]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/blob - kernel/sysctl.c
projects / mirror_ubuntu-bionic-kernel.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
UBUNTU: link-to-tracker: update tracking bug
[mirror_ubuntu-bionic-kernel.git] / kernel / sysctl.c
1 /*
2 * sysctl.c: General linux system control interface
3 *
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12 * Horn.
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16 * Wendling.
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19 */
20
21 #include <linux/module.h>
22 #include <linux/aio.h>
23 #include <linux/mm.h>
24 #include <linux/swap.h>
25 #include <linux/slab.h>
26 #include <linux/sysctl.h>
27 #include <linux/bitmap.h>
28 #include <linux/signal.h>
29 #include <linux/printk.h>
30 #include <linux/proc_fs.h>
31 #include <linux/security.h>
32 #include <linux/ctype.h>
33 #include <linux/kmemleak.h>
34 #include <linux/fs.h>
35 #include <linux/init.h>
36 #include <linux/kernel.h>
37 #include <linux/kobject.h>
38 #include <linux/net.h>
39 #include <linux/sysrq.h>
40 #include <linux/highuid.h>
41 #include <linux/writeback.h>
42 #include <linux/ratelimit.h>
43 #include <linux/compaction.h>
44 #include <linux/hugetlb.h>
45 #include <linux/initrd.h>
46 #include <linux/key.h>
47 #include <linux/times.h>
48 #include <linux/limits.h>
49 #include <linux/dcache.h>
50 #include <linux/dnotify.h>
51 #include <linux/syscalls.h>
52 #include <linux/vmstat.h>
53 #include <linux/nfs_fs.h>
54 #include <linux/acpi.h>
55 #include <linux/reboot.h>
56 #include <linux/ftrace.h>
57 #include <linux/perf_event.h>
58 #include <linux/kprobes.h>
59 #include <linux/pipe_fs_i.h>
60 #include <linux/oom.h>
61 #include <linux/kmod.h>
62 #include <linux/capability.h>
63 #include <linux/binfmts.h>
64 #include <linux/sched/sysctl.h>
65 #include <linux/sched/coredump.h>
66 #include <linux/kexec.h>
67 #include <linux/bpf.h>
68 #include <linux/mount.h>
69 #include <linux/pipe_fs_i.h>
70
71 #include <linux/uaccess.h>
72 #include <asm/processor.h>
73
74 #ifdef CONFIG_X86
75 #include <asm/nmi.h>
76 #include <asm/stacktrace.h>
77 #include <asm/io.h>
78 #endif
79 #ifdef CONFIG_SPARC
80 #include <asm/setup.h>
81 #endif
82 #ifdef CONFIG_BSD_PROCESS_ACCT
83 #include <linux/acct.h>
84 #endif
85 #ifdef CONFIG_RT_MUTEXES
86 #include <linux/rtmutex.h>
87 #endif
88 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
89 #include <linux/lockdep.h>
90 #endif
91 #ifdef CONFIG_CHR_DEV_SG
92 #include <scsi/sg.h>
93 #endif
94
95 #ifdef CONFIG_LOCKUP_DETECTOR
96 #include <linux/nmi.h>
97 #endif
98
99 #if defined(CONFIG_SYSCTL)
100
101 /* External variables not in a header file. */
102 extern int suid_dumpable;
103 #ifdef CONFIG_COREDUMP
104 extern int core_uses_pid;
105 extern char core_pattern[];
106 extern unsigned int core_pipe_limit;
107 #endif
108 #ifdef CONFIG_USER_NS
109 extern int unprivileged_userns_clone;
110 #endif
111 extern int pid_max;
112 extern int pid_max_min, pid_max_max;
113 extern int percpu_pagelist_fraction;
114 extern int latencytop_enabled;
115 extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
116 #ifndef CONFIG_MMU
117 extern int sysctl_nr_trim_pages;
118 #endif
119
120 /* Constants used for minimum and maximum */
121 #ifdef CONFIG_LOCKUP_DETECTOR
122 static int sixty = 60;
123 #endif
124
125 static int __maybe_unused neg_one = -1;
126
127 static int zero;
128 static int __maybe_unused one = 1;
129 static int __maybe_unused two = 2;
130 static int __maybe_unused four = 4;
131 static unsigned long one_ul = 1;
132 static int one_hundred = 100;
133 static int one_thousand = 1000;
134 #ifdef CONFIG_PRINTK
135 static int ten_thousand = 10000;
136 #endif
137 #ifdef CONFIG_PERF_EVENTS
138 static int six_hundred_forty_kb = 640 * 1024;
139 #endif
140
141 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
142 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
143
144 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
145 static int maxolduid = 65535;
146 static int minolduid;
147
148 static int ngroups_max = NGROUPS_MAX;
149 static const int cap_last_cap = CAP_LAST_CAP;
150
151 /*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */
152 #ifdef CONFIG_DETECT_HUNG_TASK
153 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
154 #endif
155
156 #ifdef CONFIG_INOTIFY_USER
157 #include <linux/inotify.h>
158 #endif
159 #ifdef CONFIG_SPARC
160 #endif
161
162 #ifdef __hppa__
163 extern int pwrsw_enabled;
164 #endif
165
166 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
167 extern int unaligned_enabled;
168 #endif
169
170 #ifdef CONFIG_IA64
171 extern int unaligned_dump_stack;
172 #endif
173
174 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
175 extern int no_unaligned_warning;
176 #endif
177
178 #ifdef CONFIG_PROC_SYSCTL
179
180 /**
181 * enum sysctl_writes_mode - supported sysctl write modes
182 *
183 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
184 * to be written, and multiple writes on the same sysctl file descriptor
185 * will rewrite the sysctl value, regardless of file position. No warning
186 * is issued when the initial position is not 0.
187 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
188 * not 0.
189 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
190 * file position 0 and the value must be fully contained in the buffer
191 * sent to the write syscall. If dealing with strings respect the file
192 * position, but restrict this to the max length of the buffer, anything
193 * passed the max lenght will be ignored. Multiple writes will append
194 * to the buffer.
195 *
196 * These write modes control how current file position affects the behavior of
197 * updating sysctl values through the proc interface on each write.
198 */
199 enum sysctl_writes_mode {
200 SYSCTL_WRITES_LEGACY = -1,
201 SYSCTL_WRITES_WARN = 0,
202 SYSCTL_WRITES_STRICT = 1,
203 };
204
205 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
206
207 static int proc_do_cad_pid(struct ctl_table *table, int write,
208 void __user *buffer, size_t *lenp, loff_t *ppos);
209 static int proc_taint(struct ctl_table *table, int write,
210 void __user *buffer, size_t *lenp, loff_t *ppos);
211 #endif
212
213 #ifdef CONFIG_PRINTK
214 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
215 void __user *buffer, size_t *lenp, loff_t *ppos);
216 #endif
217
218 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
219 void __user *buffer, size_t *lenp, loff_t *ppos);
220 #ifdef CONFIG_COREDUMP
221 static int proc_dostring_coredump(struct ctl_table *table, int write,
222 void __user *buffer, size_t *lenp, loff_t *ppos);
223 #endif
224
225 #ifdef CONFIG_MAGIC_SYSRQ
226 /* Note: sysrq code uses it's own private copy */
227 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
228
229 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
230 void __user *buffer, size_t *lenp,
231 loff_t *ppos)
232 {
233 int error;
234
235 error = proc_dointvec(table, write, buffer, lenp, ppos);
236 if (error)
237 return error;
238
239 if (write)
240 sysrq_toggle_support(__sysrq_enabled);
241
242 return 0;
243 }
244
245 #endif
246
247 static struct ctl_table kern_table[];
248 static struct ctl_table vm_table[];
249 static struct ctl_table fs_table[];
250 static struct ctl_table debug_table[];
251 static struct ctl_table dev_table[];
252 extern struct ctl_table random_table[];
253 #ifdef CONFIG_EPOLL
254 extern struct ctl_table epoll_table[];
255 #endif
256
257 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
258 int sysctl_legacy_va_layout;
259 #endif
260
261 /* The default sysctl tables: */
262
263 static struct ctl_table sysctl_base_table[] = {
264 {
265 .procname = "kernel",
266 .mode = 0555,
267 .child = kern_table,
268 },
269 {
270 .procname = "vm",
271 .mode = 0555,
272 .child = vm_table,
273 },
274 {
275 .procname = "fs",
276 .mode = 0555,
277 .child = fs_table,
278 },
279 {
280 .procname = "debug",
281 .mode = 0555,
282 .child = debug_table,
283 },
284 {
285 .procname = "dev",
286 .mode = 0555,
287 .child = dev_table,
288 },
289 { }
290 };
291
292 #ifdef CONFIG_SCHED_DEBUG
293 static int min_sched_granularity_ns = 100000; /* 100 usecs */
294 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
295 static int min_wakeup_granularity_ns; /* 0 usecs */
296 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
297 #ifdef CONFIG_SMP
298 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
299 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
300 #endif /* CONFIG_SMP */
301 #endif /* CONFIG_SCHED_DEBUG */
302
303 #ifdef CONFIG_COMPACTION
304 static int min_extfrag_threshold;
305 static int max_extfrag_threshold = 1000;
306 #endif
307
308 static struct ctl_table kern_table[] = {
309 {
310 .procname = "sched_child_runs_first",
311 .data = &sysctl_sched_child_runs_first,
312 .maxlen = sizeof(unsigned int),
313 .mode = 0644,
314 .proc_handler = proc_dointvec,
315 },
316 #ifdef CONFIG_SCHED_DEBUG
317 {
318 .procname = "sched_min_granularity_ns",
319 .data = &sysctl_sched_min_granularity,
320 .maxlen = sizeof(unsigned int),
321 .mode = 0644,
322 .proc_handler = sched_proc_update_handler,
323 .extra1 = &min_sched_granularity_ns,
324 .extra2 = &max_sched_granularity_ns,
325 },
326 {
327 .procname = "sched_latency_ns",
328 .data = &sysctl_sched_latency,
329 .maxlen = sizeof(unsigned int),
330 .mode = 0644,
331 .proc_handler = sched_proc_update_handler,
332 .extra1 = &min_sched_granularity_ns,
333 .extra2 = &max_sched_granularity_ns,
334 },
335 {
336 .procname = "sched_wakeup_granularity_ns",
337 .data = &sysctl_sched_wakeup_granularity,
338 .maxlen = sizeof(unsigned int),
339 .mode = 0644,
