2870static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2871 int *valp,
2872 int write, void *data)
2873{
2874 if (write) {
2875 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2876 return 1;
2877 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2878 } else {
2879 int val = *valp;
2880 unsigned long lval;
2881 if (val < 0) {
2882 *negp = true;
2883 lval = -(unsigned long)val;
2884 } else {
2885 *negp = false;
2886 lval = (unsigned long)val;
2887 }
2888 *lvalp = jiffies_to_clock_t(lval);
2889 }
2890 return 0;
2891}
2892
2893static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2894 int *valp,
2895 int write, void *data)
2896{
2897 if (write) {
2898 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2899
2900 if (jif > INT_MAX)
2901 return 1;
2902 *valp = (int)jif;
2903 } else {
2904 int val = *valp;
2905 unsigned long lval;
2906 if (val < 0) {
2907 *negp = true;
2908 lval = -(unsigned long)val;
2909 } else {
2910 *negp = false;
2911 lval = (unsigned long)val;
2912 }
2913 *lvalp = jiffies_to_msecs(lval);
2914 }
2915 return 0;
2916}
2917
2918/**
2919 * proc_dointvec_jiffies - read a vector of integers as seconds
2920 * @table: the sysctl table
2921 * @write: %TRUE if this is a write to the sysctl file
2922 * @buffer: the user buffer
2923 * @lenp: the size of the user buffer
2924 * @ppos: file position
2925 *
2926 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2927 * values from/to the user buffer, treated as an ASCII string.
2928 * The values read are assumed to be in seconds, and are converted into
2929 * jiffies.
2930 *
2931 * Returns 0 on success.
2932 */
2933int proc_dointvec_jiffies(struct ctl_table *table, int write,
2934 void __user *buffer, size_t *lenp, loff_t *ppos)
2935{
2936 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2937 do_proc_dointvec_jiffies_conv,NULL);
2938}
2939
2940/**
2941 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2942 * @table: the sysctl table
2943 * @write: %TRUE if this is a write to the sysctl file
2944 * @buffer: the user buffer
2945 * @lenp: the size of the user buffer
2946 * @ppos: pointer to the file position
2947 *
2948 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2949 * values from/to the user buffer, treated as an ASCII string.
2950 * The values read are assumed to be in 1/USER_HZ seconds, and
2951 * are converted into jiffies.
2952 *
2953 * Returns 0 on success.
2954 */
2955int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2956 void __user *buffer, size_t *lenp, loff_t *ppos)
2957{
2958 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2959 do_proc_dointvec_userhz_jiffies_conv,NULL);
2960}
2961
2962/**
2963 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2964 * @table: the sysctl table
2965 * @write: %TRUE if this is a write to the sysctl file
2966 * @buffer: the user buffer
2967 * @lenp: the size of the user buffer
2968 * @ppos: file position
2969 * @ppos: the current position in the file
2970 *
2971 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2972 * values from/to the user buffer, treated as an ASCII string.
2973 * The values read are assumed to be in 1/1000 seconds, and
2974 * are converted into jiffies.
2975 *
2976 * Returns 0 on success.
2977 */
2978int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2979 void __user *buffer, size_t *lenp, loff_t *ppos)
2980{
2981 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2982 do_proc_dointvec_ms_jiffies_conv, NULL);
2983}
2984
2985static int proc_do_cad_pid(struct ctl_table *table, int write,
2986 void __user *buffer, size_t *lenp, loff_t *ppos)
2987{
2988 struct pid *new_pid;
2989 pid_t tmp;
2990 int r;
2991
2992 tmp = pid_vnr(cad_pid);
2993
2994 r = __do_proc_dointvec(&tmp, table, write, buffer,
2995 lenp, ppos, NULL, NULL);
2996 if (r || !write)
2997 return r;
2998
2999 new_pid = find_get_pid(tmp);
3000 if (!new_pid)
3001 return -ESRCH;
3002
3003 put_pid(xchg(&cad_pid, new_pid));
3004 return 0;
3005}
3006
3007/**
3008 * proc_do_large_bitmap - read/write from/to a large bitmap
3009 * @table: the sysctl table
3010 * @write: %TRUE if this is a write to the sysctl file
3011 * @buffer: the user buffer
3012 * @lenp: the size of the user buffer
3013 * @ppos: file position
3014 *
3015 * The bitmap is stored at table->data and the bitmap length (in bits)
3016 * in table->maxlen.
3017 *
3018 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
3019 * large bitmaps may be represented in a compact manner. Writing into
3020 * the file will clear the bitmap then update it with the given input.
3021 *
3022 * Returns 0 on success.
3023 */
3024int proc_do_large_bitmap(struct ctl_table *table, int write,
3025 void __user *buffer, size_t *lenp, loff_t *ppos)
3026{
3027 int err = 0;
3028 bool first = 1;
3029 size_t left = *lenp;
3030 unsigned long bitmap_len = table->maxlen;
3031 unsigned long *bitmap = *(unsigned long **) table->data;
3032 unsigned long *tmp_bitmap = NULL;
3033 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
3034
3035 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
3036 *lenp = 0;
3037 return 0;
3038 }
3039
3040 if (write) {
3041 char *kbuf, *p;
3042
3043 if (left > PAGE_SIZE - 1)
3044 left = PAGE_SIZE - 1;
3045
3046 p = kbuf = memdup_user_nul(buffer, left);
3047 if (IS_ERR(kbuf))
3048 return PTR_ERR(kbuf);
3049
3050 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
3051 GFP_KERNEL);
3052 if (!tmp_bitmap) {
3053 kfree(kbuf);
3054 return -ENOMEM;
3055 }
3056 proc_skip_char(&p, &left, '\n');
3057 while (!err && left) {
3058 unsigned long val_a, val_b;
3059 bool neg;
3060
3061 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3062 sizeof(tr_a), &c);
3063 if (err)
3064 break;
3065 if (val_a >= bitmap_len || neg) {
3066 err = -EINVAL;
3067 break;
3068 }
3069
3070 val_b = val_a;
3071 if (left) {
3072 p++;
3073 left--;
3074 }
3075
3076 if (c == '-') {
3077 err = proc_get_long(&p, &left, &val_b,
3078 &neg, tr_b, sizeof(tr_b),
3079 &c);
3080 if (err)
3081 break;
3082 if (val_b >= bitmap_len || neg ||
3083 val_a > val_b) {
3084 err = -EINVAL;
3085 break;
3086 }
3087 if (left) {
3088 p++;
3089 left--;
3090 }
3091 }
3092
3093 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3094 first = 0;
3095 proc_skip_char(&p, &left, '\n');
3096 }
3097 kfree(kbuf);
3098 } else {
3099 unsigned long bit_a, bit_b = 0;
3100
3101 while (left) {
3102 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3103 if (bit_a >= bitmap_len)
3104 break;
3105 bit_b = find_next_zero_bit(bitmap, bitmap_len,
3106 bit_a + 1) - 1;
3107
3108 if (!first) {
3109 err = proc_put_char(&buffer, &left, ',');
3110 if (err)
3111 break;
3112 }
3113 err = proc_put_long(&buffer, &left, bit_a, false);
3114 if (err)
3115 break;
3116 if (bit_a != bit_b) {
3117 err = proc_put_char(&buffer, &left, '-');
3118 if (err)
3119 break;
3120 err = proc_put_long(&buffer, &left, bit_b, false);
3121 if (err)
3122 break;
3123 }
3124
3125 first = 0; bit_b++;
3126 }
3127 if (!err)
3128 err = proc_put_char(&buffer, &left, '\n');
3129 }
3130
3131 if (!err) {
3132 if (write) {
3133 if (*ppos)
3134 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3135 else
3136 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3137 }
3138 *lenp -= left;
3139 *ppos += *lenp;
3140 }
3141
3142 kfree(tmp_bitmap);
3143 return err;
3144}
3145
3146#else /* CONFIG_PROC_SYSCTL */
3147
3148int proc_dostring(struct ctl_table *table, int write,
3149 void __user *buffer, size_t *lenp, loff_t *ppos)
3150{
3151 return -ENOSYS;
3152}
3153
3154int proc_dointvec(struct ctl_table *table, int write,
3155 void __user *buffer, size_t *lenp, loff_t *ppos)
3156{
3157 return -ENOSYS;
3158}
3159
3160int proc_douintvec(struct ctl_table *table, int write,
3161 void __user *buffer, size_t *lenp, loff_t *ppos)
3162{
3163 return -ENOSYS;
3164}
3165
3166int proc_dointvec_minmax(struct ctl_table *table, int write,
3167 void __user *buffer, size_t *lenp, loff_t *ppos)
3168{
3169 return -ENOSYS;
3170}
3171
3172int proc_douintvec_minmax(struct ctl_table *table, int write,
3173 void __user *buffer, size_t *lenp, loff_t *ppos)
3174{
3175 return -ENOSYS;
3176}
3177
3178int proc_dointvec_jiffies(struct ctl_table *table, int write,
3179 void __user *buffer, size_t *lenp, loff_t *ppos)
3180{
3181 return -ENOSYS;
3182}
3183
3184int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3185 void __user *buffer, size_t *lenp, loff_t *ppos)
3186{
3187 return -ENOSYS;
3188}
3189
3190int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3191 void __user *buffer, size_t *lenp, loff_t *ppos)
3192{
3193 return -ENOSYS;
3194}
3195
3196int proc_doulongvec_minmax(struct ctl_table *table, int write,
3197 void __user *buffer, size_t *lenp, loff_t *ppos)
3198{
3199 return -ENOSYS;
3200}
3201
3202int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3203 void __user *buffer,
3204 size_t *lenp, loff_t *ppos)
3205{
3206 return -ENOSYS;
3207}
3208
3209
3210#endif /* CONFIG_PROC_SYSCTL */
3211
3212/*
3213 * No sense putting this after each symbol definition, twice,
3214 * exception granted :-)
3215 */
3216EXPORT_SYMBOL(proc_dointvec);
3217EXPORT_SYMBOL(proc_douintvec);
3218EXPORT_SYMBOL(proc_dointvec_jiffies);
3219EXPORT_SYMBOL(proc_dointvec_minmax);
3220EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3221EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3222EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3223EXPORT_SYMBOL(proc_dostring);
3224EXPORT_SYMBOL(proc_doulongvec_minmax);
3225EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
1/*
2 * sysctl.c: General linux system control interface
3 *
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12 * Horn.
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16 * Wendling.
