]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blob - kernel/sysctl.c
projects / mirror_ubuntu-artful-kernel.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
UBUNTU: Start new release
[mirror_ubuntu-artful-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/kmemcheck.h>
34 #include <linux/kmemleak.h>
35 #include <linux/fs.h>
36 #include <linux/init.h>
37 #include <linux/kernel.h>
38 #include <linux/kobject.h>
39 #include <linux/net.h>
40 #include <linux/sysrq.h>
41 #include <linux/highuid.h>
42 #include <linux/writeback.h>
43 #include <linux/ratelimit.h>
44 #include <linux/compaction.h>
45 #include <linux/hugetlb.h>
46 #include <linux/initrd.h>
47 #include <linux/key.h>
48 #include <linux/times.h>
49 #include <linux/limits.h>
50 #include <linux/dcache.h>
51 #include <linux/dnotify.h>
52 #include <linux/syscalls.h>
53 #include <linux/vmstat.h>
54 #include <linux/nfs_fs.h>
55 #include <linux/acpi.h>
56 #include <linux/reboot.h>
57 #include <linux/ftrace.h>
58 #include <linux/perf_event.h>
59 #include <linux/kprobes.h>
60 #include <linux/pipe_fs_i.h>
61 #include <linux/oom.h>
62 #include <linux/kmod.h>
63 #include <linux/capability.h>
64 #include <linux/binfmts.h>
65 #include <linux/sched/sysctl.h>
66 #include <linux/kexec.h>
67 #include <linux/bpf.h>
68 #include <linux/efi.h>
69 #include <linux/mount.h>
70
71 #include <asm/uaccess.h>
72 #include <linux/mutex.h>
73 #include <asm/processor.h>
74
75 #ifdef CONFIG_X86
76 #include <asm/msr.h>
77 #include <asm/nmi.h>
78 #include <asm/stacktrace.h>
79 #include <asm/io.h>
80 #endif
81 #ifdef CONFIG_SPARC
82 #include <asm/setup.h>
83 #endif
84 #ifdef CONFIG_BSD_PROCESS_ACCT
85 #include <linux/acct.h>
86 #endif
87 #ifdef CONFIG_RT_MUTEXES
88 #include <linux/rtmutex.h>
89 #endif
90 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
91 #include <linux/lockdep.h>
92 #endif
93 #ifdef CONFIG_CHR_DEV_SG
94 #include <scsi/sg.h>
95 #endif
96
97 #ifdef CONFIG_LOCKUP_DETECTOR
98 #include <linux/nmi.h>
99 #endif
100
101 #if defined(CONFIG_SYSCTL)
102
103 /* External variables not in a header file. */
104 extern int suid_dumpable;
105 #ifdef CONFIG_COREDUMP
106 extern int core_uses_pid;
107 extern char core_pattern[];
108 extern unsigned int core_pipe_limit;
109 #endif
110 #ifdef CONFIG_USER_NS
111 extern int unprivileged_userns_clone;
112 #endif
113 extern int pid_max;
114 extern int pid_max_min, pid_max_max;
115 extern int percpu_pagelist_fraction;
116 extern int compat_log;
117 extern int latencytop_enabled;
118 extern int sysctl_nr_open_min, sysctl_nr_open_max;
119 #ifndef CONFIG_MMU
120 extern int sysctl_nr_trim_pages;
121 #endif
122
123 /* Constants used for minimum and maximum */
124 #ifdef CONFIG_LOCKUP_DETECTOR
125 static int sixty = 60;
126 #endif
127
128 static int __maybe_unused neg_one = -1;
129
130 static int zero;
131 static int __maybe_unused one = 1;
132 static int __maybe_unused two = 2;
133 static int __maybe_unused four = 4;
134 static unsigned long one_ul = 1;
135 static int one_hundred = 100;
136 #ifdef CONFIG_PRINTK
137 static int ten_thousand = 10000;
138 #endif
139
140 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
141 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
142
143 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
144 static int maxolduid = 65535;
145 static int minolduid;
146
147 static int ngroups_max = NGROUPS_MAX;
148 static const int cap_last_cap = CAP_LAST_CAP;
149
150 /*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */
151 #ifdef CONFIG_DETECT_HUNG_TASK
152 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
153 #endif
154
155 #ifdef CONFIG_INOTIFY_USER
156 #include <linux/inotify.h>
157 #endif
158 #ifdef CONFIG_SPARC
159 #endif
160
161 #ifdef __hppa__
162 extern int pwrsw_enabled;
163 #endif
164
165 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
166 extern int unaligned_enabled;
167 #endif
168
169 #ifdef CONFIG_IA64
170 extern int unaligned_dump_stack;
171 #endif
172
173 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
174 extern int no_unaligned_warning;
175 #endif
176
177 #ifdef CONFIG_PROC_SYSCTL
178
179 #define SYSCTL_WRITES_LEGACY -1
180 #define SYSCTL_WRITES_WARN 0
181 #define SYSCTL_WRITES_STRICT 1
182
183 static int sysctl_writes_strict = SYSCTL_WRITES_STRICT;
184
185 static int proc_do_cad_pid(struct ctl_table *table, int write,
186 void __user *buffer, size_t *lenp, loff_t *ppos);
187 static int proc_taint(struct ctl_table *table, int write,
188 void __user *buffer, size_t *lenp, loff_t *ppos);
189 #endif
190
191 #ifdef CONFIG_PRINTK
192 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
193 void __user *buffer, size_t *lenp, loff_t *ppos);
194 #endif
195
196 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
197 void __user *buffer, size_t *lenp, loff_t *ppos);
198 #ifdef CONFIG_COREDUMP
199 static int proc_dostring_coredump(struct ctl_table *table, int write,
200 void __user *buffer, size_t *lenp, loff_t *ppos);
201 #endif
202
203 #ifdef CONFIG_X86
204 int proc_dointvec_ibrs_ctrl(struct ctl_table *table, int write,
205 void __user *buffer, size_t *lenp, loff_t *ppos);
206 int proc_dointvec_ibpb_ctrl(struct ctl_table *table, int write,
207 void __user *buffer, size_t *lenp, loff_t *ppos);
208 int proc_dointvec_ibrs_dump(struct ctl_table *table, int write,
209 void __user *buffer, size_t *lenp, loff_t *ppos);
210 #endif
211
212 #ifdef CONFIG_MAGIC_SYSRQ
213 /* Note: sysrq code uses it's own private copy */
214 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
215
216 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
217 void __user *buffer, size_t *lenp,
218 loff_t *ppos)
219 {
220 int error;
221
222 error = proc_dointvec(table, write, buffer, lenp, ppos);
223 if (error)
224 return error;
225
226 if (write)
227 sysrq_toggle_support(__sysrq_enabled);
228
229 return 0;
230 }
231
232 #endif
233
234 static struct ctl_table kern_table[];
235 static struct ctl_table vm_table[];
236 static struct ctl_table fs_table[];
237 static struct ctl_table debug_table[];
238 static struct ctl_table dev_table[];
239 extern struct ctl_table random_table[];
240 #ifdef CONFIG_EPOLL
241 extern struct ctl_table epoll_table[];
242 #endif
243
244 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
245 int sysctl_legacy_va_layout;
246 #endif
247
248 u32 sysctl_ibrs_dump = 0;
249 u32 sysctl_ibrs_enabled = 0;
250 EXPORT_SYMBOL(sysctl_ibrs_enabled);
251 u32 sysctl_ibpb_enabled = 0;
252 EXPORT_SYMBOL(sysctl_ibpb_enabled);
253
254 /* The default sysctl tables: */
255
256 static struct ctl_table sysctl_base_table[] = {
257 {
258 .procname = "kernel",
259 .mode = 0555,
260 .child = kern_table,
261 },
262 {
263 .procname = "vm",
264 .mode = 0555,
265 .child = vm_table,
266 },
267 {
268 .procname = "fs",
269 .mode = 0555,
270 .child = fs_table,
271 },
272 {
273 .procname = "debug",
274 .mode = 0555,
275 .child = debug_table,
276 },
277 {
278 .procname = "dev",
279 .mode = 0555,
280 .child = dev_table,
281 },
282 { }
283 };
284
285 #ifdef CONFIG_SCHED_DEBUG
286 static int min_sched_granularity_ns = 100000; /* 100 usecs */
287 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
288 static int min_wakeup_granularity_ns; /* 0 usecs */
289 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
290 #ifdef CONFIG_SMP
291 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
292 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
293 #endif /* CONFIG_SMP */
294 #endif /* CONFIG_SCHED_DEBUG */
295
296 #ifdef CONFIG_COMPACTION
297 static int min_extfrag_threshold;
298 static int max_extfrag_threshold = 1000;
299 #endif
300
301 static unsigned int secure_boot_enabled;
302 int secure_boot_proc_handler(struct ctl_table *table, int write,
303 void __user *buffer, size_t *lenp, loff_t *ppos)
304 {
305 secure_boot_enabled = efi_enabled(EFI_SECURE_BOOT);
306 return proc_dointvec(table, write, buffer, lenp, ppos);
307 }
308
309 static unsigned int moksbstate_disabled;
310 int moksbstate_disabled_proc_handler(struct ctl_table *table, int write,
311 void __user *buffer, size_t *lenp, loff_t *ppos)
312 {
313 moksbstate_disabled = efi_enabled(EFI_MOKSBSTATE_DISABLED);
314 return proc_dointvec(table, write, buffer, lenp, ppos);
315 }
316
317 static struct ctl_table kern_table[] = {
318 {
319 .procname = "secure_boot",
320 .data = &secure_boot_enabled,
321 .maxlen = sizeof(unsigned int),
322 .mode = 0444,
323 .proc_handler = secure_boot_proc_handler,
324 },
325 {
326 .procname = "moksbstate_disabled",
327 .data = &moksbstate_disabled,
328 .maxlen = sizeof(unsigned int),
329 .mode = 0444,
330 .proc_handler = moksbstate_disabled_proc_handler,
331 },
332 {
333 .procname = "sched_child_runs_first",
334 .data = &sysctl_sched_child_runs_first,
335 .maxlen = sizeof(unsigned int),
336 .mode = 0644,
337 .proc_handler = proc_dointvec,
338 },
339 #ifdef CONFIG_SCHED_DEBUG
340 {
341 .procname = "sched_min_granularity_ns",
342 .data = &sysctl_sched_min_granularity,
343 .maxlen = sizeof(unsigned int),
344 .mode = 0644,
345 .proc_handler = sched_proc_update_handler,
346 .extra1 = &min_sched_granularity_ns,
347 .extra2 = &max_sched_granularity_ns,
348 },
349 {
350 .procname = "sched_latency_ns",
351 .data = &sysctl_sched_latency,
352 .maxlen = sizeof(unsigned int),
353 .mode = 0644,
354 .proc_handler = sched_proc_update_handler,
355 .extra1 = &min_sched_granularity_ns,
356 .extra2 = &max_sched_granularity_ns,
357 },
358 {
359 .procname = "sched_wakeup_granularity_ns",
360 .data = &sysctl_sched_wakeup_granularity,
361 .maxlen = sizeof(unsigned int),
362 .mode = 0644,
363 .proc_handler = sched_proc_update_handler,
364 .extra1 = &min_wakeup_granularity_ns,