340 .proc_handler = sched_proc_update_handler,
341 .extra1 = &min_wakeup_granularity_ns,
342 .extra2 = &max_wakeup_granularity_ns,
343 },
344 #ifdef CONFIG_SMP
345 {
346 .procname = "sched_tunable_scaling",
347 .data = &sysctl_sched_tunable_scaling,
348 .maxlen = sizeof(enum sched_tunable_scaling),
349 .mode = 0644,
350 .proc_handler = sched_proc_update_handler,
351 .extra1 = &min_sched_tunable_scaling,
352 .extra2 = &max_sched_tunable_scaling,
353 },
354 {
355 .procname = "sched_migration_cost_ns",
356 .data = &sysctl_sched_migration_cost,
357 .maxlen = sizeof(unsigned int),
358 .mode = 0644,
359 .proc_handler = proc_dointvec,
360 },
361 {
362 .procname = "sched_nr_migrate",
363 .data = &sysctl_sched_nr_migrate,
364 .maxlen = sizeof(unsigned int),
365 .mode = 0644,
366 .proc_handler = proc_dointvec,
367 },
368 {
369 .procname = "sched_time_avg_ms",
370 .data = &sysctl_sched_time_avg,
371 .maxlen = sizeof(unsigned int),
372 .mode = 0644,
373 .proc_handler = proc_dointvec_minmax,
374 .extra1 = &one,
375 },
376 #ifdef CONFIG_SCHEDSTATS
377 {
378 .procname = "sched_schedstats",
379 .data = NULL,
380 .maxlen = sizeof(unsigned int),
381 .mode = 0644,
382 .proc_handler = sysctl_schedstats,
383 .extra1 = &zero,
384 .extra2 = &one,
385 },
386 #endif /* CONFIG_SCHEDSTATS */
387 #endif /* CONFIG_SMP */
388 #ifdef CONFIG_NUMA_BALANCING
389 {
390 .procname = "numa_balancing_scan_delay_ms",
391 .data = &sysctl_numa_balancing_scan_delay,
392 .maxlen = sizeof(unsigned int),
393 .mode = 0644,
394 .proc_handler = proc_dointvec,
395 },
396 {
397 .procname = "numa_balancing_scan_period_min_ms",
398 .data = &sysctl_numa_balancing_scan_period_min,
399 .maxlen = sizeof(unsigned int),
400 .mode = 0644,
401 .proc_handler = proc_dointvec,
402 },
403 {
404 .procname = "numa_balancing_scan_period_max_ms",
405 .data = &sysctl_numa_balancing_scan_period_max,
406 .maxlen = sizeof(unsigned int),
407 .mode = 0644,
408 .proc_handler = proc_dointvec,
409 },
410 {
411 .procname = "numa_balancing_scan_size_mb",
412 .data = &sysctl_numa_balancing_scan_size,
413 .maxlen = sizeof(unsigned int),
414 .mode = 0644,
415 .proc_handler = proc_dointvec_minmax,
416 .extra1 = &one,
417 },
418 {
419 .procname = "numa_balancing",
420 .data = NULL, /* filled in by handler */
421 .maxlen = sizeof(unsigned int),
422 .mode = 0644,
423 .proc_handler = sysctl_numa_balancing,
424 .extra1 = &zero,
425 .extra2 = &one,
426 },
427 #endif /* CONFIG_NUMA_BALANCING */
428 #endif /* CONFIG_SCHED_DEBUG */
429 {
430 .procname = "sched_rt_period_us",
431 .data = &sysctl_sched_rt_period,
432 .maxlen = sizeof(unsigned int),
433 .mode = 0644,
434 .proc_handler = sched_rt_handler,
435 },
436 {
437 .procname = "sched_rt_runtime_us",
438 .data = &sysctl_sched_rt_runtime,
439 .maxlen = sizeof(int),
440 .mode = 0644,
441 .proc_handler = sched_rt_handler,
442 },
443 {
444 .procname = "sched_rr_timeslice_ms",
445 .data = &sysctl_sched_rr_timeslice,
446 .maxlen = sizeof(int),
447 .mode = 0644,
448 .proc_handler = sched_rr_handler,
449 },
450 #ifdef CONFIG_SCHED_AUTOGROUP
451 {
452 .procname = "sched_autogroup_enabled",
453 .data = &sysctl_sched_autogroup_enabled,
454 .maxlen = sizeof(unsigned int),
455 .mode = 0644,
456 .proc_handler = proc_dointvec_minmax,
457 .extra1 = &zero,
458 .extra2 = &one,
459 },
460 #endif
461 #ifdef CONFIG_CFS_BANDWIDTH
462 {
463 .procname = "sched_cfs_bandwidth_slice_us",
464 .data = &sysctl_sched_cfs_bandwidth_slice,
465 .maxlen = sizeof(unsigned int),
466 .mode = 0644,
467 .proc_handler = proc_dointvec_minmax,
468 .extra1 = &one,
469 },
470 #endif
471 #ifdef CONFIG_PROVE_LOCKING
472 {
473 .procname = "prove_locking",
474 .data = &prove_locking,
475 .maxlen = sizeof(int),
476 .mode = 0644,
477 .proc_handler = proc_dointvec,
478 },
479 #endif
480 #ifdef CONFIG_LOCK_STAT
481 {
482 .procname = "lock_stat",
483 .data = &lock_stat,
484 .maxlen = sizeof(int),
485 .mode = 0644,
486 .proc_handler = proc_dointvec,
487 },
488 #endif
489 {
490 .procname = "panic",
491 .data = &panic_timeout,
492 .maxlen = sizeof(int),
493 .mode = 0644,
494 .proc_handler = proc_dointvec,
495 },
496 #ifdef CONFIG_COREDUMP
497 {
498 .procname = "core_uses_pid",
499 .data = &core_uses_pid,
500 .maxlen = sizeof(int),
501 .mode = 0644,
502 .proc_handler = proc_dointvec,
503 },
504 {
505 .procname = "core_pattern",
506 .data = core_pattern,
507 .maxlen = CORENAME_MAX_SIZE,
508 .mode = 0644,
509 .proc_handler = proc_dostring_coredump,
510 },
511 {
512 .procname = "core_pipe_limit",
513 .data = &core_pipe_limit,
514 .maxlen = sizeof(unsigned int),
515 .mode = 0644,
516 .proc_handler = proc_dointvec,
517 },
518 #endif
519 #ifdef CONFIG_USER_NS
520 {
521 .procname = "unprivileged_userns_clone",
522 .data = &unprivileged_userns_clone,
523 .maxlen = sizeof(int),
524 .mode = 0644,
525 .proc_handler = proc_dointvec,
526 },
527 #endif
528 #ifdef CONFIG_PROC_SYSCTL
529 {
530 .procname = "tainted",
531 .maxlen = sizeof(long),
532 .mode = 0644,
533 .proc_handler = proc_taint,
534 },
535 {
536 .procname = "sysctl_writes_strict",
537 .data = &sysctl_writes_strict,
538 .maxlen = sizeof(int),
539 .mode = 0644,
540 .proc_handler = proc_dointvec_minmax,
541 .extra1 = &neg_one,
542 .extra2 = &one,
543 },
544 #endif
545 #ifdef CONFIG_LATENCYTOP
546 {
547 .procname = "latencytop",
548 .data = &latencytop_enabled,
549 .maxlen = sizeof(int),
550 .mode = 0644,
551 .proc_handler = sysctl_latencytop,
552 },
553 #endif
554 #ifdef CONFIG_BLK_DEV_INITRD
555 {
556 .procname = "real-root-dev",
557 .data = &real_root_dev,
558 .maxlen = sizeof(int),
559 .mode = 0644,
560 .proc_handler = proc_dointvec,
561 },
562 #endif
563 {
564 .procname = "print-fatal-signals",
565 .data = &print_fatal_signals,
566 .maxlen = sizeof(int),
567 .mode = 0644,
568 .proc_handler = proc_dointvec,
569 },
570 #ifdef CONFIG_SPARC
571 {
572 .procname = "reboot-cmd",
573 .data = reboot_command,
574 .maxlen = 256,
575 .mode = 0644,
576 .proc_handler = proc_dostring,
577 },
578 {
579 .procname = "stop-a",
580 .data = &stop_a_enabled,
581 .maxlen = sizeof (int),
582 .mode = 0644,
583 .proc_handler = proc_dointvec,
584 },
585 {
586 .procname = "scons-poweroff",
587 .data = &scons_pwroff,
588 .maxlen = sizeof (int),
589 .mode = 0644,
590 .proc_handler = proc_dointvec,
591 },
592 #endif
593 #ifdef CONFIG_SPARC64
594 {
595 .procname = "tsb-ratio",
596 .data = &sysctl_tsb_ratio,
597 .maxlen = sizeof (int),
598 .mode = 0644,
599 .proc_handler = proc_dointvec,
600 },
601 #endif
602 #ifdef __hppa__
603 {
604 .procname = "soft-power",
605 .data = &pwrsw_enabled,
606 .maxlen = sizeof (int),
607 .mode = 0644,
608 .proc_handler = proc_dointvec,
609 },
610 #endif
611 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
612 {
613 .procname = "unaligned-trap",
614 .data = &unaligned_enabled,
615 .maxlen = sizeof (int),
616 .mode = 0644,
617 .proc_handler = proc_dointvec,
618 },
619 #endif
620 {
621 .procname = "ctrl-alt-del",
622 .data = &C_A_D,
623 .maxlen = sizeof(int),
624 .mode = 0644,
625 .proc_handler = proc_dointvec,
626 },
627 #ifdef CONFIG_FUNCTION_TRACER
628 {
629 .procname = "ftrace_enabled",
630 .data = &ftrace_enabled,
631 .maxlen = sizeof(int),
632 .mode = 0644,
633 .proc_handler = ftrace_enable_sysctl,
634 },
635 #endif
636 #ifdef CONFIG_STACK_TRACER
637 {
638 .procname = "stack_tracer_enabled",
639 .data = &stack_tracer_enabled,
640 .maxlen = sizeof(int),
641 .mode = 0644,
642 .proc_handler = stack_trace_sysctl,
643 },
644 #endif
645 #ifdef CONFIG_TRACING
646 {
647 .procname = "ftrace_dump_on_oops",
648 .data = &ftrace_dump_on_oops,
649 .maxlen = sizeof(int),
650 .mode = 0644,
651 .proc_handler = proc_dointvec,
652 },
653 {
654 .procname = "traceoff_on_warning",
655 .data = &__disable_trace_on_warning,
656 .maxlen = sizeof(__disable_trace_on_warning),
657 .mode = 0644,
658 .proc_handler = proc_dointvec,
659 },
660 {
661 .procname = "tracepoint_printk",
662 .data = &tracepoint_printk,
663 .maxlen = sizeof(tracepoint_printk),
664 .mode = 0644,
665 .proc_handler = tracepoint_printk_sysctl,
666 },
667 #endif
668 #ifdef CONFIG_KEXEC_CORE
669 {
670 .procname = "kexec_load_disabled",
671 .data = &kexec_load_disabled,
672 .maxlen = sizeof(int),
673 .mode = 0644,
674 /* only handle a transition from default "0" to "1" */
675 .proc_handler = proc_dointvec_minmax,
676 .extra1 = &one,
677 .extra2 = &one,
678 },
679 #endif
680 #ifdef CONFIG_MODULES
681 {
682 .procname = "modprobe",
683 .data = &modprobe_path,
684 .maxlen = KMOD_PATH_LEN,
685 .mode = 0644,
686 .proc_handler = proc_dostring,
687 },
688 {
689 .procname = "modules_disabled",
690 .data = &modules_disabled,
691 .maxlen = sizeof(int),
692 .mode = 0644,
693 /* only handle a transition from default "0" to "1" */
694 .proc_handler = proc_dointvec_minmax,
695 .extra1 = &one,
696 .extra2 = &one,
697 },
698 #endif
699 #ifdef CONFIG_UEVENT_HELPER
700 {
701 .procname = "hotplug",
702 .data = &uevent_helper,
703 .maxlen = UEVENT_HELPER_PATH_LEN,
704 .mode = 0644,
705 .proc_handler = proc_dostring,
706 },
707 #endif
708 #ifdef CONFIG_CHR_DEV_SG
709 {
710 .procname = "sg-big-buff",
711 .data = &sg_big_buff,
712 .maxlen = sizeof (int),
713 .mode = 0444,
714 .proc_handler = proc_dointvec,
715 },
716 #endif
717 #ifdef CONFIG_BSD_PROCESS_ACCT
718 {
719 .procname = "acct",
720 .data = &acct_parm,
721 .maxlen = 3*sizeof(int),
722 .mode = 0644,
723 .proc_handler = proc_dointvec,
724 },
725 #endif
726 #ifdef CONFIG_MAGIC_SYSRQ
727 {
728 .procname = "sysrq",
729 .data = &__sysrq_enabled,
730 .maxlen = sizeof (int),
731 .mode = 0644,
732 .proc_handler = sysrq_sysctl_handler,
733 },
734 #endif
735 #ifdef CONFIG_PROC_SYSCTL
736 {
737 .procname = "cad_pid",
738 .data = NULL,
