]> git.proxmox.com Git - mirror_ubuntu-jammy-kernel.git/blob - kernel/sysctl.c
Merge tag 'ovl-update-4.19' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/vfs
[mirror_ubuntu-jammy-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 extern int pid_max;
109 extern int pid_max_min, pid_max_max;
110 extern int percpu_pagelist_fraction;
111 extern int latencytop_enabled;
112 extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
113 #ifndef CONFIG_MMU
114 extern int sysctl_nr_trim_pages;
115 #endif
116
117 /* Constants used for minimum and maximum */
118 #ifdef CONFIG_LOCKUP_DETECTOR
119 static int sixty = 60;
120 #endif
121
122 static int __maybe_unused neg_one = -1;
123
124 static int zero;
125 static int __maybe_unused one = 1;
126 static int __maybe_unused two = 2;
127 static int __maybe_unused four = 4;
128 static unsigned long one_ul = 1;
129 static int one_hundred = 100;
130 static int one_thousand = 1000;
131 #ifdef CONFIG_PRINTK
132 static int ten_thousand = 10000;
133 #endif
134 #ifdef CONFIG_PERF_EVENTS
135 static int six_hundred_forty_kb = 640 * 1024;
136 #endif
137
138 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
139 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
140
141 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
142 static int maxolduid = 65535;
143 static int minolduid;
144
145 static int ngroups_max = NGROUPS_MAX;
146 static const int cap_last_cap = CAP_LAST_CAP;
147
148 /*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */
149 #ifdef CONFIG_DETECT_HUNG_TASK
150 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
151 #endif
152
153 #ifdef CONFIG_INOTIFY_USER
154 #include <linux/inotify.h>
155 #endif
156 #ifdef CONFIG_SPARC
157 #endif
158
159 #ifdef __hppa__
160 extern int pwrsw_enabled;
161 #endif
162
163 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
164 extern int unaligned_enabled;
165 #endif
166
167 #ifdef CONFIG_IA64
168 extern int unaligned_dump_stack;
169 #endif
170
171 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
172 extern int no_unaligned_warning;
173 #endif
174
175 #ifdef CONFIG_PROC_SYSCTL
176
177 /**
178 * enum sysctl_writes_mode - supported sysctl write modes
179 *
180 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
181 * to be written, and multiple writes on the same sysctl file descriptor
182 * will rewrite the sysctl value, regardless of file position. No warning
183 * is issued when the initial position is not 0.
184 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
185 * not 0.
186 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
187 * file position 0 and the value must be fully contained in the buffer
188 * sent to the write syscall. If dealing with strings respect the file
189 * position, but restrict this to the max length of the buffer, anything
190 * passed the max lenght will be ignored. Multiple writes will append
191 * to the buffer.
192 *
193 * These write modes control how current file position affects the behavior of
194 * updating sysctl values through the proc interface on each write.