365 .extra2 = &max_wakeup_granularity_ns,
366 },
367 #ifdef CONFIG_SMP
368 {
369 .procname = "sched_tunable_scaling",
370 .data = &sysctl_sched_tunable_scaling,
371 .maxlen = sizeof(enum sched_tunable_scaling),
372 .mode = 0644,
373 .proc_handler = sched_proc_update_handler,
374 .extra1 = &min_sched_tunable_scaling,
375 .extra2 = &max_sched_tunable_scaling,
376 },
377 {
378 .procname = "sched_migration_cost_ns",
379 .data = &sysctl_sched_migration_cost,
380 .maxlen = sizeof(unsigned int),
381 .mode = 0644,
382 .proc_handler = proc_dointvec,
383 },
384 {
385 .procname = "sched_nr_migrate",
386 .data = &sysctl_sched_nr_migrate,
387 .maxlen = sizeof(unsigned int),
388 .mode = 0644,
389 .proc_handler = proc_dointvec,
390 },
391 {
392 .procname = "sched_time_avg_ms",
393 .data = &sysctl_sched_time_avg,
394 .maxlen = sizeof(unsigned int),
395 .mode = 0644,
396 .proc_handler = proc_dointvec,
397 },
398 {
399 .procname = "sched_shares_window_ns",
400 .data = &sysctl_sched_shares_window,
401 .maxlen = sizeof(unsigned int),
402 .mode = 0644,
403 .proc_handler = proc_dointvec,
404 },
405 #endif /* CONFIG_SMP */
406 #ifdef CONFIG_NUMA_BALANCING
407 {
408 .procname = "numa_balancing_scan_delay_ms",
409 .data = &sysctl_numa_balancing_scan_delay,
410 .maxlen = sizeof(unsigned int),
411 .mode = 0644,
412 .proc_handler = proc_dointvec,
413 },
414 {
415 .procname = "numa_balancing_scan_period_min_ms",
416 .data = &sysctl_numa_balancing_scan_period_min,
417 .maxlen = sizeof(unsigned int),
418 .mode = 0644,
419 .proc_handler = proc_dointvec,
420 },
421 {
422 .procname = "numa_balancing_scan_period_max_ms",
423 .data = &sysctl_numa_balancing_scan_period_max,
424 .maxlen = sizeof(unsigned int),
425 .mode = 0644,
426 .proc_handler = proc_dointvec,
427 },
428 {
429 .procname = "numa_balancing_scan_size_mb",
430 .data = &sysctl_numa_balancing_scan_size,
431 .maxlen = sizeof(unsigned int),
432 .mode = 0644,
433 .proc_handler = proc_dointvec_minmax,
434 .extra1 = &one,
435 },
436 {
437 .procname = "numa_balancing",
438 .data = NULL, /* filled in by handler */
439 .maxlen = sizeof(unsigned int),
440 .mode = 0644,
441 .proc_handler = sysctl_numa_balancing,
442 .extra1 = &zero,
443 .extra2 = &one,
444 },
445 #endif /* CONFIG_NUMA_BALANCING */
446 #endif /* CONFIG_SCHED_DEBUG */
447 {
448 .procname = "sched_rt_period_us",
449 .data = &sysctl_sched_rt_period,
450 .maxlen = sizeof(unsigned int),
451 .mode = 0644,
452 .proc_handler = sched_rt_handler,
453 },
454 {
455 .procname = "sched_rt_runtime_us",
456 .data = &sysctl_sched_rt_runtime,
457 .maxlen = sizeof(int),
458 .mode = 0644,
459 .proc_handler = sched_rt_handler,
460 },
461 {
462 .procname = "sched_rr_timeslice_ms",
463 .data = &sched_rr_timeslice,
464 .maxlen = sizeof(int),
465 .mode = 0644,
466 .proc_handler = sched_rr_handler,
467 },
468 #ifdef CONFIG_SCHED_AUTOGROUP
469 {
470 .procname = "sched_autogroup_enabled",
471 .data = &sysctl_sched_autogroup_enabled,
472 .maxlen = sizeof(unsigned int),
473 .mode = 0644,
474 .proc_handler = proc_dointvec_minmax,
475 .extra1 = &zero,
476 .extra2 = &one,
477 },
478 #endif
479 #ifdef CONFIG_CFS_BANDWIDTH
480 {
481 .procname = "sched_cfs_bandwidth_slice_us",
482 .data = &sysctl_sched_cfs_bandwidth_slice,
483 .maxlen = sizeof(unsigned int),
484 .mode = 0644,
485 .proc_handler = proc_dointvec_minmax,
486 .extra1 = &one,
487 },
488 #endif
489 #ifdef CONFIG_PROVE_LOCKING
490 {
491 .procname = "prove_locking",
492 .data = &prove_locking,
493 .maxlen = sizeof(int),
494 .mode = 0644,
495 .proc_handler = proc_dointvec,
496 },
497 #endif
498 #ifdef CONFIG_LOCK_STAT
499 {
500 .procname = "lock_stat",
501 .data = &lock_stat,
502 .maxlen = sizeof(int),
503 .mode = 0644,
504 .proc_handler = proc_dointvec,
505 },
506 #endif
507 {
508 .procname = "panic",
509 .data = &panic_timeout,
510 .maxlen = sizeof(int),
511 .mode = 0644,
512 .proc_handler = proc_dointvec,
513 },
514 #ifdef CONFIG_COREDUMP
515 {
516 .procname = "core_uses_pid",
517 .data = &core_uses_pid,
518 .maxlen = sizeof(int),
519 .mode = 0644,
520 .proc_handler = proc_dointvec,
521 },
522 {
523 .procname = "core_pattern",
524 .data = core_pattern,
525 .maxlen = CORENAME_MAX_SIZE,
526 .mode = 0644,
527 .proc_handler = proc_dostring_coredump,
528 },
529 {
530 .procname = "core_pipe_limit",
531 .data = &core_pipe_limit,
532 .maxlen = sizeof(unsigned int),
533 .mode = 0644,
534 .proc_handler = proc_dointvec,
535 },
536 #endif
537 #ifdef CONFIG_USER_NS
538 {
539 .procname = "unprivileged_userns_clone",
540 .data = &unprivileged_userns_clone,
541 .maxlen = sizeof(int),
542 .mode = 0644,
543 .proc_handler = proc_dointvec,
544 },
545 #endif
546 #ifdef CONFIG_PROC_SYSCTL
547 {
548 .procname = "tainted",
549 .maxlen = sizeof(long),
550 .mode = 0644,
551 .proc_handler = proc_taint,
552 },
553 {
554 .procname = "sysctl_writes_strict",
555 .data = &sysctl_writes_strict,
556 .maxlen = sizeof(int),
557 .mode = 0644,
558 .proc_handler = proc_dointvec_minmax,
559 .extra1 = &neg_one,
560 .extra2 = &one,
561 },
562 #endif
563 #ifdef CONFIG_LATENCYTOP
564 {
565 .procname = "latencytop",
566 .data = &latencytop_enabled,
567 .maxlen = sizeof(int),
568 .mode = 0644,
569 .proc_handler = proc_dointvec,
570 },
571 #endif
572 #ifdef CONFIG_BLK_DEV_INITRD
573 {
574 .procname = "real-root-dev",
575 .data = &real_root_dev,
576 .maxlen = sizeof(int),
577 .mode = 0644,
578 .proc_handler = proc_dointvec,
579 },
580 #endif
581 {
582 .procname = "print-fatal-signals",
583 .data = &print_fatal_signals,
584 .maxlen = sizeof(int),
585 .mode = 0644,
586 .proc_handler = proc_dointvec,
587 },
588 #ifdef CONFIG_SPARC
589 {
590 .procname = "reboot-cmd",
591 .data = reboot_command,
592 .maxlen = 256,
593 .mode = 0644,
594 .proc_handler = proc_dostring,
595 },
596 {
597 .procname = "stop-a",
598 .data = &stop_a_enabled,
599 .maxlen = sizeof (int),
600 .mode = 0644,
601 .proc_handler = proc_dointvec,
602 },
603 {
604 .procname = "scons-poweroff",
605 .data = &scons_pwroff,
606 .maxlen = sizeof (int),
607 .mode = 0644,
608 .proc_handler = proc_dointvec,
609 },
610 #endif
611 #ifdef CONFIG_SPARC64
612 {
613 .procname = "tsb-ratio",
614 .data = &sysctl_tsb_ratio,
615 .maxlen = sizeof (int),
616 .mode = 0644,
617 .proc_handler = proc_dointvec,
618 },
619 #endif
620 #ifdef __hppa__
621 {
622 .procname = "soft-power",
623 .data = &pwrsw_enabled,
624 .maxlen = sizeof (int),
625 .mode = 0644,
626 .proc_handler = proc_dointvec,
627 },
628 #endif
629 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
630 {
631 .procname = "unaligned-trap",
632 .data = &unaligned_enabled,
633 .maxlen = sizeof (int),
634 .mode = 0644,
635 .proc_handler = proc_dointvec,
636 },
637 #endif
638 {
639 .procname = "ctrl-alt-del",
640 .data = &C_A_D,
641 .maxlen = sizeof(int),
642 .mode = 0644,
643 .proc_handler = proc_dointvec,
644 },
645 #ifdef CONFIG_FUNCTION_TRACER
646 {
647 .procname = "ftrace_enabled",
648 .data = &ftrace_enabled,
649 .maxlen = sizeof(int),
650 .mode = 0644,
651 .proc_handler = ftrace_enable_sysctl,
652 },
653 #endif
654 #ifdef CONFIG_STACK_TRACER
655 {
656 .procname = "stack_tracer_enabled",
657 .data = &stack_tracer_enabled,
658 .maxlen = sizeof(int),
659 .mode = 0644,
660 .proc_handler = stack_trace_sysctl,
661 },
662 #endif
663 #ifdef CONFIG_TRACING
664 {
665 .procname = "ftrace_dump_on_oops",
666 .data = &ftrace_dump_on_oops,
667 .maxlen = sizeof(int),
668 .mode = 0644,
669 .proc_handler = proc_dointvec,
670 },
671 {
672 .procname = "traceoff_on_warning",
673 .data = &__disable_trace_on_warning,
674 .maxlen = sizeof(__disable_trace_on_warning),
675 .mode = 0644,
676 .proc_handler = proc_dointvec,
677 },
678 {
679 .procname = "tracepoint_printk",
680 .data = &tracepoint_printk,
681 .maxlen = sizeof(tracepoint_printk),
682 .mode = 0644,
683 .proc_handler = proc_dointvec,
684 },
685 #endif
686 #ifdef CONFIG_KEXEC_CORE
687 {
688 .procname = "kexec_load_disabled",
689 .data = &kexec_load_disabled,
690 .maxlen = sizeof(int),
691 .mode = 0644,
692 /* only handle a transition from default "0" to "1" */
693 .proc_handler = proc_dointvec_minmax,
694 .extra1 = &one,
695 .extra2 = &one,
696 },
697 #endif
698 #ifdef CONFIG_MODULES
699 {
700 .procname = "modprobe",
701 .data = &modprobe_path,
702 .maxlen = KMOD_PATH_LEN,
703 .mode = 0644,
704 .proc_handler = proc_dostring,
705 },
706 {
707 .procname = "modules_disabled",
708 .data = &modules_disabled,
709 .maxlen = sizeof(int),
710 .mode = 0644,
711 /* only handle a transition from default "0" to "1" */
712 .proc_handler = proc_dointvec_minmax,
713 .extra1 = &one,
714 .extra2 = &one,
715 },
716 #endif
717 #ifdef CONFIG_UEVENT_HELPER
718 {
719 .procname = "hotplug",
720 .data = &uevent_helper,
721 .maxlen = UEVENT_HELPER_PATH_LEN,
722 .mode = 0644,
723 .proc_handler = proc_dostring,
724 },
725 #endif
726 #ifdef CONFIG_CHR_DEV_SG
727 {
728 .procname = "sg-big-buff",
729 .data = &sg_big_buff,
730 .maxlen = sizeof (int),
731 .mode = 0444,
732 .proc_handler = proc_dointvec,
733 },
734 #endif
735 #ifdef CONFIG_BSD_PROCESS_ACCT
736 {
737 .procname = "acct",
738 .data = &acct_parm,
739 .maxlen = 3*sizeof(int),
740 .mode = 0644,
741 .proc_handler = proc_dointvec,
742 },
743 #endif
744 #ifdef CONFIG_MAGIC_SYSRQ
745 {
746 .procname = "sysrq",
747 .data = &__sysrq_enabled,
748 .maxlen = sizeof (int),
749 .mode = 0644,
750 .proc_handler = sysrq_sysctl_handler,
751 },
752 #endif
753 #ifdef CONFIG_PROC_SYSCTL
754 {
755 .procname = "cad_pid",
756 .data = NULL,
757 .maxlen = sizeof (int),
758 .mode = 0600,
759 .proc_handler = proc_do_cad_pid,
760 },
761 #endif
762 {
763 .procname = "threads-max",
764 .data = NULL,