739 .maxlen = sizeof (int),
740 .mode = 0600,
741 .proc_handler = proc_do_cad_pid,
742 },
743 #endif
744 {
745 .procname = "threads-max",
746 .data = NULL,
747 .maxlen = sizeof(int),
748 .mode = 0644,
749 .proc_handler = sysctl_max_threads,
750 },
751 {
752 .procname = "random",
753 .mode = 0555,
754 .child = random_table,
755 },
756 {
757 .procname = "usermodehelper",
758 .mode = 0555,
759 .child = usermodehelper_table,
760 },
761 {
762 .procname = "overflowuid",
763 .data = &overflowuid,
764 .maxlen = sizeof(int),
765 .mode = 0644,
766 .proc_handler = proc_dointvec_minmax,
767 .extra1 = &minolduid,
768 .extra2 = &maxolduid,
769 },
770 {
771 .procname = "overflowgid",
772 .data = &overflowgid,
773 .maxlen = sizeof(int),
774 .mode = 0644,
775 .proc_handler = proc_dointvec_minmax,
776 .extra1 = &minolduid,
777 .extra2 = &maxolduid,
778 },
779 #ifdef CONFIG_S390
780 #ifdef CONFIG_MATHEMU
781 {
782 .procname = "ieee_emulation_warnings",
783 .data = &sysctl_ieee_emulation_warnings,
784 .maxlen = sizeof(int),
785 .mode = 0644,
786 .proc_handler = proc_dointvec,
787 },
788 #endif
789 {
790 .procname = "userprocess_debug",
791 .data = &show_unhandled_signals,
792 .maxlen = sizeof(int),
793 .mode = 0644,
794 .proc_handler = proc_dointvec,
795 },
796 #endif
797 {
798 .procname = "pid_max",
799 .data = &pid_max,
800 .maxlen = sizeof (int),
801 .mode = 0644,
802 .proc_handler = proc_dointvec_minmax,
803 .extra1 = &pid_max_min,
804 .extra2 = &pid_max_max,
805 },
806 {
807 .procname = "panic_on_oops",
808 .data = &panic_on_oops,
809 .maxlen = sizeof(int),
810 .mode = 0644,
811 .proc_handler = proc_dointvec,
812 },
813 #if defined CONFIG_PRINTK
814 {
815 .procname = "printk",
816 .data = &console_loglevel,
817 .maxlen = 4*sizeof(int),
818 .mode = 0644,
819 .proc_handler = proc_dointvec,
820 },
821 {
822 .procname = "printk_ratelimit",
823 .data = &printk_ratelimit_state.interval,
824 .maxlen = sizeof(int),
825 .mode = 0644,
826 .proc_handler = proc_dointvec_jiffies,
827 },
828 {
829 .procname = "printk_ratelimit_burst",
830 .data = &printk_ratelimit_state.burst,
831 .maxlen = sizeof(int),
832 .mode = 0644,
833 .proc_handler = proc_dointvec,
834 },
835 {
836 .procname = "printk_delay",
837 .data = &printk_delay_msec,
838 .maxlen = sizeof(int),
839 .mode = 0644,
840 .proc_handler = proc_dointvec_minmax,
841 .extra1 = &zero,
842 .extra2 = &ten_thousand,
843 },
844 {
845 .procname = "printk_devkmsg",
846 .data = devkmsg_log_str,
847 .maxlen = DEVKMSG_STR_MAX_SIZE,
848 .mode = 0644,
849 .proc_handler = devkmsg_sysctl_set_loglvl,
850 },
851 {
852 .procname = "dmesg_restrict",
853 .data = &dmesg_restrict,
854 .maxlen = sizeof(int),
855 .mode = 0644,
856 .proc_handler = proc_dointvec_minmax_sysadmin,
857 .extra1 = &zero,
858 .extra2 = &one,
859 },
860 {
861 .procname = "kptr_restrict",
862 .data = &kptr_restrict,
863 .maxlen = sizeof(int),
864 .mode = 0644,
865 .proc_handler = proc_dointvec_minmax_sysadmin,
866 .extra1 = &zero,
867 .extra2 = &two,
868 },
869 #endif
870 {
871 .procname = "ngroups_max",
872 .data = &ngroups_max,
873 .maxlen = sizeof (int),
874 .mode = 0444,
875 .proc_handler = proc_dointvec,
876 },
877 {
878 .procname = "cap_last_cap",
879 .data = (void *)&cap_last_cap,
880 .maxlen = sizeof(int),
881 .mode = 0444,
882 .proc_handler = proc_dointvec,
883 },
884 #if defined(CONFIG_LOCKUP_DETECTOR)
885 {
886 .procname = "watchdog",
887 .data = &watchdog_user_enabled,
888 .maxlen = sizeof(int),
889 .mode = 0644,
890 .proc_handler = proc_watchdog,
891 .extra1 = &zero,
892 .extra2 = &one,
893 },
894 {
895 .procname = "watchdog_thresh",
896 .data = &watchdog_thresh,
897 .maxlen = sizeof(int),
898 .mode = 0644,
899 .proc_handler = proc_watchdog_thresh,
900 .extra1 = &zero,
901 .extra2 = &sixty,
902 },
903 {
904 .procname = "nmi_watchdog",
905 .data = &nmi_watchdog_user_enabled,
906 .maxlen = sizeof(int),
907 .mode = NMI_WATCHDOG_SYSCTL_PERM,
908 .proc_handler = proc_nmi_watchdog,
909 .extra1 = &zero,
910 .extra2 = &one,
911 },
912 {
913 .procname = "watchdog_cpumask",
914 .data = &watchdog_cpumask_bits,
915 .maxlen = NR_CPUS,
916 .mode = 0644,
917 .proc_handler = proc_watchdog_cpumask,
918 },
919 #ifdef CONFIG_SOFTLOCKUP_DETECTOR
920 {
921 .procname = "soft_watchdog",
922 .data = &soft_watchdog_user_enabled,
923 .maxlen = sizeof(int),
924 .mode = 0644,
925 .proc_handler = proc_soft_watchdog,
926 .extra1 = &zero,
927 .extra2 = &one,
928 },
929 {
930 .procname = "softlockup_panic",
931 .data = &softlockup_panic,
932 .maxlen = sizeof(int),
933 .mode = 0644,
934 .proc_handler = proc_dointvec_minmax,
935 .extra1 = &zero,
936 .extra2 = &one,
937 },
938 #ifdef CONFIG_SMP
939 {
940 .procname = "softlockup_all_cpu_backtrace",
941 .data = &sysctl_softlockup_all_cpu_backtrace,
942 .maxlen = sizeof(int),
943 .mode = 0644,
944 .proc_handler = proc_dointvec_minmax,
945 .extra1 = &zero,
946 .extra2 = &one,
947 },
948 #endif /* CONFIG_SMP */
949 #endif
950 #ifdef CONFIG_HARDLOCKUP_DETECTOR
951 {
952 .procname = "hardlockup_panic",
953 .data = &hardlockup_panic,
954 .maxlen = sizeof(int),
955 .mode = 0644,
956 .proc_handler = proc_dointvec_minmax,
957 .extra1 = &zero,
958 .extra2 = &one,
959 },
960 #ifdef CONFIG_SMP
961 {
962 .procname = "hardlockup_all_cpu_backtrace",
963 .data = &sysctl_hardlockup_all_cpu_backtrace,
964 .maxlen = sizeof(int),
965 .mode = 0644,
966 .proc_handler = proc_dointvec_minmax,
967 .extra1 = &zero,
968 .extra2 = &one,
969 },
970 #endif /* CONFIG_SMP */
971 #endif
972 #endif
973
974 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
975 {
976 .procname = "unknown_nmi_panic",
977 .data = &unknown_nmi_panic,
978 .maxlen = sizeof (int),
979 .mode = 0644,
980 .proc_handler = proc_dointvec,
981 },
982 #endif
983 #if defined(CONFIG_X86)
984 {
985 .procname = "panic_on_unrecovered_nmi",
986 .data = &panic_on_unrecovered_nmi,
987 .maxlen = sizeof(int),
988 .mode = 0644,
989 .proc_handler = proc_dointvec,
990 },
991 {
992 .procname = "panic_on_io_nmi",
993 .data = &panic_on_io_nmi,
994 .maxlen = sizeof(int),
995 .mode = 0644,
996 .proc_handler = proc_dointvec,
997 },
998 #ifdef CONFIG_DEBUG_STACKOVERFLOW
999 {
1000 .procname = "panic_on_stackoverflow",
1001 .data = &sysctl_panic_on_stackoverflow,
1002 .maxlen = sizeof(int),
1003 .mode = 0644,
1004 .proc_handler = proc_dointvec,
1005 },
1006 #endif
1007 {
1008 .procname = "bootloader_type",
1009 .data = &bootloader_type,
1010 .maxlen = sizeof (int),
1011 .mode = 0444,
1012 .proc_handler = proc_dointvec,
1013 },
1014 {
1015 .procname = "bootloader_version",
1016 .data = &bootloader_version,
1017 .maxlen = sizeof (int),
1018 .mode = 0444,
1019 .proc_handler = proc_dointvec,
1020 },
1021 {
1022 .procname = "io_delay_type",
1023 .data = &io_delay_type,
1024 .maxlen = sizeof(int),
1025 .mode = 0644,
1026 .proc_handler = proc_dointvec,
1027 },
1028 #endif
1029 #if defined(CONFIG_MMU)
1030 {
1031 .procname = "randomize_va_space",
1032 .data = &randomize_va_space,
1033 .maxlen = sizeof(int),
1034 .mode = 0644,
1035 .proc_handler = proc_dointvec,
1036 },
1037 #endif
1038 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1039 {
1040 .procname = "spin_retry",
1041 .data = &spin_retry,
1042 .maxlen = sizeof (int),
1043 .mode = 0644,
1044 .proc_handler = proc_dointvec,
1045 },
1046 #endif
1047 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1048 {
1049 .procname = "acpi_video_flags",
1050 .data = &acpi_realmode_flags,
1051 .maxlen = sizeof (unsigned long),
1052 .mode = 0644,
1053 .proc_handler = proc_doulongvec_minmax,
1054 },
1055 #endif
1056 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1057 {
1058 .procname = "ignore-unaligned-usertrap",
1059 .data = &no_unaligned_warning,
1060 .maxlen = sizeof (int),
1061 .mode = 0644,
1062 .proc_handler = proc_dointvec,
1063 },
1064 #endif
1065 #ifdef CONFIG_IA64
1066 {
1067 .procname = "unaligned-dump-stack",
1068 .data = &unaligned_dump_stack,
1069 .maxlen = sizeof (int),
1070 .mode = 0644,
1071 .proc_handler = proc_dointvec,
1072 },
1073 #endif
1074 #ifdef CONFIG_DETECT_HUNG_TASK
1075 {
1076 .procname = "hung_task_panic",
1077 .data = &sysctl_hung_task_panic,
1078 .maxlen = sizeof(int),
1079 .mode = 0644,
1080 .proc_handler = proc_dointvec_minmax,
1081 .extra1 = &zero,
1082 .extra2 = &one,
1083 },
1084 {
1085 .procname = "hung_task_check_count",
1086 .data = &sysctl_hung_task_check_count,
1087 .maxlen = sizeof(int),
1088 .mode = 0644,
1089 .proc_handler = proc_dointvec_minmax,
1090 .extra1 = &zero,
1091 },
1092 {
1093 .procname = "hung_task_timeout_secs",
1094 .data = &sysctl_hung_task_timeout_secs,
1095 .maxlen = sizeof(unsigned long),
1096 .mode = 0644,
1097 .proc_handler = proc_dohung_task_timeout_secs,
1098 .extra2 = &hung_task_timeout_max,
1099 },
1100 {
1101 .procname = "hung_task_warnings",
1102 .data = &sysctl_hung_task_warnings,
1103 .maxlen = sizeof(int),
1104 .mode = 0644,
1105 .proc_handler = proc_dointvec_minmax,
1106 .extra1 = &neg_one,
1107 },
1108 #endif
1109 #ifdef CONFIG_RT_MUTEXES
1110 {
1111 .procname = "max_lock_depth",
1112 .data = &max_lock_depth,
1113 .maxlen = sizeof(int),
1114 .mode = 0644,
1115 .proc_handler = proc_dointvec,
1116 },
1117 #endif
1118 {
1119 .procname = "poweroff_cmd",
1120 .data = &poweroff_cmd,
1121 .maxlen = POWEROFF_CMD_PATH_LEN,
1122 .mode = 0644,
1123 .proc_handler = proc_dostring,
1124 },
1125 #ifdef CONFIG_KEYS
1126 {
1127 .procname = "keys",
1128 .mode = 0555,
1129 .child = key_sysctls,
1130 },
1131 #endif
1132 #ifdef CONFIG_PERF_EVENTS
1133 /*
1134 * User-space scripts rely on the existence of this file
1135 * as a feature check for perf_events being enabled.
1136 *
1137 * So it's an ABI, do not remove!