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19 */
20
21#include <linux/module.h>
22#include <linux/mm.h>
23#include <linux/swap.h>
24#include <linux/slab.h>
25#include <linux/sysctl.h>
26#include <linux/signal.h>
27#include <linux/printk.h>
28#include <linux/proc_fs.h>
29#include <linux/security.h>
30#include <linux/ctype.h>
31#include <linux/kmemcheck.h>
32#include <linux/fs.h>
33#include <linux/init.h>
34#include <linux/kernel.h>
35#include <linux/kobject.h>
36#include <linux/net.h>
37#include <linux/sysrq.h>
38#include <linux/highuid.h>
39#include <linux/writeback.h>
40#include <linux/ratelimit.h>
41#include <linux/compaction.h>
42#include <linux/hugetlb.h>
43#include <linux/initrd.h>
44#include <linux/key.h>
45#include <linux/times.h>
46#include <linux/limits.h>
47#include <linux/dcache.h>
48#include <linux/dnotify.h>
49#include <linux/syscalls.h>
50#include <linux/vmstat.h>
51#include <linux/nfs_fs.h>
52#include <linux/acpi.h>
53#include <linux/reboot.h>
54#include <linux/ftrace.h>
55#include <linux/perf_event.h>
56#include <linux/kprobes.h>
57#include <linux/pipe_fs_i.h>
58#include <linux/oom.h>
59#include <linux/kmod.h>
60
61#include <asm/uaccess.h>
62#include <asm/processor.h>
63
64#ifdef CONFIG_X86
65#include <asm/nmi.h>
66#include <asm/stacktrace.h>
67#include <asm/io.h>
68#endif
69#ifdef CONFIG_BSD_PROCESS_ACCT
70#include <linux/acct.h>
71#endif
72#ifdef CONFIG_RT_MUTEXES
73#include <linux/rtmutex.h>
74#endif
75#if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
76#include <linux/lockdep.h>
77#endif
78#ifdef CONFIG_CHR_DEV_SG
79#include <scsi/sg.h>
80#endif
81
82#ifdef CONFIG_LOCKUP_DETECTOR
83#include <linux/nmi.h>
84#endif
85
86
87#if defined(CONFIG_SYSCTL)
88
89/* External variables not in a header file. */
90extern int sysctl_overcommit_memory;
91extern int sysctl_overcommit_ratio;
92extern int max_threads;
93extern int core_uses_pid;
94extern int suid_dumpable;
95extern char core_pattern[];
96extern unsigned int core_pipe_limit;
97extern int pid_max;
98extern int min_free_kbytes;
99extern int pid_max_min, pid_max_max;
100extern int sysctl_drop_caches;
101extern int percpu_pagelist_fraction;
102extern int compat_log;
103extern int latencytop_enabled;
104extern int sysctl_nr_open_min, sysctl_nr_open_max;
105#ifndef CONFIG_MMU
106extern int sysctl_nr_trim_pages;
107#endif
108#ifdef CONFIG_BLOCK
109extern int blk_iopoll_enabled;
110#endif
111
112/* Constants used for minimum and maximum */
113#ifdef CONFIG_LOCKUP_DETECTOR
114static int sixty = 60;
115static int neg_one = -1;
116#endif
117
118static int zero;
119static int __maybe_unused one = 1;
120static int __maybe_unused two = 2;
121static int __maybe_unused three = 3;
122static unsigned long one_ul = 1;
123static int one_hundred = 100;
124#ifdef CONFIG_PRINTK
125static int ten_thousand = 10000;
126#endif
127
128/* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
129static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
130
131/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
132static int maxolduid = 65535;
133static int minolduid;
134static int min_percpu_pagelist_fract = 8;
135
136static int ngroups_max = NGROUPS_MAX;
137
138#ifdef CONFIG_INOTIFY_USER
139#include <linux/inotify.h>
140#endif
141#ifdef CONFIG_SPARC
142#include <asm/system.h>
143#endif
144
145#ifdef CONFIG_SPARC64
146extern int sysctl_tsb_ratio;
147#endif
148
149#ifdef __hppa__
150extern int pwrsw_enabled;
151extern int unaligned_enabled;
152#endif
153
154#ifdef CONFIG_S390
155#ifdef CONFIG_MATHEMU
156extern int sysctl_ieee_emulation_warnings;
157#endif
158extern int sysctl_userprocess_debug;
159extern int spin_retry;
160#endif
161
162#ifdef CONFIG_IA64
163extern int no_unaligned_warning;
164extern int unaligned_dump_stack;
165#endif
166
167#ifdef CONFIG_PROC_SYSCTL
168static int proc_do_cad_pid(struct ctl_table *table, int write,
169 void __user *buffer, size_t *lenp, loff_t *ppos);
170static int proc_taint(struct ctl_table *table, int write,
171 void __user *buffer, size_t *lenp, loff_t *ppos);
172#endif
173
174#ifdef CONFIG_PRINTK
175static int proc_dmesg_restrict(struct ctl_table *table, int write,
176 void __user *buffer, size_t *lenp, loff_t *ppos);
177#endif
178
179#ifdef CONFIG_MAGIC_SYSRQ
180/* Note: sysrq code uses it's own private copy */
181static int __sysrq_enabled = SYSRQ_DEFAULT_ENABLE;
182
183static int sysrq_sysctl_handler(ctl_table *table, int write,
184 void __user *buffer, size_t *lenp,
185 loff_t *ppos)
186{
187 int error;
188
189 error = proc_dointvec(table, write, buffer, lenp, ppos);
190 if (error)
191 return error;
192
193 if (write)
194 sysrq_toggle_support(__sysrq_enabled);
195
196 return 0;
197}
198
199#endif
200
201static struct ctl_table root_table[];
202static struct ctl_table_root sysctl_table_root;
203static struct ctl_table_header root_table_header = {
204 {{.count = 1,
205 .ctl_table = root_table,
206 .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),}},
207 .root = &sysctl_table_root,
208 .set = &sysctl_table_root.default_set,
209};
210static struct ctl_table_root sysctl_table_root = {
211 .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
212 .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
213};
214
215static struct ctl_table kern_table[];
216static struct ctl_table vm_table[];
217static struct ctl_table fs_table[];
218static struct ctl_table debug_table[];
219static struct ctl_table dev_table[];
220extern struct ctl_table random_table[];
221#ifdef CONFIG_EPOLL
222extern struct ctl_table epoll_table[];
223#endif
224
225#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
226int sysctl_legacy_va_layout;
227#endif
228
229/* The default sysctl tables: */
230
231static struct ctl_table root_table[] = {
232 {
233 .procname = "kernel",
234 .mode = 0555,
235 .child = kern_table,
236 },
237 {
238 .procname = "vm",
239 .mode = 0555,
240 .child = vm_table,
241 },
242 {
243 .procname = "fs",
244 .mode = 0555,
245 .child = fs_table,
246 },
247 {
248 .procname = "debug",
249 .mode = 0555,
250 .child = debug_table,
251 },
252 {
253 .procname = "dev",
254 .mode = 0555,
255 .child = dev_table,
256 },
257 { }
258};
259
260#ifdef CONFIG_SCHED_DEBUG
261static int min_sched_granularity_ns = 100000; /* 100 usecs */
262static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
263static int min_wakeup_granularity_ns; /* 0 usecs */
264static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
265static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
266static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
267#endif
268
269#ifdef CONFIG_COMPACTION
270static int min_extfrag_threshold;
271static int max_extfrag_threshold = 1000;
272#endif
273
274static struct ctl_table kern_table[] = {
275 {
276 .procname = "sched_child_runs_first",
277 .data = &sysctl_sched_child_runs_first,
278 .maxlen = sizeof(unsigned int),
279 .mode = 0644,
280 .proc_handler = proc_dointvec,
281 },
282#ifdef CONFIG_SCHED_DEBUG
283 {
284 .procname = "sched_min_granularity_ns",
285 .data = &sysctl_sched_min_granularity,
286 .maxlen = sizeof(unsigned int),
287 .mode = 0644,
288 .proc_handler = sched_proc_update_handler,
289 .extra1 = &min_sched_granularity_ns,
290 .extra2 = &max_sched_granularity_ns,
291 },
292 {
293 .procname = "sched_latency_ns",
294 .data = &sysctl_sched_latency,
295 .maxlen = sizeof(unsigned int),
296 .mode = 0644,
297 .proc_handler = sched_proc_update_handler,
298 .extra1 = &min_sched_granularity_ns,
299 .extra2 = &max_sched_granularity_ns,
300 },
301 {
302 .procname = "sched_wakeup_granularity_ns",
303 .data = &sysctl_sched_wakeup_granularity,
304 .maxlen = sizeof(unsigned int),
305 .mode = 0644,
306 .proc_handler = sched_proc_update_handler,
307 .extra1 = &min_wakeup_granularity_ns,
308 .extra2 = &max_wakeup_granularity_ns,
309 },
310 {
311 .procname = "sched_tunable_scaling",
312 .data = &sysctl_sched_tunable_scaling,
313 .maxlen = sizeof(enum sched_tunable_scaling),
314 .mode = 0644,
315 .proc_handler = sched_proc_update_handler,
316 .extra1 = &min_sched_tunable_scaling,
317 .extra2 = &max_sched_tunable_scaling,
318 },
319 {
320 .procname = "sched_migration_cost",
321 .data = &sysctl_sched_migration_cost,
322 .maxlen = sizeof(unsigned int),
323 .mode = 0644,
324 .proc_handler = proc_dointvec,
325 },
326 {
327 .procname = "sched_nr_migrate",
328 .data = &sysctl_sched_nr_migrate,
329 .maxlen = sizeof(unsigned int),
330 .mode = 0644,
331 .proc_handler = proc_dointvec,
332 },
333 {
334 .procname = "sched_time_avg",
335 .data = &sysctl_sched_time_avg,
336 .maxlen = sizeof(unsigned int),
337 .mode = 0644,
338 .proc_handler = proc_dointvec,
339 },
340 {
341 .procname = "sched_shares_window",
342 .data = &sysctl_sched_shares_window,
343 .maxlen = sizeof(unsigned int),
344 .mode = 0644,
345 .proc_handler = proc_dointvec,
346 },
347 {
348 .procname = "timer_migration",
349 .data = &sysctl_timer_migration,
350 .maxlen = sizeof(unsigned int),
351 .mode = 0644,
352 .proc_handler = proc_dointvec_minmax,
353 .extra1 = &zero,
354 .extra2 = &one,
355 },
356#endif
357 {
358 .procname = "sched_rt_period_us",
359 .data = &sysctl_sched_rt_period,
360 .maxlen = sizeof(unsigned int),
361 .mode = 0644,
362 .proc_handler = sched_rt_handler,
363 },
364 {
365 .procname = "sched_rt_runtime_us",
366 .data = &sysctl_sched_rt_runtime,
367 .maxlen = sizeof(int),
368 .mode = 0644,
369 .proc_handler = sched_rt_handler,
370 },
371#ifdef CONFIG_SCHED_AUTOGROUP
372 {
373 .procname = "sched_autogroup_enabled",
374 .data = &sysctl_sched_autogroup_enabled,
375 .maxlen = sizeof(unsigned int),
376 .mode = 0644,
377 .proc_handler = proc_dointvec_minmax,
378 .extra1 = &zero,
379 .extra2 = &one,
380 },
381#endif
382#ifdef CONFIG_PROVE_LOCKING
383 {
384 .procname = "prove_locking",
385 .data = &prove_locking,
386 .maxlen = sizeof(int),
387 .mode = 0644,
388 .proc_handler = proc_dointvec,
389 },
390#endif
391#ifdef CONFIG_LOCK_STAT
392 {
393 .procname = "lock_stat",
394 .data = &lock_stat,
395 .maxlen = sizeof(int),
396 .mode = 0644,
397 .proc_handler = proc_dointvec,
398 },
399#endif
400 {
401 .procname = "panic",
402 .data = &panic_timeout,
403 .maxlen = sizeof(int),
404 .mode = 0644,
405 .proc_handler = proc_dointvec,
406 },
407 {
408 .procname = "core_uses_pid",
409 .data = &core_uses_pid,
410 .maxlen = sizeof(int),
411 .mode = 0644,
412 .proc_handler = proc_dointvec,
413 },
414 {
415 .procname = "core_pattern",
416 .data = core_pattern,
417 .maxlen = CORENAME_MAX_SIZE,
418 .mode = 0644,
419 .proc_handler = proc_dostring,
420 },
421 {
422 .procname = "core_pipe_limit",
423 .data = &core_pipe_limit,
424 .maxlen = sizeof(unsigned int),
425 .mode = 0644,
426 .proc_handler = proc_dointvec,
427 },
428#ifdef CONFIG_PROC_SYSCTL
429 {
430 .procname = "tainted",
431 .maxlen = sizeof(long),
432 .mode = 0644,
433 .proc_handler = proc_taint,
434 },
435#endif
436#ifdef CONFIG_LATENCYTOP
437 {
438 .procname = "latencytop",
439 .data = &latencytop_enabled,
440 .maxlen = sizeof(int),
441 .mode = 0644,
442 .proc_handler = proc_dointvec,
443 },
444#endif
445#ifdef CONFIG_BLK_DEV_INITRD
446 {
447 .procname = "real-root-dev",
448 .data = &real_root_dev,
449 .maxlen = sizeof(int),
450 .mode = 0644,
451 .proc_handler = proc_dointvec,
452 },
453#endif