765 .maxlen = sizeof(int),
766 .mode = 0644,
767 .proc_handler = sysctl_max_threads,
768 },
769 {
770 .procname = "random",
771 .mode = 0555,
772 .child = random_table,
773 },
774 {
775 .procname = "usermodehelper",
776 .mode = 0555,
777 .child = usermodehelper_table,
778 },
779 {
780 .procname = "overflowuid",
781 .data = &overflowuid,
782 .maxlen = sizeof(int),
783 .mode = 0644,
784 .proc_handler = proc_dointvec_minmax,
785 .extra1 = &minolduid,
786 .extra2 = &maxolduid,
787 },
788 {
789 .procname = "overflowgid",
790 .data = &overflowgid,
791 .maxlen = sizeof(int),
792 .mode = 0644,
793 .proc_handler = proc_dointvec_minmax,
794 .extra1 = &minolduid,
795 .extra2 = &maxolduid,
796 },
797 #ifdef CONFIG_S390
798 #ifdef CONFIG_MATHEMU
799 {
800 .procname = "ieee_emulation_warnings",
801 .data = &sysctl_ieee_emulation_warnings,
802 .maxlen = sizeof(int),
803 .mode = 0644,
804 .proc_handler = proc_dointvec,
805 },
806 #endif
807 {
808 .procname = "userprocess_debug",
809 .data = &show_unhandled_signals,
810 .maxlen = sizeof(int),
811 .mode = 0644,
812 .proc_handler = proc_dointvec,
813 },
814 #endif
815 {
816 .procname = "pid_max",
817 .data = &pid_max,
818 .maxlen = sizeof (int),
819 .mode = 0644,
820 .proc_handler = proc_dointvec_minmax,
821 .extra1 = &pid_max_min,
822 .extra2 = &pid_max_max,
823 },
824 {
825 .procname = "panic_on_oops",
826 .data = &panic_on_oops,
827 .maxlen = sizeof(int),
828 .mode = 0644,
829 .proc_handler = proc_dointvec,
830 },
831 #if defined CONFIG_PRINTK
832 {
833 .procname = "printk",
834 .data = &console_loglevel,
835 .maxlen = 4*sizeof(int),
836 .mode = 0644,
837 .proc_handler = proc_dointvec,
838 },
839 {
840 .procname = "printk_ratelimit",
841 .data = &printk_ratelimit_state.interval,
842 .maxlen = sizeof(int),
843 .mode = 0644,
844 .proc_handler = proc_dointvec_jiffies,
845 },
846 {
847 .procname = "printk_ratelimit_burst",
848 .data = &printk_ratelimit_state.burst,
849 .maxlen = sizeof(int),
850 .mode = 0644,
851 .proc_handler = proc_dointvec,
852 },
853 {
854 .procname = "printk_delay",
855 .data = &printk_delay_msec,
856 .maxlen = sizeof(int),
857 .mode = 0644,
858 .proc_handler = proc_dointvec_minmax,
859 .extra1 = &zero,
860 .extra2 = &ten_thousand,
861 },
862 {
863 .procname = "dmesg_restrict",
864 .data = &dmesg_restrict,
865 .maxlen = sizeof(int),
866 .mode = 0644,
867 .proc_handler = proc_dointvec_minmax_sysadmin,
868 .extra1 = &zero,
869 .extra2 = &one,
870 },
871 {
872 .procname = "kptr_restrict",
873 .data = &kptr_restrict,
874 .maxlen = sizeof(int),
875 .mode = 0644,
876 .proc_handler = proc_dointvec_minmax_sysadmin,
877 .extra1 = &zero,
878 .extra2 = &two,
879 },
880 #endif
881 {
882 .procname = "ngroups_max",
883 .data = &ngroups_max,
884 .maxlen = sizeof (int),
885 .mode = 0444,
886 .proc_handler = proc_dointvec,
887 },
888 {
889 .procname = "cap_last_cap",
890 .data = (void *)&cap_last_cap,
891 .maxlen = sizeof(int),
892 .mode = 0444,
893 .proc_handler = proc_dointvec,
894 },
895 #if defined(CONFIG_LOCKUP_DETECTOR)
896 {
897 .procname = "watchdog",
898 .data = &watchdog_user_enabled,
899 .maxlen = sizeof (int),
900 .mode = 0644,
901 .proc_handler = proc_watchdog,
902 .extra1 = &zero,
903 .extra2 = &one,
904 },
905 {
906 .procname = "watchdog_thresh",
907 .data = &watchdog_thresh,
908 .maxlen = sizeof(int),
909 .mode = 0644,
910 .proc_handler = proc_watchdog_thresh,
911 .extra1 = &zero,
912 .extra2 = &sixty,
913 },
914 {
915 .procname = "nmi_watchdog",
916 .data = &nmi_watchdog_enabled,
917 .maxlen = sizeof (int),
918 .mode = 0644,
919 .proc_handler = proc_nmi_watchdog,
920 .extra1 = &zero,
921 #if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR)
922 .extra2 = &one,
923 #else
924 .extra2 = &zero,
925 #endif
926 },
927 {
928 .procname = "soft_watchdog",
929 .data = &soft_watchdog_enabled,
930 .maxlen = sizeof (int),
931 .mode = 0644,
932 .proc_handler = proc_soft_watchdog,
933 .extra1 = &zero,
934 .extra2 = &one,
935 },
936 {
937 .procname = "watchdog_cpumask",
938 .data = &watchdog_cpumask_bits,
939 .maxlen = NR_CPUS,
940 .mode = 0644,
941 .proc_handler = proc_watchdog_cpumask,
942 },
943 {
944 .procname = "softlockup_panic",
945 .data = &softlockup_panic,
946 .maxlen = sizeof(int),
947 .mode = 0644,
948 .proc_handler = proc_dointvec_minmax,
949 .extra1 = &zero,
950 .extra2 = &one,
951 },
952 #ifdef CONFIG_HARDLOCKUP_DETECTOR
953 {
954 .procname = "hardlockup_panic",
955 .data = &hardlockup_panic,
956 .maxlen = sizeof(int),
957 .mode = 0644,
958 .proc_handler = proc_dointvec_minmax,
959 .extra1 = &zero,
960 .extra2 = &one,
961 },
962 #endif
963 #ifdef CONFIG_SMP
964 {
965 .procname = "softlockup_all_cpu_backtrace",
966 .data = &sysctl_softlockup_all_cpu_backtrace,
967 .maxlen = sizeof(int),
968 .mode = 0644,
969 .proc_handler = proc_dointvec_minmax,
970 .extra1 = &zero,
971 .extra2 = &one,
972 },
973 {
974 .procname = "hardlockup_all_cpu_backtrace",
975 .data = &sysctl_hardlockup_all_cpu_backtrace,
976 .maxlen = sizeof(int),
977 .mode = 0644,
978 .proc_handler = proc_dointvec_minmax,
979 .extra1 = &zero,
980 .extra2 = &one,
981 },
982 #endif /* CONFIG_SMP */
983 #endif
984 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
985 {
986 .procname = "unknown_nmi_panic",
987 .data = &unknown_nmi_panic,
988 .maxlen = sizeof (int),
989 .mode = 0644,
990 .proc_handler = proc_dointvec,
991 },
992 #endif
993 #if defined(CONFIG_X86)
994 {
995 .procname = "panic_on_unrecovered_nmi",
996 .data = &panic_on_unrecovered_nmi,
997 .maxlen = sizeof(int),
998 .mode = 0644,
999 .proc_handler = proc_dointvec,
1000 },
1001 {
1002 .procname = "panic_on_io_nmi",
1003 .data = &panic_on_io_nmi,
1004 .maxlen = sizeof(int),
1005 .mode = 0644,
1006 .proc_handler = proc_dointvec,
1007 },
1008 #ifdef CONFIG_DEBUG_STACKOVERFLOW
1009 {
1010 .procname = "panic_on_stackoverflow",
1011 .data = &sysctl_panic_on_stackoverflow,
1012 .maxlen = sizeof(int),
1013 .mode = 0644,
1014 .proc_handler = proc_dointvec,
1015 },
1016 #endif
1017 {
1018 .procname = "bootloader_type",
1019 .data = &bootloader_type,
1020 .maxlen = sizeof (int),
1021 .mode = 0444,
1022 .proc_handler = proc_dointvec,
1023 },
1024 {
1025 .procname = "bootloader_version",
1026 .data = &bootloader_version,
1027 .maxlen = sizeof (int),
1028 .mode = 0444,
1029 .proc_handler = proc_dointvec,
1030 },
1031 {
1032 .procname = "kstack_depth_to_print",
1033 .data = &kstack_depth_to_print,
1034 .maxlen = sizeof(int),
1035 .mode = 0644,
1036 .proc_handler = proc_dointvec,
1037 },
1038 {
1039 .procname = "io_delay_type",
1040 .data = &io_delay_type,
1041 .maxlen = sizeof(int),
1042 .mode = 0644,
1043 .proc_handler = proc_dointvec,
1044 },
1045 #endif
1046 #if defined(CONFIG_MMU)
1047 {
1048 .procname = "randomize_va_space",
1049 .data = &randomize_va_space,
1050 .maxlen = sizeof(int),
1051 .mode = 0644,
1052 .proc_handler = proc_dointvec,
1053 },
1054 #endif
1055 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1056 {
1057 .procname = "spin_retry",
1058 .data = &spin_retry,
1059 .maxlen = sizeof (int),
1060 .mode = 0644,
1061 .proc_handler = proc_dointvec,
1062 },
1063 #endif
1064 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1065 {
1066 .procname = "acpi_video_flags",
1067 .data = &acpi_realmode_flags,
1068 .maxlen = sizeof (unsigned long),
1069 .mode = 0644,
1070 .proc_handler = proc_doulongvec_minmax,
1071 },
1072 #endif
1073 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1074 {
1075 .procname = "ignore-unaligned-usertrap",
1076 .data = &no_unaligned_warning,
1077 .maxlen = sizeof (int),
1078 .mode = 0644,
1079 .proc_handler = proc_dointvec,
1080 },
1081 #endif
1082 #ifdef CONFIG_IA64
1083 {
1084 .procname = "unaligned-dump-stack",
1085 .data = &unaligned_dump_stack,
1086 .maxlen = sizeof (int),
1087 .mode = 0644,
1088 .proc_handler = proc_dointvec,
1089 },
1090 #endif
1091 #ifdef CONFIG_DETECT_HUNG_TASK
1092 {
1093 .procname = "hung_task_panic",
1094 .data = &sysctl_hung_task_panic,
1095 .maxlen = sizeof(int),
1096 .mode = 0644,
1097 .proc_handler = proc_dointvec_minmax,
1098 .extra1 = &zero,
1099 .extra2 = &one,
1100 },
1101 {
1102 .procname = "hung_task_check_count",
1103 .data = &sysctl_hung_task_check_count,
1104 .maxlen = sizeof(int),
1105 .mode = 0644,
1106 .proc_handler = proc_dointvec_minmax,
1107 .extra1 = &zero,
1108 },
1109 {
1110 .procname = "hung_task_timeout_secs",
1111 .data = &sysctl_hung_task_timeout_secs,
1112 .maxlen = sizeof(unsigned long),
1113 .mode = 0644,
1114 .proc_handler = proc_dohung_task_timeout_secs,
1115 .extra2 = &hung_task_timeout_max,
1116 },
1117 {
1118 .procname = "hung_task_warnings",
1119 .data = &sysctl_hung_task_warnings,
1120 .maxlen = sizeof(int),
1121 .mode = 0644,
1122 .proc_handler = proc_dointvec_minmax,
1123 .extra1 = &neg_one,
1124 },
1125 #endif
1126 #ifdef CONFIG_COMPAT
1127 {
1128 .procname = "compat-log",
1129 .data = &compat_log,
1130 .maxlen = sizeof (int),
1131 .mode = 0644,
1132 .proc_handler = proc_dointvec,
1133 },
1134 #endif
1135 #ifdef CONFIG_RT_MUTEXES
1136 {
1137 .procname = "max_lock_depth",
1138 .data = &max_lock_depth,
1139 .maxlen = sizeof(int),
1140 .mode = 0644,
1141 .proc_handler = proc_dointvec,
1142 },
1143 #endif
1144 {
1145 .procname = "poweroff_cmd",
1146 .data = &poweroff_cmd,
1147 .maxlen = POWEROFF_CMD_PATH_LEN,
1148 .mode = 0644,
1149 .proc_handler = proc_dostring,
1150 },
1151 #ifdef CONFIG_KEYS
1152 {
1153 .procname = "keys",
1154 .mode = 0555,
1155 .child = key_sysctls,
1156 },
1157 #endif
1158 #ifdef CONFIG_PERF_EVENTS
1159 /*
1160 * User-space scripts rely on the existence of this file
1161 * as a feature check for perf_events being enabled.