1138 */
1139 {
1140 .procname = "perf_event_paranoid",
1141 .data = &sysctl_perf_event_paranoid,
1142 .maxlen = sizeof(sysctl_perf_event_paranoid),
1143 .mode = 0644,
1144 .proc_handler = proc_dointvec,
1145 },
1146 {
1147 .procname = "perf_event_mlock_kb",
1148 .data = &sysctl_perf_event_mlock,
1149 .maxlen = sizeof(sysctl_perf_event_mlock),
1150 .mode = 0644,
1151 .proc_handler = proc_dointvec,
1152 },
1153 {
1154 .procname = "perf_event_max_sample_rate",
1155 .data = &sysctl_perf_event_sample_rate,
1156 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1157 .mode = 0644,
1158 .proc_handler = perf_proc_update_handler,
1159 .extra1 = &one,
1160 },
1161 {
1162 .procname = "perf_cpu_time_max_percent",
1163 .data = &sysctl_perf_cpu_time_max_percent,
1164 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1165 .mode = 0644,
1166 .proc_handler = perf_cpu_time_max_percent_handler,
1167 .extra1 = &zero,
1168 .extra2 = &one_hundred,
1169 },
1170 {
1171 .procname = "perf_event_max_stack",
1172 .data = &sysctl_perf_event_max_stack,
1173 .maxlen = sizeof(sysctl_perf_event_max_stack),
1174 .mode = 0644,
1175 .proc_handler = perf_event_max_stack_handler,
1176 .extra1 = &zero,
1177 .extra2 = &six_hundred_forty_kb,
1178 },
1179 {
1180 .procname = "perf_event_max_contexts_per_stack",
1181 .data = &sysctl_perf_event_max_contexts_per_stack,
1182 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
1183 .mode = 0644,
1184 .proc_handler = perf_event_max_stack_handler,
1185 .extra1 = &zero,
1186 .extra2 = &one_thousand,
1187 },
1188 #endif
1189 {
1190 .procname = "panic_on_warn",
1191 .data = &panic_on_warn,
1192 .maxlen = sizeof(int),
1193 .mode = 0644,
1194 .proc_handler = proc_dointvec_minmax,
1195 .extra1 = &zero,
1196 .extra2 = &one,
1197 },
1198 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1199 {
1200 .procname = "timer_migration",
1201 .data = &sysctl_timer_migration,
1202 .maxlen = sizeof(unsigned int),
1203 .mode = 0644,
1204 .proc_handler = timer_migration_handler,
1205 .extra1 = &zero,
1206 .extra2 = &one,
1207 },
1208 #endif
1209 #ifdef CONFIG_BPF_SYSCALL
1210 {
1211 .procname = "unprivileged_bpf_disabled",
1212 .data = &sysctl_unprivileged_bpf_disabled,
1213 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
1214 .mode = 0644,
1215 /* only handle a transition from default "0" to "1" */
1216 .proc_handler = proc_dointvec_minmax,
1217 .extra1 = &one,
1218 .extra2 = &one,
1219 },
1220 #endif
1221 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1222 {
1223 .procname = "panic_on_rcu_stall",
1224 .data = &sysctl_panic_on_rcu_stall,
1225 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
1226 .mode = 0644,
1227 .proc_handler = proc_dointvec_minmax,
1228 .extra1 = &zero,
1229 .extra2 = &one,
1230 },
1231 #endif
1232 { }
1233 };
1234
1235 static struct ctl_table vm_table[] = {
1236 {
1237 .procname = "overcommit_memory",
1238 .data = &sysctl_overcommit_memory,
1239 .maxlen = sizeof(sysctl_overcommit_memory),
1240 .mode = 0644,
1241 .proc_handler = proc_dointvec_minmax,
1242 .extra1 = &zero,
1243 .extra2 = &two,
1244 },
1245 {
1246 .procname = "panic_on_oom",
1247 .data = &sysctl_panic_on_oom,
1248 .maxlen = sizeof(sysctl_panic_on_oom),
1249 .mode = 0644,
1250 .proc_handler = proc_dointvec_minmax,
1251 .extra1 = &zero,
1252 .extra2 = &two,
1253 },
1254 {
1255 .procname = "oom_kill_allocating_task",
1256 .data = &sysctl_oom_kill_allocating_task,
1257 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1258 .mode = 0644,
1259 .proc_handler = proc_dointvec,
1260 },
1261 {
1262 .procname = "oom_dump_tasks",
1263 .data = &sysctl_oom_dump_tasks,
1264 .maxlen = sizeof(sysctl_oom_dump_tasks),
1265 .mode = 0644,
1266 .proc_handler = proc_dointvec,
1267 },
1268 {
1269 .procname = "overcommit_ratio",
1270 .data = &sysctl_overcommit_ratio,
1271 .maxlen = sizeof(sysctl_overcommit_ratio),
1272 .mode = 0644,
1273 .proc_handler = overcommit_ratio_handler,
1274 },
1275 {
1276 .procname = "overcommit_kbytes",
1277 .data = &sysctl_overcommit_kbytes,
1278 .maxlen = sizeof(sysctl_overcommit_kbytes),
1279 .mode = 0644,
1280 .proc_handler = overcommit_kbytes_handler,
1281 },
1282 {
1283 .procname = "page-cluster",
1284 .data = &page_cluster,
1285 .maxlen = sizeof(int),
1286 .mode = 0644,
1287 .proc_handler = proc_dointvec_minmax,
1288 .extra1 = &zero,
1289 },
1290 {
1291 .procname = "dirty_background_ratio",
1292 .data = &dirty_background_ratio,
1293 .maxlen = sizeof(dirty_background_ratio),
1294 .mode = 0644,
1295 .proc_handler = dirty_background_ratio_handler,
1296 .extra1 = &zero,
1297 .extra2 = &one_hundred,
1298 },
1299 {
1300 .procname = "dirty_background_bytes",
1301 .data = &dirty_background_bytes,
1302 .maxlen = sizeof(dirty_background_bytes),
1303 .mode = 0644,
1304 .proc_handler = dirty_background_bytes_handler,
1305 .extra1 = &one_ul,
1306 },
1307 {
1308 .procname = "dirty_ratio",
1309 .data = &vm_dirty_ratio,
1310 .maxlen = sizeof(vm_dirty_ratio),
1311 .mode = 0644,
1312 .proc_handler = dirty_ratio_handler,
1313 .extra1 = &zero,
1314 .extra2 = &one_hundred,
1315 },
1316 {
1317 .procname = "dirty_bytes",
1318 .data = &vm_dirty_bytes,
1319 .maxlen = sizeof(vm_dirty_bytes),
1320 .mode = 0644,
1321 .proc_handler = dirty_bytes_handler,
1322 .extra1 = &dirty_bytes_min,
1323 },
1324 {
1325 .procname = "dirty_writeback_centisecs",
1326 .data = &dirty_writeback_interval,
1327 .maxlen = sizeof(dirty_writeback_interval),
1328 .mode = 0644,
1329 .proc_handler = dirty_writeback_centisecs_handler,
1330 },
1331 {
1332 .procname = "dirty_expire_centisecs",
1333 .data = &dirty_expire_interval,
1334 .maxlen = sizeof(dirty_expire_interval),
1335 .mode = 0644,
1336 .proc_handler = proc_dointvec_minmax,
1337 .extra1 = &zero,
1338 },
1339 {
1340 .procname = "dirtytime_expire_seconds",
1341 .data = &dirtytime_expire_interval,
1342 .maxlen = sizeof(dirty_expire_interval),
1343 .mode = 0644,
1344 .proc_handler = dirtytime_interval_handler,
1345 .extra1 = &zero,
1346 },
1347 {
1348 .procname = "swappiness",
1349 .data = &vm_swappiness,
1350 .maxlen = sizeof(vm_swappiness),
1351 .mode = 0644,
1352 .proc_handler = proc_dointvec_minmax,
1353 .extra1 = &zero,
1354 .extra2 = &one_hundred,
1355 },
1356 #ifdef CONFIG_HUGETLB_PAGE
1357 {
1358 .procname = "nr_hugepages",
1359 .data = NULL,
1360 .maxlen = sizeof(unsigned long),
1361 .mode = 0644,
1362 .proc_handler = hugetlb_sysctl_handler,
1363 },
1364 #ifdef CONFIG_NUMA
1365 {
1366 .procname = "nr_hugepages_mempolicy",
1367 .data = NULL,
1368 .maxlen = sizeof(unsigned long),
1369 .mode = 0644,
1370 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1371 },
1372 {
1373 .procname = "numa_stat",
1374 .data = &sysctl_vm_numa_stat,
1375 .maxlen = sizeof(int),
1376 .mode = 0644,
1377 .proc_handler = sysctl_vm_numa_stat_handler,
1378 .extra1 = &zero,
1379 .extra2 = &one,
1380 },
1381 #endif
1382 {
1383 .procname = "hugetlb_shm_group",
1384 .data = &sysctl_hugetlb_shm_group,
1385 .maxlen = sizeof(gid_t),
1386 .mode = 0644,
1387 .proc_handler = proc_dointvec,
1388 },
1389 {
1390 .procname = "hugepages_treat_as_movable",
1391 .data = &hugepages_treat_as_movable,
1392 .maxlen = sizeof(int),
1393 .mode = 0644,
1394 .proc_handler = proc_dointvec,
1395 },
1396 {
1397 .procname = "nr_overcommit_hugepages",
1398 .data = NULL,
1399 .maxlen = sizeof(unsigned long),
1400 .mode = 0644,
1401 .proc_handler = hugetlb_overcommit_handler,
1402 },
1403 #endif
1404 {
1405 .procname = "lowmem_reserve_ratio",
1406 .data = &sysctl_lowmem_reserve_ratio,
1407 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1408 .mode = 0644,
1409 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1410 },
1411 {
1412 .procname = "drop_caches",
1413 .data = &sysctl_drop_caches,
1414 .maxlen = sizeof(int),
1415 .mode = 0644,
1416 .proc_handler = drop_caches_sysctl_handler,
1417 .extra1 = &one,
1418 .extra2 = &four,
1419 },
1420 #ifdef CONFIG_COMPACTION
1421 {
1422 .procname = "compact_memory",
1423 .data = &sysctl_compact_memory,
1424 .maxlen = sizeof(int),
1425 .mode = 0200,
1426 .proc_handler = sysctl_compaction_handler,
1427 },
1428 {
1429 .procname = "extfrag_threshold",
1430 .data = &sysctl_extfrag_threshold,
1431 .maxlen = sizeof(int),
1432 .mode = 0644,
1433 .proc_handler = sysctl_extfrag_handler,
1434 .extra1 = &min_extfrag_threshold,
1435 .extra2 = &max_extfrag_threshold,
1436 },
1437 {
1438 .procname = "compact_unevictable_allowed",
1439 .data = &sysctl_compact_unevictable_allowed,
1440 .maxlen = sizeof(int),
1441 .mode = 0644,
1442 .proc_handler = proc_dointvec,
1443 .extra1 = &zero,
1444 .extra2 = &one,
1445 },
1446
1447 #endif /* CONFIG_COMPACTION */
1448 {
1449 .procname = "min_free_kbytes",
1450 .data = &min_free_kbytes,
1451 .maxlen = sizeof(min_free_kbytes),
1452 .mode = 0644,
1453 .proc_handler = min_free_kbytes_sysctl_handler,
1454 .extra1 = &zero,
1455 },
1456 {
1457 .procname = "watermark_scale_factor",
1458 .data = &watermark_scale_factor,
1459 .maxlen = sizeof(watermark_scale_factor),
1460 .mode = 0644,
1461 .proc_handler = watermark_scale_factor_sysctl_handler,
1462 .extra1 = &one,
1463 .extra2 = &one_thousand,
1464 },
1465 {
1466 .procname = "percpu_pagelist_fraction",
1467 .data = &percpu_pagelist_fraction,
1468 .maxlen = sizeof(percpu_pagelist_fraction),
1469 .mode = 0644,
1470 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1471 .extra1 = &zero,
1472 },
1473 #ifdef CONFIG_MMU
1474 {
1475 .procname = "max_map_count",
1476 .data = &sysctl_max_map_count,
1477 .maxlen = sizeof(sysctl_max_map_count),
1478 .mode = 0644,
1479 .proc_handler = proc_dointvec_minmax,
1480 .extra1 = &zero,
1481 },
1482 #else
1483 {
1484 .procname = "nr_trim_pages",
1485 .data = &sysctl_nr_trim_pages,
1486 .maxlen = sizeof(sysctl_nr_trim_pages),
1487 .mode = 0644,
1488 .proc_handler = proc_dointvec_minmax,
1489 .extra1 = &zero,
1490 },
1491 #endif
1492 {
1493 .procname = "laptop_mode",
1494 .data = &laptop_mode,
1495 .maxlen = sizeof(laptop_mode),
1496 .mode = 0644,
1497 .proc_handler = proc_dointvec_jiffies,
1498 },
1499 {
1500 .procname = "block_dump",
1501 .data = &block_dump,
1502 .maxlen = sizeof(block_dump),
1503 .mode = 0644,
1504 .proc_handler = proc_dointvec,
1505 .extra1 = &zero,
1506 },
1507 {
1508 .procname = "vfs_cache_pressure",
1509 .data = &sysctl_vfs_cache_pressure,
1510 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1511 .mode = 0644,
1512 .proc_handler = proc_dointvec,
1513 .extra1 = &zero,
1514 },
1515 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1516 {
1517 .procname = "legacy_va_layout",
1518 .data = &sysctl_legacy_va_layout,
1519 .maxlen = sizeof(sysctl_legacy_va_layout),
1520 .mode = 0644,
1521 .proc_handler = proc_dointvec,
1522 .extra1 = &zero,
1523 },
1524 #endif
1525 #ifdef CONFIG_NUMA
1526 {
1527 .procname = "zone_reclaim_mode",
1528 .data = &node_reclaim_mode,
1529 .maxlen = sizeof(node_reclaim_mode),
1530 .mode = 0644,
1531 .proc_handler = proc_dointvec,
1532 .extra1 = &zero,
1533 },
1534 {
1535 .procname = "min_unmapped_ratio",
1536 .data = &sysctl_min_unmapped_ratio,
1537 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1538 .mode = 0644,
1539 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1540 .extra1 = &zero,