454 {
455 .procname = "print-fatal-signals",
456 .data = &print_fatal_signals,
457 .maxlen = sizeof(int),
458 .mode = 0644,
459 .proc_handler = proc_dointvec,
460 },
461#ifdef CONFIG_SPARC
462 {
463 .procname = "reboot-cmd",
464 .data = reboot_command,
465 .maxlen = 256,
466 .mode = 0644,
467 .proc_handler = proc_dostring,
468 },
469 {
470 .procname = "stop-a",
471 .data = &stop_a_enabled,
472 .maxlen = sizeof (int),
473 .mode = 0644,
474 .proc_handler = proc_dointvec,
475 },
476 {
477 .procname = "scons-poweroff",
478 .data = &scons_pwroff,
479 .maxlen = sizeof (int),
480 .mode = 0644,
481 .proc_handler = proc_dointvec,
482 },
483#endif
484#ifdef CONFIG_SPARC64
485 {
486 .procname = "tsb-ratio",
487 .data = &sysctl_tsb_ratio,
488 .maxlen = sizeof (int),
489 .mode = 0644,
490 .proc_handler = proc_dointvec,
491 },
492#endif
493#ifdef __hppa__
494 {
495 .procname = "soft-power",
496 .data = &pwrsw_enabled,
497 .maxlen = sizeof (int),
498 .mode = 0644,
499 .proc_handler = proc_dointvec,
500 },
501 {
502 .procname = "unaligned-trap",
503 .data = &unaligned_enabled,
504 .maxlen = sizeof (int),
505 .mode = 0644,
506 .proc_handler = proc_dointvec,
507 },
508#endif
509 {
510 .procname = "ctrl-alt-del",
511 .data = &C_A_D,
512 .maxlen = sizeof(int),
513 .mode = 0644,
514 .proc_handler = proc_dointvec,
515 },
516#ifdef CONFIG_FUNCTION_TRACER
517 {
518 .procname = "ftrace_enabled",
519 .data = &ftrace_enabled,
520 .maxlen = sizeof(int),
521 .mode = 0644,
522 .proc_handler = ftrace_enable_sysctl,
523 },
524#endif
525#ifdef CONFIG_STACK_TRACER
526 {
527 .procname = "stack_tracer_enabled",
528 .data = &stack_tracer_enabled,
529 .maxlen = sizeof(int),
530 .mode = 0644,
531 .proc_handler = stack_trace_sysctl,
532 },
533#endif
534#ifdef CONFIG_TRACING
535 {
536 .procname = "ftrace_dump_on_oops",
537 .data = &ftrace_dump_on_oops,
538 .maxlen = sizeof(int),
539 .mode = 0644,
540 .proc_handler = proc_dointvec,
541 },
542#endif
543#ifdef CONFIG_MODULES
544 {
545 .procname = "modprobe",
546 .data = &modprobe_path,
547 .maxlen = KMOD_PATH_LEN,
548 .mode = 0644,
549 .proc_handler = proc_dostring,
550 },
551 {
552 .procname = "modules_disabled",
553 .data = &modules_disabled,
554 .maxlen = sizeof(int),
555 .mode = 0644,
556 /* only handle a transition from default "0" to "1" */
557 .proc_handler = proc_dointvec_minmax,
558 .extra1 = &one,
559 .extra2 = &one,
560 },
561#endif
562#ifdef CONFIG_HOTPLUG
563 {
564 .procname = "hotplug",
565 .data = &uevent_helper,
566 .maxlen = UEVENT_HELPER_PATH_LEN,
567 .mode = 0644,
568 .proc_handler = proc_dostring,
569 },
570#endif
571#ifdef CONFIG_CHR_DEV_SG
572 {
573 .procname = "sg-big-buff",
574 .data = &sg_big_buff,
575 .maxlen = sizeof (int),
576 .mode = 0444,
577 .proc_handler = proc_dointvec,
578 },
579#endif
580#ifdef CONFIG_BSD_PROCESS_ACCT
581 {
582 .procname = "acct",
583 .data = &acct_parm,
584 .maxlen = 3*sizeof(int),
585 .mode = 0644,
586 .proc_handler = proc_dointvec,
587 },
588#endif
589#ifdef CONFIG_MAGIC_SYSRQ
590 {
591 .procname = "sysrq",
592 .data = &__sysrq_enabled,
593 .maxlen = sizeof (int),
594 .mode = 0644,
595 .proc_handler = sysrq_sysctl_handler,
596 },
597#endif
598#ifdef CONFIG_PROC_SYSCTL
599 {
600 .procname = "cad_pid",
601 .data = NULL,
602 .maxlen = sizeof (int),
603 .mode = 0600,
604 .proc_handler = proc_do_cad_pid,
605 },
606#endif
607 {
608 .procname = "threads-max",
609 .data = &max_threads,
610 .maxlen = sizeof(int),
611 .mode = 0644,
612 .proc_handler = proc_dointvec,
613 },
614 {
615 .procname = "random",
616 .mode = 0555,
617 .child = random_table,
618 },
619 {
620 .procname = "usermodehelper",
621 .mode = 0555,
622 .child = usermodehelper_table,
623 },
624 {
625 .procname = "overflowuid",
626 .data = &overflowuid,
627 .maxlen = sizeof(int),
628 .mode = 0644,
629 .proc_handler = proc_dointvec_minmax,
630 .extra1 = &minolduid,
631 .extra2 = &maxolduid,
632 },
633 {
634 .procname = "overflowgid",
635 .data = &overflowgid,
636 .maxlen = sizeof(int),
637 .mode = 0644,
638 .proc_handler = proc_dointvec_minmax,
639 .extra1 = &minolduid,
640 .extra2 = &maxolduid,
641 },
642#ifdef CONFIG_S390
643#ifdef CONFIG_MATHEMU
644 {
645 .procname = "ieee_emulation_warnings",
646 .data = &sysctl_ieee_emulation_warnings,
647 .maxlen = sizeof(int),
648 .mode = 0644,
649 .proc_handler = proc_dointvec,
650 },
651#endif
652 {
653 .procname = "userprocess_debug",
654 .data = &show_unhandled_signals,
655 .maxlen = sizeof(int),
656 .mode = 0644,
657 .proc_handler = proc_dointvec,
658 },
659#endif
660 {
661 .procname = "pid_max",
662 .data = &pid_max,
663 .maxlen = sizeof (int),
664 .mode = 0644,
665 .proc_handler = proc_dointvec_minmax,
666 .extra1 = &pid_max_min,
667 .extra2 = &pid_max_max,
668 },
669 {
670 .procname = "panic_on_oops",
671 .data = &panic_on_oops,
672 .maxlen = sizeof(int),
673 .mode = 0644,
674 .proc_handler = proc_dointvec,
675 },
676#if defined CONFIG_PRINTK
677 {
678 .procname = "printk",
679 .data = &console_loglevel,
680 .maxlen = 4*sizeof(int),
681 .mode = 0644,
682 .proc_handler = proc_dointvec,
683 },
684 {
685 .procname = "printk_ratelimit",
686 .data = &printk_ratelimit_state.interval,
687 .maxlen = sizeof(int),
688 .mode = 0644,
689 .proc_handler = proc_dointvec_jiffies,
690 },
691 {
692 .procname = "printk_ratelimit_burst",
693 .data = &printk_ratelimit_state.burst,
694 .maxlen = sizeof(int),
695 .mode = 0644,
696 .proc_handler = proc_dointvec,
697 },
698 {
699 .procname = "printk_delay",
700 .data = &printk_delay_msec,
701 .maxlen = sizeof(int),
702 .mode = 0644,
703 .proc_handler = proc_dointvec_minmax,
704 .extra1 = &zero,
705 .extra2 = &ten_thousand,
706 },
707 {
708 .procname = "dmesg_restrict",
709 .data = &dmesg_restrict,
710 .maxlen = sizeof(int),
711 .mode = 0644,
712 .proc_handler = proc_dointvec_minmax,
713 .extra1 = &zero,
714 .extra2 = &one,
715 },
716 {
717 .procname = "kptr_restrict",
718 .data = &kptr_restrict,
719 .maxlen = sizeof(int),
720 .mode = 0644,
721 .proc_handler = proc_dmesg_restrict,
722 .extra1 = &zero,
723 .extra2 = &two,
724 },
725#endif
726 {
727 .procname = "ngroups_max",
728 .data = &ngroups_max,
729 .maxlen = sizeof (int),
730 .mode = 0444,
731 .proc_handler = proc_dointvec,
732 },
733#if defined(CONFIG_LOCKUP_DETECTOR)
734 {
735 .procname = "watchdog",
736 .data = &watchdog_enabled,
737 .maxlen = sizeof (int),
738 .mode = 0644,
739 .proc_handler = proc_dowatchdog,
740 .extra1 = &zero,
741 .extra2 = &one,
742 },
743 {
744 .procname = "watchdog_thresh",
745 .data = &watchdog_thresh,
746 .maxlen = sizeof(int),
747 .mode = 0644,
748 .proc_handler = proc_dowatchdog,
749 .extra1 = &neg_one,
750 .extra2 = &sixty,
751 },
752 {
753 .procname = "softlockup_panic",
754 .data = &softlockup_panic,
755 .maxlen = sizeof(int),
756 .mode = 0644,
757 .proc_handler = proc_dointvec_minmax,
758 .extra1 = &zero,
759 .extra2 = &one,
760 },
761 {
762 .procname = "nmi_watchdog",
763 .data = &watchdog_enabled,
764 .maxlen = sizeof (int),
765 .mode = 0644,
766 .proc_handler = proc_dowatchdog,
767 .extra1 = &zero,
768 .extra2 = &one,
769 },
770#endif
771#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
772 {
773 .procname = "unknown_nmi_panic",
774 .data = &unknown_nmi_panic,
775 .maxlen = sizeof (int),
776 .mode = 0644,
777 .proc_handler = proc_dointvec,
778 },
779#endif
780#if defined(CONFIG_X86)
781 {
782 .procname = "panic_on_unrecovered_nmi",
783 .data = &panic_on_unrecovered_nmi,
784 .maxlen = sizeof(int),
785 .mode = 0644,
786 .proc_handler = proc_dointvec,
787 },
788 {
789 .procname = "panic_on_io_nmi",
790 .data = &panic_on_io_nmi,
791 .maxlen = sizeof(int),
792 .mode = 0644,
793 .proc_handler = proc_dointvec,
794 },
795 {
796 .procname = "bootloader_type",
797 .data = &bootloader_type,
798 .maxlen = sizeof (int),
799 .mode = 0444,
800 .proc_handler = proc_dointvec,
801 },
802 {
803 .procname = "bootloader_version",
804 .data = &bootloader_version,
805 .maxlen = sizeof (int),
806 .mode = 0444,
807 .proc_handler = proc_dointvec,
808 },
809 {
810 .procname = "kstack_depth_to_print",
811 .data = &kstack_depth_to_print,
812 .maxlen = sizeof(int),
813 .mode = 0644,
814 .proc_handler = proc_dointvec,
815 },
816 {
817 .procname = "io_delay_type",
818 .data = &io_delay_type,
819 .maxlen = sizeof(int),
820 .mode = 0644,
821 .proc_handler = proc_dointvec,
822 },
823#endif
824#if defined(CONFIG_MMU)
825 {
826 .procname = "randomize_va_space",
827 .data = &randomize_va_space,
828 .maxlen = sizeof(int),
829 .mode = 0644,
830 .proc_handler = proc_dointvec,
831 },
832#endif
833#if defined(CONFIG_S390) && defined(CONFIG_SMP)
834 {
835 .procname = "spin_retry",
836 .data = &spin_retry,
837 .maxlen = sizeof (int),
838 .mode = 0644,
839 .proc_handler = proc_dointvec,
840 },
841#endif
842#if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
843 {
844 .procname = "acpi_video_flags",
845 .data = &acpi_realmode_flags,
846 .maxlen = sizeof (unsigned long),
847 .mode = 0644,
848 .proc_handler = proc_doulongvec_minmax,
849 },
850#endif
851#ifdef CONFIG_IA64
852 {
853 .procname = "ignore-unaligned-usertrap",
854 .data = &no_unaligned_warning,
855 .maxlen = sizeof (int),
856 .mode = 0644,
857 .proc_handler = proc_dointvec,
858 },
859 {
860 .procname = "unaligned-dump-stack",
861 .data = &unaligned_dump_stack,
862 .maxlen = sizeof (int),
863 .mode = 0644,
864 .proc_handler = proc_dointvec,
865 },
866#endif
867#ifdef CONFIG_DETECT_HUNG_TASK
868 {
869 .procname = "hung_task_panic",
870 .data = &sysctl_hung_task_panic,
871 .maxlen = sizeof(int),
872 .mode = 0644,
873 .proc_handler = proc_dointvec_minmax,
874 .extra1 = &zero,
875 .extra2 = &one,
876 },
877 {
878 .procname = "hung_task_check_count",
879 .data = &sysctl_hung_task_check_count,
880 .maxlen = sizeof(unsigned long),
881 .mode = 0644,
882 .proc_handler = proc_doulongvec_minmax,
883 },
884 {
885 .procname = "hung_task_timeout_secs",
886 .data = &sysctl_hung_task_timeout_secs,
887 .maxlen = sizeof(unsigned long),
888 .mode = 0644,
889 .proc_handler = proc_dohung_task_timeout_secs,
890 },
891 {
892 .procname = "hung_task_warnings",
893 .data = &sysctl_hung_task_warnings,
894 .maxlen = sizeof(unsigned long),
895 .mode = 0644,
896 .proc_handler = proc_doulongvec_minmax,
897 },
898#endif
899#ifdef CONFIG_COMPAT
900 {
901 .procname = "compat-log",
902 .data = &compat_log,
903 .maxlen = sizeof (int),
904 .mode = 0644,
905 .proc_handler = proc_dointvec,
906 },
907#endif
908#ifdef CONFIG_RT_MUTEXES
909 {
910 .procname = "max_lock_depth",
911 .data = &max_lock_depth,
912 .maxlen = sizeof(int),
913 .mode = 0644,
914 .proc_handler = proc_dointvec,
915 },
916#endif
917 {
918 .procname = "poweroff_cmd",
919 .data = &poweroff_cmd,
920 .maxlen = POWEROFF_CMD_PATH_LEN,
921 .mode = 0644,
922 .proc_handler = proc_dostring,
923 },
924#ifdef CONFIG_KEYS
925 {
926 .procname = "keys",
927 .mode = 0555,
928 .child = key_sysctls,
929 },
930#endif
931#ifdef CONFIG_RCU_TORTURE_TEST
932 {
933 .procname = "rcutorture_runnable",
934 .data = &rcutorture_runnable,
935 .maxlen = sizeof(int),
936 .mode = 0644,
937 .proc_handler = proc_dointvec,
938 },
939#endif
940#ifdef CONFIG_PERF_EVENTS
941 /*
942 * User-space scripts rely on the existence of this file
943 * as a feature check for perf_events being enabled.