1162 *
1163 * So it's an ABI, do not remove!
1164 */
1165 {
1166 .procname = "perf_event_paranoid",
1167 .data = &sysctl_perf_event_paranoid,
1168 .maxlen = sizeof(sysctl_perf_event_paranoid),
1169 .mode = 0644,
1170 .proc_handler = proc_dointvec,
1171 },
1172 {
1173 .procname = "perf_event_mlock_kb",
1174 .data = &sysctl_perf_event_mlock,
1175 .maxlen = sizeof(sysctl_perf_event_mlock),
1176 .mode = 0644,
1177 .proc_handler = proc_dointvec,
1178 },
1179 {
1180 .procname = "perf_event_max_sample_rate",
1181 .data = &sysctl_perf_event_sample_rate,
1182 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1183 .mode = 0644,
1184 .proc_handler = perf_proc_update_handler,
1185 .extra1 = &one,
1186 },
1187 {
1188 .procname = "perf_cpu_time_max_percent",
1189 .data = &sysctl_perf_cpu_time_max_percent,
1190 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1191 .mode = 0644,
1192 .proc_handler = perf_cpu_time_max_percent_handler,
1193 .extra1 = &zero,
1194 .extra2 = &one_hundred,
1195 },
1196 #endif
1197 #ifdef CONFIG_KMEMCHECK
1198 {
1199 .procname = "kmemcheck",
1200 .data = &kmemcheck_enabled,
1201 .maxlen = sizeof(int),
1202 .mode = 0644,
1203 .proc_handler = proc_dointvec,
1204 },
1205 #endif
1206 {
1207 .procname = "panic_on_warn",
1208 .data = &panic_on_warn,
1209 .maxlen = sizeof(int),
1210 .mode = 0644,
1211 .proc_handler = proc_dointvec_minmax,
1212 .extra1 = &zero,
1213 .extra2 = &one,
1214 },
1215 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1216 {
1217 .procname = "timer_migration",
1218 .data = &sysctl_timer_migration,
1219 .maxlen = sizeof(unsigned int),
1220 .mode = 0644,
1221 .proc_handler = timer_migration_handler,
1222 .extra1 = &zero,
1223 .extra2 = &one,
1224 },
1225 #endif
1226 #ifdef CONFIG_BPF_SYSCALL
1227 {
1228 .procname = "unprivileged_bpf_disabled",
1229 .data = &sysctl_unprivileged_bpf_disabled,
1230 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
1231 .mode = 0644,
1232 /* only handle a transition from default "0" to "1" */
1233 .proc_handler = proc_dointvec_minmax,
1234 .extra1 = &one,
1235 .extra2 = &one,
1236 },
1237 #endif
1238 #ifdef CONFIG_X86
1239 {
1240 .procname = "ibrs_enabled",
1241 .data = &sysctl_ibrs_enabled,
1242 .maxlen = sizeof(unsigned int),
1243 .mode = 0644,
1244 .proc_handler = proc_dointvec_ibrs_ctrl,
1245 .extra1 = &zero,
1246 .extra2 = &two,
1247 },
1248 {
1249 .procname = "ibpb_enabled",
1250 .data = &sysctl_ibpb_enabled,
1251 .maxlen = sizeof(unsigned int),
1252 .mode = 0644,
1253 .proc_handler = proc_dointvec_ibpb_ctrl,
1254 .extra1 = &zero,
1255 .extra2 = &one,
1256 },
1257 {
1258 .procname = "ibrs_dump",
1259 .data = &sysctl_ibrs_dump,
1260 .maxlen = sizeof(unsigned int),
1261 .mode = 0644,
1262 .proc_handler = proc_dointvec_ibrs_dump,
1263 .extra1 = &zero,
1264 .extra2 = &one,
1265 },
1266 #endif
1267 { }
1268 };
1269
1270 static struct ctl_table vm_table[] = {
1271 {
1272 .procname = "overcommit_memory",
1273 .data = &sysctl_overcommit_memory,
1274 .maxlen = sizeof(sysctl_overcommit_memory),
1275 .mode = 0644,
1276 .proc_handler = proc_dointvec_minmax,
1277 .extra1 = &zero,
1278 .extra2 = &two,
1279 },
1280 {
1281 .procname = "panic_on_oom",
1282 .data = &sysctl_panic_on_oom,
1283 .maxlen = sizeof(sysctl_panic_on_oom),
1284 .mode = 0644,
1285 .proc_handler = proc_dointvec_minmax,
1286 .extra1 = &zero,
1287 .extra2 = &two,
1288 },
1289 {
1290 .procname = "oom_kill_allocating_task",
1291 .data = &sysctl_oom_kill_allocating_task,
1292 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1293 .mode = 0644,
1294 .proc_handler = proc_dointvec,
1295 },
1296 {
1297 .procname = "oom_dump_tasks",
1298 .data = &sysctl_oom_dump_tasks,
1299 .maxlen = sizeof(sysctl_oom_dump_tasks),
1300 .mode = 0644,
1301 .proc_handler = proc_dointvec,
1302 },
1303 {
1304 .procname = "overcommit_ratio",
1305 .data = &sysctl_overcommit_ratio,
1306 .maxlen = sizeof(sysctl_overcommit_ratio),
1307 .mode = 0644,
1308 .proc_handler = overcommit_ratio_handler,
1309 },
1310 {
1311 .procname = "overcommit_kbytes",
1312 .data = &sysctl_overcommit_kbytes,
1313 .maxlen = sizeof(sysctl_overcommit_kbytes),
1314 .mode = 0644,
1315 .proc_handler = overcommit_kbytes_handler,
1316 },
1317 {
1318 .procname = "page-cluster",
1319 .data = &page_cluster,
1320 .maxlen = sizeof(int),
1321 .mode = 0644,
1322 .proc_handler = proc_dointvec_minmax,
1323 .extra1 = &zero,
1324 },
1325 {
1326 .procname = "dirty_background_ratio",
1327 .data = &dirty_background_ratio,
1328 .maxlen = sizeof(dirty_background_ratio),
1329 .mode = 0644,
1330 .proc_handler = dirty_background_ratio_handler,
1331 .extra1 = &zero,
1332 .extra2 = &one_hundred,
1333 },
1334 {
1335 .procname = "dirty_background_bytes",
1336 .data = &dirty_background_bytes,
1337 .maxlen = sizeof(dirty_background_bytes),
1338 .mode = 0644,
1339 .proc_handler = dirty_background_bytes_handler,
1340 .extra1 = &one_ul,
1341 },
1342 {
1343 .procname = "dirty_ratio",
1344 .data = &vm_dirty_ratio,
1345 .maxlen = sizeof(vm_dirty_ratio),
1346 .mode = 0644,
1347 .proc_handler = dirty_ratio_handler,
1348 .extra1 = &zero,
1349 .extra2 = &one_hundred,
1350 },
1351 {
1352 .procname = "dirty_bytes",
1353 .data = &vm_dirty_bytes,
1354 .maxlen = sizeof(vm_dirty_bytes),
1355 .mode = 0644,
1356 .proc_handler = dirty_bytes_handler,
1357 .extra1 = &dirty_bytes_min,
1358 },
1359 {
1360 .procname = "dirty_writeback_centisecs",
1361 .data = &dirty_writeback_interval,
1362 .maxlen = sizeof(dirty_writeback_interval),
1363 .mode = 0644,
1364 .proc_handler = dirty_writeback_centisecs_handler,
1365 },
1366 {
1367 .procname = "dirty_expire_centisecs",
1368 .data = &dirty_expire_interval,
1369 .maxlen = sizeof(dirty_expire_interval),
1370 .mode = 0644,
1371 .proc_handler = proc_dointvec_minmax,
1372 .extra1 = &zero,
1373 },
1374 {
1375 .procname = "dirtytime_expire_seconds",
1376 .data = &dirtytime_expire_interval,
1377 .maxlen = sizeof(dirty_expire_interval),
1378 .mode = 0644,
1379 .proc_handler = dirtytime_interval_handler,
1380 .extra1 = &zero,
1381 },
1382 {
1383 .procname = "nr_pdflush_threads",
1384 .mode = 0444 /* read-only */,
1385 .proc_handler = pdflush_proc_obsolete,
1386 },
1387 {
1388 .procname = "swappiness",
1389 .data = &vm_swappiness,
1390 .maxlen = sizeof(vm_swappiness),
1391 .mode = 0644,
1392 .proc_handler = proc_dointvec_minmax,
1393 .extra1 = &zero,
1394 .extra2 = &one_hundred,
1395 },
1396 #ifdef CONFIG_HUGETLB_PAGE
1397 {
1398 .procname = "nr_hugepages",
1399 .data = NULL,
1400 .maxlen = sizeof(unsigned long),
1401 .mode = 0644,
1402 .proc_handler = hugetlb_sysctl_handler,
1403 },
1404 #ifdef CONFIG_NUMA
1405 {
1406 .procname = "nr_hugepages_mempolicy",
1407 .data = NULL,
1408 .maxlen = sizeof(unsigned long),
1409 .mode = 0644,
1410 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1411 },
1412 #endif
1413 {
1414 .procname = "hugetlb_shm_group",
1415 .data = &sysctl_hugetlb_shm_group,
1416 .maxlen = sizeof(gid_t),
1417 .mode = 0644,
1418 .proc_handler = proc_dointvec,
1419 },
1420 {
1421 .procname = "hugepages_treat_as_movable",
1422 .data = &hugepages_treat_as_movable,
1423 .maxlen = sizeof(int),
1424 .mode = 0644,
1425 .proc_handler = proc_dointvec,
1426 },
1427 {
1428 .procname = "nr_overcommit_hugepages",
1429 .data = NULL,
1430 .maxlen = sizeof(unsigned long),
1431 .mode = 0644,
1432 .proc_handler = hugetlb_overcommit_handler,
1433 },
1434 #endif
1435 {
1436 .procname = "lowmem_reserve_ratio",
1437 .data = &sysctl_lowmem_reserve_ratio,
1438 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1439 .mode = 0644,
1440 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1441 },
1442 {
1443 .procname = "drop_caches",
1444 .data = &sysctl_drop_caches,
1445 .maxlen = sizeof(int),
1446 .mode = 0644,
1447 .proc_handler = drop_caches_sysctl_handler,
1448 .extra1 = &one,
1449 .extra2 = &four,
1450 },
1451 #ifdef CONFIG_COMPACTION
1452 {
1453 .procname = "compact_memory",
1454 .data = &sysctl_compact_memory,
1455 .maxlen = sizeof(int),
1456 .mode = 0200,
1457 .proc_handler = sysctl_compaction_handler,
1458 },
1459 {
1460 .procname = "extfrag_threshold",
1461 .data = &sysctl_extfrag_threshold,
1462 .maxlen = sizeof(int),
1463 .mode = 0644,
1464 .proc_handler = sysctl_extfrag_handler,
1465 .extra1 = &min_extfrag_threshold,
1466 .extra2 = &max_extfrag_threshold,
1467 },
1468 {
1469 .procname = "compact_unevictable_allowed",
1470 .data = &sysctl_compact_unevictable_allowed,
1471 .maxlen = sizeof(int),
1472 .mode = 0644,
1473 .proc_handler = proc_dointvec,
1474 .extra1 = &zero,
1475 .extra2 = &one,
1476 },
1477
1478 #endif /* CONFIG_COMPACTION */
1479 {
1480 .procname = "min_free_kbytes",
1481 .data = &min_free_kbytes,
1482 .maxlen = sizeof(min_free_kbytes),
1483 .mode = 0644,
1484 .proc_handler = min_free_kbytes_sysctl_handler,
1485 .extra1 = &zero,
1486 },
1487 {
1488 .procname = "percpu_pagelist_fraction",
1489 .data = &percpu_pagelist_fraction,
1490 .maxlen = sizeof(percpu_pagelist_fraction),
1491 .mode = 0644,
1492 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1493 .extra1 = &zero,
1494 },
1495 #ifdef CONFIG_MMU
1496 {
1497 .procname = "max_map_count",
1498 .data = &sysctl_max_map_count,
1499 .maxlen = sizeof(sysctl_max_map_count),
1500 .mode = 0644,
1501 .proc_handler = proc_dointvec_minmax,
1502 .extra1 = &zero,
1503 },
1504 #else
1505 {
1506 .procname = "nr_trim_pages",
1507 .data = &sysctl_nr_trim_pages,
1508 .maxlen = sizeof(sysctl_nr_trim_pages),
1509 .mode = 0644,
1510 .proc_handler = proc_dointvec_minmax,
1511 .extra1 = &zero,
1512 },
1513 #endif
1514 {
1515 .procname = "laptop_mode",
1516 .data = &laptop_mode,
1517 .maxlen = sizeof(laptop_mode),
1518 .mode = 0644,
1519 .proc_handler = proc_dointvec_jiffies,
1520 },
1521 {
1522 .procname = "block_dump",
1523 .data = &block_dump,
1524 .maxlen = sizeof(block_dump),
1525 .mode = 0644,
1526 .proc_handler = proc_dointvec,
1527 .extra1 = &zero,
1528 },
1529 {
1530 .procname = "vfs_cache_pressure",
1531 .data = &sysctl_vfs_cache_pressure,
1532 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1533 .mode = 0644,
1534 .proc_handler = proc_dointvec,
1535 .extra1 = &zero,
1536 },
1537 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1538 {
1539 .procname = "legacy_va_layout",
1540 .data = &sysctl_legacy_va_layout,
1541 .maxlen = sizeof(sysctl_legacy_va_layout),
1542 .mode = 0644,
1543 .proc_handler = proc_dointvec,
1544 .extra1 = &zero,
1545 },
1546 #endif
1547 #ifdef CONFIG_NUMA
1548 {
1549 .procname = "zone_reclaim_mode",
1550 .data = &zone_reclaim_mode,
1551 .maxlen = sizeof(zone_reclaim_mode),
1552 .mode = 0644,
1553 .proc_handler = proc_dointvec,
1554 .extra1 = &zero,
1555 },
1556 {