1541 .extra2 = &one_hundred,
1542 },
1543 {
1544 .procname = "min_slab_ratio",
1545 .data = &sysctl_min_slab_ratio,
1546 .maxlen = sizeof(sysctl_min_slab_ratio),
1547 .mode = 0644,
1548 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1549 .extra1 = &zero,
1550 .extra2 = &one_hundred,
1551 },
1552 #endif
1553 #ifdef CONFIG_SMP
1554 {
1555 .procname = "stat_interval",
1556 .data = &sysctl_stat_interval,
1557 .maxlen = sizeof(sysctl_stat_interval),
1558 .mode = 0644,
1559 .proc_handler = proc_dointvec_jiffies,
1560 },
1561 {
1562 .procname = "stat_refresh",
1563 .data = NULL,
1564 .maxlen = 0,
1565 .mode = 0600,
1566 .proc_handler = vmstat_refresh,
1567 },
1568 #endif
1569 #ifdef CONFIG_MMU
1570 {
1571 .procname = "mmap_min_addr",
1572 .data = &dac_mmap_min_addr,
1573 .maxlen = sizeof(unsigned long),
1574 .mode = 0644,
1575 .proc_handler = mmap_min_addr_handler,
1576 },
1577 #endif
1578 #ifdef CONFIG_NUMA
1579 {
1580 .procname = "numa_zonelist_order",
1581 .data = &numa_zonelist_order,
1582 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1583 .mode = 0644,
1584 .proc_handler = numa_zonelist_order_handler,
1585 },
1586 #endif
1587 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1588 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1589 {
1590 .procname = "vdso_enabled",
1591 #ifdef CONFIG_X86_32
1592 .data = &vdso32_enabled,
1593 .maxlen = sizeof(vdso32_enabled),
1594 #else
1595 .data = &vdso_enabled,
1596 .maxlen = sizeof(vdso_enabled),
1597 #endif
1598 .mode = 0644,
1599 .proc_handler = proc_dointvec,
1600 .extra1 = &zero,
1601 },
1602 #endif
1603 #ifdef CONFIG_HIGHMEM
1604 {
1605 .procname = "highmem_is_dirtyable",
1606 .data = &vm_highmem_is_dirtyable,
1607 .maxlen = sizeof(vm_highmem_is_dirtyable),
1608 .mode = 0644,
1609 .proc_handler = proc_dointvec_minmax,
1610 .extra1 = &zero,
1611 .extra2 = &one,
1612 },
1613 #endif
1614 #ifdef CONFIG_MEMORY_FAILURE
1615 {
1616 .procname = "memory_failure_early_kill",
1617 .data = &sysctl_memory_failure_early_kill,
1618 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1619 .mode = 0644,
1620 .proc_handler = proc_dointvec_minmax,
1621 .extra1 = &zero,
1622 .extra2 = &one,
1623 },
1624 {
1625 .procname = "memory_failure_recovery",
1626 .data = &sysctl_memory_failure_recovery,
1627 .maxlen = sizeof(sysctl_memory_failure_recovery),
1628 .mode = 0644,
1629 .proc_handler = proc_dointvec_minmax,
1630 .extra1 = &zero,
1631 .extra2 = &one,
1632 },
1633 #endif
1634 {
1635 .procname = "user_reserve_kbytes",
1636 .data = &sysctl_user_reserve_kbytes,
1637 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1638 .mode = 0644,
1639 .proc_handler = proc_doulongvec_minmax,
1640 },
1641 {
1642 .procname = "admin_reserve_kbytes",
1643 .data = &sysctl_admin_reserve_kbytes,
1644 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1645 .mode = 0644,
1646 .proc_handler = proc_doulongvec_minmax,
1647 },
1648 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1649 {
1650 .procname = "mmap_rnd_bits",
1651 .data = &mmap_rnd_bits,
1652 .maxlen = sizeof(mmap_rnd_bits),
1653 .mode = 0600,
1654 .proc_handler = proc_dointvec_minmax,
1655 .extra1 = (void *)&mmap_rnd_bits_min,
1656 .extra2 = (void *)&mmap_rnd_bits_max,
1657 },
1658 #endif
1659 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1660 {
1661 .procname = "mmap_rnd_compat_bits",
1662 .data = &mmap_rnd_compat_bits,
1663 .maxlen = sizeof(mmap_rnd_compat_bits),
1664 .mode = 0600,
1665 .proc_handler = proc_dointvec_minmax,
1666 .extra1 = (void *)&mmap_rnd_compat_bits_min,
1667 .extra2 = (void *)&mmap_rnd_compat_bits_max,
1668 },
1669 #endif
1670 { }
1671 };
1672
1673 static struct ctl_table fs_table[] = {
1674 {
1675 .procname = "inode-nr",
1676 .data = &inodes_stat,
1677 .maxlen = 2*sizeof(long),
1678 .mode = 0444,
1679 .proc_handler = proc_nr_inodes,
1680 },
1681 {
1682 .procname = "inode-state",
1683 .data = &inodes_stat,
1684 .maxlen = 7*sizeof(long),
1685 .mode = 0444,
1686 .proc_handler = proc_nr_inodes,
1687 },
1688 {
1689 .procname = "file-nr",
1690 .data = &files_stat,
1691 .maxlen = sizeof(files_stat),
1692 .mode = 0444,
1693 .proc_handler = proc_nr_files,
1694 },
1695 {
1696 .procname = "file-max",
1697 .data = &files_stat.max_files,
1698 .maxlen = sizeof(files_stat.max_files),
1699 .mode = 0644,
1700 .proc_handler = proc_doulongvec_minmax,
1701 },
1702 {
1703 .procname = "nr_open",
1704 .data = &sysctl_nr_open,
1705 .maxlen = sizeof(unsigned int),
1706 .mode = 0644,
1707 .proc_handler = proc_dointvec_minmax,
1708 .extra1 = &sysctl_nr_open_min,
1709 .extra2 = &sysctl_nr_open_max,
1710 },
1711 {
1712 .procname = "dentry-state",
1713 .data = &dentry_stat,
1714 .maxlen = 6*sizeof(long),
1715 .mode = 0444,
1716 .proc_handler = proc_nr_dentry,
1717 },
1718 {
1719 .procname = "overflowuid",
1720 .data = &fs_overflowuid,
1721 .maxlen = sizeof(int),
1722 .mode = 0644,
1723 .proc_handler = proc_dointvec_minmax,
1724 .extra1 = &minolduid,
1725 .extra2 = &maxolduid,
1726 },
1727 {
1728 .procname = "overflowgid",
1729 .data = &fs_overflowgid,
1730 .maxlen = sizeof(int),
1731 .mode = 0644,
1732 .proc_handler = proc_dointvec_minmax,
1733 .extra1 = &minolduid,
1734 .extra2 = &maxolduid,
1735 },
1736 #ifdef CONFIG_FILE_LOCKING
1737 {
1738 .procname = "leases-enable",
1739 .data = &leases_enable,
1740 .maxlen = sizeof(int),
1741 .mode = 0644,
1742 .proc_handler = proc_dointvec,
1743 },
1744 #endif
1745 #ifdef CONFIG_DNOTIFY
1746 {
1747 .procname = "dir-notify-enable",
1748 .data = &dir_notify_enable,
1749 .maxlen = sizeof(int),
1750 .mode = 0644,
1751 .proc_handler = proc_dointvec,
1752 },
1753 #endif
1754 #ifdef CONFIG_MMU
1755 #ifdef CONFIG_FILE_LOCKING
1756 {
1757 .procname = "lease-break-time",
1758 .data = &lease_break_time,
1759 .maxlen = sizeof(int),
1760 .mode = 0644,
1761 .proc_handler = proc_dointvec,
1762 },
1763 #endif
1764 #ifdef CONFIG_AIO
1765 {
1766 .procname = "aio-nr",
1767 .data = &aio_nr,
1768 .maxlen = sizeof(aio_nr),
1769 .mode = 0444,
1770 .proc_handler = proc_doulongvec_minmax,
1771 },
1772 {
1773 .procname = "aio-max-nr",
1774 .data = &aio_max_nr,
1775 .maxlen = sizeof(aio_max_nr),
1776 .mode = 0644,
1777 .proc_handler = proc_doulongvec_minmax,
1778 },
1779 #endif /* CONFIG_AIO */
1780 #ifdef CONFIG_INOTIFY_USER
1781 {
1782 .procname = "inotify",
1783 .mode = 0555,
1784 .child = inotify_table,
1785 },
1786 #endif
1787 #ifdef CONFIG_EPOLL
1788 {
1789 .procname = "epoll",
1790 .mode = 0555,
1791 .child = epoll_table,
1792 },
1793 #endif
1794 #endif
1795 {
1796 .procname = "protected_symlinks",
1797 .data = &sysctl_protected_symlinks,
1798 .maxlen = sizeof(int),
1799 .mode = 0600,
1800 .proc_handler = proc_dointvec_minmax,
1801 .extra1 = &zero,
1802 .extra2 = &one,
1803 },
1804 {
1805 .procname = "protected_hardlinks",
1806 .data = &sysctl_protected_hardlinks,
1807 .maxlen = sizeof(int),
1808 .mode = 0600,
1809 .proc_handler = proc_dointvec_minmax,
1810 .extra1 = &zero,
1811 .extra2 = &one,
1812 },
1813 {
1814 .procname = "suid_dumpable",
1815 .data = &suid_dumpable,
1816 .maxlen = sizeof(int),
1817 .mode = 0644,
1818 .proc_handler = proc_dointvec_minmax_coredump,
1819 .extra1 = &zero,
1820 .extra2 = &two,
1821 },
1822 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1823 {
1824 .procname = "binfmt_misc",
1825 .mode = 0555,
1826 .child = sysctl_mount_point,
1827 },
1828 #endif
1829 {
1830 .procname = "pipe-max-size",
1831 .data = &pipe_max_size,
1832 .maxlen = sizeof(pipe_max_size),
1833 .mode = 0644,
1834 .proc_handler = &pipe_proc_fn,
1835 .extra1 = &pipe_min_size,
1836 },
1837 {
1838 .procname = "pipe-user-pages-hard",
1839 .data = &pipe_user_pages_hard,
1840 .maxlen = sizeof(pipe_user_pages_hard),
1841 .mode = 0644,
1842 .proc_handler = proc_doulongvec_minmax,
1843 },
1844 {
1845 .procname = "pipe-user-pages-soft",
1846 .data = &pipe_user_pages_soft,
1847 .maxlen = sizeof(pipe_user_pages_soft),
1848 .mode = 0644,
1849 .proc_handler = proc_doulongvec_minmax,
1850 },
1851 {
1852 .procname = "mount-max",
1853 .data = &sysctl_mount_max,
1854 .maxlen = sizeof(unsigned int),
1855 .mode = 0644,
1856 .proc_handler = proc_dointvec_minmax,
1857 .extra1 = &one,
1858 },
1859 { }
1860 };
1861
1862 static struct ctl_table debug_table[] = {
1863 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1864 {
1865 .procname = "exception-trace",
1866 .data = &show_unhandled_signals,
1867 .maxlen = sizeof(int),
1868 .mode = 0644,
1869 .proc_handler = proc_dointvec
1870 },
1871 #endif
1872 #if defined(CONFIG_OPTPROBES)
1873 {
1874 .procname = "kprobes-optimization",
1875 .data = &sysctl_kprobes_optimization,
1876 .maxlen = sizeof(int),
1877 .mode = 0644,
1878 .proc_handler = proc_kprobes_optimization_handler,
1879 .extra1 = &zero,
1880 .extra2 = &one,
1881 },
1882 #endif
1883 { }
1884 };
1885
1886 static struct ctl_table dev_table[] = {
1887 { }
1888 };
1889
1890 int __init sysctl_init(void)
1891 {
1892 struct ctl_table_header *hdr;
1893
1894 hdr = register_sysctl_table(sysctl_base_table);
1895 kmemleak_not_leak(hdr);
1896 return 0;
1897 }
1898
1899 #endif /* CONFIG_SYSCTL */
1900
1901 /*
1902 * /proc/sys support
1903 */
1904
1905 #ifdef CONFIG_PROC_SYSCTL
1906
1907 static int _proc_do_string(char *data, int maxlen, int write,
1908 char __user *buffer,
1909 size_t *lenp, loff_t *ppos)
1910 {
1911 size_t len;
1912 char __user *p;
1913 char c;
1914
1915 if (!data || !maxlen || !*lenp) {
1916 *lenp = 0;
1917 return 0;
1918 }
1919
1920 if (write) {
1921 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1922 /* Only continue writes not past the end of buffer. */
1923 len = strlen(data);
1924 if (len > maxlen - 1)
1925 len = maxlen - 1;
1926
1927 if (*ppos > len)
1928 return 0;
1929 len = *ppos;
1930 } else {
1931 /* Start writing from beginning of buffer. */
1932 len = 0;
1933 }
1934
1935 *ppos += *lenp;
1936 p = buffer;
1937 while ((p - buffer) < *lenp && len < maxlen - 1) {
1938 if (get_user(c, p++))
1939 return -EFAULT;
1940 if (c == 0 || c == '\n')
1941 break;
1942 data[len++] = c;
1943 }
1944 data[len] = 0;
1945 } else {
1946 len = strlen(data);
1947 if (len > maxlen)
1948 len = maxlen;
1949
1950 if (*ppos > len) {
1951 *lenp = 0;
1952 return 0;
1953 }
1954
1955 data += *ppos;
1956 len -= *ppos;
1957
1958 if (len > *lenp)
1959 len = *lenp;
1960 if (len)
1961 if (copy_to_user(buffer, data, len))
1962 return -EFAULT;
1963 if (len < *lenp) {
1964 if (put_user('\n', buffer + len))
1965 return -EFAULT;
1966 len++;
1967 }
1968 *lenp = len;
1969 *ppos += len;
1970 }
1971 return 0;
1972 }
1973
1974 static void warn_sysctl_write(struct ctl_table *table)
1975 {
1976 pr_warn_once("%s wrote to %s when file position was not 0!\n"
1977 "This will not be supported in the future. To silence this\n"
1978 "warning, set kernel.sysctl_writes_strict = -1\n",
1979 current->comm, table->procname);
1980 }
1981
1982 /**
1983 * proc_first_pos_non_zero_ignore - check if firs position is allowed
1984 * @ppos: file position
1985 * @table: the sysctl table
1986 *
1987 * Returns true if the first position is non-zero and the sysctl_writes_strict
1988 * mode indicates this is not allowed for numeric input types. String proc
1989 * hadlers can ignore the return value.