944 *
945 * So it's an ABI, do not remove!
946 */
947 {
948 .procname = "perf_event_paranoid",
949 .data = &sysctl_perf_event_paranoid,
950 .maxlen = sizeof(sysctl_perf_event_paranoid),
951 .mode = 0644,
952 .proc_handler = proc_dointvec,
953 },
954 {
955 .procname = "perf_event_mlock_kb",
956 .data = &sysctl_perf_event_mlock,
957 .maxlen = sizeof(sysctl_perf_event_mlock),
958 .mode = 0644,
959 .proc_handler = proc_dointvec,
960 },
961 {
962 .procname = "perf_event_max_sample_rate",
963 .data = &sysctl_perf_event_sample_rate,
964 .maxlen = sizeof(sysctl_perf_event_sample_rate),
965 .mode = 0644,
966 .proc_handler = perf_proc_update_handler,
967 },
968#endif
969#ifdef CONFIG_KMEMCHECK
970 {
971 .procname = "kmemcheck",
972 .data = &kmemcheck_enabled,
973 .maxlen = sizeof(int),
974 .mode = 0644,
975 .proc_handler = proc_dointvec,
976 },
977#endif
978#ifdef CONFIG_BLOCK
979 {
980 .procname = "blk_iopoll",
981 .data = &blk_iopoll_enabled,
982 .maxlen = sizeof(int),
983 .mode = 0644,
984 .proc_handler = proc_dointvec,
985 },
986#endif
987 { }
988};
989
990static struct ctl_table vm_table[] = {
991 {
992 .procname = "overcommit_memory",
993 .data = &sysctl_overcommit_memory,
994 .maxlen = sizeof(sysctl_overcommit_memory),
995 .mode = 0644,
996 .proc_handler = proc_dointvec_minmax,
997 .extra1 = &zero,
998 .extra2 = &two,
999 },
1000 {
1001 .procname = "panic_on_oom",
1002 .data = &sysctl_panic_on_oom,
1003 .maxlen = sizeof(sysctl_panic_on_oom),
1004 .mode = 0644,
1005 .proc_handler = proc_dointvec_minmax,
1006 .extra1 = &zero,
1007 .extra2 = &two,
1008 },
1009 {
1010 .procname = "oom_kill_allocating_task",
1011 .data = &sysctl_oom_kill_allocating_task,
1012 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1013 .mode = 0644,
1014 .proc_handler = proc_dointvec,
1015 },
1016 {
1017 .procname = "oom_dump_tasks",
1018 .data = &sysctl_oom_dump_tasks,
1019 .maxlen = sizeof(sysctl_oom_dump_tasks),
1020 .mode = 0644,
1021 .proc_handler = proc_dointvec,
1022 },
1023 {
1024 .procname = "overcommit_ratio",
1025 .data = &sysctl_overcommit_ratio,
1026 .maxlen = sizeof(sysctl_overcommit_ratio),
1027 .mode = 0644,
1028 .proc_handler = proc_dointvec,
1029 },
1030 {
1031 .procname = "page-cluster",
1032 .data = &page_cluster,
1033 .maxlen = sizeof(int),
1034 .mode = 0644,
1035 .proc_handler = proc_dointvec_minmax,
1036 .extra1 = &zero,
1037 },
1038 {
1039 .procname = "dirty_background_ratio",
1040 .data = &dirty_background_ratio,
1041 .maxlen = sizeof(dirty_background_ratio),
1042 .mode = 0644,
1043 .proc_handler = dirty_background_ratio_handler,
1044 .extra1 = &zero,
1045 .extra2 = &one_hundred,
1046 },
1047 {
1048 .procname = "dirty_background_bytes",
1049 .data = &dirty_background_bytes,
1050 .maxlen = sizeof(dirty_background_bytes),
1051 .mode = 0644,
1052 .proc_handler = dirty_background_bytes_handler,
1053 .extra1 = &one_ul,
1054 },
1055 {
1056 .procname = "dirty_ratio",
1057 .data = &vm_dirty_ratio,
1058 .maxlen = sizeof(vm_dirty_ratio),
1059 .mode = 0644,
1060 .proc_handler = dirty_ratio_handler,
1061 .extra1 = &zero,
1062 .extra2 = &one_hundred,
1063 },
1064 {
1065 .procname = "dirty_bytes",
1066 .data = &vm_dirty_bytes,
1067 .maxlen = sizeof(vm_dirty_bytes),
1068 .mode = 0644,
1069 .proc_handler = dirty_bytes_handler,
1070 .extra1 = &dirty_bytes_min,
1071 },
1072 {
1073 .procname = "dirty_writeback_centisecs",
1074 .data = &dirty_writeback_interval,
1075 .maxlen = sizeof(dirty_writeback_interval),
1076 .mode = 0644,
1077 .proc_handler = dirty_writeback_centisecs_handler,
1078 },
1079 {
1080 .procname = "dirty_expire_centisecs",
1081 .data = &dirty_expire_interval,
1082 .maxlen = sizeof(dirty_expire_interval),
1083 .mode = 0644,
1084 .proc_handler = proc_dointvec_minmax,
1085 .extra1 = &zero,
1086 },
1087 {
1088 .procname = "nr_pdflush_threads",
1089 .data = &nr_pdflush_threads,
1090 .maxlen = sizeof nr_pdflush_threads,
1091 .mode = 0444 /* read-only*/,
1092 .proc_handler = proc_dointvec,
1093 },
1094 {
1095 .procname = "swappiness",
1096 .data = &vm_swappiness,
1097 .maxlen = sizeof(vm_swappiness),
1098 .mode = 0644,
1099 .proc_handler = proc_dointvec_minmax,
1100 .extra1 = &zero,
1101 .extra2 = &one_hundred,
1102 },
1103#ifdef CONFIG_HUGETLB_PAGE
1104 {
1105 .procname = "nr_hugepages",
1106 .data = NULL,
1107 .maxlen = sizeof(unsigned long),
1108 .mode = 0644,
1109 .proc_handler = hugetlb_sysctl_handler,
1110 .extra1 = (void *)&hugetlb_zero,
1111 .extra2 = (void *)&hugetlb_infinity,
1112 },
1113#ifdef CONFIG_NUMA
1114 {
1115 .procname = "nr_hugepages_mempolicy",
1116 .data = NULL,
1117 .maxlen = sizeof(unsigned long),
1118 .mode = 0644,
1119 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1120 .extra1 = (void *)&hugetlb_zero,
1121 .extra2 = (void *)&hugetlb_infinity,
1122 },
1123#endif
1124 {
1125 .procname = "hugetlb_shm_group",
1126 .data = &sysctl_hugetlb_shm_group,
1127 .maxlen = sizeof(gid_t),
1128 .mode = 0644,
1129 .proc_handler = proc_dointvec,
1130 },
1131 {
1132 .procname = "hugepages_treat_as_movable",
1133 .data = &hugepages_treat_as_movable,
1134 .maxlen = sizeof(int),
1135 .mode = 0644,
1136 .proc_handler = hugetlb_treat_movable_handler,
1137 },
1138 {
1139 .procname = "nr_overcommit_hugepages",
1140 .data = NULL,
1141 .maxlen = sizeof(unsigned long),
1142 .mode = 0644,
1143 .proc_handler = hugetlb_overcommit_handler,
1144 .extra1 = (void *)&hugetlb_zero,
1145 .extra2 = (void *)&hugetlb_infinity,
1146 },
1147#endif
1148 {
1149 .procname = "lowmem_reserve_ratio",
1150 .data = &sysctl_lowmem_reserve_ratio,
1151 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1152 .mode = 0644,
1153 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1154 },
1155 {
1156 .procname = "drop_caches",
1157 .data = &sysctl_drop_caches,
1158 .maxlen = sizeof(int),
1159 .mode = 0644,
1160 .proc_handler = drop_caches_sysctl_handler,
1161 .extra1 = &one,
1162 .extra2 = &three,
1163 },
1164#ifdef CONFIG_COMPACTION
1165 {
1166 .procname = "compact_memory",
1167 .data = &sysctl_compact_memory,
1168 .maxlen = sizeof(int),
1169 .mode = 0200,
1170 .proc_handler = sysctl_compaction_handler,
1171 },
1172 {
1173 .procname = "extfrag_threshold",
1174 .data = &sysctl_extfrag_threshold,
1175 .maxlen = sizeof(int),
1176 .mode = 0644,
1177 .proc_handler = sysctl_extfrag_handler,
1178 .extra1 = &min_extfrag_threshold,
1179 .extra2 = &max_extfrag_threshold,
1180 },
1181
1182#endif /* CONFIG_COMPACTION */
1183 {
1184 .procname = "min_free_kbytes",
1185 .data = &min_free_kbytes,
1186 .maxlen = sizeof(min_free_kbytes),
1187 .mode = 0644,
1188 .proc_handler = min_free_kbytes_sysctl_handler,
1189 .extra1 = &zero,
1190 },
1191 {
1192 .procname = "percpu_pagelist_fraction",
1193 .data = &percpu_pagelist_fraction,
1194 .maxlen = sizeof(percpu_pagelist_fraction),
1195 .mode = 0644,
1196 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1197 .extra1 = &min_percpu_pagelist_fract,
1198 },
1199#ifdef CONFIG_MMU
1200 {
1201 .procname = "max_map_count",
1202 .data = &sysctl_max_map_count,
1203 .maxlen = sizeof(sysctl_max_map_count),
1204 .mode = 0644,
1205 .proc_handler = proc_dointvec_minmax,
1206 .extra1 = &zero,
1207 },
1208#else
1209 {
1210 .procname = "nr_trim_pages",
1211 .data = &sysctl_nr_trim_pages,
1212 .maxlen = sizeof(sysctl_nr_trim_pages),
1213 .mode = 0644,
1214 .proc_handler = proc_dointvec_minmax,
1215 .extra1 = &zero,
1216 },
1217#endif
1218 {
1219 .procname = "laptop_mode",
1220 .data = &laptop_mode,
1221 .maxlen = sizeof(laptop_mode),
1222 .mode = 0644,
1223 .proc_handler = proc_dointvec_jiffies,
1224 },
1225 {
1226 .procname = "block_dump",
1227 .data = &block_dump,
1228 .maxlen = sizeof(block_dump),
1229 .mode = 0644,
1230 .proc_handler = proc_dointvec,
1231 .extra1 = &zero,
1232 },
1233 {
1234 .procname = "vfs_cache_pressure",
1235 .data = &sysctl_vfs_cache_pressure,
1236 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1237 .mode = 0644,
1238 .proc_handler = proc_dointvec,
1239 .extra1 = &zero,
1240 },
1241#ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1242 {
1243 .procname = "legacy_va_layout",
1244 .data = &sysctl_legacy_va_layout,
1245 .maxlen = sizeof(sysctl_legacy_va_layout),
1246 .mode = 0644,
1247 .proc_handler = proc_dointvec,
1248 .extra1 = &zero,
1249 },
1250#endif
1251#ifdef CONFIG_NUMA
1252 {
1253 .procname = "zone_reclaim_mode",
1254 .data = &zone_reclaim_mode,
1255 .maxlen = sizeof(zone_reclaim_mode),
1256 .mode = 0644,
1257 .proc_handler = proc_dointvec,
1258 .extra1 = &zero,
1259 },
1260 {
1261 .procname = "min_unmapped_ratio",
1262 .data = &sysctl_min_unmapped_ratio,
1263 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1264 .mode = 0644,
1265 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1266 .extra1 = &zero,
1267 .extra2 = &one_hundred,
1268 },
1269 {
1270 .procname = "min_slab_ratio",
1271 .data = &sysctl_min_slab_ratio,
1272 .maxlen = sizeof(sysctl_min_slab_ratio),
1273 .mode = 0644,
1274 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1275 .extra1 = &zero,
1276 .extra2 = &one_hundred,
1277 },
1278#endif
1279#ifdef CONFIG_SMP
1280 {
1281 .procname = "stat_interval",
1282 .data = &sysctl_stat_interval,
1283 .maxlen = sizeof(sysctl_stat_interval),
1284 .mode = 0644,
1285 .proc_handler = proc_dointvec_jiffies,
1286 },
1287#endif
1288#ifdef CONFIG_MMU
1289 {
1290 .procname = "mmap_min_addr",
1291 .data = &dac_mmap_min_addr,
1292 .maxlen = sizeof(unsigned long),
1293 .mode = 0644,
1294 .proc_handler = mmap_min_addr_handler,
1295 },
1296#endif
1297#ifdef CONFIG_NUMA
1298 {
1299 .procname = "numa_zonelist_order",
1300 .data = &numa_zonelist_order,
1301 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1302 .mode = 0644,
1303 .proc_handler = numa_zonelist_order_handler,
1304 },
1305#endif