1557 .procname = "min_unmapped_ratio",
1558 .data = &sysctl_min_unmapped_ratio,
1559 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1560 .mode = 0644,
1561 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1562 .extra1 = &zero,
1563 .extra2 = &one_hundred,
1564 },
1565 {
1566 .procname = "min_slab_ratio",
1567 .data = &sysctl_min_slab_ratio,
1568 .maxlen = sizeof(sysctl_min_slab_ratio),
1569 .mode = 0644,
1570 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1571 .extra1 = &zero,
1572 .extra2 = &one_hundred,
1573 },
1574 #endif
1575 #ifdef CONFIG_SMP
1576 {
1577 .procname = "stat_interval",
1578 .data = &sysctl_stat_interval,
1579 .maxlen = sizeof(sysctl_stat_interval),
1580 .mode = 0644,
1581 .proc_handler = proc_dointvec_jiffies,
1582 },
1583 #endif
1584 #ifdef CONFIG_MMU
1585 {
1586 .procname = "mmap_min_addr",
1587 .data = &dac_mmap_min_addr,
1588 .maxlen = sizeof(unsigned long),
1589 .mode = 0644,
1590 .proc_handler = mmap_min_addr_handler,
1591 },
1592 #endif
1593 #ifdef CONFIG_NUMA
1594 {
1595 .procname = "numa_zonelist_order",
1596 .data = &numa_zonelist_order,
1597 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1598 .mode = 0644,
1599 .proc_handler = numa_zonelist_order_handler,
1600 },
1601 #endif
1602 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1603 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1604 {
1605 .procname = "vdso_enabled",
1606 #ifdef CONFIG_X86_32
1607 .data = &vdso32_enabled,
1608 .maxlen = sizeof(vdso32_enabled),
1609 #else
1610 .data = &vdso_enabled,
1611 .maxlen = sizeof(vdso_enabled),
1612 #endif
1613 .mode = 0644,
1614 .proc_handler = proc_dointvec,
1615 .extra1 = &zero,
1616 },
1617 #endif
1618 #ifdef CONFIG_HIGHMEM
1619 {
1620 .procname = "highmem_is_dirtyable",
1621 .data = &vm_highmem_is_dirtyable,
1622 .maxlen = sizeof(vm_highmem_is_dirtyable),
1623 .mode = 0644,
1624 .proc_handler = proc_dointvec_minmax,
1625 .extra1 = &zero,
1626 .extra2 = &one,
1627 },
1628 #endif
1629 #ifdef CONFIG_MEMORY_FAILURE
1630 {
1631 .procname = "memory_failure_early_kill",
1632 .data = &sysctl_memory_failure_early_kill,
1633 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1634 .mode = 0644,
1635 .proc_handler = proc_dointvec_minmax,
1636 .extra1 = &zero,
1637 .extra2 = &one,
1638 },
1639 {
1640 .procname = "memory_failure_recovery",
1641 .data = &sysctl_memory_failure_recovery,
1642 .maxlen = sizeof(sysctl_memory_failure_recovery),
1643 .mode = 0644,
1644 .proc_handler = proc_dointvec_minmax,
1645 .extra1 = &zero,
1646 .extra2 = &one,
1647 },
1648 #endif
1649 {
1650 .procname = "user_reserve_kbytes",
1651 .data = &sysctl_user_reserve_kbytes,
1652 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1653 .mode = 0644,
1654 .proc_handler = proc_doulongvec_minmax,
1655 },
1656 {
1657 .procname = "admin_reserve_kbytes",
1658 .data = &sysctl_admin_reserve_kbytes,
1659 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1660 .mode = 0644,
1661 .proc_handler = proc_doulongvec_minmax,
1662 },
1663 { }
1664 };
1665
1666 static struct ctl_table fs_table[] = {
1667 {
1668 .procname = "inode-nr",
1669 .data = &inodes_stat,
1670 .maxlen = 2*sizeof(long),
1671 .mode = 0444,
1672 .proc_handler = proc_nr_inodes,
1673 },
1674 {
1675 .procname = "inode-state",
1676 .data = &inodes_stat,
1677 .maxlen = 7*sizeof(long),
1678 .mode = 0444,
1679 .proc_handler = proc_nr_inodes,
1680 },
1681 {
1682 .procname = "file-nr",
1683 .data = &files_stat,
1684 .maxlen = sizeof(files_stat),
1685 .mode = 0444,
1686 .proc_handler = proc_nr_files,
1687 },
1688 {
1689 .procname = "file-max",
1690 .data = &files_stat.max_files,
1691 .maxlen = sizeof(files_stat.max_files),
1692 .mode = 0644,
1693 .proc_handler = proc_doulongvec_minmax,
1694 },
1695 {
1696 .procname = "nr_open",
1697 .data = &sysctl_nr_open,
1698 .maxlen = sizeof(int),
1699 .mode = 0644,
1700 .proc_handler = proc_dointvec_minmax,
1701 .extra1 = &sysctl_nr_open_min,
1702 .extra2 = &sysctl_nr_open_max,
1703 },
1704 {
1705 .procname = "dentry-state",
1706 .data = &dentry_stat,
1707 .maxlen = 6*sizeof(long),
1708 .mode = 0444,
1709 .proc_handler = proc_nr_dentry,
1710 },
1711 {
1712 .procname = "overflowuid",
1713 .data = &fs_overflowuid,
1714 .maxlen = sizeof(int),
1715 .mode = 0644,
1716 .proc_handler = proc_dointvec_minmax,
1717 .extra1 = &minolduid,
1718 .extra2 = &maxolduid,
1719 },
1720 {
1721 .procname = "overflowgid",
1722 .data = &fs_overflowgid,
1723 .maxlen = sizeof(int),
1724 .mode = 0644,
1725 .proc_handler = proc_dointvec_minmax,
1726 .extra1 = &minolduid,
1727 .extra2 = &maxolduid,
1728 },
1729 #ifdef CONFIG_FILE_LOCKING
1730 {
1731 .procname = "leases-enable",
1732 .data = &leases_enable,
1733 .maxlen = sizeof(int),
1734 .mode = 0644,
1735 .proc_handler = proc_dointvec,
1736 },
1737 #endif
1738 #ifdef CONFIG_DNOTIFY
1739 {
1740 .procname = "dir-notify-enable",
1741 .data = &dir_notify_enable,
1742 .maxlen = sizeof(int),
1743 .mode = 0644,
1744 .proc_handler = proc_dointvec,
1745 },
1746 #endif
1747 #ifdef CONFIG_MMU
1748 #ifdef CONFIG_FILE_LOCKING
1749 {
1750 .procname = "lease-break-time",
1751 .data = &lease_break_time,
1752 .maxlen = sizeof(int),
1753 .mode = 0644,
1754 .proc_handler = proc_dointvec,
1755 },
1756 #endif
1757 #ifdef CONFIG_AIO
1758 {
1759 .procname = "aio-nr",
1760 .data = &aio_nr,
1761 .maxlen = sizeof(aio_nr),
1762 .mode = 0444,
1763 .proc_handler = proc_doulongvec_minmax,
1764 },
1765 {
1766 .procname = "aio-max-nr",
1767 .data = &aio_max_nr,
1768 .maxlen = sizeof(aio_max_nr),
1769 .mode = 0644,
1770 .proc_handler = proc_doulongvec_minmax,
1771 },
1772 #endif /* CONFIG_AIO */
1773 #ifdef CONFIG_INOTIFY_USER
1774 {
1775 .procname = "inotify",
1776 .mode = 0555,
1777 .child = inotify_table,
1778 },
1779 #endif
1780 #ifdef CONFIG_EPOLL
1781 {
1782 .procname = "epoll",
1783 .mode = 0555,
1784 .child = epoll_table,
1785 },
1786 #endif
1787 #endif
1788 {
1789 .procname = "protected_symlinks",
1790 .data = &sysctl_protected_symlinks,
1791 .maxlen = sizeof(int),
1792 .mode = 0600,
1793 .proc_handler = proc_dointvec_minmax,
1794 .extra1 = &zero,
1795 .extra2 = &one,
1796 },
1797 {
1798 .procname = "protected_hardlinks",
1799 .data = &sysctl_protected_hardlinks,
1800 .maxlen = sizeof(int),
1801 .mode = 0600,
1802 .proc_handler = proc_dointvec_minmax,
1803 .extra1 = &zero,
1804 .extra2 = &one,
1805 },
1806 {
1807 .procname = "suid_dumpable",
1808 .data = &suid_dumpable,
1809 .maxlen = sizeof(int),
1810 .mode = 0644,
1811 .proc_handler = proc_dointvec_minmax_coredump,
1812 .extra1 = &zero,
1813 .extra2 = &two,
1814 },
1815 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1816 {
1817 .procname = "binfmt_misc",
1818 .mode = 0555,
1819 .child = sysctl_mount_point,
1820 },
1821 #endif
1822 {
1823 .procname = "pipe-max-size",
1824 .data = &pipe_max_size,
1825 .maxlen = sizeof(int),
1826 .mode = 0644,
1827 .proc_handler = &pipe_proc_fn,
1828 .extra1 = &pipe_min_size,
1829 },
1830 {
1831 .procname = "pipe-user-pages-hard",
1832 .data = &pipe_user_pages_hard,
1833 .maxlen = sizeof(pipe_user_pages_hard),
1834 .mode = 0644,
1835 .proc_handler = proc_doulongvec_minmax,
1836 },
1837 {
1838 .procname = "pipe-user-pages-soft",
1839 .data = &pipe_user_pages_soft,
1840 .maxlen = sizeof(pipe_user_pages_soft),
1841 .mode = 0644,
1842 .proc_handler = proc_doulongvec_minmax,
1843 },
1844 {
1845 .procname = "mount-max",
1846 .data = &sysctl_mount_max,
1847 .maxlen = sizeof(unsigned int),
1848 .mode = 0644,
1849 .proc_handler = proc_dointvec_minmax,
1850 .extra1 = &one,
1851 },
1852 { }
1853 };
1854
1855 static struct ctl_table debug_table[] = {
1856 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1857 {
1858 .procname = "exception-trace",
1859 .data = &show_unhandled_signals,
1860 .maxlen = sizeof(int),
1861 .mode = 0644,
1862 .proc_handler = proc_dointvec
1863 },
1864 #endif
1865 #if defined(CONFIG_OPTPROBES)
1866 {
1867 .procname = "kprobes-optimization",
1868 .data = &sysctl_kprobes_optimization,
1869 .maxlen = sizeof(int),
1870 .mode = 0644,
1871 .proc_handler = proc_kprobes_optimization_handler,
1872 .extra1 = &zero,
1873 .extra2 = &one,
1874 },
1875 #endif
1876 { }
1877 };
1878
1879 static struct ctl_table dev_table[] = {
1880 { }
1881 };
1882
1883 int __init sysctl_init(void)
1884 {
1885 struct ctl_table_header *hdr;
1886
1887 hdr = register_sysctl_table(sysctl_base_table);
1888 kmemleak_not_leak(hdr);
1889 return 0;
1890 }
1891
1892 #endif /* CONFIG_SYSCTL */
1893
1894 /*
1895 * /proc/sys support
1896 */
1897
1898 #ifdef CONFIG_PROC_SYSCTL
1899
1900 static int _proc_do_string(char *data, int maxlen, int write,
1901 char __user *buffer,
1902 size_t *lenp, loff_t *ppos)
1903 {
1904 size_t len;
1905 char __user *p;
1906 char c;
1907
1908 if (!data || !maxlen || !*lenp) {
1909 *lenp = 0;
1910 return 0;
1911 }
1912
1913 if (write) {
1914 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1915 /* Only continue writes not past the end of buffer. */
1916 len = strlen(data);
1917 if (len > maxlen - 1)
1918 len = maxlen - 1;
1919
1920 if (*ppos > len)
1921 return 0;
1922 len = *ppos;
1923 } else {
1924 /* Start writing from beginning of buffer. */
1925 len = 0;
1926 }
1927
1928 *ppos += *lenp;
1929 p = buffer;
1930 while ((p - buffer) < *lenp && len < maxlen - 1) {
1931 if (get_user(c, p++))
1932 return -EFAULT;
1933 if (c == 0 || c == '\n')
1934 break;
1935 data[len++] = c;
1936 }
1937 data[len] = 0;
1938 } else {
1939 len = strlen(data);
1940 if (len > maxlen)
1941 len = maxlen;
1942
1943 if (*ppos > len) {
1944 *lenp = 0;
1945 return 0;
1946 }
1947
1948 data += *ppos;
1949 len -= *ppos;
1950
1951 if (len > *lenp)
1952 len = *lenp;
1953 if (len)
1954 if (copy_to_user(buffer, data, len))
1955 return -EFAULT;
1956 if (len < *lenp) {
1957 if (put_user('\n', buffer + len))
1958 return -EFAULT;
1959 len++;
1960 }
1961 *lenp = len;
1962 *ppos += len;
1963 }
1964 return 0;
1965 }
1966
1967 static void warn_sysctl_write(struct ctl_table *table)
1968 {
1969 pr_warn_once("%s wrote to %s when file position was not 0!\n"
1970 "This will not be supported in the future. To silence this\n"
1971 "warning, set kernel.sysctl_writes_strict = -1\n",
1972 current->comm, table->procname);
1973 }
1974
1975 /**
1976 * proc_dostring - read a string sysctl
1977 * @table: the sysctl table
1978 * @write: %TRUE if this is a write to the sysctl file
1979 * @buffer: the user buffer
1980 * @lenp: the size of the user buffer
1981 * @ppos: file position
1982 *
1983 * Reads/writes a string from/to the user buffer. If the kernel
1984 * buffer provided is not large enough to hold the string, the
1985 * string is truncated. The copied string is %NULL-terminated.