1990 */
1991 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
1992 struct ctl_table *table)
1993 {
1994 if (!*ppos)
1995 return false;
1996
1997 switch (sysctl_writes_strict) {
1998 case SYSCTL_WRITES_STRICT:
1999 return true;
2000 case SYSCTL_WRITES_WARN:
2001 warn_sysctl_write(table);
2002 return false;
2003 default:
2004 return false;
2005 }
2006 }
2007
2008 /**
2009 * proc_dostring - read a string sysctl
2010 * @table: the sysctl table
2011 * @write: %TRUE if this is a write to the sysctl file
2012 * @buffer: the user buffer
2013 * @lenp: the size of the user buffer
2014 * @ppos: file position
2015 *
2016 * Reads/writes a string from/to the user buffer. If the kernel
2017 * buffer provided is not large enough to hold the string, the
2018 * string is truncated. The copied string is %NULL-terminated.
2019 * If the string is being read by the user process, it is copied
2020 * and a newline '\n' is added. It is truncated if the buffer is
2021 * not large enough.
2022 *
2023 * Returns 0 on success.
2024 */
2025 int proc_dostring(struct ctl_table *table, int write,
2026 void __user *buffer, size_t *lenp, loff_t *ppos)
2027 {
2028 if (write)
2029 proc_first_pos_non_zero_ignore(ppos, table);
2030
2031 return _proc_do_string((char *)(table->data), table->maxlen, write,
2032 (char __user *)buffer, lenp, ppos);
2033 }
2034
2035 static size_t proc_skip_spaces(char **buf)
2036 {
2037 size_t ret;
2038 char *tmp = skip_spaces(*buf);
2039 ret = tmp - *buf;
2040 *buf = tmp;
2041 return ret;
2042 }
2043
2044 static void proc_skip_char(char **buf, size_t *size, const char v)
2045 {
2046 while (*size) {
2047 if (**buf != v)
2048 break;
2049 (*size)--;
2050 (*buf)++;
2051 }
2052 }
2053
2054 #define TMPBUFLEN 22
2055 /**
2056 * proc_get_long - reads an ASCII formatted integer from a user buffer
2057 *
2058 * @buf: a kernel buffer
2059 * @size: size of the kernel buffer
2060 * @val: this is where the number will be stored
2061 * @neg: set to %TRUE if number is negative
2062 * @perm_tr: a vector which contains the allowed trailers
2063 * @perm_tr_len: size of the perm_tr vector
2064 * @tr: pointer to store the trailer character
2065 *
2066 * In case of success %0 is returned and @buf and @size are updated with
2067 * the amount of bytes read. If @tr is non-NULL and a trailing
2068 * character exists (size is non-zero after returning from this
2069 * function), @tr is updated with the trailing character.
2070 */
2071 static int proc_get_long(char **buf, size_t *size,
2072 unsigned long *val, bool *neg,
2073 const char *perm_tr, unsigned perm_tr_len, char *tr)
2074 {
2075 int len;
2076 char *p, tmp[TMPBUFLEN];
2077
2078 if (!*size)
2079 return -EINVAL;
2080
2081 len = *size;
2082 if (len > TMPBUFLEN - 1)
2083 len = TMPBUFLEN - 1;
2084
2085 memcpy(tmp, *buf, len);
2086
2087 tmp[len] = 0;
2088 p = tmp;
2089 if (*p == '-' && *size > 1) {
2090 *neg = true;
2091 p++;
2092 } else
2093 *neg = false;
2094 if (!isdigit(*p))
2095 return -EINVAL;
2096
2097 *val = simple_strtoul(p, &p, 0);
2098
2099 len = p - tmp;
2100
2101 /* We don't know if the next char is whitespace thus we may accept
2102 * invalid integers (e.g. 1234...a) or two integers instead of one
2103 * (e.g. 123...1). So lets not allow such large numbers. */
2104 if (len == TMPBUFLEN - 1)
2105 return -EINVAL;
2106
2107 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2108 return -EINVAL;
2109
2110 if (tr && (len < *size))
2111 *tr = *p;
2112
2113 *buf += len;
2114 *size -= len;
2115
2116 return 0;
2117 }
2118
2119 /**
2120 * proc_put_long - converts an integer to a decimal ASCII formatted string
2121 *
2122 * @buf: the user buffer
2123 * @size: the size of the user buffer
2124 * @val: the integer to be converted
2125 * @neg: sign of the number, %TRUE for negative
2126 *
2127 * In case of success %0 is returned and @buf and @size are updated with
2128 * the amount of bytes written.
2129 */
2130 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2131 bool neg)
2132 {
2133 int len;
2134 char tmp[TMPBUFLEN], *p = tmp;
2135
2136 sprintf(p, "%s%lu", neg ? "-" : "", val);
2137 len = strlen(tmp);
2138 if (len > *size)
2139 len = *size;
2140 if (copy_to_user(*buf, tmp, len))
2141 return -EFAULT;
2142 *size -= len;
2143 *buf += len;
2144 return 0;
2145 }
2146 #undef TMPBUFLEN
2147
2148 static int proc_put_char(void __user **buf, size_t *size, char c)
2149 {
2150 if (*size) {
2151 char __user **buffer = (char __user **)buf;
2152 if (put_user(c, *buffer))
2153 return -EFAULT;
2154 (*size)--, (*buffer)++;
2155 *buf = *buffer;
2156 }
2157 return 0;
2158 }
2159
2160 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2161 int *valp,
2162 int write, void *data)
2163 {
2164 if (write) {
2165 if (*negp) {
2166 if (*lvalp > (unsigned long) INT_MAX + 1)
2167 return -EINVAL;
2168 *valp = -*lvalp;
2169 } else {
2170 if (*lvalp > (unsigned long) INT_MAX)
2171 return -EINVAL;
2172 *valp = *lvalp;
2173 }
2174 } else {
2175 int val = *valp;
2176 if (val < 0) {
2177 *negp = true;
2178 *lvalp = -(unsigned long)val;
2179 } else {
2180 *negp = false;
2181 *lvalp = (unsigned long)val;
2182 }
2183 }
2184 return 0;
2185 }
2186
2187 static int do_proc_douintvec_conv(unsigned long *lvalp,
2188 unsigned int *valp,
2189 int write, void *data)
2190 {
2191 if (write) {
2192 if (*lvalp > UINT_MAX)
2193 return -EINVAL;
2194 *valp = *lvalp;
2195 } else {
2196 unsigned int val = *valp;
2197 *lvalp = (unsigned long)val;
2198 }
2199 return 0;
2200 }
2201
2202 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2203
2204 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2205 int write, void __user *buffer,
2206 size_t *lenp, loff_t *ppos,
2207 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2208 int write, void *data),
2209 void *data)
2210 {
2211 int *i, vleft, first = 1, err = 0;
2212 size_t left;
2213 char *kbuf = NULL, *p;
2214
2215 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2216 *lenp = 0;
2217 return 0;
2218 }
2219
2220 i = (int *) tbl_data;
2221 vleft = table->maxlen / sizeof(*i);
2222 left = *lenp;
2223
2224 if (!conv)
2225 conv = do_proc_dointvec_conv;
2226
2227 if (write) {
2228 if (proc_first_pos_non_zero_ignore(ppos, table))
2229 goto out;
2230
2231 if (left > PAGE_SIZE - 1)
2232 left = PAGE_SIZE - 1;
2233 p = kbuf = memdup_user_nul(buffer, left);
2234 if (IS_ERR(kbuf))
2235 return PTR_ERR(kbuf);
2236 }
2237
2238 for (; left && vleft--; i++, first=0) {
2239 unsigned long lval;
2240 bool neg;
2241
2242 if (write) {
2243 left -= proc_skip_spaces(&p);
2244
2245 if (!left)
2246 break;
2247 err = proc_get_long(&p, &left, &lval, &neg,
2248 proc_wspace_sep,
2249 sizeof(proc_wspace_sep), NULL);
2250 if (err)
2251 break;
2252 if (conv(&neg, &lval, i, 1, data)) {
2253 err = -EINVAL;
2254 break;
2255 }
2256 } else {
2257 if (conv(&neg, &lval, i, 0, data)) {
2258 err = -EINVAL;
2259 break;
2260 }
2261 if (!first)
2262 err = proc_put_char(&buffer, &left, '\t');
2263 if (err)
2264 break;
2265 err = proc_put_long(&buffer, &left, lval, neg);
2266 if (err)
2267 break;
2268 }
2269 }
2270
2271 if (!write && !first && left && !err)
2272 err = proc_put_char(&buffer, &left, '\n');
2273 if (write && !err && left)
2274 left -= proc_skip_spaces(&p);
2275 if (write) {
2276 kfree(kbuf);
2277 if (first)
2278 return err ? : -EINVAL;
2279 }
2280 *lenp -= left;
2281 out:
2282 *ppos += *lenp;
2283 return err;
2284 }
2285
2286 static int do_proc_dointvec(struct ctl_table *table, int write,
2287 void __user *buffer, size_t *lenp, loff_t *ppos,
2288 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2289 int write, void *data),
2290 void *data)
2291 {
2292 return __do_proc_dointvec(table->data, table, write,
2293 buffer, lenp, ppos, conv, data);
2294 }
2295
2296 static int do_proc_douintvec_w(unsigned int *tbl_data,
2297 struct ctl_table *table,
2298 void __user *buffer,
2299 size_t *lenp, loff_t *ppos,
2300 int (*conv)(unsigned long *lvalp,
2301 unsigned int *valp,
2302 int write, void *data),
2303 void *data)
2304 {
2305 unsigned long lval;
2306 int err = 0;
2307 size_t left;
2308 bool neg;
2309 char *kbuf = NULL, *p;
2310
2311 left = *lenp;
2312
2313 if (proc_first_pos_non_zero_ignore(ppos, table))
2314 goto bail_early;
2315
2316 if (left > PAGE_SIZE - 1)
2317 left = PAGE_SIZE - 1;
2318
2319 p = kbuf = memdup_user_nul(buffer, left);
2320 if (IS_ERR(kbuf))
2321 return -EINVAL;
2322
2323 left -= proc_skip_spaces(&p);
2324 if (!left) {
2325 err = -EINVAL;
2326 goto out_free;
2327 }
2328
2329 err = proc_get_long(&p, &left, &lval, &neg,
2330 proc_wspace_sep,
2331 sizeof(proc_wspace_sep), NULL);
2332 if (err || neg) {
2333 err = -EINVAL;
2334 goto out_free;
2335 }
2336
2337 if (conv(&lval, tbl_data, 1, data)) {
2338 err = -EINVAL;
2339 goto out_free;
2340 }
2341
2342 if (!err && left)
2343 left -= proc_skip_spaces(&p);
2344
2345 out_free:
2346 kfree(kbuf);
2347 if (err)
2348 return -EINVAL;
2349
2350 return 0;
2351
2352 /* This is in keeping with old __do_proc_dointvec() */
2353 bail_early:
2354 *ppos += *lenp;
2355 return err;
2356 }
2357
2358 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2359 size_t *lenp, loff_t *ppos,
2360 int (*conv)(unsigned long *lvalp,
2361 unsigned int *valp,
2362 int write, void *data),
2363 void *data)
2364 {
2365 unsigned long lval;
2366 int err = 0;
2367 size_t left;
2368
2369 left = *lenp;
2370
2371 if (conv(&lval, tbl_data, 0, data)) {
2372 err = -EINVAL;
2373 goto out;
2374 }
2375
2376 err = proc_put_long(&buffer, &left, lval, false);
2377 if (err || !left)
2378 goto out;
2379
2380 err = proc_put_char(&buffer, &left, '\n');
2381
2382 out:
2383 *lenp -= left;
2384 *ppos += *lenp;
2385
2386 return err;
2387 }
2388
2389 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2390 int write, void __user *buffer,
2391 size_t *lenp, loff_t *ppos,
2392 int (*conv)(unsigned long *lvalp,
2393 unsigned int *valp,
2394 int write, void *data),
2395 void *data)
2396 {
2397 unsigned int *i, vleft;
2398
2399 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2400 *lenp = 0;
2401 return 0;
2402 }
2403
2404 i = (unsigned int *) tbl_data;
2405 vleft = table->maxlen / sizeof(*i);
2406
2407 /*
2408 * Arrays are not supported, keep this simple. *Do not* add
2409 * support for them.
