?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
mm: close PageTail race
[linux/fpc-iii.git] / kernel / sysctl.c
blob49e13e1f8fe6a5e481edb3026ae20360918986c3
1 /*
2 * sysctl.c: General linux system control interface
3 *
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12 * Horn.
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16 * Wendling.
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19 */
20
21 #include <linux/module.h>
22 #include <linux/mm.h>
23 #include <linux/swap.h>
24 #include <linux/slab.h>
25 #include <linux/sysctl.h>
26 #include <linux/bitmap.h>
27 #include <linux/signal.h>
28 #include <linux/printk.h>
29 #include <linux/proc_fs.h>
30 #include <linux/security.h>
31 #include <linux/ctype.h>
32 #include <linux/kmemcheck.h>
33 #include <linux/kmemleak.h>
34 #include <linux/fs.h>
35 #include <linux/init.h>
36 #include <linux/kernel.h>
37 #include <linux/kobject.h>
38 #include <linux/net.h>
39 #include <linux/sysrq.h>
40 #include <linux/highuid.h>
41 #include <linux/writeback.h>
42 #include <linux/ratelimit.h>
43 #include <linux/compaction.h>
44 #include <linux/hugetlb.h>
45 #include <linux/initrd.h>
46 #include <linux/key.h>
47 #include <linux/times.h>
48 #include <linux/limits.h>
49 #include <linux/dcache.h>
50 #include <linux/dnotify.h>
51 #include <linux/syscalls.h>
52 #include <linux/vmstat.h>
53 #include <linux/nfs_fs.h>
54 #include <linux/acpi.h>
55 #include <linux/reboot.h>
56 #include <linux/ftrace.h>
57 #include <linux/perf_event.h>
58 #include <linux/kprobes.h>
59 #include <linux/pipe_fs_i.h>
60 #include <linux/oom.h>
61 #include <linux/kmod.h>
62 #include <linux/capability.h>
63 #include <linux/binfmts.h>
64 #include <linux/sched/sysctl.h>
65 #include <linux/kexec.h>
66
67 #include <asm/uaccess.h>
68 #include <asm/processor.h>
69
70 #ifdef CONFIG_X86
71 #include <asm/nmi.h>
72 #include <asm/stacktrace.h>
73 #include <asm/io.h>
74 #endif
75 #ifdef CONFIG_SPARC
76 #include <asm/setup.h>
77 #endif
78 #ifdef CONFIG_BSD_PROCESS_ACCT
79 #include <linux/acct.h>
80 #endif
81 #ifdef CONFIG_RT_MUTEXES
82 #include <linux/rtmutex.h>
83 #endif
84 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
85 #include <linux/lockdep.h>
86 #endif
87 #ifdef CONFIG_CHR_DEV_SG
88 #include <scsi/sg.h>
89 #endif
90
91 #ifdef CONFIG_LOCKUP_DETECTOR
92 #include <linux/nmi.h>
93 #endif
94
95
96 #if defined(CONFIG_SYSCTL)
97
98 /* External variables not in a header file. */
99 extern int max_threads;
100 extern int suid_dumpable;
101 #ifdef CONFIG_COREDUMP
102 extern int core_uses_pid;
103 extern char core_pattern[];
104 extern unsigned int core_pipe_limit;
105 #endif
106 extern int pid_max;
107 extern int pid_max_min, pid_max_max;
108 extern int percpu_pagelist_fraction;
109 extern int compat_log;
110 extern int latencytop_enabled;
111 extern int sysctl_nr_open_min, sysctl_nr_open_max;
112 #ifndef CONFIG_MMU
113 extern int sysctl_nr_trim_pages;
114 #endif
115 #ifdef CONFIG_BLOCK
116 extern int blk_iopoll_enabled;
117 #endif
118
119 /* Constants used for minimum and maximum */
120 #ifdef CONFIG_LOCKUP_DETECTOR
121 static int sixty = 60;
122 #endif
123
124 static int __maybe_unused neg_one = -1;
125
126 static int zero;
127 static int __maybe_unused one = 1;
128 static int __maybe_unused two = 2;
129 static int __maybe_unused three = 3;
130 static unsigned long one_ul = 1;
131 static int one_hundred = 100;
132 #ifdef CONFIG_PRINTK
133 static int ten_thousand = 10000;
134 #endif
135
136 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
137 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
138
139 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
140 static int maxolduid = 65535;
141 static int minolduid;
142 static int min_percpu_pagelist_fract = 8;
143
144 static int ngroups_max = NGROUPS_MAX;
145 static const int cap_last_cap = CAP_LAST_CAP;
146
147 #ifdef CONFIG_INOTIFY_USER
148 #include <linux/inotify.h>
149 #endif
150 #ifdef CONFIG_SPARC
151 #endif
152
153 #ifdef CONFIG_SPARC64
154 extern int sysctl_tsb_ratio;
155 #endif
156
157 #ifdef __hppa__
158 extern int pwrsw_enabled;
159 #endif
160
161 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
162 extern int unaligned_enabled;
163 #endif
164
165 #ifdef CONFIG_IA64
166 extern int unaligned_dump_stack;
167 #endif
168
169 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
170 extern int no_unaligned_warning;
171 #endif
172
173 #ifdef CONFIG_PROC_SYSCTL
174 static int proc_do_cad_pid(struct ctl_table *table, int write,
175 void __user *buffer, size_t *lenp, loff_t *ppos);
176 static int proc_taint(struct ctl_table *table, int write,
177 void __user *buffer, size_t *lenp, loff_t *ppos);
178 #endif
179
180 #ifdef CONFIG_PRINTK
181 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
182 void __user *buffer, size_t *lenp, loff_t *ppos);
183 #endif
184
185 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
186 void __user *buffer, size_t *lenp, loff_t *ppos);
187 #ifdef CONFIG_COREDUMP
188 static int proc_dostring_coredump(struct ctl_table *table, int write,
189 void __user *buffer, size_t *lenp, loff_t *ppos);
190 #endif
191
192 #ifdef CONFIG_MAGIC_SYSRQ
193 /* Note: sysrq code uses it's own private copy */
194 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
195
196 static int sysrq_sysctl_handler(ctl_table *table, int write,
197 void __user *buffer, size_t *lenp,
198 loff_t *ppos)
199 {
200 int error;
201
202 error = proc_dointvec(table, write, buffer, lenp, ppos);
203 if (error)
204 return error;
205
206 if (write)
207 sysrq_toggle_support(__sysrq_enabled);
208
209 return 0;
210 }
211
212 #endif
213
214 static struct ctl_table kern_table[];
215 static struct ctl_table vm_table[];
216 static struct ctl_table fs_table[];
217 static struct ctl_table debug_table[];
218 static struct ctl_table dev_table[];
219 extern struct ctl_table random_table[];
220 #ifdef CONFIG_EPOLL
221 extern struct ctl_table epoll_table[];
222 #endif
223
224 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
225 int sysctl_legacy_va_layout;
226 #endif
227
228 /* The default sysctl tables: */
229
230 static struct ctl_table sysctl_base_table[] = {
231 {
232 .procname = "kernel",
233 .mode = 0555,
234 .child = kern_table,
235 },
236 {
237 .procname = "vm",
238 .mode = 0555,
239 .child = vm_table,
240 },
241 {
242 .procname = "fs",
243 .mode = 0555,
244 .child = fs_table,
245 },
246 {
247 .procname = "debug",
248 .mode = 0555,
249 .child = debug_table,
250 },
251 {
252 .procname = "dev",
253 .mode = 0555,
254 .child = dev_table,
255 },
256 { }
257 };
258
259 #ifdef CONFIG_SCHED_DEBUG
260 static int min_sched_granularity_ns = 100000; /* 100 usecs */
261 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
262 static int min_wakeup_granularity_ns; /* 0 usecs */
263 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
264 #ifdef CONFIG_SMP
265 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
266 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
267 #endif /* CONFIG_SMP */
268 #endif /* CONFIG_SCHED_DEBUG */
269
270 #ifdef CONFIG_COMPACTION
271 static int min_extfrag_threshold;
272 static int max_extfrag_threshold = 1000;
273 #endif
274
275 static struct ctl_table kern_table[] = {
276 {
277 .procname = "sched_child_runs_first",
278 .data = &sysctl_sched_child_runs_first,
279 .maxlen = sizeof(unsigned int),
280 .mode = 0644,
281 .proc_handler = proc_dointvec,
282 },
283 #ifdef CONFIG_SCHED_DEBUG
284 {
285 .procname = "sched_min_granularity_ns",
286 .data = &sysctl_sched_min_granularity,
287 .maxlen = sizeof(unsigned int),
288 .mode = 0644,
289 .proc_handler = sched_proc_update_handler,
290 .extra1 = &min_sched_granularity_ns,
291 .extra2 = &max_sched_granularity_ns,
292 },
293 {
294 .procname = "sched_latency_ns",
295 .data = &sysctl_sched_latency,
296 .maxlen = sizeof(unsigned int),
297 .mode = 0644,
298 .proc_handler = sched_proc_update_handler,
299 .extra1 = &min_sched_granularity_ns,
300 .extra2 = &max_sched_granularity_ns,
301 },
302 {
303 .procname = "sched_wakeup_granularity_ns",
304 .data = &sysctl_sched_wakeup_granularity,
305 .maxlen = sizeof(unsigned int),
306 .mode = 0644,
307 .proc_handler = sched_proc_update_handler,
308 .extra1 = &min_wakeup_granularity_ns,
309 .extra2 = &max_wakeup_granularity_ns,
310 },
311 #ifdef CONFIG_SMP
312 {
313 .procname = "sched_tunable_scaling",
314 .data = &sysctl_sched_tunable_scaling,
315 .maxlen = sizeof(enum sched_tunable_scaling),
316 .mode = 0644,
317 .proc_handler = sched_proc_update_handler,
318 .extra1 = &min_sched_tunable_scaling,
319 .extra2 = &max_sched_tunable_scaling,
320 },
321 {
322 .procname = "sched_migration_cost_ns",
323 .data = &sysctl_sched_migration_cost,