1986 * If the string is being read by the user process, it is copied
1987 * and a newline '\n' is added. It is truncated if the buffer is
1988 * not large enough.
1989 *
1990 * Returns 0 on success.
1991 */
1992 int proc_dostring(struct ctl_table *table, int write,
1993 void __user *buffer, size_t *lenp, loff_t *ppos)
1994 {
1995 if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN)
1996 warn_sysctl_write(table);
1997
1998 return _proc_do_string((char *)(table->data), table->maxlen, write,
1999 (char __user *)buffer, lenp, ppos);
2000 }
2001
2002 static size_t proc_skip_spaces(char **buf)
2003 {
2004 size_t ret;
2005 char *tmp = skip_spaces(*buf);
2006 ret = tmp - *buf;
2007 *buf = tmp;
2008 return ret;
2009 }
2010
2011 static void proc_skip_char(char **buf, size_t *size, const char v)
2012 {
2013 while (*size) {
2014 if (**buf != v)
2015 break;
2016 (*size)--;
2017 (*buf)++;
2018 }
2019 }
2020
2021 #define TMPBUFLEN 22
2022 /**
2023 * proc_get_long - reads an ASCII formatted integer from a user buffer
2024 *
2025 * @buf: a kernel buffer
2026 * @size: size of the kernel buffer
2027 * @val: this is where the number will be stored
2028 * @neg: set to %TRUE if number is negative
2029 * @perm_tr: a vector which contains the allowed trailers
2030 * @perm_tr_len: size of the perm_tr vector
2031 * @tr: pointer to store the trailer character
2032 *
2033 * In case of success %0 is returned and @buf and @size are updated with
2034 * the amount of bytes read. If @tr is non-NULL and a trailing
2035 * character exists (size is non-zero after returning from this
2036 * function), @tr is updated with the trailing character.
2037 */
2038 static int proc_get_long(char **buf, size_t *size,
2039 unsigned long *val, bool *neg,
2040 const char *perm_tr, unsigned perm_tr_len, char *tr)
2041 {
2042 int len;
2043 char *p, tmp[TMPBUFLEN];
2044
2045 if (!*size)
2046 return -EINVAL;
2047
2048 len = *size;
2049 if (len > TMPBUFLEN - 1)
2050 len = TMPBUFLEN - 1;
2051
2052 memcpy(tmp, *buf, len);
2053
2054 tmp[len] = 0;
2055 p = tmp;
2056 if (*p == '-' && *size > 1) {
2057 *neg = true;
2058 p++;
2059 } else
2060 *neg = false;
2061 if (!isdigit(*p))
2062 return -EINVAL;
2063
2064 *val = simple_strtoul(p, &p, 0);
2065
2066 len = p - tmp;
2067
2068 /* We don't know if the next char is whitespace thus we may accept
2069 * invalid integers (e.g. 1234...a) or two integers instead of one
2070 * (e.g. 123...1). So lets not allow such large numbers. */
2071 if (len == TMPBUFLEN - 1)
2072 return -EINVAL;
2073
2074 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2075 return -EINVAL;
2076
2077 if (tr && (len < *size))
2078 *tr = *p;
2079
2080 *buf += len;
2081 *size -= len;
2082
2083 return 0;
2084 }
2085
2086 /**
2087 * proc_put_long - converts an integer to a decimal ASCII formatted string
2088 *
2089 * @buf: the user buffer
2090 * @size: the size of the user buffer
2091 * @val: the integer to be converted
2092 * @neg: sign of the number, %TRUE for negative
2093 *
2094 * In case of success %0 is returned and @buf and @size are updated with
2095 * the amount of bytes written.
2096 */
2097 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2098 bool neg)
2099 {
2100 int len;
2101 char tmp[TMPBUFLEN], *p = tmp;
2102
2103 sprintf(p, "%s%lu", neg ? "-" : "", val);
2104 len = strlen(tmp);
2105 if (len > *size)
2106 len = *size;
2107 if (copy_to_user(*buf, tmp, len))
2108 return -EFAULT;
2109 *size -= len;
2110 *buf += len;
2111 return 0;
2112 }
2113 #undef TMPBUFLEN
2114
2115 static int proc_put_char(void __user **buf, size_t *size, char c)
2116 {
2117 if (*size) {
2118 char __user **buffer = (char __user **)buf;
2119 if (put_user(c, *buffer))
2120 return -EFAULT;
2121 (*size)--, (*buffer)++;
2122 *buf = *buffer;
2123 }
2124 return 0;
2125 }
2126
2127 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2128 int *valp,
2129 int write, void *data)
2130 {
2131 if (write) {
2132 if (*negp) {
2133 if (*lvalp > (unsigned long) INT_MAX + 1)
2134 return -EINVAL;
2135 *valp = -*lvalp;
2136 } else {
2137 if (*lvalp > (unsigned long) INT_MAX)
2138 return -EINVAL;
2139 *valp = *lvalp;
2140 }
2141 } else {
2142 int val = *valp;
2143 if (val < 0) {
2144 *negp = true;
2145 *lvalp = -(unsigned long)val;
2146 } else {
2147 *negp = false;
2148 *lvalp = (unsigned long)val;
2149 }
2150 }
2151 return 0;
2152 }
2153
2154 static int do_proc_douintvec_conv(bool *negp, unsigned long *lvalp,
2155 int *valp,
2156 int write, void *data)
2157 {
2158 if (write) {
2159 if (*negp)
2160 return -EINVAL;
2161 if (*lvalp > UINT_MAX)
2162 return -EINVAL;
2163 *valp = *lvalp;
2164 } else {
2165 unsigned int val = *valp;
2166 *negp = false;
2167 *lvalp = (unsigned long)val;
2168 }
2169 return 0;
2170 }
2171
2172 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2173
2174 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2175 int write, void __user *buffer,
2176 size_t *lenp, loff_t *ppos,
2177 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2178 int write, void *data),
2179 void *data)
2180 {
2181 int *i, vleft, first = 1, err = 0;
2182 unsigned long page = 0;
2183 size_t left;
2184 char *kbuf;
2185
2186 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2187 *lenp = 0;
2188 return 0;
2189 }
2190
2191 i = (int *) tbl_data;
2192 vleft = table->maxlen / sizeof(*i);
2193 left = *lenp;
2194
2195 if (!conv)
2196 conv = do_proc_dointvec_conv;
2197
2198 if (write) {
2199 if (*ppos) {
2200 switch (sysctl_writes_strict) {
2201 case SYSCTL_WRITES_STRICT:
2202 goto out;
2203 case SYSCTL_WRITES_WARN:
2204 warn_sysctl_write(table);
2205 break;
2206 default:
2207 break;
2208 }
2209 }
2210
2211 if (left > PAGE_SIZE - 1)
2212 left = PAGE_SIZE - 1;
2213 page = __get_free_page(GFP_TEMPORARY);
2214 kbuf = (char *) page;
2215 if (!kbuf)
2216 return -ENOMEM;
2217 if (copy_from_user(kbuf, buffer, left)) {
2218 err = -EFAULT;
2219 goto free;
2220 }
2221 kbuf[left] = 0;
2222 }
2223
2224 for (; left && vleft--; i++, first=0) {
2225 unsigned long lval;
2226 bool neg;
2227
2228 if (write) {
2229 left -= proc_skip_spaces(&kbuf);
2230
2231 if (!left)
2232 break;
2233 err = proc_get_long(&kbuf, &left, &lval, &neg,
2234 proc_wspace_sep,
2235 sizeof(proc_wspace_sep), NULL);
2236 if (err)
2237 break;
2238 if (conv(&neg, &lval, i, 1, data)) {
2239 err = -EINVAL;
2240 break;
2241 }
2242 } else {
2243 if (conv(&neg, &lval, i, 0, data)) {
2244 err = -EINVAL;
2245 break;
2246 }
2247 if (!first)
2248 err = proc_put_char(&buffer, &left, '\t');
2249 if (err)
2250 break;
2251 err = proc_put_long(&buffer, &left, lval, neg);
2252 if (err)
2253 break;
2254 }
2255 }
2256
2257 if (!write && !first && left && !err)
2258 err = proc_put_char(&buffer, &left, '\n');
2259 if (write && !err && left)
2260 left -= proc_skip_spaces(&kbuf);
2261 free:
2262 if (write) {
2263 free_page(page);
2264 if (first)
2265 return err ? : -EINVAL;
2266 }
2267 *lenp -= left;
2268 out:
2269 *ppos += *lenp;
2270 return err;
2271 }
2272
2273 static int do_proc_dointvec(struct ctl_table *table, int write,
2274 void __user *buffer, size_t *lenp, loff_t *ppos,
2275 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2276 int write, void *data),
2277 void *data)
2278 {
2279 return __do_proc_dointvec(table->data, table, write,
2280 buffer, lenp, ppos, conv, data);
2281 }
2282
2283 /**
2284 * proc_dointvec - read a vector of integers
2285 * @table: the sysctl table
2286 * @write: %TRUE if this is a write to the sysctl file
2287 * @buffer: the user buffer
2288 * @lenp: the size of the user buffer
2289 * @ppos: file position
2290 *
2291 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2292 * values from/to the user buffer, treated as an ASCII string.
2293 *
2294 * Returns 0 on success.