195 */
196 enum sysctl_writes_mode {
197 SYSCTL_WRITES_LEGACY = -1,
198 SYSCTL_WRITES_WARN = 0,
199 SYSCTL_WRITES_STRICT = 1,
200 };
201
202 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
203
204 static int proc_do_cad_pid(struct ctl_table *table, int write,
205 void __user *buffer, size_t *lenp, loff_t *ppos);
206 static int proc_taint(struct ctl_table *table, int write,
207 void __user *buffer, size_t *lenp, loff_t *ppos);
208 #endif
209
210 #ifdef CONFIG_PRINTK
211 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
212 void __user *buffer, size_t *lenp, loff_t *ppos);
213 #endif
214
215 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
216 void __user *buffer, size_t *lenp, loff_t *ppos);
217 #ifdef CONFIG_COREDUMP
218 static int proc_dostring_coredump(struct ctl_table *table, int write,
219 void __user *buffer, size_t *lenp, loff_t *ppos);
220 #endif
221 static int proc_dopipe_max_size(struct ctl_table *table, int write,
222 void __user *buffer, size_t *lenp, loff_t *ppos);
223
224 #ifdef CONFIG_MAGIC_SYSRQ
225 /* Note: sysrq code uses it's own private copy */
226 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
227
228 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
229 void __user *buffer, size_t *lenp,
230 loff_t *ppos)
231 {
232 int error;
233
234 error = proc_dointvec(table, write, buffer, lenp, ppos);
235 if (error)
236 return error;
237
238 if (write)
239 sysrq_toggle_support(__sysrq_enabled);
240
241 return 0;
242 }
243
244 #endif
245
246 static struct ctl_table kern_table[];
247 static struct ctl_table vm_table[];
248 static struct ctl_table fs_table[];
249 static struct ctl_table debug_table[];
250 static struct ctl_table dev_table[];
251 extern struct ctl_table random_table[];
252 #ifdef CONFIG_EPOLL
253 extern struct ctl_table epoll_table[];
254 #endif
255
256 #ifdef CONFIG_FW_LOADER_USER_HELPER
257 extern struct ctl_table firmware_config_table[];
258 #endif
259
260 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
261 int sysctl_legacy_va_layout;
262 #endif
263
264 /* The default sysctl tables: */
265
266 static struct ctl_table sysctl_base_table[] = {
267 {
268 .procname = "kernel",
269 .mode = 0555,
270 .child = kern_table,
271 },
272 {
273 .procname = "vm",
274 .mode = 0555,
275 .child = vm_table,
276 },
277 {
278 .procname = "fs",
279 .mode = 0555,
280 .child = fs_table,
281 },
282 {
283 .procname = "debug",
284 .mode = 0555,
285 .child = debug_table,
286 },
287 {
288 .procname = "dev",
289 .mode = 0555,
290 .child = dev_table,
291 },
292 { }
293 };
294
295 #ifdef CONFIG_SCHED_DEBUG
296 static int min_sched_granularity_ns = 100000; /* 100 usecs */
297 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
298 static int min_wakeup_granularity_ns; /* 0 usecs */
299 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
300 #ifdef CONFIG_SMP
301 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
302 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
303 #endif /* CONFIG_SMP */
304 #endif /* CONFIG_SCHED_DEBUG */
305
306 #ifdef CONFIG_COMPACTION
307 static int min_extfrag_threshold;
308 static int max_extfrag_threshold = 1000;
309 #endif
310
311 static struct ctl_table kern_table[] = {
312 {
313 .procname = "sched_child_runs_first",
314 .data = &sysctl_sched_child_runs_first,
315 .maxlen = sizeof(unsigned int),
316 .mode = 0644,
317 .proc_handler = proc_dointvec,
318 },
319 #ifdef CONFIG_SCHED_DEBUG
320 {
321 .procname = "sched_min_granularity_ns",
322 .data = &sysctl_sched_min_granularity,
323 .maxlen = sizeof(unsigned int),
324 .mode = 0644,
325 .proc_handler = sched_proc_update_handler,