324 .maxlen = sizeof(unsigned int),
325 .mode = 0644,
326 .proc_handler = proc_dointvec,
327 },
328 {
329 .procname = "sched_nr_migrate",
330 .data = &sysctl_sched_nr_migrate,
331 .maxlen = sizeof(unsigned int),
332 .mode = 0644,
333 .proc_handler = proc_dointvec,
334 },
335 {
336 .procname = "sched_time_avg_ms",
337 .data = &sysctl_sched_time_avg,
338 .maxlen = sizeof(unsigned int),
339 .mode = 0644,
340 .proc_handler = proc_dointvec,
341 },
342 {
343 .procname = "sched_shares_window_ns",
344 .data = &sysctl_sched_shares_window,
345 .maxlen = sizeof(unsigned int),
346 .mode = 0644,
347 .proc_handler = proc_dointvec,
348 },
349 {
350 .procname = "timer_migration",
351 .data = &sysctl_timer_migration,
352 .maxlen = sizeof(unsigned int),
353 .mode = 0644,
354 .proc_handler = proc_dointvec_minmax,
355 .extra1 = &zero,
356 .extra2 = &one,
357 },
358 #endif /* CONFIG_SMP */
359 #ifdef CONFIG_NUMA_BALANCING
360 {
361 .procname = "numa_balancing_scan_delay_ms",
362 .data = &sysctl_numa_balancing_scan_delay,
363 .maxlen = sizeof(unsigned int),
364 .mode = 0644,
365 .proc_handler = proc_dointvec,
366 },
367 {
368 .procname = "numa_balancing_scan_period_min_ms",
369 .data = &sysctl_numa_balancing_scan_period_min,
370 .maxlen = sizeof(unsigned int),
371 .mode = 0644,
372 .proc_handler = proc_dointvec,
373 },
374 {
375 .procname = "numa_balancing_scan_period_max_ms",
376 .data = &sysctl_numa_balancing_scan_period_max,
377 .maxlen = sizeof(unsigned int),
378 .mode = 0644,
379 .proc_handler = proc_dointvec,
380 },
381 {
382 .procname = "numa_balancing_scan_size_mb",
383 .data = &sysctl_numa_balancing_scan_size,
384 .maxlen = sizeof(unsigned int),
385 .mode = 0644,
386 .proc_handler = proc_dointvec,
387 },
388 {
389 .procname = "numa_balancing_migrate_deferred",
390 .data = &sysctl_numa_balancing_migrate_deferred,
391 .maxlen = sizeof(unsigned int),
392 .mode = 0644,
393 .proc_handler = proc_dointvec,
394 },
395 {
396 .procname = "numa_balancing",
397 .data = NULL, /* filled in by handler */
398 .maxlen = sizeof(unsigned int),
399 .mode = 0644,
400 .proc_handler = sysctl_numa_balancing,
401 .extra1 = &zero,
402 .extra2 = &one,
403 },
404 #endif /* CONFIG_NUMA_BALANCING */
405 #endif /* CONFIG_SCHED_DEBUG */
406 {
407 .procname = "sched_rt_period_us",
408 .data = &sysctl_sched_rt_period,
409 .maxlen = sizeof(unsigned int),
410 .mode = 0644,
411 .proc_handler = sched_rt_handler,
412 },
413 {
414 .procname = "sched_rt_runtime_us",
415 .data = &sysctl_sched_rt_runtime,
416 .maxlen = sizeof(int),
417 .mode = 0644,
418 .proc_handler = sched_rt_handler,
419 },
420 {
421 .procname = "sched_rr_timeslice_ms",
422 .data = &sched_rr_timeslice,
423 .maxlen = sizeof(int),
424 .mode = 0644,
425 .proc_handler = sched_rr_handler,
426 },
427 #ifdef CONFIG_SCHED_AUTOGROUP
428 {
429 .procname = "sched_autogroup_enabled",
430 .data = &sysctl_sched_autogroup_enabled,
431 .maxlen = sizeof(unsigned int),
432 .mode = 0644,
433 .proc_handler = proc_dointvec_minmax,
434 .extra1 = &zero,
435 .extra2 = &one,
436 },
437 #endif
438 #ifdef CONFIG_CFS_BANDWIDTH
439 {
440 .procname = "sched_cfs_bandwidth_slice_us",
441 .data = &sysctl_sched_cfs_bandwidth_slice,
442 .maxlen = sizeof(unsigned int),
443 .mode = 0644,
444 .proc_handler = proc_dointvec_minmax,
445 .extra1 = &one,
446 },
447 #endif
448 #ifdef CONFIG_PROVE_LOCKING
449 {
450 .procname = "prove_locking",
451 .data = &prove_locking,
452 .maxlen = sizeof(int),
453 .mode = 0644,
454 .proc_handler = proc_dointvec,
455 },
456 #endif
457 #ifdef CONFIG_LOCK_STAT
458 {
459 .procname = "lock_stat",
460 .data = &lock_stat,
461 .maxlen = sizeof(int),
462 .mode = 0644,
463 .proc_handler = proc_dointvec,
464 },
465 #endif
466 {
467 .procname = "panic",
468 .data = &panic_timeout,
469 .maxlen = sizeof(int),
470 .mode = 0644,
471 .proc_handler = proc_dointvec,
472 },
473 #ifdef CONFIG_COREDUMP
474 {
475 .procname = "core_uses_pid",
476 .data = &core_uses_pid,
477 .maxlen = sizeof(int),
478 .mode = 0644,
479 .proc_handler = proc_dointvec,
480 },
481 {
482 .procname = "core_pattern",
483 .data = core_pattern,
484 .maxlen = CORENAME_MAX_SIZE,
485 .mode = 0644,
486 .proc_handler = proc_dostring_coredump,
487 },
488 {
489 .procname = "core_pipe_limit",
490 .data = &core_pipe_limit,
491 .maxlen = sizeof(unsigned int),
492 .mode = 0644,
493 .proc_handler = proc_dointvec,
494 },
495 #endif
496 #ifdef CONFIG_PROC_SYSCTL
497 {
498 .procname = "tainted",
499 .maxlen = sizeof(long),
500 .mode = 0644,
501 .proc_handler = proc_taint,
502 },
503 #endif
504 #ifdef CONFIG_LATENCYTOP
505 {
506 .procname = "latencytop",
507 .data = &latencytop_enabled,
508 .maxlen = sizeof(int),
509 .mode = 0644,
510 .proc_handler = proc_dointvec,
511 },
512 #endif
513 #ifdef CONFIG_BLK_DEV_INITRD
514 {
515 .procname = "real-root-dev",
516 .data = &real_root_dev,
517 .maxlen = sizeof(int),
518 .mode = 0644,
519 .proc_handler = proc_dointvec,
520 },
521 #endif
522 {
523 .procname = "print-fatal-signals",
524 .data = &print_fatal_signals,
525 .maxlen = sizeof(int),
526 .mode = 0644,
527 .proc_handler = proc_dointvec,
528 },
529 #ifdef CONFIG_SPARC
530 {
531 .procname = "reboot-cmd",
532 .data = reboot_command,
533 .maxlen = 256,
534 .mode = 0644,
535 .proc_handler = proc_dostring,
536 },
537 {
538 .procname = "stop-a",
539 .data = &stop_a_enabled,
540 .maxlen = sizeof (int),
541 .mode = 0644,
542 .proc_handler = proc_dointvec,
543 },
544 {
545 .procname = "scons-poweroff",
546 .data = &scons_pwroff,
547 .maxlen = sizeof (int),
548 .mode = 0644,
549 .proc_handler = proc_dointvec,
550 },
551 #endif
552 #ifdef CONFIG_SPARC64
553 {
554 .procname = "tsb-ratio",
555 .data = &sysctl_tsb_ratio,
556 .maxlen = sizeof (int),
557 .mode = 0644,
558 .proc_handler = proc_dointvec,
559 },
560 #endif
561 #ifdef __hppa__
562 {
563 .procname = "soft-power",
564 .data = &pwrsw_enabled,
565 .maxlen = sizeof (int),
566 .mode = 0644,
567 .proc_handler = proc_dointvec,
568 },
569 #endif
570 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
571 {
572 .procname = "unaligned-trap",
573 .data = &unaligned_enabled,
574 .maxlen = sizeof (int),
575 .mode = 0644,
576 .proc_handler = proc_dointvec,
577 },
578 #endif
579 {
580 .procname = "ctrl-alt-del",
581 .data = &C_A_D,
582 .maxlen = sizeof(int),
583 .mode = 0644,
584 .proc_handler = proc_dointvec,
585 },
586 #ifdef CONFIG_FUNCTION_TRACER
587 {
588 .procname = "ftrace_enabled",
589 .data = &ftrace_enabled,
590 .maxlen = sizeof(int),
591 .mode = 0644,
592 .proc_handler = ftrace_enable_sysctl,
593 },
594 #endif
595 #ifdef CONFIG_STACK_TRACER
596 {
597 .procname = "stack_tracer_enabled",
598 .data = &stack_tracer_enabled,
599 .maxlen = sizeof(int),
600 .mode = 0644,
601 .proc_handler = stack_trace_sysctl,
602 },
603 #endif
604 #ifdef CONFIG_TRACING
605 {
606 .procname = "ftrace_dump_on_oops",
607 .data = &ftrace_dump_on_oops,
608 .maxlen = sizeof(int),
609 .mode = 0644,
610 .proc_handler = proc_dointvec,
611 },
612 {
613 .procname = "traceoff_on_warning",
614 .data = &__disable_trace_on_warning,
615 .maxlen = sizeof(__disable_trace_on_warning),
616 .mode = 0644,
617 .proc_handler = proc_dointvec,
618 },
619 #endif
620 #ifdef CONFIG_KEXEC
621 {
622 .procname = "kexec_load_disabled",
623 .data = &kexec_load_disabled,
624 .maxlen = sizeof(int),
625 .mode = 0644,
626 /* only handle a transition from default "0" to "1" */
627 .proc_handler = proc_dointvec_minmax,
628 .extra1 = &one,
629 .extra2 = &one,
630 },
631 #endif
632 #ifdef CONFIG_MODULES
633 {
634 .procname = "modprobe",
635 .data = &modprobe_path,
636 .maxlen = KMOD_PATH_LEN,
637 .mode = 0644,
638 .proc_handler = proc_dostring,
639 },
640 {
641 .procname = "modules_disabled",
642 .data = &modules_disabled,
643 .maxlen = sizeof(int),
644 .mode = 0644,
645 /* only handle a transition from default "0" to "1" */
646 .proc_handler = proc_dointvec_minmax,
647 .extra1 = &one,
648 .extra2 = &one,
649 },
650 #endif
651
652 {
653 .procname = "hotplug",
654 .data = &uevent_helper,
655 .maxlen = UEVENT_HELPER_PATH_LEN,
656 .mode = 0644,
657 .proc_handler = proc_dostring,
658 },
659
660 #ifdef CONFIG_CHR_DEV_SG
661 {
662 .procname = "sg-big-buff",
663 .data = &sg_big_buff,
664 .maxlen = sizeof (int),
665 .mode = 0444,
666 .proc_handler = proc_dointvec,
667 },
668 #endif
669 #ifdef CONFIG_BSD_PROCESS_ACCT
670 {
671 .procname = "acct",
672 .data = &acct_parm,
673 .maxlen = 3*sizeof(int),
674 .mode = 0644,
675 .proc_handler = proc_dointvec,
676 },
677 #endif
678 #ifdef CONFIG_MAGIC_SYSRQ
679 {
680 .procname = "sysrq",
681 .data = &__sysrq_enabled,
682 .maxlen = sizeof (int),
683 .mode = 0644,
684 .proc_handler = sysrq_sysctl_handler,
685 },