1306#if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1307 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1308 {
1309 .procname = "vdso_enabled",
1310 .data = &vdso_enabled,
1311 .maxlen = sizeof(vdso_enabled),
1312 .mode = 0644,
1313 .proc_handler = proc_dointvec,
1314 .extra1 = &zero,
1315 },
1316#endif
1317#ifdef CONFIG_HIGHMEM
1318 {
1319 .procname = "highmem_is_dirtyable",
1320 .data = &vm_highmem_is_dirtyable,
1321 .maxlen = sizeof(vm_highmem_is_dirtyable),
1322 .mode = 0644,
1323 .proc_handler = proc_dointvec_minmax,
1324 .extra1 = &zero,
1325 .extra2 = &one,
1326 },
1327#endif
1328 {
1329 .procname = "scan_unevictable_pages",
1330 .data = &scan_unevictable_pages,
1331 .maxlen = sizeof(scan_unevictable_pages),
1332 .mode = 0644,
1333 .proc_handler = scan_unevictable_handler,
1334 },
1335#ifdef CONFIG_MEMORY_FAILURE
1336 {
1337 .procname = "memory_failure_early_kill",
1338 .data = &sysctl_memory_failure_early_kill,
1339 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1340 .mode = 0644,
1341 .proc_handler = proc_dointvec_minmax,
1342 .extra1 = &zero,
1343 .extra2 = &one,
1344 },
1345 {
1346 .procname = "memory_failure_recovery",
1347 .data = &sysctl_memory_failure_recovery,
1348 .maxlen = sizeof(sysctl_memory_failure_recovery),
1349 .mode = 0644,
1350 .proc_handler = proc_dointvec_minmax,
1351 .extra1 = &zero,
1352 .extra2 = &one,
1353 },
1354#endif
1355 { }
1356};
1357
1358#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1359static struct ctl_table binfmt_misc_table[] = {
1360 { }
1361};
1362#endif
1363
1364static struct ctl_table fs_table[] = {
1365 {
1366 .procname = "inode-nr",
1367 .data = &inodes_stat,
1368 .maxlen = 2*sizeof(int),
1369 .mode = 0444,
1370 .proc_handler = proc_nr_inodes,
1371 },
1372 {
1373 .procname = "inode-state",
1374 .data = &inodes_stat,
1375 .maxlen = 7*sizeof(int),
1376 .mode = 0444,
1377 .proc_handler = proc_nr_inodes,
1378 },
1379 {
1380 .procname = "file-nr",
1381 .data = &files_stat,
1382 .maxlen = sizeof(files_stat),
1383 .mode = 0444,
1384 .proc_handler = proc_nr_files,
1385 },
1386 {
1387 .procname = "file-max",
1388 .data = &files_stat.max_files,
1389 .maxlen = sizeof(files_stat.max_files),
1390 .mode = 0644,
1391 .proc_handler = proc_doulongvec_minmax,
1392 },
1393 {
1394 .procname = "nr_open",
1395 .data = &sysctl_nr_open,
1396 .maxlen = sizeof(int),
1397 .mode = 0644,
1398 .proc_handler = proc_dointvec_minmax,
1399 .extra1 = &sysctl_nr_open_min,
1400 .extra2 = &sysctl_nr_open_max,
1401 },
1402 {
1403 .procname = "dentry-state",
1404 .data = &dentry_stat,
1405 .maxlen = 6*sizeof(int),
1406 .mode = 0444,
1407 .proc_handler = proc_nr_dentry,
1408 },
1409 {
1410 .procname = "overflowuid",
1411 .data = &fs_overflowuid,
1412 .maxlen = sizeof(int),
1413 .mode = 0644,
1414 .proc_handler = proc_dointvec_minmax,
1415 .extra1 = &minolduid,
1416 .extra2 = &maxolduid,
1417 },
1418 {
1419 .procname = "overflowgid",
1420 .data = &fs_overflowgid,
1421 .maxlen = sizeof(int),
1422 .mode = 0644,
1423 .proc_handler = proc_dointvec_minmax,
1424 .extra1 = &minolduid,
1425 .extra2 = &maxolduid,
1426 },
1427#ifdef CONFIG_FILE_LOCKING
1428 {
1429 .procname = "leases-enable",
1430 .data = &leases_enable,
1431 .maxlen = sizeof(int),
1432 .mode = 0644,
1433 .proc_handler = proc_dointvec,
1434 },
1435#endif
1436#ifdef CONFIG_DNOTIFY
1437 {
1438 .procname = "dir-notify-enable",
1439 .data = &dir_notify_enable,
1440 .maxlen = sizeof(int),
1441 .mode = 0644,
1442 .proc_handler = proc_dointvec,
1443 },
1444#endif
1445#ifdef CONFIG_MMU
1446#ifdef CONFIG_FILE_LOCKING
1447 {
1448 .procname = "lease-break-time",
1449 .data = &lease_break_time,
1450 .maxlen = sizeof(int),
1451 .mode = 0644,
1452 .proc_handler = proc_dointvec,
1453 },
1454#endif
1455#ifdef CONFIG_AIO
1456 {
1457 .procname = "aio-nr",
1458 .data = &aio_nr,
1459 .maxlen = sizeof(aio_nr),
1460 .mode = 0444,
1461 .proc_handler = proc_doulongvec_minmax,
1462 },
1463 {
1464 .procname = "aio-max-nr",
1465 .data = &aio_max_nr,
1466 .maxlen = sizeof(aio_max_nr),
1467 .mode = 0644,
1468 .proc_handler = proc_doulongvec_minmax,
1469 },
1470#endif /* CONFIG_AIO */
1471#ifdef CONFIG_INOTIFY_USER
1472 {
1473 .procname = "inotify",
1474 .mode = 0555,
1475 .child = inotify_table,
1476 },
1477#endif
1478#ifdef CONFIG_EPOLL
1479 {
1480 .procname = "epoll",
1481 .mode = 0555,
1482 .child = epoll_table,
1483 },
1484#endif
1485#endif
1486 {
1487 .procname = "suid_dumpable",
1488 .data = &suid_dumpable,
1489 .maxlen = sizeof(int),
1490 .mode = 0644,
1491 .proc_handler = proc_dointvec_minmax,
1492 .extra1 = &zero,
1493 .extra2 = &two,
1494 },
1495#if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1496 {
1497 .procname = "binfmt_misc",
1498 .mode = 0555,
1499 .child = binfmt_misc_table,
1500 },
1501#endif
1502 {
1503 .procname = "pipe-max-size",
1504 .data = &pipe_max_size,
1505 .maxlen = sizeof(int),
1506 .mode = 0644,
1507 .proc_handler = &pipe_proc_fn,
1508 .extra1 = &pipe_min_size,
1509 },
1510 { }
1511};
1512
1513static struct ctl_table debug_table[] = {
1514#if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
1515 defined(CONFIG_S390) || defined(CONFIG_TILE)
1516 {
1517 .procname = "exception-trace",
1518 .data = &show_unhandled_signals,
1519 .maxlen = sizeof(int),
1520 .mode = 0644,
1521 .proc_handler = proc_dointvec
1522 },
1523#endif
1524#if defined(CONFIG_OPTPROBES)
1525 {
1526 .procname = "kprobes-optimization",
1527 .data = &sysctl_kprobes_optimization,
1528 .maxlen = sizeof(int),
1529 .mode = 0644,
1530 .proc_handler = proc_kprobes_optimization_handler,
1531 .extra1 = &zero,
1532 .extra2 = &one,
1533 },
1534#endif
1535 { }
1536};
1537
1538static struct ctl_table dev_table[] = {
1539 { }
1540};
1541
1542static DEFINE_SPINLOCK(sysctl_lock);
1543
1544/* called under sysctl_lock */
1545static int use_table(struct ctl_table_header *p)
1546{
1547 if (unlikely(p->unregistering))
1548 return 0;
1549 p->used++;
1550 return 1;
1551}
1552
1553/* called under sysctl_lock */
1554static void unuse_table(struct ctl_table_header *p)
1555{
1556 if (!--p->used)
1557 if (unlikely(p->unregistering))
1558 complete(p->unregistering);
1559}
1560
1561/* called under sysctl_lock, will reacquire if has to wait */
1562static void start_unregistering(struct ctl_table_header *p)
1563{
1564 /*
1565 * if p->used is 0, nobody will ever touch that entry again;
1566 * we'll eliminate all paths to it before dropping sysctl_lock
1567 */
1568 if (unlikely(p->used)) {
1569 struct completion wait;
1570 init_completion(&wait);
1571 p->unregistering = &wait;
1572 spin_unlock(&sysctl_lock);
1573 wait_for_completion(&wait);
1574 spin_lock(&sysctl_lock);
1575 } else {
1576 /* anything non-NULL; we'll never dereference it */
1577 p->unregistering = ERR_PTR(-EINVAL);
1578 }
1579 /*
1580 * do not remove from the list until nobody holds it; walking the
1581 * list in do_sysctl() relies on that.
1582 */
1583 list_del_init(&p->ctl_entry);
1584}
1585
1586void sysctl_head_get(struct ctl_table_header *head)
1587{
1588 spin_lock(&sysctl_lock);
1589 head->count++;
1590 spin_unlock(&sysctl_lock);
1591}
1592
1593void sysctl_head_put(struct ctl_table_header *head)
1594{
1595 spin_lock(&sysctl_lock);
1596 if (!--head->count)
1597 kfree_rcu(head, rcu);
1598 spin_unlock(&sysctl_lock);
1599}
1600
1601struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1602{
1603 if (!head)
1604 BUG();
1605 spin_lock(&sysctl_lock);
1606 if (!use_table(head))
1607 head = ERR_PTR(-ENOENT);
1608 spin_unlock(&sysctl_lock);
1609 return head;
1610}
1611
1612void sysctl_head_finish(struct ctl_table_header *head)
1613{
1614 if (!head)
1615 return;
1616 spin_lock(&sysctl_lock);
1617 unuse_table(head);
1618 spin_unlock(&sysctl_lock);
1619}
1620
1621static struct ctl_table_set *
1622lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1623{
1624 struct ctl_table_set *set = &root->default_set;
1625 if (root->lookup)
1626 set = root->lookup(root, namespaces);
1627 return set;
1628}
1629
1630static struct list_head *
1631lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1632{
1633 struct ctl_table_set *set = lookup_header_set(root, namespaces);
1634 return &set->list;
1635}
1636
1637struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1638 struct ctl_table_header *prev)
1639{
1640 struct ctl_table_root *root;
1641 struct list_head *header_list;
1642 struct ctl_table_header *head;
1643 struct list_head *tmp;
1644
1645 spin_lock(&sysctl_lock);
1646 if (prev) {
1647 head = prev;
1648 tmp = &prev->ctl_entry;
1649 unuse_table(prev);
1650 goto next;
1651 }
1652 tmp = &root_table_header.ctl_entry;
1653 for (;;) {
1654 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1655
1656 if (!use_table(head))
1657 goto next;
1658 spin_unlock(&sysctl_lock);
1659 return head;
1660 next:
1661 root = head->root;
1662 tmp = tmp->next;
1663 header_list = lookup_header_list(root, namespaces);
1664 if (tmp != header_list)
1665 continue;
1666
1667 do {
1668 root = list_entry(root->root_list.next,
1669 struct ctl_table_root, root_list);
1670 if (root == &sysctl_table_root)
1671 goto out;
1672 header_list = lookup_header_list(root, namespaces);
1673 } while (list_empty(header_list));
1674 tmp = header_list->next;
1675 }
1676out:
1677 spin_unlock(&sysctl_lock);
1678 return NULL;
1679}
1680
1681struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1682{
1683 return __sysctl_head_next(current->nsproxy, prev);
1684}
1685
1686void register_sysctl_root(struct ctl_table_root *root)
1687{
1688 spin_lock(&sysctl_lock);
1689 list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1690 spin_unlock(&sysctl_lock);
1691}
1692
1693/*
1694 * sysctl_perm does NOT grant the superuser all rights automatically, because
1695 * some sysctl variables are readonly even to root.