2295 */
2296 int proc_dointvec(struct ctl_table *table, int write,
2297 void __user *buffer, size_t *lenp, loff_t *ppos)
2298 {
2299 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2300 }
2301
2302 /**
2303 * proc_douintvec - read a vector of unsigned integers
2304 * @table: the sysctl table
2305 * @write: %TRUE if this is a write to the sysctl file
2306 * @buffer: the user buffer
2307 * @lenp: the size of the user buffer
2308 * @ppos: file position
2309 *
2310 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2311 * values from/to the user buffer, treated as an ASCII string.
2312 *
2313 * Returns 0 on success.
2314 */
2315 int proc_douintvec(struct ctl_table *table, int write,
2316 void __user *buffer, size_t *lenp, loff_t *ppos)
2317 {
2318 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2319 do_proc_douintvec_conv, NULL);
2320 }
2321
2322 /*
2323 * Taint values can only be increased
2324 * This means we can safely use a temporary.
2325 */
2326 static int proc_taint(struct ctl_table *table, int write,
2327 void __user *buffer, size_t *lenp, loff_t *ppos)
2328 {
2329 struct ctl_table t;
2330 unsigned long tmptaint = get_taint();
2331 int err;
2332
2333 if (write && !capable(CAP_SYS_ADMIN))
2334 return -EPERM;
2335
2336 t = *table;
2337 t.data = &tmptaint;
2338 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2339 if (err < 0)
2340 return err;
2341
2342 if (write) {
2343 /*
2344 * Poor man's atomic or. Not worth adding a primitive
2345 * to everyone's atomic.h for this
2346 */
2347 int i;
2348 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2349 if ((tmptaint >> i) & 1)
2350 add_taint(i, LOCKDEP_STILL_OK);
2351 }
2352 }
2353
2354 return err;
2355 }
2356
2357 #ifdef CONFIG_PRINTK
2358 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2359 void __user *buffer, size_t *lenp, loff_t *ppos)
2360 {
2361 if (write && !capable(CAP_SYS_ADMIN))
2362 return -EPERM;
2363
2364 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2365 }
2366 #endif
2367
2368 struct do_proc_dointvec_minmax_conv_param {
2369 int *min;
2370 int *max;
2371 };
2372
2373 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2374 int *valp,
2375 int write, void *data)
2376 {
2377 struct do_proc_dointvec_minmax_conv_param *param = data;
2378 if (write) {
2379 int val = *negp ? -*lvalp : *lvalp;
2380 if ((param->min && *param->min > val) ||
2381 (param->max && *param->max < val))
2382 return -EINVAL;
2383 *valp = val;
2384 } else {
2385 int val = *valp;
2386 if (val < 0) {
2387 *negp = true;
2388 *lvalp = -(unsigned long)val;
2389 } else {
2390 *negp = false;
2391 *lvalp = (unsigned long)val;
2392 }
2393 }
2394 return 0;
2395 }
2396
2397 /**
2398 * proc_dointvec_minmax - read a vector of integers with min/max values
2399 * @table: the sysctl table
2400 * @write: %TRUE if this is a write to the sysctl file
2401 * @buffer: the user buffer
2402 * @lenp: the size of the user buffer
2403 * @ppos: file position
2404 *
2405 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2406 * values from/to the user buffer, treated as an ASCII string.
2407 *
2408 * This routine will ensure the values are within the range specified by
2409 * table->extra1 (min) and table->extra2 (max).
2410 *
2411 * Returns 0 on success.
2412 */
2413 int proc_dointvec_minmax(struct ctl_table *table, int write,
2414 void __user *buffer, size_t *lenp, loff_t *ppos)
2415 {
2416 struct do_proc_dointvec_minmax_conv_param param = {
2417 .min = (int *) table->extra1,
2418 .max = (int *) table->extra2,
2419 };
2420 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2421 do_proc_dointvec_minmax_conv, &param);
2422 }
2423
2424 #ifdef CONFIG_X86
2425 int proc_dointvec_ibrs_dump(struct ctl_table *table, int write,
2426 void __user *buffer, size_t *lenp, loff_t *ppos)
2427 {
2428 int ret, orig_inuse;
2429 unsigned int cpu;
2430
2431
2432 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2433 printk("sysctl_ibrs_enabled = %u, sysctl_ibpb_enabled = %u\n", sysctl_ibrs_enabled, sysctl_ibpb_enabled);
2434 printk("use_ibrs = %d, use_ibpb = %d\n", use_ibrs, use_ibpb);
2435 mutex_lock(&spec_ctrl_mutex);
2436 orig_inuse = use_ibrs;
2437 /* temporary halt to ibrs usage to dump ibrs values */
2438 clear_ibrs_inuse();
2439 for_each_online_cpu(cpu) {
2440 u64 val;
2441
2442 if (boot_cpu_has(X86_FEATURE_SPEC_CTRL))
2443 rdmsrl_on_cpu(cpu, MSR_IA32_SPEC_CTRL, &val);
2444 else
2445 val = 0;
2446 printk("read cpu %d ibrs val %lu\n", cpu, (unsigned long) val);
2447 }
2448 use_ibrs = orig_inuse;
2449 mutex_unlock(&spec_ctrl_mutex);
2450 return ret;
2451 }
2452
2453 int proc_dointvec_ibrs_ctrl(struct ctl_table *table, int write,
2454 void __user *buffer, size_t *lenp, loff_t *ppos)
2455 {
2456 int ret;
2457 unsigned int cpu;
2458
2459 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2460 pr_debug("sysctl_ibrs_enabled = %u, sysctl_ibpb_enabled = %u\n", sysctl_ibrs_enabled, sysctl_ibpb_enabled);
2461 pr_debug("before:use_ibrs = %d, use_ibpb = %d\n", use_ibrs, use_ibpb);
2462 mutex_lock(&spec_ctrl_mutex);
2463 if (sysctl_ibrs_enabled == 0) {
2464 /* always set IBRS off */
2465 set_ibrs_disabled();
2466 if (ibrs_supported) {
2467 for_each_online_cpu(cpu)
2468 wrmsrl_on_cpu(cpu, MSR_IA32_SPEC_CTRL, 0x0);
2469 }
2470 } else if (sysctl_ibrs_enabled == 2) {
2471 /* always set IBRS on, even in user space */
2472 clear_ibrs_disabled();
2473 if (ibrs_supported) {
2474 for_each_online_cpu(cpu)
2475 wrmsrl_on_cpu(cpu, MSR_IA32_SPEC_CTRL, FEATURE_ENABLE_IBRS);
2476 } else {
2477 sysctl_ibrs_enabled = 0;
2478 }
2479 } else if (sysctl_ibrs_enabled == 1) {
2480 /* use IBRS in kernel */
2481 clear_ibrs_disabled();
2482 if (!ibrs_inuse)
2483 /* platform don't support ibrs */
2484 sysctl_ibrs_enabled = 0;
2485 }
2486 mutex_unlock(&spec_ctrl_mutex);
2487 pr_debug("after:use_ibrs = %d, use_ibpb = %d\n", use_ibrs, use_ibpb);
2488 return ret;
2489 }
2490
2491 int proc_dointvec_ibpb_ctrl(struct ctl_table *table, int write,
2492 void __user *buffer, size_t *lenp, loff_t *ppos)
2493 {
2494 int ret;
2495
2496 ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2497 pr_debug("sysctl_ibrs_enabled = %u, sysctl_ibpb_enabled = %u\n", sysctl_ibrs_enabled, sysctl_ibpb_enabled);
2498 pr_debug("before:use_ibrs = %d, use_ibpb = %d\n", use_ibrs, use_ibpb);
2499 mutex_lock(&spec_ctrl_mutex);
2500 if (sysctl_ibpb_enabled == 0)
2501 set_ibpb_disabled();
2502 else if (sysctl_ibpb_enabled == 1) {
2503 clear_ibpb_disabled();
2504 if (!ibpb_inuse)
2505 /* platform don't support ibpb */
2506 sysctl_ibpb_enabled = 0;
2507 }
2508 mutex_unlock(&spec_ctrl_mutex);
2509 pr_debug("after:use_ibrs = %d, use_ibpb = %d\n", use_ibrs, use_ibpb);
2510 return ret;
2511 }
2512 #endif
2513
2514 static void validate_coredump_safety(void)
2515 {
2516 #ifdef CONFIG_COREDUMP
2517 if (suid_dumpable == SUID_DUMP_ROOT &&
2518 core_pattern[0] != '/' && core_pattern[0] != '|') {
2519 printk(KERN_WARNING "Unsafe core_pattern used with "\
2520 "suid_dumpable=2. Pipe handler or fully qualified "\
2521 "core dump path required.\n");
2522 }
2523 #endif
2524 }
2525
2526 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2527 void __user *buffer, size_t *lenp, loff_t *ppos)
2528 {
2529 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2530 if (!error)
2531 validate_coredump_safety();
2532 return error;
2533 }
2534
2535 #ifdef CONFIG_COREDUMP
2536 static int proc_dostring_coredump(struct ctl_table *table, int write,
2537 void __user *buffer, size_t *lenp, loff_t *ppos)
2538 {
2539 int error = proc_dostring(table, write, buffer, lenp, ppos);
2540 if (!error)
2541 validate_coredump_safety();
2542 return error;
2543 }
2544 #endif
2545
2546 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2547 void __user *buffer,
2548 size_t *lenp, loff_t *ppos,
2549 unsigned long convmul,
2550 unsigned long convdiv)
2551 {
2552 unsigned long *i, *min, *max;
2553 int vleft, first = 1, err = 0;
2554 unsigned long page = 0;
2555 size_t left;
2556 char *kbuf;
2557
2558 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2559 *lenp = 0;
2560 return 0;
2561 }
2562
2563 i = (unsigned long *) data;
2564 min = (unsigned long *) table->extra1;
2565 max = (unsigned long *) table->extra2;
2566 vleft = table->maxlen / sizeof(unsigned long);
2567 left = *lenp;
2568
2569 if (write) {
2570 if (*ppos) {
2571 switch (sysctl_writes_strict) {
2572 case SYSCTL_WRITES_STRICT:
2573 goto out;
2574 case SYSCTL_WRITES_WARN:
2575 warn_sysctl_write(table);
2576 break;
2577 default:
2578 break;
2579 }
2580 }
2581
2582 if (left > PAGE_SIZE - 1)
2583 left = PAGE_SIZE - 1;
2584 page = __get_free_page(GFP_TEMPORARY);
2585 kbuf = (char *) page;
2586 if (!kbuf)
2587 return -ENOMEM;
2588 if (copy_from_user(kbuf, buffer, left)) {
2589 err = -EFAULT;
2590 goto free;
2591 }
2592 kbuf[left] = 0;
2593 }
2594
2595 for (; left && vleft--; i++, first = 0) {
2596 unsigned long val;
2597
2598 if (write) {
2599 bool neg;
2600
2601 left -= proc_skip_spaces(&kbuf);
2602
2603 err = proc_get_long(&kbuf, &left, &val, &neg,
2604 proc_wspace_sep,
2605 sizeof(proc_wspace_sep), NULL);
2606 if (err)
2607 break;
2608 if (neg)
2609 continue;
2610 val = convmul * val / convdiv;
2611 if ((min && val < *min) || (max && val > *max))
2612 continue;
2613 *i = val;
2614 } else {
2615 val = convdiv * (*i) / convmul;
2616 if (!first) {
2617 err = proc_put_char(&buffer, &left, '\t');
2618 if (err)
2619 break;
2620 }
2621 err = proc_put_long(&buffer, &left, val, false);
2622 if (err)
2623 break;
2624 }
2625 }
2626
2627 if (!write && !first && left && !err)
2628 err = proc_put_char(&buffer, &left, '\n');
2629 if (write && !err)
2630 left -= proc_skip_spaces(&kbuf);
2631 free:
2632 if (write) {
2633 free_page(page);
2634 if (first)
2635 return err ? : -EINVAL;
2636 }
2637 *lenp -= left;
2638 out:
2639 *ppos += *lenp;
2640 return err;
2641 }
2642
2643 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2644 void __user *buffer,
2645 size_t *lenp, loff_t *ppos,
2646 unsigned long convmul,
2647 unsigned long convdiv)
2648 {
2649 return __do_proc_doulongvec_minmax(table->data, table, write,
2650 buffer, lenp, ppos, convmul, convdiv);
2651 }
2652
2653 /**
2654 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2655 * @table: the sysctl table
2656 * @write: %TRUE if this is a write to the sysctl file
2657 * @buffer: the user buffer
2658 * @lenp: the size of the user buffer
2659 * @ppos: file position
2660 *
2661 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2662 * values from/to the user buffer, treated as an ASCII string.