2410 */
2411 if (vleft != 1) {
2412 *lenp = 0;
2413 return -EINVAL;
2414 }
2415
2416 if (!conv)
2417 conv = do_proc_douintvec_conv;
2418
2419 if (write)
2420 return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2421 conv, data);
2422 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2423 }
2424
2425 static int do_proc_douintvec(struct ctl_table *table, int write,
2426 void __user *buffer, size_t *lenp, loff_t *ppos,
2427 int (*conv)(unsigned long *lvalp,
2428 unsigned int *valp,
2429 int write, void *data),
2430 void *data)
2431 {
2432 return __do_proc_douintvec(table->data, table, write,
2433 buffer, lenp, ppos, conv, data);
2434 }
2435
2436 /**
2437 * proc_dointvec - read a vector of integers
2438 * @table: the sysctl table
2439 * @write: %TRUE if this is a write to the sysctl file
2440 * @buffer: the user buffer
2441 * @lenp: the size of the user buffer
2442 * @ppos: file position
2443 *
2444 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2445 * values from/to the user buffer, treated as an ASCII string.
2446 *
2447 * Returns 0 on success.
2448 */
2449 int proc_dointvec(struct ctl_table *table, int write,
2450 void __user *buffer, size_t *lenp, loff_t *ppos)
2451 {
2452 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2453 }
2454
2455 /**
2456 * proc_douintvec - read a vector of unsigned integers
2457 * @table: the sysctl table
2458 * @write: %TRUE if this is a write to the sysctl file
2459 * @buffer: the user buffer
2460 * @lenp: the size of the user buffer
2461 * @ppos: file position
2462 *
2463 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2464 * values from/to the user buffer, treated as an ASCII string.
2465 *
2466 * Returns 0 on success.
2467 */
2468 int proc_douintvec(struct ctl_table *table, int write,
2469 void __user *buffer, size_t *lenp, loff_t *ppos)
2470 {
2471 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2472 do_proc_douintvec_conv, NULL);
2473 }
2474
2475 /*
2476 * Taint values can only be increased
2477 * This means we can safely use a temporary.
2478 */
2479 static int proc_taint(struct ctl_table *table, int write,
2480 void __user *buffer, size_t *lenp, loff_t *ppos)
2481 {
2482 struct ctl_table t;
2483 unsigned long tmptaint = get_taint();
2484 int err;
2485
2486 if (write && !capable(CAP_SYS_ADMIN))
2487 return -EPERM;
2488
2489 t = *table;
2490 t.data = &tmptaint;
2491 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2492 if (err < 0)
2493 return err;
2494
2495 if (write) {
2496 /*
2497 * Poor man's atomic or. Not worth adding a primitive
2498 * to everyone's atomic.h for this
2499 */
2500 int i;
2501 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2502 if ((tmptaint >> i) & 1)
2503 add_taint(i, LOCKDEP_STILL_OK);
2504 }
2505 }
2506
2507 return err;
2508 }
2509
2510 #ifdef CONFIG_PRINTK
2511 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2512 void __user *buffer, size_t *lenp, loff_t *ppos)
2513 {
2514 if (write && !capable(CAP_SYS_ADMIN))
2515 return -EPERM;
2516
2517 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2518 }
2519 #endif
2520
2521 struct do_proc_dointvec_minmax_conv_param {
2522 int *min;
2523 int *max;
2524 };
2525
2526 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2527 int *valp,
2528 int write, void *data)
2529 {
2530 struct do_proc_dointvec_minmax_conv_param *param = data;
2531 if (write) {
2532 int val = *negp ? -*lvalp : *lvalp;
2533 if ((param->min && *param->min > val) ||
2534 (param->max && *param->max < val))
2535 return -EINVAL;
2536 *valp = val;
2537 } else {
2538 int val = *valp;
2539 if (val < 0) {
2540 *negp = true;
2541 *lvalp = -(unsigned long)val;
2542 } else {
2543 *negp = false;
2544 *lvalp = (unsigned long)val;
2545 }
2546 }
2547 return 0;
2548 }
2549
2550 /**
2551 * proc_dointvec_minmax - read a vector of integers with min/max values
2552 * @table: the sysctl table
2553 * @write: %TRUE if this is a write to the sysctl file
2554 * @buffer: the user buffer
2555 * @lenp: the size of the user buffer
2556 * @ppos: file position
2557 *
2558 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2559 * values from/to the user buffer, treated as an ASCII string.
2560 *
2561 * This routine will ensure the values are within the range specified by
2562 * table->extra1 (min) and table->extra2 (max).
2563 *
2564 * Returns 0 on success.
2565 */
2566 int proc_dointvec_minmax(struct ctl_table *table, int write,
2567 void __user *buffer, size_t *lenp, loff_t *ppos)
2568 {
2569 struct do_proc_dointvec_minmax_conv_param param = {
2570 .min = (int *) table->extra1,
2571 .max = (int *) table->extra2,
2572 };
2573 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2574 do_proc_dointvec_minmax_conv, &param);
2575 }
2576
2577 struct do_proc_douintvec_minmax_conv_param {
2578 unsigned int *min;
2579 unsigned int *max;
2580 };
2581
2582 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2583 unsigned int *valp,
2584 int write, void *data)
2585 {
2586 struct do_proc_douintvec_minmax_conv_param *param = data;
2587
2588 if (write) {
2589 unsigned int val = *lvalp;
2590
2591 if (*lvalp > UINT_MAX)
2592 return -EINVAL;
2593
2594 if ((param->min && *param->min > val) ||
2595 (param->max && *param->max < val))
2596 return -ERANGE;
2597
2598 *valp = val;
2599 } else {
2600 unsigned int val = *valp;
2601 *lvalp = (unsigned long) val;
2602 }
2603
2604 return 0;
2605 }
2606
2607 /**
2608 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2609 * @table: the sysctl table
2610 * @write: %TRUE if this is a write to the sysctl file
2611 * @buffer: the user buffer
2612 * @lenp: the size of the user buffer
2613 * @ppos: file position
2614 *
2615 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2616 * values from/to the user buffer, treated as an ASCII string. Negative
2617 * strings are not allowed.
2618 *
2619 * This routine will ensure the values are within the range specified by
2620 * table->extra1 (min) and table->extra2 (max). There is a final sanity
2621 * check for UINT_MAX to avoid having to support wrap around uses from
2622 * userspace.
2623 *
2624 * Returns 0 on success.
2625 */
2626 int proc_douintvec_minmax(struct ctl_table *table, int write,
2627 void __user *buffer, size_t *lenp, loff_t *ppos)
2628 {
2629 struct do_proc_douintvec_minmax_conv_param param = {
2630 .min = (unsigned int *) table->extra1,
2631 .max = (unsigned int *) table->extra2,
2632 };
2633 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2634 do_proc_douintvec_minmax_conv, &param);
2635 }
2636
2637 struct do_proc_dopipe_max_size_conv_param {
2638 unsigned int *min;
2639 };
2640
2641 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
2642 unsigned int *valp,
2643 int write, void *data)
2644 {
2645 struct do_proc_dopipe_max_size_conv_param *param = data;
2646
2647 if (write) {
2648 unsigned int val;
2649
2650 if (*lvalp > UINT_MAX)
2651 return -EINVAL;
2652
2653 val = round_pipe_size(*lvalp);
2654 if (val == 0)
2655 return -EINVAL;
2656
2657 if (param->min && *param->min > val)
2658 return -ERANGE;
2659
2660 *valp = val;
2661 } else {
2662 unsigned int val = *valp;
2663 *lvalp = (unsigned long) val;
2664 }
2665
2666 return 0;
2667 }
2668
2669 int proc_dopipe_max_size(struct ctl_table *table, int write,
2670 void __user *buffer, size_t *lenp, loff_t *ppos)
2671 {
2672 struct do_proc_dopipe_max_size_conv_param param = {
2673 .min = (unsigned int *) table->extra1,
2674 };
2675 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2676 do_proc_dopipe_max_size_conv, &param);
2677 }
2678
2679 static void validate_coredump_safety(void)
2680 {
2681 #ifdef CONFIG_COREDUMP
2682 if (suid_dumpable == SUID_DUMP_ROOT &&
2683 core_pattern[0] != '/' && core_pattern[0] != '|') {
2684 printk(KERN_WARNING
2685 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2686 "Pipe handler or fully qualified core dump path required.\n"
2687 "Set kernel.core_pattern before fs.suid_dumpable.\n"
2688 );
2689 }
2690 #endif
2691 }
2692
2693 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2694 void __user *buffer, size_t *lenp, loff_t *ppos)
2695 {
2696 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2697 if (!error)
2698 validate_coredump_safety();
2699 return error;
2700 }
2701
2702 #ifdef CONFIG_COREDUMP
2703 static int proc_dostring_coredump(struct ctl_table *table, int write,
2704 void __user *buffer, size_t *lenp, loff_t *ppos)
2705 {
2706 int error = proc_dostring(table, write, buffer, lenp, ppos);
2707 if (!error)
2708 validate_coredump_safety();
2709 return error;
2710 }
2711 #endif
2712
2713 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2714 void __user *buffer,
2715 size_t *lenp, loff_t *ppos,
2716 unsigned long convmul,
2717 unsigned long convdiv)
2718 {
2719 unsigned long *i, *min, *max;
2720 int vleft, first = 1, err = 0;
2721 size_t left;
2722 char *kbuf = NULL, *p;
2723
2724 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2725 *lenp = 0;
2726 return 0;
2727 }
2728
2729 i = (unsigned long *) data;
2730 min = (unsigned long *) table->extra1;
2731 max = (unsigned long *) table->extra2;
2732 vleft = table->maxlen / sizeof(unsigned long);
2733 left = *lenp;
2734
2735 if (write) {
2736 if (proc_first_pos_non_zero_ignore(ppos, table))
2737 goto out;
2738
2739 if (left > PAGE_SIZE - 1)
2740 left = PAGE_SIZE - 1;
2741 p = kbuf = memdup_user_nul(buffer, left);
2742 if (IS_ERR(kbuf))
2743 return PTR_ERR(kbuf);
2744 }
2745
2746 for (; left && vleft--; i++, first = 0) {
2747 unsigned long val;
2748
2749 if (write) {
2750 bool neg;
2751
2752 left -= proc_skip_spaces(&p);
2753
2754 err = proc_get_long(&p, &left, &val, &neg,
2755 proc_wspace_sep,
2756 sizeof(proc_wspace_sep), NULL);
2757 if (err)
2758 break;
2759 if (neg)
2760 continue;
2761 val = convmul * val / convdiv;
2762 if ((min && val < *min) || (max && val > *max))
2763 continue;
2764 *i = val;
2765 } else {
2766 val = convdiv * (*i) / convmul;
2767 if (!first) {
2768 err = proc_put_char(&buffer, &left, '\t');
2769 if (err)
2770 break;
2771 }
2772 err = proc_put_long(&buffer, &left, val, false);
2773 if (err)
2774 break;
2775 }
2776 }
2777
2778 if (!write && !first && left && !err)
2779 err = proc_put_char(&buffer, &left, '\n');
2780 if (write && !err)
2781 left -= proc_skip_spaces(&p);
2782 if (write) {
2783 kfree(kbuf);
2784 if (first)
2785 return err ? : -EINVAL;
2786 }
2787 *lenp -= left;
2788 out:
2789 *ppos += *lenp;
2790 return err;
2791 }
2792
2793 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2794 void __user *buffer,
2795 size_t *lenp, loff_t *ppos,
2796 unsigned long convmul,
2797 unsigned long convdiv)
2798 {
2799 return __do_proc_doulongvec_minmax(table->data, table, write,
2800 buffer, lenp, ppos, convmul, convdiv);
2801 }
2802
2803 /**
2804 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2805 * @table: the sysctl table
2806 * @write: %TRUE if this is a write to the sysctl file
2807 * @buffer: the user buffer
2808 * @lenp: the size of the user buffer
2809 * @ppos: file position
2810 *
2811 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2812 * values from/to the user buffer, treated as an ASCII string.