1696 */
1697
1698static int test_perm(int mode, int op)
1699{
1700 if (!current_euid())
1701 mode >>= 6;
1702 else if (in_egroup_p(0))
1703 mode >>= 3;
1704 if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1705 return 0;
1706 return -EACCES;
1707}
1708
1709int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1710{
1711 int mode;
1712
1713 if (root->permissions)
1714 mode = root->permissions(root, current->nsproxy, table);
1715 else
1716 mode = table->mode;
1717
1718 return test_perm(mode, op);
1719}
1720
1721static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1722{
1723 for (; table->procname; table++) {
1724 table->parent = parent;
1725 if (table->child)
1726 sysctl_set_parent(table, table->child);
1727 }
1728}
1729
1730static __init int sysctl_init(void)
1731{
1732 sysctl_set_parent(NULL, root_table);
1733#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1734 sysctl_check_table(current->nsproxy, root_table);
1735#endif
1736 return 0;
1737}
1738
1739core_initcall(sysctl_init);
1740
1741static struct ctl_table *is_branch_in(struct ctl_table *branch,
1742 struct ctl_table *table)
1743{
1744 struct ctl_table *p;
1745 const char *s = branch->procname;
1746
1747 /* branch should have named subdirectory as its first element */
1748 if (!s || !branch->child)
1749 return NULL;
1750
1751 /* ... and nothing else */
1752 if (branch[1].procname)
1753 return NULL;
1754
1755 /* table should contain subdirectory with the same name */
1756 for (p = table; p->procname; p++) {
1757 if (!p->child)
1758 continue;
1759 if (p->procname && strcmp(p->procname, s) == 0)
1760 return p;
1761 }
1762 return NULL;
1763}
1764
1765/* see if attaching q to p would be an improvement */
1766static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1767{
1768 struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1769 struct ctl_table *next;
1770 int is_better = 0;
1771 int not_in_parent = !p->attached_by;
1772
1773 while ((next = is_branch_in(by, to)) != NULL) {
1774 if (by == q->attached_by)
1775 is_better = 1;
1776 if (to == p->attached_by)
1777 not_in_parent = 1;
1778 by = by->child;
1779 to = next->child;
1780 }
1781
1782 if (is_better && not_in_parent) {
1783 q->attached_by = by;
1784 q->attached_to = to;
1785 q->parent = p;
1786 }
1787}
1788
1789/**
1790 * __register_sysctl_paths - register a sysctl hierarchy
1791 * @root: List of sysctl headers to register on
1792 * @namespaces: Data to compute which lists of sysctl entries are visible
1793 * @path: The path to the directory the sysctl table is in.
1794 * @table: the top-level table structure
1795 *
1796 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1797 * array. A completely 0 filled entry terminates the table.
1798 *
1799 * The members of the &struct ctl_table structure are used as follows:
1800 *
1801 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1802 * enter a sysctl file
1803 *
1804 * data - a pointer to data for use by proc_handler
1805 *
1806 * maxlen - the maximum size in bytes of the data
1807 *
1808 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1809 *
1810 * child - a pointer to the child sysctl table if this entry is a directory, or
1811 * %NULL.
1812 *
1813 * proc_handler - the text handler routine (described below)
1814 *
1815 * de - for internal use by the sysctl routines
1816 *
1817 * extra1, extra2 - extra pointers usable by the proc handler routines
1818 *
1819 * Leaf nodes in the sysctl tree will be represented by a single file
1820 * under /proc; non-leaf nodes will be represented by directories.
1821 *
1822 * sysctl(2) can automatically manage read and write requests through
1823 * the sysctl table. The data and maxlen fields of the ctl_table
1824 * struct enable minimal validation of the values being written to be
1825 * performed, and the mode field allows minimal authentication.
1826 *
1827 * There must be a proc_handler routine for any terminal nodes
1828 * mirrored under /proc/sys (non-terminals are handled by a built-in
1829 * directory handler). Several default handlers are available to
1830 * cover common cases -
1831 *
1832 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1833 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1834 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1835 *
1836 * It is the handler's job to read the input buffer from user memory
1837 * and process it. The handler should return 0 on success.
1838 *
1839 * This routine returns %NULL on a failure to register, and a pointer
1840 * to the table header on success.
1841 */
1842struct ctl_table_header *__register_sysctl_paths(
1843 struct ctl_table_root *root,
1844 struct nsproxy *namespaces,
1845 const struct ctl_path *path, struct ctl_table *table)
1846{
1847 struct ctl_table_header *header;
1848 struct ctl_table *new, **prevp;
1849 unsigned int n, npath;
1850 struct ctl_table_set *set;
1851
1852 /* Count the path components */
1853 for (npath = 0; path[npath].procname; ++npath)
1854 ;
1855
1856 /*
1857 * For each path component, allocate a 2-element ctl_table array.
1858 * The first array element will be filled with the sysctl entry
1859 * for this, the second will be the sentinel (procname == 0).
1860 *
1861 * We allocate everything in one go so that we don't have to
1862 * worry about freeing additional memory in unregister_sysctl_table.
1863 */
1864 header = kzalloc(sizeof(struct ctl_table_header) +
1865 (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1866 if (!header)
1867 return NULL;
1868
1869 new = (struct ctl_table *) (header + 1);
1870
1871 /* Now connect the dots */
1872 prevp = &header->ctl_table;
1873 for (n = 0; n < npath; ++n, ++path) {
1874 /* Copy the procname */
1875 new->procname = path->procname;
1876 new->mode = 0555;
1877
1878 *prevp = new;
1879 prevp = &new->child;
1880
1881 new += 2;
1882 }
1883 *prevp = table;
1884 header->ctl_table_arg = table;
1885
1886 INIT_LIST_HEAD(&header->ctl_entry);
1887 header->used = 0;
1888 header->unregistering = NULL;
1889 header->root = root;
1890 sysctl_set_parent(NULL, header->ctl_table);
1891 header->count = 1;
1892#ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1893 if (sysctl_check_table(namespaces, header->ctl_table)) {
1894 kfree(header);
1895 return NULL;
1896 }
1897#endif
1898 spin_lock(&sysctl_lock);
1899 header->set = lookup_header_set(root, namespaces);
1900 header->attached_by = header->ctl_table;
1901 header->attached_to = root_table;
1902 header->parent = &root_table_header;
1903 for (set = header->set; set; set = set->parent) {
1904 struct ctl_table_header *p;
1905 list_for_each_entry(p, &set->list, ctl_entry) {
1906 if (p->unregistering)
1907 continue;
1908 try_attach(p, header);
1909 }
1910 }
1911 header->parent->count++;
1912 list_add_tail(&header->ctl_entry, &header->set->list);
1913 spin_unlock(&sysctl_lock);
1914
1915 return header;
1916}
1917
1918/**
1919 * register_sysctl_table_path - register a sysctl table hierarchy
1920 * @path: The path to the directory the sysctl table is in.
1921 * @table: the top-level table structure
1922 *
1923 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1924 * array. A completely 0 filled entry terminates the table.
1925 *
1926 * See __register_sysctl_paths for more details.
1927 */
1928struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1929 struct ctl_table *table)
1930{
1931 return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1932 path, table);
1933}
1934
1935/**
1936 * register_sysctl_table - register a sysctl table hierarchy
1937 * @table: the top-level table structure
1938 *
1939 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1940 * array. A completely 0 filled entry terminates the table.
1941 *
1942 * See register_sysctl_paths for more details.
1943 */
1944struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1945{
1946 static const struct ctl_path null_path[] = { {} };
1947
1948 return register_sysctl_paths(null_path, table);
1949}
1950
1951/**
1952 * unregister_sysctl_table - unregister a sysctl table hierarchy
1953 * @header: the header returned from register_sysctl_table
1954 *
1955 * Unregisters the sysctl table and all children. proc entries may not
1956 * actually be removed until they are no longer used by anyone.
1957 */
1958void unregister_sysctl_table(struct ctl_table_header * header)
1959{
1960 might_sleep();
1961
1962 if (header == NULL)
1963 return;
1964
1965 spin_lock(&sysctl_lock);
1966 start_unregistering(header);
1967 if (!--header->parent->count) {
1968 WARN_ON(1);
1969 kfree_rcu(header->parent, rcu);
1970 }
1971 if (!--header->count)
1972 kfree_rcu(header, rcu);
1973 spin_unlock(&sysctl_lock);
1974}
1975
1976int sysctl_is_seen(struct ctl_table_header *p)
1977{
1978 struct ctl_table_set *set = p->set;
1979 int res;
1980 spin_lock(&sysctl_lock);
1981 if (p->unregistering)
1982 res = 0;
1983 else if (!set->is_seen)
1984 res = 1;
1985 else
1986 res = set->is_seen(set);
1987 spin_unlock(&sysctl_lock);
1988 return res;
1989}
1990
1991void setup_sysctl_set(struct ctl_table_set *p,
1992 struct ctl_table_set *parent,
1993 int (*is_seen)(struct ctl_table_set *))
1994{
1995 INIT_LIST_HEAD(&p->list);
1996 p->parent = parent ? parent : &sysctl_table_root.default_set;
1997 p->is_seen = is_seen;
1998}
1999
2000#else /* !CONFIG_SYSCTL */
2001struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
2002{
2003 return NULL;
2004}
2005
2006struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
2007 struct ctl_table *table)
2008{
2009 return NULL;
2010}
2011
2012void unregister_sysctl_table(struct ctl_table_header * table)
2013{
2014}
2015
2016void setup_sysctl_set(struct ctl_table_set *p,
2017 struct ctl_table_set *parent,
2018 int (*is_seen)(struct ctl_table_set *))
2019{
2020}
2021
2022void sysctl_head_put(struct ctl_table_header *head)
2023{
2024}
2025
2026#endif /* CONFIG_SYSCTL */
2027
2028/*
2029 * /proc/sys support
2030 */
2031
2032#ifdef CONFIG_PROC_SYSCTL
2033
2034static int _proc_do_string(void* data, int maxlen, int write,
2035 void __user *buffer,
2036 size_t *lenp, loff_t *ppos)
2037{
2038 size_t len;
2039 char __user *p;
2040 char c;
2041
2042 if (!data || !maxlen || !*lenp) {
2043 *lenp = 0;
2044 return 0;
2045 }
2046
2047 if (write) {
2048 len = 0;
2049 p = buffer;
2050 while (len < *lenp) {
2051 if (get_user(c, p++))
2052 return -EFAULT;
2053 if (c == 0 || c == '\n')
2054 break;
2055 len++;
2056 }
2057 if (len >= maxlen)
2058 len = maxlen-1;
2059 if(copy_from_user(data, buffer, len))
2060 return -EFAULT;
2061 ((char *) data)[len] = 0;
2062 *ppos += *lenp;
2063 } else {
2064 len = strlen(data);
2065 if (len > maxlen)
2066 len = maxlen;
2067
2068 if (*ppos > len) {
2069 *lenp = 0;
2070 return 0;
2071 }
2072
2073 data += *ppos;
2074 len -= *ppos;
2075
2076 if (len > *lenp)
2077 len = *lenp;
2078 if (len)
2079 if(copy_to_user(buffer, data, len))
2080 return -EFAULT;
2081 if (len < *lenp) {
2082 if(put_user('\n', ((char __user *) buffer) + len))
2083 return -EFAULT;
2084 len++;
2085 }
2086 *lenp = len;
2087 *ppos += len;
2088 }
2089 return 0;
2090}
2091
2092/**
2093 * proc_dostring - read a string sysctl
2094 * @table: the sysctl table
2095 * @write: %TRUE if this is a write to the sysctl file
2096 * @buffer: the user buffer
2097 * @lenp: the size of the user buffer
2098 * @ppos: file position
2099 *
2100 * Reads/writes a string from/to the user buffer. If the kernel
2101 * buffer provided is not large enough to hold the string, the
2102 * string is truncated. The copied string is %NULL-terminated.
2103 * If the string is being read by the user process, it is copied
2104 * and a newline '\n' is added. It is truncated if the buffer is
2105 * not large enough.
2106 *
2107 * Returns 0 on success.