686 #endif
687 #ifdef CONFIG_PROC_SYSCTL
688 {
689 .procname = "cad_pid",
690 .data = NULL,
691 .maxlen = sizeof (int),
692 .mode = 0600,
693 .proc_handler = proc_do_cad_pid,
694 },
695 #endif
696 {
697 .procname = "threads-max",
698 .data = &max_threads,
699 .maxlen = sizeof(int),
700 .mode = 0644,
701 .proc_handler = proc_dointvec,
702 },
703 {
704 .procname = "random",
705 .mode = 0555,
706 .child = random_table,
707 },
708 {
709 .procname = "usermodehelper",
710 .mode = 0555,
711 .child = usermodehelper_table,
712 },
713 {
714 .procname = "overflowuid",
715 .data = &overflowuid,
716 .maxlen = sizeof(int),
717 .mode = 0644,
718 .proc_handler = proc_dointvec_minmax,
719 .extra1 = &minolduid,
720 .extra2 = &maxolduid,
721 },
722 {
723 .procname = "overflowgid",
724 .data = &overflowgid,
725 .maxlen = sizeof(int),
726 .mode = 0644,
727 .proc_handler = proc_dointvec_minmax,
728 .extra1 = &minolduid,
729 .extra2 = &maxolduid,
730 },
731 #ifdef CONFIG_S390
732 #ifdef CONFIG_MATHEMU
733 {
734 .procname = "ieee_emulation_warnings",
735 .data = &sysctl_ieee_emulation_warnings,
736 .maxlen = sizeof(int),
737 .mode = 0644,
738 .proc_handler = proc_dointvec,
739 },
740 #endif
741 {
742 .procname = "userprocess_debug",
743 .data = &show_unhandled_signals,
744 .maxlen = sizeof(int),
745 .mode = 0644,
746 .proc_handler = proc_dointvec,
747 },
748 #endif
749 {
750 .procname = "pid_max",
751 .data = &pid_max,
752 .maxlen = sizeof (int),
753 .mode = 0644,
754 .proc_handler = proc_dointvec_minmax,
755 .extra1 = &pid_max_min,
756 .extra2 = &pid_max_max,
757 },
758 {
759 .procname = "panic_on_oops",
760 .data = &panic_on_oops,
761 .maxlen = sizeof(int),
762 .mode = 0644,
763 .proc_handler = proc_dointvec,
764 },
765 #if defined CONFIG_PRINTK
766 {
767 .procname = "printk",
768 .data = &console_loglevel,
769 .maxlen = 4*sizeof(int),
770 .mode = 0644,
771 .proc_handler = proc_dointvec,
772 },
773 {
774 .procname = "printk_ratelimit",
775 .data = &printk_ratelimit_state.interval,
776 .maxlen = sizeof(int),
777 .mode = 0644,
778 .proc_handler = proc_dointvec_jiffies,
779 },
780 {
781 .procname = "printk_ratelimit_burst",
782 .data = &printk_ratelimit_state.burst,
783 .maxlen = sizeof(int),
784 .mode = 0644,
785 .proc_handler = proc_dointvec,
786 },
787 {
788 .procname = "printk_delay",
789 .data = &printk_delay_msec,
790 .maxlen = sizeof(int),
791 .mode = 0644,
792 .proc_handler = proc_dointvec_minmax,
793 .extra1 = &zero,
794 .extra2 = &ten_thousand,
795 },
796 {
797 .procname = "dmesg_restrict",
798 .data = &dmesg_restrict,
799 .maxlen = sizeof(int),
800 .mode = 0644,
801 .proc_handler = proc_dointvec_minmax_sysadmin,
802 .extra1 = &zero,
803 .extra2 = &one,
804 },
805 {
806 .procname = "kptr_restrict",
807 .data = &kptr_restrict,
808 .maxlen = sizeof(int),
809 .mode = 0644,
810 .proc_handler = proc_dointvec_minmax_sysadmin,
811 .extra1 = &zero,
812 .extra2 = &two,
813 },
814 #endif
815 {
816 .procname = "ngroups_max",
817 .data = &ngroups_max,
818 .maxlen = sizeof (int),
819 .mode = 0444,
820 .proc_handler = proc_dointvec,
821 },
822 {
823 .procname = "cap_last_cap",
824 .data = (void *)&cap_last_cap,
825 .maxlen = sizeof(int),
826 .mode = 0444,
827 .proc_handler = proc_dointvec,
828 },
829 #if defined(CONFIG_LOCKUP_DETECTOR)
830 {
831 .procname = "watchdog",
832 .data = &watchdog_user_enabled,
833 .maxlen = sizeof (int),
834 .mode = 0644,
835 .proc_handler = proc_dowatchdog,
836 .extra1 = &zero,
837 .extra2 = &one,
838 },
839 {
840 .procname = "watchdog_thresh",
841 .data = &watchdog_thresh,
842 .maxlen = sizeof(int),
843 .mode = 0644,
844 .proc_handler = proc_dowatchdog,
845 .extra1 = &zero,
846 .extra2 = &sixty,
847 },
848 {
849 .procname = "softlockup_panic",
850 .data = &softlockup_panic,
851 .maxlen = sizeof(int),
852 .mode = 0644,
853 .proc_handler = proc_dointvec_minmax,
854 .extra1 = &zero,
855 .extra2 = &one,
856 },
857 {
858 .procname = "nmi_watchdog",
859 .data = &watchdog_user_enabled,
860 .maxlen = sizeof (int),
861 .mode = 0644,
862 .proc_handler = proc_dowatchdog,
863 .extra1 = &zero,
864 .extra2 = &one,
865 },
866 #endif
867 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
868 {
869 .procname = "unknown_nmi_panic",
870 .data = &unknown_nmi_panic,
871 .maxlen = sizeof (int),
872 .mode = 0644,
873 .proc_handler = proc_dointvec,
874 },
875 #endif
876 #if defined(CONFIG_X86)
877 {
878 .procname = "panic_on_unrecovered_nmi",
879 .data = &panic_on_unrecovered_nmi,
880 .maxlen = sizeof(int),
881 .mode = 0644,
882 .proc_handler = proc_dointvec,
883 },
884 {
885 .procname = "panic_on_io_nmi",
886 .data = &panic_on_io_nmi,
887 .maxlen = sizeof(int),
888 .mode = 0644,
889 .proc_handler = proc_dointvec,
890 },
891 #ifdef CONFIG_DEBUG_STACKOVERFLOW
892 {
893 .procname = "panic_on_stackoverflow",
894 .data = &sysctl_panic_on_stackoverflow,
895 .maxlen = sizeof(int),
896 .mode = 0644,
897 .proc_handler = proc_dointvec,
898 },
899 #endif
900 {
901 .procname = "bootloader_type",
902 .data = &bootloader_type,
903 .maxlen = sizeof (int),
904 .mode = 0444,
905 .proc_handler = proc_dointvec,
906 },
907 {
908 .procname = "bootloader_version",
909 .data = &bootloader_version,
910 .maxlen = sizeof (int),
911 .mode = 0444,
912 .proc_handler = proc_dointvec,
913 },
914 {
915 .procname = "kstack_depth_to_print",
916 .data = &kstack_depth_to_print,
917 .maxlen = sizeof(int),
918 .mode = 0644,
919 .proc_handler = proc_dointvec,
920 },
921 {
922 .procname = "io_delay_type",
923 .data = &io_delay_type,
924 .maxlen = sizeof(int),
925 .mode = 0644,
926 .proc_handler = proc_dointvec,
927 },
928 #endif
929 #if defined(CONFIG_MMU)
930 {
931 .procname = "randomize_va_space",
932 .data = &randomize_va_space,
933 .maxlen = sizeof(int),
934 .mode = 0644,
935 .proc_handler = proc_dointvec,
936 },
937 #endif
938 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
939 {
940 .procname = "spin_retry",
941 .data = &spin_retry,
942 .maxlen = sizeof (int),
943 .mode = 0644,
944 .proc_handler = proc_dointvec,
945 },
946 #endif
947 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
948 {
949 .procname = "acpi_video_flags",
950 .data = &acpi_realmode_flags,
951 .maxlen = sizeof (unsigned long),
952 .mode = 0644,
953 .proc_handler = proc_doulongvec_minmax,
954 },
955 #endif
956 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
957 {
958 .procname = "ignore-unaligned-usertrap",
959 .data = &no_unaligned_warning,
960 .maxlen = sizeof (int),
961 .mode = 0644,
962 .proc_handler = proc_dointvec,
963 },
964 #endif
965 #ifdef CONFIG_IA64
966 {
967 .procname = "unaligned-dump-stack",
968 .data = &unaligned_dump_stack,
969 .maxlen = sizeof (int),
970 .mode = 0644,
971 .proc_handler = proc_dointvec,
972 },
973 #endif
974 #ifdef CONFIG_DETECT_HUNG_TASK
975 {
976 .procname = "hung_task_panic",
977 .data = &sysctl_hung_task_panic,
978 .maxlen = sizeof(int),
979 .mode = 0644,
980 .proc_handler = proc_dointvec_minmax,
981 .extra1 = &zero,
982 .extra2 = &one,
983 },
984 {
985 .procname = "hung_task_check_count",
986 .data = &sysctl_hung_task_check_count,
987 .maxlen = sizeof(int),
988 .mode = 0644,
989 .proc_handler = proc_dointvec_minmax,
990 .extra1 = &zero,
991 },
992 {
993 .procname = "hung_task_timeout_secs",
994 .data = &sysctl_hung_task_timeout_secs,
995 .maxlen = sizeof(unsigned long),
996 .mode = 0644,
997 .proc_handler = proc_dohung_task_timeout_secs,
998 },
999 {
1000 .procname = "hung_task_warnings",
1001 .data = &sysctl_hung_task_warnings,
1002 .maxlen = sizeof(int),
1003 .mode = 0644,
1004 .proc_handler = proc_dointvec_minmax,
1005 .extra1 = &neg_one,
1006 },
1007 #endif
1008 #ifdef CONFIG_COMPAT
1009 {
1010 .procname = "compat-log",
1011 .data = &compat_log,
1012 .maxlen = sizeof (int),
1013 .mode = 0644,
1014 .proc_handler = proc_dointvec,
1015 },
1016 #endif
1017 #ifdef CONFIG_RT_MUTEXES
1018 {
1019 .procname = "max_lock_depth",
1020 .data = &max_lock_depth,
1021 .maxlen = sizeof(int),
1022 .mode = 0644,
1023 .proc_handler = proc_dointvec,
1024 },
1025 #endif
1026 {
1027 .procname = "poweroff_cmd",
1028 .data = &poweroff_cmd,
1029 .maxlen = POWEROFF_CMD_PATH_LEN,
1030 .mode = 0644,
1031 .proc_handler = proc_dostring,
1032 },
1033 #ifdef CONFIG_KEYS
1034 {
1035 .procname = "keys",
1036 .mode = 0555,
1037 .child = key_sysctls,
1038 },
1039 #endif
1040 #ifdef CONFIG_RCU_TORTURE_TEST
1041 {
1042 .procname = "rcutorture_runnable",
1043 .data = &rcutorture_runnable,
1044 .maxlen = sizeof(int),
1045 .mode = 0644,
1046 .proc_handler = proc_dointvec,
1047 },
1048 #endif
1049 #ifdef CONFIG_PERF_EVENTS
1050 /*
1051 * User-space scripts rely on the existence of this file
1052 * as a feature check for perf_events being enabled.