2663 *
2664 * This routine will ensure the values are within the range specified by
2665 * table->extra1 (min) and table->extra2 (max).
2666 *
2667 * Returns 0 on success.
2668 */
2669 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2670 void __user *buffer, size_t *lenp, loff_t *ppos)
2671 {
2672 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2673 }
2674
2675 /**
2676 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2677 * @table: the sysctl table
2678 * @write: %TRUE if this is a write to the sysctl file
2679 * @buffer: the user buffer
2680 * @lenp: the size of the user buffer
2681 * @ppos: file position
2682 *
2683 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2684 * values from/to the user buffer, treated as an ASCII string. The values
2685 * are treated as milliseconds, and converted to jiffies when they are stored.
2686 *
2687 * This routine will ensure the values are within the range specified by
2688 * table->extra1 (min) and table->extra2 (max).
2689 *
2690 * Returns 0 on success.
2691 */
2692 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2693 void __user *buffer,
2694 size_t *lenp, loff_t *ppos)
2695 {
2696 return do_proc_doulongvec_minmax(table, write, buffer,
2697 lenp, ppos, HZ, 1000l);
2698 }
2699
2700
2701 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2702 int *valp,
2703 int write, void *data)
2704 {
2705 if (write) {
2706 if (*lvalp > LONG_MAX / HZ)
2707 return 1;
2708 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2709 } else {
2710 int val = *valp;
2711 unsigned long lval;
2712 if (val < 0) {
2713 *negp = true;
2714 lval = -(unsigned long)val;
2715 } else {
2716 *negp = false;
2717 lval = (unsigned long)val;
2718 }
2719 *lvalp = lval / HZ;
2720 }
2721 return 0;
2722 }
2723
2724 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2725 int *valp,
2726 int write, void *data)
2727 {
2728 if (write) {
2729 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2730 return 1;
2731 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2732 } else {
2733 int val = *valp;
2734 unsigned long lval;
2735 if (val < 0) {
2736 *negp = true;
2737 lval = -(unsigned long)val;
2738 } else {
2739 *negp = false;
2740 lval = (unsigned long)val;
2741 }
2742 *lvalp = jiffies_to_clock_t(lval);
2743 }
2744 return 0;
2745 }
2746
2747 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2748 int *valp,
2749 int write, void *data)
2750 {
2751 if (write) {
2752 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2753
2754 if (jif > INT_MAX)
2755 return 1;
2756 *valp = (int)jif;
2757 } else {
2758 int val = *valp;
2759 unsigned long lval;
2760 if (val < 0) {
2761 *negp = true;
2762 lval = -(unsigned long)val;
2763 } else {
2764 *negp = false;
2765 lval = (unsigned long)val;
2766 }
2767 *lvalp = jiffies_to_msecs(lval);
2768 }
2769 return 0;
2770 }
2771
2772 /**
2773 * proc_dointvec_jiffies - read a vector of integers as seconds
2774 * @table: the sysctl table
2775 * @write: %TRUE if this is a write to the sysctl file
2776 * @buffer: the user buffer
2777 * @lenp: the size of the user buffer
2778 * @ppos: file position
2779 *
2780 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2781 * values from/to the user buffer, treated as an ASCII string.
2782 * The values read are assumed to be in seconds, and are converted into
2783 * jiffies.
2784 *
2785 * Returns 0 on success.
2786 */
2787 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2788 void __user *buffer, size_t *lenp, loff_t *ppos)
2789 {
2790 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2791 do_proc_dointvec_jiffies_conv,NULL);
2792 }
2793
2794 /**
2795 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2796 * @table: the sysctl table
2797 * @write: %TRUE if this is a write to the sysctl file
2798 * @buffer: the user buffer
2799 * @lenp: the size of the user buffer
2800 * @ppos: pointer to the file position
2801 *
2802 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2803 * values from/to the user buffer, treated as an ASCII string.
2804 * The values read are assumed to be in 1/USER_HZ seconds, and
2805 * are converted into jiffies.
2806 *
2807 * Returns 0 on success.
2808 */
2809 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2810 void __user *buffer, size_t *lenp, loff_t *ppos)
2811 {
2812 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2813 do_proc_dointvec_userhz_jiffies_conv,NULL);
2814 }
2815
2816 /**
2817 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2818 * @table: the sysctl table
2819 * @write: %TRUE if this is a write to the sysctl file
2820 * @buffer: the user buffer
2821 * @lenp: the size of the user buffer
2822 * @ppos: file position
2823 * @ppos: the current position in the file
2824 *
2825 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2826 * values from/to the user buffer, treated as an ASCII string.
2827 * The values read are assumed to be in 1/1000 seconds, and
2828 * are converted into jiffies.
2829 *
2830 * Returns 0 on success.
2831 */
2832 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2833 void __user *buffer, size_t *lenp, loff_t *ppos)
2834 {
2835 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2836 do_proc_dointvec_ms_jiffies_conv, NULL);
2837 }
2838
2839 static int proc_do_cad_pid(struct ctl_table *table, int write,
2840 void __user *buffer, size_t *lenp, loff_t *ppos)
2841 {
2842 struct pid *new_pid;
2843 pid_t tmp;
2844 int r;
2845
2846 tmp = pid_vnr(cad_pid);
2847
2848 r = __do_proc_dointvec(&tmp, table, write, buffer,
2849 lenp, ppos, NULL, NULL);
2850 if (r || !write)
2851 return r;
2852
2853 new_pid = find_get_pid(tmp);
2854 if (!new_pid)
2855 return -ESRCH;
2856
2857 put_pid(xchg(&cad_pid, new_pid));
2858 return 0;
2859 }
2860
2861 /**
2862 * proc_do_large_bitmap - read/write from/to a large bitmap
2863 * @table: the sysctl table
2864 * @write: %TRUE if this is a write to the sysctl file
2865 * @buffer: the user buffer
2866 * @lenp: the size of the user buffer
2867 * @ppos: file position
2868 *
2869 * The bitmap is stored at table->data and the bitmap length (in bits)
2870 * in table->maxlen.
2871 *
2872 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2873 * large bitmaps may be represented in a compact manner. Writing into
2874 * the file will clear the bitmap then update it with the given input.
2875 *
2876 * Returns 0 on success.
2877 */
2878 int proc_do_large_bitmap(struct ctl_table *table, int write,
2879 void __user *buffer, size_t *lenp, loff_t *ppos)
2880 {
2881 int err = 0;
2882 bool first = 1;
2883 size_t left = *lenp;
2884 unsigned long bitmap_len = table->maxlen;
2885 unsigned long *bitmap = *(unsigned long **) table->data;
2886 unsigned long *tmp_bitmap = NULL;
2887 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2888
2889 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2890 *lenp = 0;
2891 return 0;
2892 }
2893
2894 if (write) {
2895 unsigned long page = 0;
2896 char *kbuf;
2897
2898 if (left > PAGE_SIZE - 1)
2899 left = PAGE_SIZE - 1;
2900
2901 page = __get_free_page(GFP_TEMPORARY);
2902 kbuf = (char *) page;
2903 if (!kbuf)
2904 return -ENOMEM;
2905 if (copy_from_user(kbuf, buffer, left)) {
2906 free_page(page);
2907 return -EFAULT;
2908 }
2909 kbuf[left] = 0;
2910
2911 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2912 GFP_KERNEL);
2913 if (!tmp_bitmap) {
2914 free_page(page);
2915 return -ENOMEM;
2916 }
2917 proc_skip_char(&kbuf, &left, '\n');
2918 while (!err && left) {
2919 unsigned long val_a, val_b;
2920 bool neg;
2921
2922 err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2923 sizeof(tr_a), &c);
2924 if (err)
2925 break;
2926 if (val_a >= bitmap_len || neg) {
2927 err = -EINVAL;
2928 break;
2929 }
2930
2931 val_b = val_a;
2932 if (left) {
2933 kbuf++;
2934 left--;
2935 }
2936
2937 if (c == '-') {
2938 err = proc_get_long(&kbuf, &left, &val_b,
2939 &neg, tr_b, sizeof(tr_b),
2940 &c);
2941 if (err)
2942 break;
2943 if (val_b >= bitmap_len || neg ||
2944 val_a > val_b) {
2945 err = -EINVAL;
2946 break;
2947 }
2948 if (left) {
2949 kbuf++;
2950 left--;
2951 }
2952 }
2953
2954 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2955 first = 0;
2956 proc_skip_char(&kbuf, &left, '\n');
2957 }
2958 free_page(page);
2959 } else {
2960 unsigned long bit_a, bit_b = 0;
2961
2962 while (left) {
2963 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2964 if (bit_a >= bitmap_len)
2965 break;
2966 bit_b = find_next_zero_bit(bitmap, bitmap_len,
2967 bit_a + 1) - 1;
2968
2969 if (!first) {
2970 err = proc_put_char(&buffer, &left, ',');
2971 if (err)
2972 break;
2973 }
2974 err = proc_put_long(&buffer, &left, bit_a, false);
2975 if (err)
2976 break;
2977 if (bit_a != bit_b) {
2978 err = proc_put_char(&buffer, &left, '-');
2979 if (err)
2980 break;
2981 err = proc_put_long(&buffer, &left, bit_b, false);
2982 if (err)
2983 break;
2984 }
2985
2986 first = 0; bit_b++;
2987 }
2988 if (!err)
2989 err = proc_put_char(&buffer, &left, '\n');
2990 }
2991
2992 if (!err) {
2993 if (write) {
2994 if (*ppos)
2995 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2996 else
2997 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2998 }
2999 kfree(tmp_bitmap);
3000 *lenp -= left;
3001 *ppos += *lenp;
3002 return 0;
3003 } else {
3004 kfree(tmp_bitmap);
3005 return err;
3006 }
3007 }
3008
3009 #else /* CONFIG_PROC_SYSCTL */
3010
3011 int proc_dostring(struct ctl_table *table, int write,
3012 void __user *buffer, size_t *lenp, loff_t *ppos)
3013 {
3014 return -ENOSYS;
3015 }
3016
3017 int proc_dointvec(struct ctl_table *table, int write,
3018 void __user *buffer, size_t *lenp, loff_t *ppos)
3019 {
3020 return -ENOSYS;
3021 }
3022
3023 int proc_douintvec(struct ctl_table *table, int write,
3024 void __user *buffer, size_t *lenp, loff_t *ppos)
3025 {
3026 return -ENOSYS;
3027 }
3028
3029 int proc_dointvec_minmax(struct ctl_table *table, int write,
3030 void __user *buffer, size_t *lenp, loff_t *ppos)
3031 {
3032 return -ENOSYS;
3033 }
3034
3035 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3036 void __user *buffer, size_t *lenp, loff_t *ppos)
3037 {
3038 return -ENOSYS;
3039 }
3040
3041 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3042 void __user *buffer, size_t *lenp, loff_t *ppos)
3043 {
3044 return -ENOSYS;
3045 }
3046
3047 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3048 void __user *buffer, size_t *lenp, loff_t *ppos)
3049 {
3050 return -ENOSYS;
3051 }
3052
3053 int proc_doulongvec_minmax(struct ctl_table *table, int write,
3054 void __user *buffer, size_t *lenp, loff_t *ppos)
3055 {
3056 return -ENOSYS;
3057 }
3058
3059 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3060 void __user *buffer,
3061 size_t *lenp, loff_t *ppos)
3062 {
3063 return -ENOSYS;
3064 }
3065
3066
3067 #endif /* CONFIG_PROC_SYSCTL */
3068
3069 /*
3070 * No sense putting this after each symbol definition, twice,
3071 * exception granted :-)
3072 */
3073 EXPORT_SYMBOL(proc_dointvec);
3074 EXPORT_SYMBOL(proc_douintvec);
3075 EXPORT_SYMBOL(proc_dointvec_jiffies);
3076 EXPORT_SYMBOL(proc_dointvec_minmax);
3077 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3078 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3079 EXPORT_SYMBOL(proc_dostring);
3080 EXPORT_SYMBOL(proc_doulongvec_minmax);
3081 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
Ubuntu Artful kernel mirror