2108 */
2109int proc_dostring(struct ctl_table *table, int write,
2110 void __user *buffer, size_t *lenp, loff_t *ppos)
2111{
2112 return _proc_do_string(table->data, table->maxlen, write,
2113 buffer, lenp, ppos);
2114}
2115
2116static size_t proc_skip_spaces(char **buf)
2117{
2118 size_t ret;
2119 char *tmp = skip_spaces(*buf);
2120 ret = tmp - *buf;
2121 *buf = tmp;
2122 return ret;
2123}
2124
2125static void proc_skip_char(char **buf, size_t *size, const char v)
2126{
2127 while (*size) {
2128 if (**buf != v)
2129 break;
2130 (*size)--;
2131 (*buf)++;
2132 }
2133}
2134
2135#define TMPBUFLEN 22
2136/**
2137 * proc_get_long - reads an ASCII formatted integer from a user buffer
2138 *
2139 * @buf: a kernel buffer
2140 * @size: size of the kernel buffer
2141 * @val: this is where the number will be stored
2142 * @neg: set to %TRUE if number is negative
2143 * @perm_tr: a vector which contains the allowed trailers
2144 * @perm_tr_len: size of the perm_tr vector
2145 * @tr: pointer to store the trailer character
2146 *
2147 * In case of success %0 is returned and @buf and @size are updated with
2148 * the amount of bytes read. If @tr is non-NULL and a trailing
2149 * character exists (size is non-zero after returning from this
2150 * function), @tr is updated with the trailing character.
2151 */
2152static int proc_get_long(char **buf, size_t *size,
2153 unsigned long *val, bool *neg,
2154 const char *perm_tr, unsigned perm_tr_len, char *tr)
2155{
2156 int len;
2157 char *p, tmp[TMPBUFLEN];
2158
2159 if (!*size)
2160 return -EINVAL;
2161
2162 len = *size;
2163 if (len > TMPBUFLEN - 1)
2164 len = TMPBUFLEN - 1;
2165
2166 memcpy(tmp, *buf, len);
2167
2168 tmp[len] = 0;
2169 p = tmp;
2170 if (*p == '-' && *size > 1) {
2171 *neg = true;
2172 p++;
2173 } else
2174 *neg = false;
2175 if (!isdigit(*p))
2176 return -EINVAL;
2177
2178 *val = simple_strtoul(p, &p, 0);
2179
2180 len = p - tmp;
2181
2182 /* We don't know if the next char is whitespace thus we may accept
2183 * invalid integers (e.g. 1234...a) or two integers instead of one
2184 * (e.g. 123...1). So lets not allow such large numbers. */
2185 if (len == TMPBUFLEN - 1)
2186 return -EINVAL;
2187
2188 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2189 return -EINVAL;
2190
2191 if (tr && (len < *size))
2192 *tr = *p;
2193
2194 *buf += len;
2195 *size -= len;
2196
2197 return 0;
2198}
2199
2200/**
2201 * proc_put_long - converts an integer to a decimal ASCII formatted string
2202 *
2203 * @buf: the user buffer
2204 * @size: the size of the user buffer
2205 * @val: the integer to be converted
2206 * @neg: sign of the number, %TRUE for negative
2207 *
2208 * In case of success %0 is returned and @buf and @size are updated with
2209 * the amount of bytes written.
2210 */
2211static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2212 bool neg)
2213{
2214 int len;
2215 char tmp[TMPBUFLEN], *p = tmp;
2216
2217 sprintf(p, "%s%lu", neg ? "-" : "", val);
2218 len = strlen(tmp);
2219 if (len > *size)
2220 len = *size;
2221 if (copy_to_user(*buf, tmp, len))
2222 return -EFAULT;
2223 *size -= len;
2224 *buf += len;
2225 return 0;
2226}
2227#undef TMPBUFLEN
2228
2229static int proc_put_char(void __user **buf, size_t *size, char c)
2230{
2231 if (*size) {
2232 char __user **buffer = (char __user **)buf;
2233 if (put_user(c, *buffer))
2234 return -EFAULT;
2235 (*size)--, (*buffer)++;
2236 *buf = *buffer;
2237 }
2238 return 0;
2239}
2240
2241static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2242 int *valp,
2243 int write, void *data)
2244{
2245 if (write) {
2246 *valp = *negp ? -*lvalp : *lvalp;
2247 } else {
2248 int val = *valp;
2249 if (val < 0) {
2250 *negp = true;
2251 *lvalp = (unsigned long)-val;
2252 } else {
2253 *negp = false;
2254 *lvalp = (unsigned long)val;
2255 }
2256 }
2257 return 0;
2258}
2259
2260static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2261
2262static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2263 int write, void __user *buffer,
2264 size_t *lenp, loff_t *ppos,
2265 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2266 int write, void *data),
2267 void *data)
2268{
2269 int *i, vleft, first = 1, err = 0;
2270 unsigned long page = 0;
2271 size_t left;
2272 char *kbuf;
2273
2274 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2275 *lenp = 0;
2276 return 0;
2277 }
2278
2279 i = (int *) tbl_data;
2280 vleft = table->maxlen / sizeof(*i);
2281 left = *lenp;
2282
2283 if (!conv)
2284 conv = do_proc_dointvec_conv;
2285
2286 if (write) {
2287 if (left > PAGE_SIZE - 1)
2288 left = PAGE_SIZE - 1;
2289 page = __get_free_page(GFP_TEMPORARY);
2290 kbuf = (char *) page;
2291 if (!kbuf)
2292 return -ENOMEM;
2293 if (copy_from_user(kbuf, buffer, left)) {
2294 err = -EFAULT;
2295 goto free;
2296 }
2297 kbuf[left] = 0;
2298 }
2299
2300 for (; left && vleft--; i++, first=0) {
2301 unsigned long lval;
2302 bool neg;
2303
2304 if (write) {
2305 left -= proc_skip_spaces(&kbuf);
2306
2307 if (!left)
2308 break;
2309 err = proc_get_long(&kbuf, &left, &lval, &neg,
2310 proc_wspace_sep,
2311 sizeof(proc_wspace_sep), NULL);
2312 if (err)
2313 break;
2314 if (conv(&neg, &lval, i, 1, data)) {
2315 err = -EINVAL;
2316 break;
2317 }
2318 } else {
2319 if (conv(&neg, &lval, i, 0, data)) {
2320 err = -EINVAL;
2321 break;
2322 }
2323 if (!first)
2324 err = proc_put_char(&buffer, &left, '\t');
2325 if (err)
2326 break;
2327 err = proc_put_long(&buffer, &left, lval, neg);
2328 if (err)
2329 break;
2330 }
2331 }
2332
2333 if (!write && !first && left && !err)
2334 err = proc_put_char(&buffer, &left, '\n');
2335 if (write && !err && left)
2336 left -= proc_skip_spaces(&kbuf);
2337free:
2338 if (write) {
2339 free_page(page);
2340 if (first)
2341 return err ? : -EINVAL;
2342 }
2343 *lenp -= left;
2344 *ppos += *lenp;
2345 return err;
2346}
2347
2348static int do_proc_dointvec(struct ctl_table *table, int write,
2349 void __user *buffer, size_t *lenp, loff_t *ppos,
2350 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2351 int write, void *data),
2352 void *data)
2353{
2354 return __do_proc_dointvec(table->data, table, write,
2355 buffer, lenp, ppos, conv, data);
2356}
2357
2358/**
2359 * proc_dointvec - read a vector of integers
2360 * @table: the sysctl table
2361 * @write: %TRUE if this is a write to the sysctl file
2362 * @buffer: the user buffer
2363 * @lenp: the size of the user buffer
2364 * @ppos: file position
2365 *
2366 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2367 * values from/to the user buffer, treated as an ASCII string.
2368 *
2369 * Returns 0 on success.
2370 */
2371int proc_dointvec(struct ctl_table *table, int write,
2372 void __user *buffer, size_t *lenp, loff_t *ppos)
2373{
2374 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2375 NULL,NULL);
2376}
2377
2378/*
2379 * Taint values can only be increased
2380 * This means we can safely use a temporary.
2381 */
2382static int proc_taint(struct ctl_table *table, int write,
2383 void __user *buffer, size_t *lenp, loff_t *ppos)
2384{
2385 struct ctl_table t;
2386 unsigned long tmptaint = get_taint();
2387 int err;
2388
2389 if (write && !capable(CAP_SYS_ADMIN))
2390 return -EPERM;
2391
2392 t = *table;
2393 t.data = &tmptaint;
2394 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2395 if (err < 0)
2396 return err;
2397
2398 if (write) {
2399 /*
2400 * Poor man's atomic or. Not worth adding a primitive
2401 * to everyone's atomic.h for this
2402 */
2403 int i;
2404 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2405 if ((tmptaint >> i) & 1)
2406 add_taint(i);
2407 }
2408 }
2409
2410 return err;
2411}
2412
2413#ifdef CONFIG_PRINTK
2414static int proc_dmesg_restrict(struct ctl_table *table, int write,
2415 void __user *buffer, size_t *lenp, loff_t *ppos)
2416{
2417 if (write && !capable(CAP_SYS_ADMIN))
2418 return -EPERM;
2419
2420 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2421}
2422#endif
2423
2424struct do_proc_dointvec_minmax_conv_param {
2425 int *min;
2426 int *max;
2427};
2428
2429static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2430 int *valp,
2431 int write, void *data)
2432{
2433 struct do_proc_dointvec_minmax_conv_param *param = data;
2434 if (write) {
2435 int val = *negp ? -*lvalp : *lvalp;
2436 if ((param->min && *param->min > val) ||
2437 (param->max && *param->max < val))
2438 return -EINVAL;
2439 *valp = val;
2440 } else {
2441 int val = *valp;
2442 if (val < 0) {
2443 *negp = true;
2444 *lvalp = (unsigned long)-val;
2445 } else {
2446 *negp = false;
2447 *lvalp = (unsigned long)val;
2448 }
2449 }
2450 return 0;
2451}
2452
2453/**
2454 * proc_dointvec_minmax - read a vector of integers with min/max values
2455 * @table: the sysctl table
2456 * @write: %TRUE if this is a write to the sysctl file
2457 * @buffer: the user buffer
2458 * @lenp: the size of the user buffer
2459 * @ppos: file position
2460 *
2461 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2462 * values from/to the user buffer, treated as an ASCII string.
2463 *
2464 * This routine will ensure the values are within the range specified by
2465 * table->extra1 (min) and table->extra2 (max).
2466 *
2467 * Returns 0 on success.
2468 */
2469int proc_dointvec_minmax(struct ctl_table *table, int write,
2470 void __user *buffer, size_t *lenp, loff_t *ppos)
2471{
2472 struct do_proc_dointvec_minmax_conv_param param = {
2473 .min = (int *) table->extra1,
2474 .max = (int *) table->extra2,
2475 };
2476 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2477 do_proc_dointvec_minmax_conv, ¶m);
2478}
2479
2480static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2481 void __user *buffer,
2482 size_t *lenp, loff_t *ppos,
2483 unsigned long convmul,
2484 unsigned long convdiv)
2485{
2486 unsigned long *i, *min, *max;
2487 int vleft, first = 1, err = 0;
2488 unsigned long page = 0;
2489 size_t left;
2490 char *kbuf;
2491
2492 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2493 *lenp = 0;
2494 return 0;
2495 }
2496
2497 i = (unsigned long *) data;
2498 min = (unsigned long *) table->extra1;
2499 max = (unsigned long *) table->extra2;
2500 vleft = table->maxlen / sizeof(unsigned long);
2501 left = *lenp;
2502
2503 if (write) {
2504 if (left > PAGE_SIZE - 1)
2505 left = PAGE_SIZE - 1;
2506 page = __get_free_page(GFP_TEMPORARY);
2507 kbuf = (char *) page;
2508 if (!kbuf)
2509 return -ENOMEM;
2510 if (copy_from_user(kbuf, buffer, left)) {
2511 err = -EFAULT;
2512 goto free;
2513 }
2514 kbuf[left] = 0;
2515 }
2516
2517 for (; left && vleft--; i++, first = 0) {
2518 unsigned long val;
2519
2520 if (write) {
2521 bool neg;
2522
2523 left -= proc_skip_spaces(&kbuf);
2524
2525 err = proc_get_long(&kbuf, &left, &val, &neg,
2526 proc_wspace_sep,
2527 sizeof(proc_wspace_sep), NULL);
2528 if (err)
2529 break;
2530 if (neg)
2531 continue;
2532 if ((min && val < *min) || (max && val > *max))
2533 continue;
2534 *i = val;
2535 } else {
2536 val = convdiv * (*i) / convmul;
2537 if (!first)
2538 err = proc_put_char(&buffer, &left, '\t');
2539 err = proc_put_long(&buffer, &left, val, false);
2540 if (err)
2541 break;
2542 }
2543 }
2544
2545 if (!write && !first && left && !err)
2546 err = proc_put_char(&buffer, &left, '\n');
2547 if (write && !err)
2548 left -= proc_skip_spaces(&kbuf);
2549free:
2550 if (write) {
2551 free_page(page);
2552 if (first)
2553 return err ? : -EINVAL;
2554 }
2555 *lenp -= left;
2556 *ppos += *lenp;
2557 return err;
2558}
2559
2560static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2561 void __user *buffer,
2562 size_t *lenp, loff_t *ppos,
2563 unsigned long convmul,
2564 unsigned long convdiv)
2565{
2566 return __do_proc_doulongvec_minmax(table->data, table, write,
2567 buffer, lenp, ppos, convmul, convdiv);
2568}
2569
2570/**
2571 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2572 * @table: the sysctl table
2573 * @write: %TRUE if this is a write to the sysctl file
2574 * @buffer: the user buffer
2575 * @lenp: the size of the user buffer
2576 * @ppos: file position
2577 *
2578 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2579 * values from/to the user buffer, treated as an ASCII string.