1053 *
1054 * So it's an ABI, do not remove!
1055 */
1056 {
1057 .procname = "perf_event_paranoid",
1058 .data = &sysctl_perf_event_paranoid,
1059 .maxlen = sizeof(sysctl_perf_event_paranoid),
1060 .mode = 0644,
1061 .proc_handler = proc_dointvec,
1062 },
1063 {
1064 .procname = "perf_event_mlock_kb",
1065 .data = &sysctl_perf_event_mlock,
1066 .maxlen = sizeof(sysctl_perf_event_mlock),
1067 .mode = 0644,
1068 .proc_handler = proc_dointvec,
1069 },
1070 {
1071 .procname = "perf_event_max_sample_rate",
1072 .data = &sysctl_perf_event_sample_rate,
1073 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1074 .mode = 0644,
1075 .proc_handler = perf_proc_update_handler,
1076 .extra1 = &one,
1077 },
1078 {
1079 .procname = "perf_cpu_time_max_percent",
1080 .data = &sysctl_perf_cpu_time_max_percent,
1081 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1082 .mode = 0644,
1083 .proc_handler = perf_cpu_time_max_percent_handler,
1084 .extra1 = &zero,
1085 .extra2 = &one_hundred,
1086 },
1087 #endif
1088 #ifdef CONFIG_KMEMCHECK
1089 {
1090 .procname = "kmemcheck",
1091 .data = &kmemcheck_enabled,
1092 .maxlen = sizeof(int),
1093 .mode = 0644,
1094 .proc_handler = proc_dointvec,
1095 },
1096 #endif
1097 #ifdef CONFIG_BLOCK
1098 {
1099 .procname = "blk_iopoll",
1100 .data = &blk_iopoll_enabled,
1101 .maxlen = sizeof(int),
1102 .mode = 0644,
1103 .proc_handler = proc_dointvec,
1104 },
1105 #endif
1106 { }
1107 };
1108
1109 static struct ctl_table vm_table[] = {
1110 {
1111 .procname = "overcommit_memory",
1112 .data = &sysctl_overcommit_memory,
1113 .maxlen = sizeof(sysctl_overcommit_memory),
1114 .mode = 0644,
1115 .proc_handler = proc_dointvec_minmax,
1116 .extra1 = &zero,
1117 .extra2 = &two,
1118 },
1119 {
1120 .procname = "panic_on_oom",
1121 .data = &sysctl_panic_on_oom,
1122 .maxlen = sizeof(sysctl_panic_on_oom),
1123 .mode = 0644,
1124 .proc_handler = proc_dointvec_minmax,
1125 .extra1 = &zero,
1126 .extra2 = &two,
1127 },
1128 {
1129 .procname = "oom_kill_allocating_task",
1130 .data = &sysctl_oom_kill_allocating_task,
1131 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1132 .mode = 0644,
1133 .proc_handler = proc_dointvec,
1134 },
1135 {
1136 .procname = "oom_dump_tasks",
1137 .data = &sysctl_oom_dump_tasks,
1138 .maxlen = sizeof(sysctl_oom_dump_tasks),
1139 .mode = 0644,
1140 .proc_handler = proc_dointvec,
1141 },
1142 {
1143 .procname = "overcommit_ratio",
1144 .data = &sysctl_overcommit_ratio,
1145 .maxlen = sizeof(sysctl_overcommit_ratio),
1146 .mode = 0644,
1147 .proc_handler = overcommit_ratio_handler,
1148 },
1149 {
1150 .procname = "overcommit_kbytes",
1151 .data = &sysctl_overcommit_kbytes,
1152 .maxlen = sizeof(sysctl_overcommit_kbytes),
1153 .mode = 0644,
1154 .proc_handler = overcommit_kbytes_handler,
1155 },
1156 {
1157 .procname = "page-cluster",
1158 .data = &page_cluster,
1159 .maxlen = sizeof(int),
1160 .mode = 0644,
1161 .proc_handler = proc_dointvec_minmax,
1162 .extra1 = &zero,
1163 },
1164 {
1165 .procname = "dirty_background_ratio",
1166 .data = &dirty_background_ratio,
1167 .maxlen = sizeof(dirty_background_ratio),
1168 .mode = 0644,
1169 .proc_handler = dirty_background_ratio_handler,
1170 .extra1 = &zero,
1171 .extra2 = &one_hundred,
1172 },
1173 {
1174 .procname = "dirty_background_bytes",
1175 .data = &dirty_background_bytes,
1176 .maxlen = sizeof(dirty_background_bytes),
1177 .mode = 0644,
1178 .proc_handler = dirty_background_bytes_handler,
1179 .extra1 = &one_ul,
1180 },
1181 {
1182 .procname = "dirty_ratio",
1183 .data = &vm_dirty_ratio,
1184 .maxlen = sizeof(vm_dirty_ratio),
1185 .mode = 0644,
1186 .proc_handler = dirty_ratio_handler,
1187 .extra1 = &zero,
1188 .extra2 = &one_hundred,
1189 },
1190 {
1191 .procname = "dirty_bytes",
1192 .data = &vm_dirty_bytes,
1193 .maxlen = sizeof(vm_dirty_bytes),
1194 .mode = 0644,
1195 .proc_handler = dirty_bytes_handler,
1196 .extra1 = &dirty_bytes_min,
1197 },
1198 {
1199 .procname = "dirty_writeback_centisecs",
1200 .data = &dirty_writeback_interval,
1201 .maxlen = sizeof(dirty_writeback_interval),
1202 .mode = 0644,
1203 .proc_handler = dirty_writeback_centisecs_handler,
1204 },
1205 {
1206 .procname = "dirty_expire_centisecs",
1207 .data = &dirty_expire_interval,
1208 .maxlen = sizeof(dirty_expire_interval),
1209 .mode = 0644,
1210 .proc_handler = proc_dointvec_minmax,
1211 .extra1 = &zero,
1212 },
1213 {
1214 .procname = "nr_pdflush_threads",
1215 .mode = 0444 /* read-only */,
1216 .proc_handler = pdflush_proc_obsolete,
1217 },
1218 {
1219 .procname = "swappiness",
1220 .data = &vm_swappiness,
1221 .maxlen = sizeof(vm_swappiness),
1222 .mode = 0644,
1223 .proc_handler = proc_dointvec_minmax,
1224 .extra1 = &zero,
1225 .extra2 = &one_hundred,
1226 },
1227 #ifdef CONFIG_HUGETLB_PAGE
1228 {
1229 .procname = "nr_hugepages",
1230 .data = NULL,
1231 .maxlen = sizeof(unsigned long),
1232 .mode = 0644,
1233 .proc_handler = hugetlb_sysctl_handler,
1234 .extra1 = (void *)&hugetlb_zero,
1235 .extra2 = (void *)&hugetlb_infinity,
1236 },
1237 #ifdef CONFIG_NUMA
1238 {
1239 .procname = "nr_hugepages_mempolicy",
1240 .data = NULL,
1241 .maxlen = sizeof(unsigned long),
1242 .mode = 0644,
1243 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1244 .extra1 = (void *)&hugetlb_zero,
1245 .extra2 = (void *)&hugetlb_infinity,
1246 },
1247 #endif
1248 {
1249 .procname = "hugetlb_shm_group",
1250 .data = &sysctl_hugetlb_shm_group,
1251 .maxlen = sizeof(gid_t),
1252 .mode = 0644,
1253 .proc_handler = proc_dointvec,
1254 },
1255 {
1256 .procname = "hugepages_treat_as_movable",
1257 .data = &hugepages_treat_as_movable,
1258 .maxlen = sizeof(int),
1259 .mode = 0644,
1260 .proc_handler = proc_dointvec,
1261 },
1262 {
1263 .procname = "nr_overcommit_hugepages",
1264 .data = NULL,
1265 .maxlen = sizeof(unsigned long),
1266 .mode = 0644,
1267 .proc_handler = hugetlb_overcommit_handler,
1268 .extra1 = (void *)&hugetlb_zero,
1269 .extra2 = (void *)&hugetlb_infinity,
1270 },
1271 #endif
1272 {
1273 .procname = "lowmem_reserve_ratio",
1274 .data = &sysctl_lowmem_reserve_ratio,
1275 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1276 .mode = 0644,
1277 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1278 },
1279 {
1280 .procname = "drop_caches",
1281 .data = &sysctl_drop_caches,
1282 .maxlen = sizeof(int),
1283 .mode = 0644,
1284 .proc_handler = drop_caches_sysctl_handler,
1285 .extra1 = &one,
1286 .extra2 = &three,
1287 },
1288 #ifdef CONFIG_COMPACTION
1289 {
1290 .procname = "compact_memory",
1291 .data = &sysctl_compact_memory,
1292 .maxlen = sizeof(int),
1293 .mode = 0200,
1294 .proc_handler = sysctl_compaction_handler,
1295 },
1296 {
1297 .procname = "extfrag_threshold",
1298 .data = &sysctl_extfrag_threshold,
1299 .maxlen = sizeof(int),
1300 .mode = 0644,
1301 .proc_handler = sysctl_extfrag_handler,
1302 .extra1 = &min_extfrag_threshold,
1303 .extra2 = &max_extfrag_threshold,
1304 },
1305
1306 #endif /* CONFIG_COMPACTION */
1307 {
1308 .procname = "min_free_kbytes",
1309 .data = &min_free_kbytes,
1310 .maxlen = sizeof(min_free_kbytes),
1311 .mode = 0644,
1312 .proc_handler = min_free_kbytes_sysctl_handler,
1313 .extra1 = &zero,
1314 },
1315 {
1316 .procname = "percpu_pagelist_fraction",
1317 .data = &percpu_pagelist_fraction,