2813 *
2814 * This routine will ensure the values are within the range specified by
2815 * table->extra1 (min) and table->extra2 (max).
2816 *
2817 * Returns 0 on success.
2818 */
2819 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2820 void __user *buffer, size_t *lenp, loff_t *ppos)
2821 {
2822 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2823 }
2824
2825 /**
2826 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2827 * @table: the sysctl table
2828 * @write: %TRUE if this is a write to the sysctl file
2829 * @buffer: the user buffer
2830 * @lenp: the size of the user buffer
2831 * @ppos: file position
2832 *
2833 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2834 * values from/to the user buffer, treated as an ASCII string. The values
2835 * are treated as milliseconds, and converted to jiffies when they are stored.
2836 *
2837 * This routine will ensure the values are within the range specified by
2838 * table->extra1 (min) and table->extra2 (max).
2839 *
2840 * Returns 0 on success.
2841 */
2842 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2843 void __user *buffer,
2844 size_t *lenp, loff_t *ppos)
2845 {
2846 return do_proc_doulongvec_minmax(table, write, buffer,
2847 lenp, ppos, HZ, 1000l);
2848 }
2849
2850
2851 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2852 int *valp,
2853 int write, void *data)
2854 {
2855 if (write) {
2856 if (*lvalp > INT_MAX / HZ)
2857 return 1;
2858 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2859 } else {
2860 int val = *valp;
2861 unsigned long lval;
2862 if (val < 0) {
2863 *negp = true;
2864 lval = -(unsigned long)val;
2865 } else {
2866 *negp = false;
2867 lval = (unsigned long)val;
2868 }
2869 *lvalp = lval / HZ;
2870 }
2871 return 0;
2872 }
2873
2874 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2875 int *valp,
2876 int write, void *data)
2877 {
2878 if (write) {
2879 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2880 return 1;
2881 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2882 } else {
2883 int val = *valp;
2884 unsigned long lval;
2885 if (val < 0) {
2886 *negp = true;
2887 lval = -(unsigned long)val;
2888 } else {
2889 *negp = false;
2890 lval = (unsigned long)val;
2891 }
2892 *lvalp = jiffies_to_clock_t(lval);
2893 }
2894 return 0;
2895 }
2896
2897 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2898 int *valp,
2899 int write, void *data)
2900 {
2901 if (write) {
2902 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2903
2904 if (jif > INT_MAX)
2905 return 1;
2906 *valp = (int)jif;
2907 } else {
2908 int val = *valp;
2909 unsigned long lval;
2910 if (val < 0) {
2911 *negp = true;
2912 lval = -(unsigned long)val;
2913 } else {
2914 *negp = false;
2915 lval = (unsigned long)val;
2916 }
2917 *lvalp = jiffies_to_msecs(lval);
2918 }
2919 return 0;
2920 }
2921
2922 /**
2923 * proc_dointvec_jiffies - read a vector of integers as seconds
2924 * @table: the sysctl table
2925 * @write: %TRUE if this is a write to the sysctl file
2926 * @buffer: the user buffer
2927 * @lenp: the size of the user buffer
2928 * @ppos: file position
2929 *
2930 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2931 * values from/to the user buffer, treated as an ASCII string.
2932 * The values read are assumed to be in seconds, and are converted into
2933 * jiffies.
2934 *
2935 * Returns 0 on success.
2936 */
2937 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2938 void __user *buffer, size_t *lenp, loff_t *ppos)
2939 {
2940 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2941 do_proc_dointvec_jiffies_conv,NULL);
2942 }
2943
2944 /**
2945 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2946 * @table: the sysctl table
2947 * @write: %TRUE if this is a write to the sysctl file
2948 * @buffer: the user buffer
2949 * @lenp: the size of the user buffer
2950 * @ppos: pointer to the file position
2951 *
2952 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2953 * values from/to the user buffer, treated as an ASCII string.
2954 * The values read are assumed to be in 1/USER_HZ seconds, and
2955 * are converted into jiffies.
2956 *
2957 * Returns 0 on success.
2958 */
2959 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2960 void __user *buffer, size_t *lenp, loff_t *ppos)
2961 {
2962 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2963 do_proc_dointvec_userhz_jiffies_conv,NULL);
2964 }
2965
2966 /**
2967 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2968 * @table: the sysctl table
2969 * @write: %TRUE if this is a write to the sysctl file
2970 * @buffer: the user buffer
2971 * @lenp: the size of the user buffer
2972 * @ppos: file position
2973 * @ppos: the current position in the file
2974 *
2975 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2976 * values from/to the user buffer, treated as an ASCII string.
2977 * The values read are assumed to be in 1/1000 seconds, and
2978 * are converted into jiffies.
2979 *
2980 * Returns 0 on success.
2981 */
2982 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2983 void __user *buffer, size_t *lenp, loff_t *ppos)
2984 {
2985 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2986 do_proc_dointvec_ms_jiffies_conv, NULL);
2987 }
2988
2989 static int proc_do_cad_pid(struct ctl_table *table, int write,
2990 void __user *buffer, size_t *lenp, loff_t *ppos)
2991 {
2992 struct pid *new_pid;
2993 pid_t tmp;
2994 int r;
2995
2996 tmp = pid_vnr(cad_pid);
2997
2998 r = __do_proc_dointvec(&tmp, table, write, buffer,
2999 lenp, ppos, NULL, NULL);
3000 if (r || !write)
3001 return r;
3002
3003 new_pid = find_get_pid(tmp);
3004 if (!new_pid)
3005 return -ESRCH;
3006
3007 put_pid(xchg(&cad_pid, new_pid));
3008 return 0;
3009 }
3010
3011 /**
3012 * proc_do_large_bitmap - read/write from/to a large bitmap
3013 * @table: the sysctl table
3014 * @write: %TRUE if this is a write to the sysctl file
3015 * @buffer: the user buffer
3016 * @lenp: the size of the user buffer
3017 * @ppos: file position
3018 *
3019 * The bitmap is stored at table->data and the bitmap length (in bits)
3020 * in table->maxlen.
3021 *
3022 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
3023 * large bitmaps may be represented in a compact manner. Writing into
3024 * the file will clear the bitmap then update it with the given input.
3025 *
3026 * Returns 0 on success.
3027 */
3028 int proc_do_large_bitmap(struct ctl_table *table, int write,
3029 void __user *buffer, size_t *lenp, loff_t *ppos)
3030 {
3031 int err = 0;
3032 bool first = 1;
3033 size_t left = *lenp;
3034 unsigned long bitmap_len = table->maxlen;
3035 unsigned long *bitmap = *(unsigned long **) table->data;
3036 unsigned long *tmp_bitmap = NULL;
3037 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
3038
3039 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
3040 *lenp = 0;
3041 return 0;
3042 }
3043
3044 if (write) {
3045 char *kbuf, *p;
3046
3047 if (left > PAGE_SIZE - 1)
3048 left = PAGE_SIZE - 1;
3049
3050 p = kbuf = memdup_user_nul(buffer, left);
3051 if (IS_ERR(kbuf))
3052 return PTR_ERR(kbuf);
3053
3054 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
3055 GFP_KERNEL);
3056 if (!tmp_bitmap) {
3057 kfree(kbuf);
3058 return -ENOMEM;
3059 }
3060 proc_skip_char(&p, &left, '\n');
3061 while (!err && left) {
3062 unsigned long val_a, val_b;
3063 bool neg;
3064
3065 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3066 sizeof(tr_a), &c);
3067 if (err)
3068 break;
3069 if (val_a >= bitmap_len || neg) {
3070 err = -EINVAL;
3071 break;
3072 }
3073
3074 val_b = val_a;
3075 if (left) {
3076 p++;
3077 left--;
3078 }
3079
3080 if (c == '-') {
3081 err = proc_get_long(&p, &left, &val_b,
3082 &neg, tr_b, sizeof(tr_b),
3083 &c);
3084 if (err)
3085 break;
3086 if (val_b >= bitmap_len || neg ||
3087 val_a > val_b) {
3088 err = -EINVAL;
3089 break;
3090 }
3091 if (left) {
3092 p++;
3093 left--;
3094 }
3095 }
3096
3097 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3098 first = 0;
3099 proc_skip_char(&p, &left, '\n');
3100 }
3101 kfree(kbuf);
3102 } else {
3103 unsigned long bit_a, bit_b = 0;
3104
3105 while (left) {
3106 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3107 if (bit_a >= bitmap_len)
3108 break;
3109 bit_b = find_next_zero_bit(bitmap, bitmap_len,
3110 bit_a + 1) - 1;
3111
3112 if (!first) {
3113 err = proc_put_char(&buffer, &left, ',');
3114 if (err)
3115 break;
3116 }
3117 err = proc_put_long(&buffer, &left, bit_a, false);
3118 if (err)
3119 break;
3120 if (bit_a != bit_b) {
3121 err = proc_put_char(&buffer, &left, '-');
3122 if (err)
3123 break;
3124 err = proc_put_long(&buffer, &left, bit_b, false);
3125 if (err)
3126 break;
3127 }
3128
3129 first = 0; bit_b++;
3130 }
3131 if (!err)
3132 err = proc_put_char(&buffer, &left, '\n');
3133 }
3134
3135 if (!err) {
3136 if (write) {
3137 if (*ppos)
3138 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3139 else
3140 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3141 }
3142 *lenp -= left;
3143 *ppos += *lenp;
3144 }
3145
3146 kfree(tmp_bitmap);
3147 return err;
3148 }
3149
3150 #else /* CONFIG_PROC_SYSCTL */
3151
3152 int proc_dostring(struct ctl_table *table, int write,
3153 void __user *buffer, size_t *lenp, loff_t *ppos)
3154 {
3155 return -ENOSYS;
3156 }
3157
3158 int proc_dointvec(struct ctl_table *table, int write,
3159 void __user *buffer, size_t *lenp, loff_t *ppos)
3160 {
3161 return -ENOSYS;
3162 }
3163
3164 int proc_douintvec(struct ctl_table *table, int write,
3165 void __user *buffer, size_t *lenp, loff_t *ppos)
3166 {
3167 return -ENOSYS;
3168 }
3169
3170 int proc_dointvec_minmax(struct ctl_table *table, int write,
3171 void __user *buffer, size_t *lenp, loff_t *ppos)
3172 {
3173 return -ENOSYS;
3174 }
3175
3176 int proc_douintvec_minmax(struct ctl_table *table, int write,
3177 void __user *buffer, size_t *lenp, loff_t *ppos)
3178 {
3179 return -ENOSYS;
3180 }
3181
3182 int proc_dopipe_max_size(struct ctl_table *table, int write,
3183 void __user *buffer, size_t *lenp, loff_t *ppos)
3184 {
3185 return -ENOSYS;
3186 }
3187
3188 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3189 void __user *buffer, size_t *lenp, loff_t *ppos)
3190 {
3191 return -ENOSYS;
3192 }
3193
3194 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3195 void __user *buffer, size_t *lenp, loff_t *ppos)
3196 {
3197 return -ENOSYS;
3198 }
3199
3200 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3201 void __user *buffer, size_t *lenp, loff_t *ppos)
3202 {
3203 return -ENOSYS;
3204 }
3205
3206 int proc_doulongvec_minmax(struct ctl_table *table, int write,
3207 void __user *buffer, size_t *lenp, loff_t *ppos)
3208 {
3209 return -ENOSYS;
3210 }
3211
3212 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3213 void __user *buffer,
3214 size_t *lenp, loff_t *ppos)
3215 {
3216 return -ENOSYS;
3217 }
3218
3219
3220 #endif /* CONFIG_PROC_SYSCTL */
3221
3222 /*
3223 * No sense putting this after each symbol definition, twice,
3224 * exception granted :-)
3225 */
3226 EXPORT_SYMBOL(proc_dointvec);
3227 EXPORT_SYMBOL(proc_douintvec);
3228 EXPORT_SYMBOL(proc_dointvec_jiffies);
3229 EXPORT_SYMBOL(proc_dointvec_minmax);
3230 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3231 EXPORT_SYMBOL_GPL(proc_dopipe_max_size);
3232 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3233 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3234 EXPORT_SYMBOL(proc_dostring);
3235 EXPORT_SYMBOL(proc_doulongvec_minmax);
3236 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
ubuntu-bionic-kernel mirror