2580 *
2581 * This routine will ensure the values are within the range specified by
2582 * table->extra1 (min) and table->extra2 (max).
2583 *
2584 * Returns 0 on success.
2585 */
2586int proc_doulongvec_minmax(struct ctl_table *table, int write,
2587 void __user *buffer, size_t *lenp, loff_t *ppos)
2588{
2589 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2590}
2591
2592/**
2593 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2594 * @table: the sysctl table
2595 * @write: %TRUE if this is a write to the sysctl file
2596 * @buffer: the user buffer
2597 * @lenp: the size of the user buffer
2598 * @ppos: file position
2599 *
2600 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2601 * values from/to the user buffer, treated as an ASCII string. The values
2602 * are treated as milliseconds, and converted to jiffies when they are stored.
2603 *
2604 * This routine will ensure the values are within the range specified by
2605 * table->extra1 (min) and table->extra2 (max).
2606 *
2607 * Returns 0 on success.
2608 */
2609int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2610 void __user *buffer,
2611 size_t *lenp, loff_t *ppos)
2612{
2613 return do_proc_doulongvec_minmax(table, write, buffer,
2614 lenp, ppos, HZ, 1000l);
2615}
2616
2617
2618static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2619 int *valp,
2620 int write, void *data)
2621{
2622 if (write) {
2623 if (*lvalp > LONG_MAX / HZ)
2624 return 1;
2625 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2626 } else {
2627 int val = *valp;
2628 unsigned long lval;
2629 if (val < 0) {
2630 *negp = true;
2631 lval = (unsigned long)-val;
2632 } else {
2633 *negp = false;
2634 lval = (unsigned long)val;
2635 }
2636 *lvalp = lval / HZ;
2637 }
2638 return 0;
2639}
2640
2641static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2642 int *valp,
2643 int write, void *data)
2644{
2645 if (write) {
2646 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2647 return 1;
2648 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2649 } else {
2650 int val = *valp;
2651 unsigned long lval;
2652 if (val < 0) {
2653 *negp = true;
2654 lval = (unsigned long)-val;
2655 } else {
2656 *negp = false;
2657 lval = (unsigned long)val;
2658 }
2659 *lvalp = jiffies_to_clock_t(lval);
2660 }
2661 return 0;
2662}
2663
2664static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2665 int *valp,
2666 int write, void *data)
2667{
2668 if (write) {
2669 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2670 } else {
2671 int val = *valp;
2672 unsigned long lval;
2673 if (val < 0) {
2674 *negp = true;
2675 lval = (unsigned long)-val;
2676 } else {
2677 *negp = false;
2678 lval = (unsigned long)val;
2679 }
2680 *lvalp = jiffies_to_msecs(lval);
2681 }
2682 return 0;
2683}
2684
2685/**
2686 * proc_dointvec_jiffies - read a vector of integers as seconds
2687 * @table: the sysctl table
2688 * @write: %TRUE if this is a write to the sysctl file
2689 * @buffer: the user buffer
2690 * @lenp: the size of the user buffer
2691 * @ppos: file position
2692 *
2693 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2694 * values from/to the user buffer, treated as an ASCII string.
2695 * The values read are assumed to be in seconds, and are converted into
2696 * jiffies.
2697 *
2698 * Returns 0 on success.
2699 */
2700int proc_dointvec_jiffies(struct ctl_table *table, int write,
2701 void __user *buffer, size_t *lenp, loff_t *ppos)
2702{
2703 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2704 do_proc_dointvec_jiffies_conv,NULL);
2705}
2706
2707/**
2708 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2709 * @table: the sysctl table
2710 * @write: %TRUE if this is a write to the sysctl file
2711 * @buffer: the user buffer
2712 * @lenp: the size of the user buffer
2713 * @ppos: pointer to the file position
2714 *
2715 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2716 * values from/to the user buffer, treated as an ASCII string.
2717 * The values read are assumed to be in 1/USER_HZ seconds, and
2718 * are converted into jiffies.
2719 *
2720 * Returns 0 on success.
2721 */
2722int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2723 void __user *buffer, size_t *lenp, loff_t *ppos)
2724{
2725 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2726 do_proc_dointvec_userhz_jiffies_conv,NULL);
2727}
2728
2729/**
2730 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2731 * @table: the sysctl table
2732 * @write: %TRUE if this is a write to the sysctl file
2733 * @buffer: the user buffer
2734 * @lenp: the size of the user buffer
2735 * @ppos: file position
2736 * @ppos: the current position in the file
2737 *
2738 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2739 * values from/to the user buffer, treated as an ASCII string.
2740 * The values read are assumed to be in 1/1000 seconds, and
2741 * are converted into jiffies.
2742 *
2743 * Returns 0 on success.
2744 */
2745int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2746 void __user *buffer, size_t *lenp, loff_t *ppos)
2747{
2748 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2749 do_proc_dointvec_ms_jiffies_conv, NULL);
2750}
2751
2752static int proc_do_cad_pid(struct ctl_table *table, int write,
2753 void __user *buffer, size_t *lenp, loff_t *ppos)
2754{
2755 struct pid *new_pid;
2756 pid_t tmp;
2757 int r;
2758
2759 tmp = pid_vnr(cad_pid);
2760
2761 r = __do_proc_dointvec(&tmp, table, write, buffer,
2762 lenp, ppos, NULL, NULL);
2763 if (r || !write)
2764 return r;
2765
2766 new_pid = find_get_pid(tmp);
2767 if (!new_pid)
2768 return -ESRCH;
2769
2770 put_pid(xchg(&cad_pid, new_pid));
2771 return 0;
2772}
2773
2774/**
2775 * proc_do_large_bitmap - read/write from/to a large bitmap
2776 * @table: the sysctl table
2777 * @write: %TRUE if this is a write to the sysctl file
2778 * @buffer: the user buffer
2779 * @lenp: the size of the user buffer
2780 * @ppos: file position
2781 *
2782 * The bitmap is stored at table->data and the bitmap length (in bits)
2783 * in table->maxlen.
2784 *
2785 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2786 * large bitmaps may be represented in a compact manner. Writing into
2787 * the file will clear the bitmap then update it with the given input.
2788 *
2789 * Returns 0 on success.
2790 */
2791int proc_do_large_bitmap(struct ctl_table *table, int write,
2792 void __user *buffer, size_t *lenp, loff_t *ppos)
2793{
2794 int err = 0;
2795 bool first = 1;
2796 size_t left = *lenp;
2797 unsigned long bitmap_len = table->maxlen;
2798 unsigned long *bitmap = (unsigned long *) table->data;
2799 unsigned long *tmp_bitmap = NULL;
2800 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2801
2802 if (!bitmap_len || !left || (*ppos && !write)) {
2803 *lenp = 0;
2804 return 0;
2805 }
2806
2807 if (write) {
2808 unsigned long page = 0;
2809 char *kbuf;
2810
2811 if (left > PAGE_SIZE - 1)
2812 left = PAGE_SIZE - 1;
2813
2814 page = __get_free_page(GFP_TEMPORARY);
2815 kbuf = (char *) page;
2816 if (!kbuf)
2817 return -ENOMEM;
2818 if (copy_from_user(kbuf, buffer, left)) {
2819 free_page(page);
2820 return -EFAULT;
2821 }
2822 kbuf[left] = 0;
2823
2824 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2825 GFP_KERNEL);
2826 if (!tmp_bitmap) {
2827 free_page(page);
2828 return -ENOMEM;
2829 }
2830 proc_skip_char(&kbuf, &left, '\n');
2831 while (!err && left) {
2832 unsigned long val_a, val_b;
2833 bool neg;
2834
2835 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2836 sizeof(tr_a), &c);
2837 if (err)
2838 break;
2839 if (val_a >= bitmap_len || neg) {
2840 err = -EINVAL;
2841 break;
2842 }
2843
2844 val_b = val_a;
2845 if (left) {
2846 kbuf++;
2847 left--;
2848 }
2849
2850 if (c == '-') {
2851 err = proc_get_long(&kbuf, &left, &val_b,
2852 &neg, tr_b, sizeof(tr_b),
2853 &c);
2854 if (err)
2855 break;
2856 if (val_b >= bitmap_len || neg ||
2857 val_a > val_b) {
2858 err = -EINVAL;
2859 break;
2860 }
2861 if (left) {
2862 kbuf++;
2863 left--;
2864 }
2865 }
2866
2867 while (val_a <= val_b)
2868 set_bit(val_a++, tmp_bitmap);
2869
2870 first = 0;
2871 proc_skip_char(&kbuf, &left, '\n');
2872 }
2873 free_page(page);
2874 } else {
2875 unsigned long bit_a, bit_b = 0;
2876
2877 while (left) {
2878 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2879 if (bit_a >= bitmap_len)
2880 break;
2881 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2882 bit_a + 1) - 1;
2883
2884 if (!first) {
2885 err = proc_put_char(&buffer, &left, ',');
2886 if (err)
2887 break;
2888 }
2889 err = proc_put_long(&buffer, &left, bit_a, false);
2890 if (err)
2891 break;
2892 if (bit_a != bit_b) {
2893 err = proc_put_char(&buffer, &left, '-');
2894 if (err)
2895 break;
2896 err = proc_put_long(&buffer, &left, bit_b, false);
2897 if (err)
2898 break;
2899 }
2900
2901 first = 0; bit_b++;
2902 }
2903 if (!err)
2904 err = proc_put_char(&buffer, &left, '\n');
2905 }
2906
2907 if (!err) {
2908 if (write) {
2909 if (*ppos)
2910 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2911 else
2912 memcpy(bitmap, tmp_bitmap,
2913 BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long));
2914 }
2915 kfree(tmp_bitmap);
2916 *lenp -= left;
2917 *ppos += *lenp;
2918 return 0;
2919 } else {
2920 kfree(tmp_bitmap);
2921 return err;
2922 }
2923}
2924
2925#else /* CONFIG_PROC_SYSCTL */
2926
2927int proc_dostring(struct ctl_table *table, int write,
2928 void __user *buffer, size_t *lenp, loff_t *ppos)
2929{
2930 return -ENOSYS;
2931}
2932
2933int proc_dointvec(struct ctl_table *table, int write,
2934 void __user *buffer, size_t *lenp, loff_t *ppos)
2935{
2936 return -ENOSYS;
2937}
2938
2939int proc_dointvec_minmax(struct ctl_table *table, int write,
2940 void __user *buffer, size_t *lenp, loff_t *ppos)
2941{
2942 return -ENOSYS;
2943}
2944
2945int proc_dointvec_jiffies(struct ctl_table *table, int write,
2946 void __user *buffer, size_t *lenp, loff_t *ppos)
2947{
2948 return -ENOSYS;
2949}
2950
2951int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2952 void __user *buffer, size_t *lenp, loff_t *ppos)
2953{
2954 return -ENOSYS;
2955}
2956
2957int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2958 void __user *buffer, size_t *lenp, loff_t *ppos)
2959{
2960 return -ENOSYS;
2961}
2962
2963int proc_doulongvec_minmax(struct ctl_table *table, int write,
2964 void __user *buffer, size_t *lenp, loff_t *ppos)
2965{
2966 return -ENOSYS;
2967}
2968
2969int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2970 void __user *buffer,
2971 size_t *lenp, loff_t *ppos)
2972{
2973 return -ENOSYS;
2974}
2975
2976
2977#endif /* CONFIG_PROC_SYSCTL */
2978
2979/*
2980 * No sense putting this after each symbol definition, twice,
2981 * exception granted :-)
2982 */
2983EXPORT_SYMBOL(proc_dointvec);
2984EXPORT_SYMBOL(proc_dointvec_jiffies);
2985EXPORT_SYMBOL(proc_dointvec_minmax);
2986EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2987EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2988EXPORT_SYMBOL(proc_dostring);
2989EXPORT_SYMBOL(proc_doulongvec_minmax);
2990EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2991EXPORT_SYMBOL(register_sysctl_table);
2992EXPORT_SYMBOL(register_sysctl_paths);
2993EXPORT_SYMBOL(unregister_sysctl_table);