1318 .maxlen = sizeof(percpu_pagelist_fraction),
1319 .mode = 0644,
1320 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1321 .extra1 = &min_percpu_pagelist_fract,
1322 },
1323 #ifdef CONFIG_MMU
1324 {
1325 .procname = "max_map_count",
1326 .data = &sysctl_max_map_count,
1327 .maxlen = sizeof(sysctl_max_map_count),
1328 .mode = 0644,
1329 .proc_handler = proc_dointvec_minmax,
1330 .extra1 = &zero,
1331 },
1332 #else
1333 {
1334 .procname = "nr_trim_pages",
1335 .data = &sysctl_nr_trim_pages,
1336 .maxlen = sizeof(sysctl_nr_trim_pages),
1337 .mode = 0644,
1338 .proc_handler = proc_dointvec_minmax,
1339 .extra1 = &zero,
1340 },
1341 #endif
1342 {
1343 .procname = "laptop_mode",
1344 .data = &laptop_mode,
1345 .maxlen = sizeof(laptop_mode),
1346 .mode = 0644,
1347 .proc_handler = proc_dointvec_jiffies,
1348 },
1349 {
1350 .procname = "block_dump",
1351 .data = &block_dump,
1352 .maxlen = sizeof(block_dump),
1353 .mode = 0644,
1354 .proc_handler = proc_dointvec,
1355 .extra1 = &zero,
1356 },
1357 {
1358 .procname = "vfs_cache_pressure",
1359 .data = &sysctl_vfs_cache_pressure,
1360 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1361 .mode = 0644,
1362 .proc_handler = proc_dointvec,
1363 .extra1 = &zero,
1364 },
1365 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1366 {
1367 .procname = "legacy_va_layout",
1368 .data = &sysctl_legacy_va_layout,
1369 .maxlen = sizeof(sysctl_legacy_va_layout),
1370 .mode = 0644,
1371 .proc_handler = proc_dointvec,
1372 .extra1 = &zero,
1373 },
1374 #endif
1375 #ifdef CONFIG_NUMA
1376 {
1377 .procname = "zone_reclaim_mode",
1378 .data = &zone_reclaim_mode,
1379 .maxlen = sizeof(zone_reclaim_mode),
1380 .mode = 0644,
1381 .proc_handler = proc_dointvec,
1382 .extra1 = &zero,
1383 },
1384 {
1385 .procname = "min_unmapped_ratio",
1386 .data = &sysctl_min_unmapped_ratio,
1387 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1388 .mode = 0644,
1389 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1390 .extra1 = &zero,
1391 .extra2 = &one_hundred,
1392 },
1393 {
1394 .procname = "min_slab_ratio",
1395 .data = &sysctl_min_slab_ratio,
1396 .maxlen = sizeof(sysctl_min_slab_ratio),
1397 .mode = 0644,
1398 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1399 .extra1 = &zero,
1400 .extra2 = &one_hundred,
1401 },
1402 #endif
1403 #ifdef CONFIG_SMP
1404 {
1405 .procname = "stat_interval",
1406 .data = &sysctl_stat_interval,
1407 .maxlen = sizeof(sysctl_stat_interval),
1408 .mode = 0644,
1409 .proc_handler = proc_dointvec_jiffies,
1410 },
1411 #endif
1412 #ifdef CONFIG_MMU
1413 {
1414 .procname = "mmap_min_addr",
1415 .data = &dac_mmap_min_addr,
1416 .maxlen = sizeof(unsigned long),
1417 .mode = 0644,
1418 .proc_handler = mmap_min_addr_handler,
1419 },
1420 #endif
1421 #ifdef CONFIG_NUMA
1422 {
1423 .procname = "numa_zonelist_order",
1424 .data = &numa_zonelist_order,
1425 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1426 .mode = 0644,
1427 .proc_handler = numa_zonelist_order_handler,
1428 },
1429 #endif
1430 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1431 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1432 {
1433 .procname = "vdso_enabled",
1434 .data = &vdso_enabled,
1435 .maxlen = sizeof(vdso_enabled),
1436 .mode = 0644,
1437 .proc_handler = proc_dointvec,
1438 .extra1 = &zero,
1439 },
1440 #endif
1441 #ifdef CONFIG_HIGHMEM
1442 {
1443 .procname = "highmem_is_dirtyable",
1444 .data = &vm_highmem_is_dirtyable,
1445 .maxlen = sizeof(vm_highmem_is_dirtyable),
1446 .mode = 0644,
1447 .proc_handler = proc_dointvec_minmax,
1448 .extra1 = &zero,
1449 .extra2 = &one,
1450 },
1451 #endif
1452 {
1453 .procname = "scan_unevictable_pages",
1454 .data = &scan_unevictable_pages,
1455 .maxlen = sizeof(scan_unevictable_pages),
1456 .mode = 0644,
1457 .proc_handler = scan_unevictable_handler,
1458 },
1459 #ifdef CONFIG_MEMORY_FAILURE
1460 {
1461 .procname = "memory_failure_early_kill",
1462 .data = &sysctl_memory_failure_early_kill,
1463 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1464 .mode = 0644,
1465 .proc_handler = proc_dointvec_minmax,
1466 .extra1 = &zero,
1467 .extra2 = &one,
1468 },
1469 {
1470 .procname = "memory_failure_recovery",
1471 .data = &sysctl_memory_failure_recovery,
1472 .maxlen = sizeof(sysctl_memory_failure_recovery),
1473 .mode = 0644,
1474 .proc_handler = proc_dointvec_minmax,
1475 .extra1 = &zero,
1476 .extra2 = &one,
1477 },
1478 #endif
1479 {
1480 .procname = "user_reserve_kbytes",
1481 .data = &sysctl_user_reserve_kbytes,
1482 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1483 .mode = 0644,
1484 .proc_handler = proc_doulongvec_minmax,
1485 },
1486 {
1487 .procname = "admin_reserve_kbytes",
1488 .data = &sysctl_admin_reserve_kbytes,
1489 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1490 .mode = 0644,
1491 .proc_handler = proc_doulongvec_minmax,
1492 },
1493 { }
1494 };
1495
1496 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1497 static struct ctl_table binfmt_misc_table[] = {
1498 { }
1499 };
1500 #endif
1501
1502 static struct ctl_table fs_table[] = {
1503 {
1504 .procname = "inode-nr",
1505 .data = &inodes_stat,
1506 .maxlen = 2*sizeof(long),
1507 .mode = 0444,
1508 .proc_handler = proc_nr_inodes,
1509 },
1510 {
1511 .procname = "inode-state",
1512 .data = &inodes_stat,
1513 .maxlen = 7*sizeof(long),
1514 .mode = 0444,
1515 .proc_handler = proc_nr_inodes,
1516 },
1517 {
1518 .procname = "file-nr",
1519 .data = &files_stat,
1520 .maxlen = sizeof(files_stat),
1521 .mode = 0444,
1522 .proc_handler = proc_nr_files,
1523 },
1524 {
1525 .procname = "file-max",
1526 .data = &files_stat.max_files,
1527 .maxlen = sizeof(files_stat.max_files),
1528 .mode = 0644,
1529 .proc_handler = proc_doulongvec_minmax,
1530 },
1531 {
1532 .procname = "nr_open",
1533 .data = &sysctl_nr_open,
1534 .maxlen = sizeof(int),
1535 .mode = 0644,
1536 .proc_handler = proc_dointvec_minmax,
1537 .extra1 = &sysctl_nr_open_min,
1538 .extra2 = &sysctl_nr_open_max,
1539 },
1540 {
1541 .procname = "dentry-state",
1542 .data = &dentry_stat,
1543 .maxlen = 6*sizeof(long),
1544 .mode = 0444,
1545 .proc_handler = proc_nr_dentry,
1546 },
1547 {
1548 .procname = "overflowuid",
1549 .data = &fs_overflowuid,
1550 .maxlen = sizeof(int),
1551 .mode = 0644,
1552 .proc_handler = proc_dointvec_minmax,
1553 .extra1 = &minolduid,
1554 .extra2 = &maxolduid,
1555 },
1556 {
1557 .procname = "overflowgid",
1558 .data = &fs_overflowgid,
1559 .maxlen = sizeof(int),
1560 .mode = 0644,
1561 .proc_handler = proc_dointvec_minmax,
1562 .extra1 = &minolduid,
1563 .extra2 = &maxolduid,
1564 },
1565 #ifdef CONFIG_FILE_LOCKING
1566 {
1567 .procname = "leases-enable",
1568 .data = &leases_enable,
1569 .maxlen = sizeof(int),
1570 .mode = 0644,
1571 .proc_handler = proc_dointvec,
1572 },
1573 #endif
1574 #ifdef CONFIG_DNOTIFY
1575 {
1576 .procname = "dir-notify-enable",
1577 .data = &dir_notify_enable,
1578 .maxlen = sizeof(int),
1579 .mode = 0644,
1580 .proc_handler = proc_dointvec,
1581 },
1582 #endif
1583 #ifdef CONFIG_MMU
1584 #ifdef CONFIG_FILE_LOCKING
1585 {
1586 .procname = "lease-break-time",
1587 .data = &lease_break_time,
1588 .maxlen = sizeof(int),
1589 .mode = 0644,
1590 .proc_handler = proc_dointvec,
1591 },
1592 #endif
1593 #ifdef CONFIG_AIO
1594 {
1595 .procname = "aio-nr",
1596 .data = &aio_nr,
1597 .maxlen = sizeof(aio_nr),
1598 .mode = 0444,
1599 .proc_handler = proc_doulongvec_minmax,
1600 },
1601 {
1602 .procname = "aio-max-nr",
1603 .data = &aio_max_nr,
1604 .maxlen = sizeof(aio_max_nr),
1605 .mode = 0644,
1606 .proc_handler = proc_doulongvec_minmax,
1607 },
1608 #endif /* CONFIG_AIO */
1609 #ifdef CONFIG_INOTIFY_USER
1610 {
1611 .procname = "inotify",
1612 .mode = 0555,
1613 .child = inotify_table,
1614 },
1615 #endif
1616 #ifdef CONFIG_EPOLL
1617 {
1618 .procname = "epoll",
1619 .mode = 0555,
1620 .child = epoll_table,
1621 },
1622 #endif
1623 #endif
1624 {
1625 .procname = "protected_symlinks",
1626 .data = &sysctl_protected_symlinks,
1627 .maxlen = sizeof(int),
1628 .mode = 0600,
1629 .proc_handler = proc_dointvec_minmax,
1630 .extra1 = &zero,
1631 .extra2 = &one,
1632 },
1633 {
1634 .procname = "protected_hardlinks",
1635 .data = &sysctl_protected_hardlinks,
1636 .maxlen = sizeof(int),
1637 .mode = 0600,
1638 .proc_handler = proc_dointvec_minmax,
1639 .extra1 = &zero,
1640 .extra2 = &one,
1641 },
1642 {
1643 .procname = "suid_dumpable",
1644 .data = &suid_dumpable,
1645 .maxlen = sizeof(int),
1646 .mode = 0644,
1647 .proc_handler = proc_dointvec_minmax_coredump,
1648 .extra1 = &zero,
1649 .extra2 = &two,
1650 },
1651 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1652 {
1653 .procname = "binfmt_misc",
1654 .mode = 0555,
1655 .child = binfmt_misc_table,
1656 },
1657 #endif
1658 {
1659 .procname = "pipe-max-size",
1660 .data = &pipe_max_size,
1661 .maxlen = sizeof(int),
1662 .mode = 0644,
1663 .proc_handler = &pipe_proc_fn,
1664 .extra1 = &pipe_min_size,
1665 },
1666 { }
1667 };
1668
1669 static struct ctl_table debug_table[] = {
1670 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1671 {
1672 .procname = "exception-trace",
1673 .data = &show_unhandled_signals,
1674 .maxlen = sizeof(int),
1675 .mode = 0644,
1676 .proc_handler = proc_dointvec
1677 },
1678 #endif
1679 #if defined(CONFIG_OPTPROBES)
1680 {
1681 .procname = "kprobes-optimization",
1682 .data = &sysctl_kprobes_optimization,
1683 .maxlen = sizeof(int),
1684 .mode = 0644,
1685 .proc_handler = proc_kprobes_optimization_handler,
1686 .extra1 = &zero,
1687 .extra2 = &one,
1688 },
1689 #endif
1690 { }
1691 };
1692
1693 static struct ctl_table dev_table[] = {
1694 { }
1695 };
1696
1697 int __init sysctl_init(void)
1698 {
1699 struct ctl_table_header *hdr;
1700
1701 hdr = register_sysctl_table(sysctl_base_table);
1702 kmemleak_not_leak(hdr);
1703 return 0;
1704 }
1705
1706 #endif /* CONFIG_SYSCTL */
1707
1708 /*
1709 * /proc/sys support
1710 */
1711
1712 #ifdef CONFIG_PROC_SYSCTL
1713
1714 static int _proc_do_string(void* data, int maxlen, int write,
1715 void __user *buffer,
1716 size_t *lenp, loff_t *ppos)
1717 {
1718 size_t len;
1719 char __user *p;
1720 char c;
1721
1722 if (!data || !maxlen || !*lenp) {
1723 *lenp = 0;
1724 return 0;
1725 }
1726
1727 if (write) {
1728 len = 0;
1729 p = buffer;
1730 while (len < *lenp) {
1731 if (get_user(c, p++))
1732 return -EFAULT;
1733 if (c == 0 || c == '\n')
1734 break;
1735 len++;
1736 }
1737 if (len >= maxlen)
1738 len = maxlen-1;
1739 if(copy_from_user(data, buffer, len))
1740 return -EFAULT;
1741 ((char *) data)[len] = 0;
1742 *ppos += *lenp;
1743 } else {
1744 len = strlen(data);
1745 if (len > maxlen)
1746 len = maxlen;
1747
1748 if (*ppos > len) {
1749 *lenp = 0;
1750 return 0;
1751 }
1752
1753 data += *ppos;
1754 len -= *ppos;
1755
1756 if (len > *lenp)
1757 len = *lenp;
1758 if (len)
1759 if(copy_to_user(buffer, data, len))
1760 return -EFAULT;
1761 if (len < *lenp) {
1762 if(put_user('\n', ((char __user *) buffer) + len))
1763 return -EFAULT;
1764 len++;
1765 }
1766 *lenp = len;
1767 *ppos += len;
1768 }
1769 return 0;
1770 }
1771
1772 /**
1773 * proc_dostring - read a string sysctl
1774 * @table: the sysctl table
1775 * @write: %TRUE if this is a write to the sysctl file
1776 * @buffer: the user buffer
1777 * @lenp: the size of the user buffer
1778 * @ppos: file position
1779 *
1780 * Reads/writes a string from/to the user buffer. If the kernel
1781 * buffer provided is not large enough to hold the string, the
1782 * string is truncated. The copied string is %NULL-terminated.
1783 * If the string is being read by the user process, it is copied
1784 * and a newline '\n' is added. It is truncated if the buffer is
1785 * not large enough.
1786 *
1787 * Returns 0 on success.
1788 */
1789 int proc_dostring(struct ctl_table *table, int write,
1790 void __user *buffer, size_t *lenp, loff_t *ppos)
1791 {
1792 return _proc_do_string(table->data, table->maxlen, write,
1793 buffer, lenp, ppos);
1794 }
1795
1796 static size_t proc_skip_spaces(char **buf)
1797 {
1798 size_t ret;
1799 char *tmp = skip_spaces(*buf);
1800 ret = tmp - *buf;
1801 *buf = tmp;
1802 return ret;
1803 }
1804
1805 static void proc_skip_char(char **buf, size_t *size, const char v)
1806 {
1807 while (*size) {
1808 if (**buf != v)
1809 break;
1810 (*size)--;
1811 (*buf)++;
1812 }
1813 }
1814
1815 #define TMPBUFLEN 22
1816 /**
1817 * proc_get_long - reads an ASCII formatted integer from a user buffer
1818 *
1819 * @buf: a kernel buffer
1820 * @size: size of the kernel buffer
1821 * @val: this is where the number will be stored
1822 * @neg: set to %TRUE if number is negative
1823 * @perm_tr: a vector which contains the allowed trailers
1824 * @perm_tr_len: size of the perm_tr vector
1825 * @tr: pointer to store the trailer character
1826 *
1827 * In case of success %0 is returned and @buf and @size are updated with
1828 * the amount of bytes read. If @tr is non-NULL and a trailing
1829 * character exists (size is non-zero after returning from this
1830 * function), @tr is updated with the trailing character.
1831 */
1832 static int proc_get_long(char **buf, size_t *size,
1833 unsigned long *val, bool *neg,
1834 const char *perm_tr, unsigned perm_tr_len, char *tr)
1835 {
1836 int len;
1837 char *p, tmp[TMPBUFLEN];
1838
1839 if (!*size)
1840 return -EINVAL;
1841
1842 len = *size;
1843 if (len > TMPBUFLEN - 1)
1844 len = TMPBUFLEN - 1;
1845
1846 memcpy(tmp, *buf, len);
1847
1848 tmp[len] = 0;
1849 p = tmp;
1850 if (*p == '-' && *size > 1) {
1851 *neg = true;
1852 p++;
1853 } else
1854 *neg = false;
1855 if (!isdigit(*p))
1856 return -EINVAL;
1857
1858 *val = simple_strtoul(p, &p, 0);
1859
1860 len = p - tmp;
1861
1862 /* We don't know if the next char is whitespace thus we may accept
1863 * invalid integers (e.g. 1234...a) or two integers instead of one
1864 * (e.g. 123...1). So lets not allow such large numbers. */
1865 if (len == TMPBUFLEN - 1)
1866 return -EINVAL;
1867
1868 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
1869 return -EINVAL;
1870
1871 if (tr && (len < *size))
1872 *tr = *p;
1873
1874 *buf += len;
1875 *size -= len;
1876
1877 return 0;
1878 }
1879
1880 /**
1881 * proc_put_long - converts an integer to a decimal ASCII formatted string
1882 *
1883 * @buf: the user buffer
1884 * @size: the size of the user buffer
1885 * @val: the integer to be converted
1886 * @neg: sign of the number, %TRUE for negative
1887 *
1888 * In case of success %0 is returned and @buf and @size are updated with
1889 * the amount of bytes written.
1890 */
1891 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
1892 bool neg)
1893 {
1894 int len;
1895 char tmp[TMPBUFLEN], *p = tmp;
1896
1897 sprintf(p, "%s%lu", neg ? "-" : "", val);
1898 len = strlen(tmp);
1899 if (len > *size)
1900 len = *size;
1901 if (copy_to_user(*buf, tmp, len))
1902 return -EFAULT;
1903 *size -= len;
1904 *buf += len;
1905 return 0;
1906 }
1907 #undef TMPBUFLEN
1908
1909 static int proc_put_char(void __user **buf, size_t *size, char c)
1910 {
1911 if (*size) {
1912 char __user **buffer = (char __user **)buf;
1913 if (put_user(c, *buffer))
1914 return -EFAULT;
1915 (*size)--, (*buffer)++;
1916 *buf = *buffer;
1917 }
1918 return 0;
1919 }
1920
1921 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
1922 int *valp,
1923 int write, void *data)
1924 {
1925 if (write) {
1926 *valp = *negp ? -*lvalp : *lvalp;
1927 } else {
1928 int val = *valp;
1929 if (val < 0) {
1930 *negp = true;
1931 *lvalp = (unsigned long)-val;
1932 } else {
1933 *negp = false;
1934 *lvalp = (unsigned long)val;
1935 }
1936 }
1937 return 0;
1938 }
1939
1940 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
1941
1942 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1943 int write, void __user *buffer,
1944 size_t *lenp, loff_t *ppos,
1945 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1946 int write, void *data),
1947 void *data)
1948 {
1949 int *i, vleft, first = 1, err = 0;
1950 unsigned long page = 0;
1951 size_t left;
1952 char *kbuf;
1953
1954 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
1955 *lenp = 0;
1956 return 0;
1957 }
1958
1959 i = (int *) tbl_data;
1960 vleft = table->maxlen / sizeof(*i);
1961 left = *lenp;
1962
1963 if (!conv)
1964 conv = do_proc_dointvec_conv;
1965
1966 if (write) {
1967 if (left > PAGE_SIZE - 1)
1968 left = PAGE_SIZE - 1;
1969 page = __get_free_page(GFP_TEMPORARY);
1970 kbuf = (char *) page;
1971 if (!kbuf)
1972 return -ENOMEM;
1973 if (copy_from_user(kbuf, buffer, left)) {
1974 err = -EFAULT;
1975 goto free;
1976 }
1977 kbuf[left] = 0;
1978 }
1979
1980 for (; left && vleft--; i++, first=0) {
1981 unsigned long lval;
1982 bool neg;
1983
1984 if (write) {
1985 left -= proc_skip_spaces(&kbuf);
1986
1987 if (!left)
1988 break;
1989 err = proc_get_long(&kbuf, &left, &lval, &neg,
1990 proc_wspace_sep,
1991 sizeof(proc_wspace_sep), NULL);
1992 if (err)
1993 break;
1994 if (conv(&neg, &lval, i, 1, data)) {
1995 err = -EINVAL;
1996 break;
1997 }
1998 } else {
1999 if (conv(&neg, &lval, i, 0, data)) {
2000 err = -EINVAL;
2001 break;
2002 }
2003 if (!first)
2004 err = proc_put_char(&buffer, &left, '\t');
2005 if (err)
2006 break;
2007 err = proc_put_long(&buffer, &left, lval, neg);
2008 if (err)
2009 break;
2010 }
2011 }
2012
2013 if (!write && !first && left && !err)
2014 err = proc_put_char(&buffer, &left, '\n');
2015 if (write && !err && left)
2016 left -= proc_skip_spaces(&kbuf);
2017 free:
2018 if (write) {
2019 free_page(page);
2020 if (first)
2021 return err ? : -EINVAL;
2022 }
2023 *lenp -= left;
2024 *ppos += *lenp;
2025 return err;
2026 }
2027
2028 static int do_proc_dointvec(struct ctl_table *table, int write,
2029 void __user *buffer, size_t *lenp, loff_t *ppos,
2030 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2031 int write, void *data),
2032 void *data)
2033 {
2034 return __do_proc_dointvec(table->data, table, write,
2035 buffer, lenp, ppos, conv, data);
2036 }
2037
2038 /**
2039 * proc_dointvec - read a vector of integers
2040 * @table: the sysctl table
2041 * @write: %TRUE if this is a write to the sysctl file
2042 * @buffer: the user buffer
2043 * @lenp: the size of the user buffer
2044 * @ppos: file position
2045 *
2046 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2047 * values from/to the user buffer, treated as an ASCII string.
2048 *
2049 * Returns 0 on success.
2050 */
2051 int proc_dointvec(struct ctl_table *table, int write,
2052 void __user *buffer, size_t *lenp, loff_t *ppos)
2053 {
2054 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2055 NULL,NULL);
2056 }
2057
2058 /*
2059 * Taint values can only be increased
2060 * This means we can safely use a temporary.
2061 */
2062 static int proc_taint(struct ctl_table *table, int write,
2063 void __user *buffer, size_t *lenp, loff_t *ppos)
2064 {
2065 struct ctl_table t;
2066 unsigned long tmptaint = get_taint();
2067 int err;
2068
2069 if (write && !capable(CAP_SYS_ADMIN))
2070 return -EPERM;
2071
2072 t = *table;
2073 t.data = &tmptaint;
2074 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2075 if (err < 0)
2076 return err;
2077
2078 if (write) {
2079 /*
2080 * Poor man's atomic or. Not worth adding a primitive
2081 * to everyone's atomic.h for this
2082 */
2083 int i;
2084 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2085 if ((tmptaint >> i) & 1)
2086 add_taint(i, LOCKDEP_STILL_OK);
2087 }
2088 }
2089
2090 return err;
2091 }
2092
2093 #ifdef CONFIG_PRINTK
2094 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2095 void __user *buffer, size_t *lenp, loff_t *ppos)
2096 {
2097 if (write && !capable(CAP_SYS_ADMIN))
2098 return -EPERM;
2099
2100 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2101 }
2102 #endif
2103
2104 struct do_proc_dointvec_minmax_conv_param {
2105 int *min;
2106 int *max;
2107 };
2108
2109 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2110 int *valp,
2111 int write, void *data)
2112 {
2113 struct do_proc_dointvec_minmax_conv_param *param = data;
2114 if (write) {
2115 int val = *negp ? -*lvalp : *lvalp;
2116 if ((param->min && *param->min > val) ||
2117 (param->max && *param->max < val))
2118 return -EINVAL;
2119 *valp = val;
2120 } else {
2121 int val = *valp;
2122 if (val < 0) {
2123 *negp = true;
2124 *lvalp = (unsigned long)-val;
2125 } else {
2126 *negp = false;
2127 *lvalp = (unsigned long)val;
2128 }
2129 }
2130 return 0;
2131 }
2132
2133 /**
2134 * proc_dointvec_minmax - read a vector of integers with min/max values
2135 * @table: the sysctl table
2136 * @write: %TRUE if this is a write to the sysctl file
2137 * @buffer: the user buffer
2138 * @lenp: the size of the user buffer
2139 * @ppos: file position
2140 *
2141 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2142 * values from/to the user buffer, treated as an ASCII string.
2143 *
2144 * This routine will ensure the values are within the range specified by
2145 * table->extra1 (min) and table->extra2 (max).
2146 *
2147 * Returns 0 on success.
2148 */
2149 int proc_dointvec_minmax(struct ctl_table *table, int write,
2150 void __user *buffer, size_t *lenp, loff_t *ppos)
2151 {
2152 struct do_proc_dointvec_minmax_conv_param param = {
2153 .min = (int *) table->extra1,
2154 .max = (int *) table->extra2,
2155 };
2156 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2157 do_proc_dointvec_minmax_conv, &param);
2158 }
2159
2160 static void validate_coredump_safety(void)
2161 {
2162 #ifdef CONFIG_COREDUMP
2163 if (suid_dumpable == SUID_DUMP_ROOT &&
2164 core_pattern[0] != '/' && core_pattern[0] != '|') {
2165 printk(KERN_WARNING "Unsafe core_pattern used with "\
2166 "suid_dumpable=2. Pipe handler or fully qualified "\
2167 "core dump path required.\n");
2168 }
2169 #endif
2170 }
2171
2172 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2173 void __user *buffer, size_t *lenp, loff_t *ppos)
2174 {
2175 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2176 if (!error)
2177 validate_coredump_safety();
2178 return error;
2179 }
2180
2181 #ifdef CONFIG_COREDUMP
2182 static int proc_dostring_coredump(struct ctl_table *table, int write,
2183 void __user *buffer, size_t *lenp, loff_t *ppos)
2184 {
2185 int error = proc_dostring(table, write, buffer, lenp, ppos);
2186 if (!error)
2187 validate_coredump_safety();
2188 return error;
2189 }
2190 #endif
2191
2192 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2193 void __user *buffer,
2194 size_t *lenp, loff_t *ppos,
2195 unsigned long convmul,
2196 unsigned long convdiv)
2197 {
2198 unsigned long *i, *min, *max;
2199 int vleft, first = 1, err = 0;
2200 unsigned long page = 0;
2201 size_t left;
2202 char *kbuf;
2203
2204 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2205 *lenp = 0;
2206 return 0;
2207 }
2208
2209 i = (unsigned long *) data;
2210 min = (unsigned long *) table->extra1;
2211 max = (unsigned long *) table->extra2;
2212 vleft = table->maxlen / sizeof(unsigned long);
2213 left = *lenp;
2214
2215 if (write) {
2216 if (left > PAGE_SIZE - 1)
2217 left = PAGE_SIZE - 1;
2218 page = __get_free_page(GFP_TEMPORARY);
2219 kbuf = (char *) page;
2220 if (!kbuf)
2221 return -ENOMEM;
2222 if (copy_from_user(kbuf, buffer, left)) {
2223 err = -EFAULT;
2224 goto free;
2225 }
2226 kbuf[left] = 0;
2227 }
2228
2229 for (; left && vleft--; i++, first = 0) {
2230 unsigned long val;
2231
2232 if (write) {
2233 bool neg;
2234
2235 left -= proc_skip_spaces(&kbuf);
2236
2237 err = proc_get_long(&kbuf, &left, &val, &neg,
2238 proc_wspace_sep,
2239 sizeof(proc_wspace_sep), NULL);
2240 if (err)
2241 break;
2242 if (neg)
2243 continue;
2244 if ((min && val < *min) || (max && val > *max))
2245 continue;
2246 *i = val;
2247 } else {
2248 val = convdiv * (*i) / convmul;
2249 if (!first) {
2250 err = proc_put_char(&buffer, &left, '\t');
2251 if (err)
2252 break;
2253 }
2254 err = proc_put_long(&buffer, &left, val, false);
2255 if (err)
2256 break;
2257 }
2258 }
2259
2260 if (!write && !first && left && !err)
2261 err = proc_put_char(&buffer, &left, '\n');
2262 if (write && !err)
2263 left -= proc_skip_spaces(&kbuf);
2264 free:
2265 if (write) {
2266 free_page(page);
2267 if (first)
2268 return err ? : -EINVAL;
2269 }
2270 *lenp -= left;
2271 *ppos += *lenp;
2272 return err;
2273 }
2274
2275 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2276 void __user *buffer,
2277 size_t *lenp, loff_t *ppos,
2278 unsigned long convmul,
2279 unsigned long convdiv)
2280 {
2281 return __do_proc_doulongvec_minmax(table->data, table, write,
2282 buffer, lenp, ppos, convmul, convdiv);
2283 }
2284
2285 /**
2286 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2287 * @table: the sysctl table
2288 * @write: %TRUE if this is a write to the sysctl file
2289 * @buffer: the user buffer
2290 * @lenp: the size of the user buffer
2291 * @ppos: file position
2292 *
2293 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2294 * values from/to the user buffer, treated as an ASCII string.
2295 *
2296 * This routine will ensure the values are within the range specified by
2297 * table->extra1 (min) and table->extra2 (max).
2298 *
2299 * Returns 0 on success.
2300 */
2301 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2302 void __user *buffer, size_t *lenp, loff_t *ppos)
2303 {
2304 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2305 }
2306
2307 /**
2308 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2309 * @table: the sysctl table
2310 * @write: %TRUE if this is a write to the sysctl file
2311 * @buffer: the user buffer
2312 * @lenp: the size of the user buffer
2313 * @ppos: file position
2314 *
2315 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2316 * values from/to the user buffer, treated as an ASCII string. The values
2317 * are treated as milliseconds, and converted to jiffies when they are stored.
2318 *
2319 * This routine will ensure the values are within the range specified by
2320 * table->extra1 (min) and table->extra2 (max).
2321 *
2322 * Returns 0 on success.
2323 */
2324 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2325 void __user *buffer,
2326 size_t *lenp, loff_t *ppos)
2327 {
2328 return do_proc_doulongvec_minmax(table, write, buffer,
2329 lenp, ppos, HZ, 1000l);
2330 }
2331
2332
2333 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2334 int *valp,
2335 int write, void *data)
2336 {
2337 if (write) {
2338 if (*lvalp > LONG_MAX / HZ)
2339 return 1;
2340 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2341 } else {
2342 int val = *valp;
2343 unsigned long lval;
2344 if (val < 0) {
2345 *negp = true;
2346 lval = (unsigned long)-val;
2347 } else {
2348 *negp = false;
2349 lval = (unsigned long)val;
2350 }
2351 *lvalp = lval / HZ;
2352 }
2353 return 0;
2354 }
2355
2356 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2357 int *valp,
2358 int write, void *data)
2359 {
2360 if (write) {
2361 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2362 return 1;
2363 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2364 } else {
2365 int val = *valp;
2366 unsigned long lval;
2367 if (val < 0) {
2368 *negp = true;
2369 lval = (unsigned long)-val;
2370 } else {
2371 *negp = false;
2372 lval = (unsigned long)val;
2373 }
2374 *lvalp = jiffies_to_clock_t(lval);
2375 }
2376 return 0;
2377 }
2378
2379 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2380 int *valp,
2381 int write, void *data)
2382 {
2383 if (write) {
2384 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2385
2386 if (jif > INT_MAX)
2387 return 1;
2388 *valp = (int)jif;
2389 } else {
2390 int val = *valp;
2391 unsigned long lval;
2392 if (val < 0) {
2393 *negp = true;
2394 lval = (unsigned long)-val;
2395 } else {
2396 *negp = false;
2397 lval = (unsigned long)val;
2398 }
2399 *lvalp = jiffies_to_msecs(lval);
2400 }
2401 return 0;
2402 }
2403
2404 /**
2405 * proc_dointvec_jiffies - read a vector of integers as seconds
2406 * @table: the sysctl table
2407 * @write: %TRUE if this is a write to the sysctl file
2408 * @buffer: the user buffer
2409 * @lenp: the size of the user buffer
2410 * @ppos: file position
2411 *
2412 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2413 * values from/to the user buffer, treated as an ASCII string.
2414 * The values read are assumed to be in seconds, and are converted into
2415 * jiffies.
2416 *
2417 * Returns 0 on success.
2418 */
2419 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2420 void __user *buffer, size_t *lenp, loff_t *ppos)
2421 {
2422 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2423 do_proc_dointvec_jiffies_conv,NULL);
2424 }
2425
2426 /**
2427 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2428 * @table: the sysctl table
2429 * @write: %TRUE if this is a write to the sysctl file
2430 * @buffer: the user buffer
2431 * @lenp: the size of the user buffer
2432 * @ppos: pointer to the file position
2433 *
2434 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2435 * values from/to the user buffer, treated as an ASCII string.
2436 * The values read are assumed to be in 1/USER_HZ seconds, and
2437 * are converted into jiffies.
2438 *
2439 * Returns 0 on success.
2440 */
2441 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2442 void __user *buffer, size_t *lenp, loff_t *ppos)
2443 {
2444 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2445 do_proc_dointvec_userhz_jiffies_conv,NULL);
2446 }
2447
2448 /**
2449 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
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 * @ppos: the current position in the file
2456 *
2457 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2458 * values from/to the user buffer, treated as an ASCII string.
2459 * The values read are assumed to be in 1/1000 seconds, and
2460 * are converted into jiffies.
2461 *
2462 * Returns 0 on success.
2463 */
2464 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2465 void __user *buffer, size_t *lenp, loff_t *ppos)
2466 {
2467 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2468 do_proc_dointvec_ms_jiffies_conv, NULL);
2469 }
2470
2471 static int proc_do_cad_pid(struct ctl_table *table, int write,
2472 void __user *buffer, size_t *lenp, loff_t *ppos)
2473 {
2474 struct pid *new_pid;
2475 pid_t tmp;
2476 int r;
2477
2478 tmp = pid_vnr(cad_pid);
2479
2480 r = __do_proc_dointvec(&tmp, table, write, buffer,
2481 lenp, ppos, NULL, NULL);
2482 if (r || !write)
2483 return r;
2484
2485 new_pid = find_get_pid(tmp);
2486 if (!new_pid)
2487 return -ESRCH;
2488
2489 put_pid(xchg(&cad_pid, new_pid));
2490 return 0;
2491 }
2492
2493 /**
2494 * proc_do_large_bitmap - read/write from/to a large bitmap
2495 * @table: the sysctl table
2496 * @write: %TRUE if this is a write to the sysctl file
2497 * @buffer: the user buffer
2498 * @lenp: the size of the user buffer
2499 * @ppos: file position
2500 *
2501 * The bitmap is stored at table->data and the bitmap length (in bits)
2502 * in table->maxlen.
2503 *
2504 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2505 * large bitmaps may be represented in a compact manner. Writing into
2506 * the file will clear the bitmap then update it with the given input.
2507 *
2508 * Returns 0 on success.
2509 */
2510 int proc_do_large_bitmap(struct ctl_table *table, int write,
2511 void __user *buffer, size_t *lenp, loff_t *ppos)
2512 {
2513 int err = 0;
2514 bool first = 1;
2515 size_t left = *lenp;
2516 unsigned long bitmap_len = table->maxlen;
2517 unsigned long *bitmap = (unsigned long *) table->data;
2518 unsigned long *tmp_bitmap = NULL;
2519 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2520
2521 if (!bitmap_len || !left || (*ppos && !write)) {
2522 *lenp = 0;
2523 return 0;
2524 }
2525
2526 if (write) {
2527 unsigned long page = 0;
2528 char *kbuf;
2529
2530 if (left > PAGE_SIZE - 1)
2531 left = PAGE_SIZE - 1;
2532
2533 page = __get_free_page(GFP_TEMPORARY);
2534 kbuf = (char *) page;
2535 if (!kbuf)
2536 return -ENOMEM;
2537 if (copy_from_user(kbuf, buffer, left)) {
2538 free_page(page);
2539 return -EFAULT;
2540 }
2541 kbuf[left] = 0;
2542
2543 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2544 GFP_KERNEL);
2545 if (!tmp_bitmap) {
2546 free_page(page);
2547 return -ENOMEM;
2548 }
2549 proc_skip_char(&kbuf, &left, '\n');
2550 while (!err && left) {
2551 unsigned long val_a, val_b;
2552 bool neg;
2553
2554 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2555 sizeof(tr_a), &c);
2556 if (err)
2557 break;
2558 if (val_a >= bitmap_len || neg) {
2559 err = -EINVAL;
2560 break;
2561 }
2562
2563 val_b = val_a;
2564 if (left) {
2565 kbuf++;
2566 left--;
2567 }
2568
2569 if (c == '-') {
2570 err = proc_get_long(&kbuf, &left, &val_b,
2571 &neg, tr_b, sizeof(tr_b),
2572 &c);
2573 if (err)
2574 break;
2575 if (val_b >= bitmap_len || neg ||
2576 val_a > val_b) {
2577 err = -EINVAL;
2578 break;
2579 }
2580 if (left) {
2581 kbuf++;
2582 left--;
2583 }
2584 }
2585
2586 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2587 first = 0;
2588 proc_skip_char(&kbuf, &left, '\n');
2589 }
2590 free_page(page);
2591 } else {
2592 unsigned long bit_a, bit_b = 0;
2593
2594 while (left) {
2595 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2596 if (bit_a >= bitmap_len)
2597 break;
2598 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2599 bit_a + 1) - 1;
2600
2601 if (!first) {
2602 err = proc_put_char(&buffer, &left, ',');
2603 if (err)
2604 break;
2605 }
2606 err = proc_put_long(&buffer, &left, bit_a, false);
2607 if (err)
2608 break;
2609 if (bit_a != bit_b) {
2610 err = proc_put_char(&buffer, &left, '-');
2611 if (err)
2612 break;
2613 err = proc_put_long(&buffer, &left, bit_b, false);
2614 if (err)
2615 break;
2616 }
2617
2618 first = 0; bit_b++;
2619 }
2620 if (!err)
2621 err = proc_put_char(&buffer, &left, '\n');
2622 }
2623
2624 if (!err) {
2625 if (write) {
2626 if (*ppos)
2627 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2628 else
2629 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2630 }
2631 kfree(tmp_bitmap);
2632 *lenp -= left;
2633 *ppos += *lenp;
2634 return 0;
2635 } else {
2636 kfree(tmp_bitmap);
2637 return err;
2638 }
2639 }
2640
2641 #else /* CONFIG_PROC_SYSCTL */
2642
2643 int proc_dostring(struct ctl_table *table, int write,
2644 void __user *buffer, size_t *lenp, loff_t *ppos)
2645 {
2646 return -ENOSYS;
2647 }
2648
2649 int proc_dointvec(struct ctl_table *table, int write,
2650 void __user *buffer, size_t *lenp, loff_t *ppos)
2651 {
2652 return -ENOSYS;
2653 }
2654
2655 int proc_dointvec_minmax(struct ctl_table *table, int write,
2656 void __user *buffer, size_t *lenp, loff_t *ppos)
2657 {
2658 return -ENOSYS;
2659 }
2660
2661 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2662 void __user *buffer, size_t *lenp, loff_t *ppos)
2663 {
2664 return -ENOSYS;
2665 }
2666
2667 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2668 void __user *buffer, size_t *lenp, loff_t *ppos)
2669 {
2670 return -ENOSYS;
2671 }
2672
2673 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2674 void __user *buffer, size_t *lenp, loff_t *ppos)
2675 {
2676 return -ENOSYS;
2677 }
2678
2679 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2680 void __user *buffer, size_t *lenp, loff_t *ppos)
2681 {
2682 return -ENOSYS;
2683 }
2684
2685 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2686 void __user *buffer,
2687 size_t *lenp, loff_t *ppos)
2688 {
2689 return -ENOSYS;
2690 }
2691
2692
2693 #endif /* CONFIG_PROC_SYSCTL */
2694
2695 /*
2696 * No sense putting this after each symbol definition, twice,
2697 * exception granted :-)
2698 */
2699 EXPORT_SYMBOL(proc_dointvec);
2700 EXPORT_SYMBOL(proc_dointvec_jiffies);
2701 EXPORT_SYMBOL(proc_dointvec_minmax);
2702 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2703 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2704 EXPORT_SYMBOL(proc_dostring);
2705 EXPORT_SYMBOL(proc_doulongvec_minmax);
2706 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
Linux with added FPC-III support