326 .extra1 = &min_sched_granularity_ns,
327 .extra2 = &max_sched_granularity_ns,
328 },
329 {
330 .procname = "sched_latency_ns",
331 .data = &sysctl_sched_latency,
332 .maxlen = sizeof(unsigned int),
333 .mode = 0644,
334 .proc_handler = sched_proc_update_handler,
335 .extra1 = &min_sched_granularity_ns,
336 .extra2 = &max_sched_granularity_ns,
337 },
338 {
339 .procname = "sched_wakeup_granularity_ns",
340 .data = &sysctl_sched_wakeup_granularity,
341 .maxlen = sizeof(unsigned int),
342 .mode = 0644,
343 .proc_handler = sched_proc_update_handler,
344 .extra1 = &min_wakeup_granularity_ns,
345 .extra2 = &max_wakeup_granularity_ns,
346 },
347 #ifdef CONFIG_SMP
348 {
349 .procname = "sched_tunable_scaling",
350 .data = &sysctl_sched_tunable_scaling,
351 .maxlen = sizeof(enum sched_tunable_scaling),
352 .mode = 0644,
353 .proc_handler = sched_proc_update_handler,
354 .extra1 = &min_sched_tunable_scaling,
355 .extra2 = &max_sched_tunable_scaling,
356 },
357 {
358 .procname = "sched_migration_cost_ns",
359 .data = &sysctl_sched_migration_cost,
360 .maxlen = sizeof(unsigned int),
361 .mode = 0644,
362 .proc_handler = proc_dointvec,
363 },
364 {
365 .procname = "sched_nr_migrate",
366 .data = &sysctl_sched_nr_migrate,
367 .maxlen = sizeof(unsigned int),
368 .mode = 0644,
369 .proc_handler = proc_dointvec,
370 },
371 #ifdef CONFIG_SCHEDSTATS
372 {
373 .procname = "sched_schedstats",
374 .data = NULL,
375 .maxlen = sizeof(unsigned int),
376 .mode = 0644,
377 .proc_handler = sysctl_schedstats,
378 .extra1 = &zero,
379 .extra2 = &one,
380 },
381 #endif /* CONFIG_SCHEDSTATS */
382 #endif /* CONFIG_SMP */
383 #ifdef CONFIG_NUMA_BALANCING
384 {
385 .procname = "numa_balancing_scan_delay_ms",
386 .data = &sysctl_numa_balancing_scan_delay,
387 .maxlen = sizeof(unsigned int),
388 .mode = 0644,
389 .proc_handler = proc_dointvec,
390 },
391 {
392 .procname = "numa_balancing_scan_period_min_ms",
393 .data = &sysctl_numa_balancing_scan_period_min,
394 .maxlen = sizeof(unsigned int),
395 .mode = 0644,
396 .proc_handler = proc_dointvec,
397 },
398 {
399 .procname = "numa_balancing_scan_period_max_ms",
400 .data = &sysctl_numa_balancing_scan_period_max,
401 .maxlen = sizeof(unsigned int),
402 .mode = 0644,
403 .proc_handler = proc_dointvec,
404 },
405 {
406 .procname = "numa_balancing_scan_size_mb",
407 .data = &sysctl_numa_balancing_scan_size,
408 .maxlen = sizeof(unsigned int),
409 .mode = 0644,
410 .proc_handler = proc_dointvec_minmax,
411 .extra1 = &one,
412 },
413 {
414 .procname = "numa_balancing",
415 .data = NULL, /* filled in by handler */
416 .maxlen = sizeof(unsigned int),
417 .mode = 0644,
418 .proc_handler = sysctl_numa_balancing,
419 .extra1 = &zero,
420 .extra2 = &one,
421 },
422 #endif /* CONFIG_NUMA_BALANCING */
423 #endif /* CONFIG_SCHED_DEBUG */
424 {
425 .procname = "sched_rt_period_us",
426 .data = &sysctl_sched_rt_period,
427 .maxlen = sizeof(unsigned int),
428 .mode = 0644,
429 .proc_handler = sched_rt_handler,
430 },
431 {
432 .procname = "sched_rt_runtime_us",
433 .data = &sysctl_sched_rt_runtime,
434 .maxlen = sizeof(int),
435 .mode = 0644,
436 .proc_handler = sched_rt_handler,
437 },
438 {
439 .procname = "sched_rr_timeslice_ms",
440 .data = &sysctl_sched_rr_timeslice,
441 .maxlen = sizeof(int),
442 .mode = 0644,
443 .proc_handler = sched_rr_handler,
444 },
445 #ifdef CONFIG_SCHED_AUTOGROUP
446 {
447 .procname = "sched_autogroup_enabled",
448 .data = &sysctl_sched_autogroup_enabled,
449 .maxlen = sizeof(unsigned int),
450 .mode = 0644,
451 .proc_handler = proc_dointvec_minmax,
452 .extra1 = &zero,
453 .extra2 = &one,
454 },
455 #endif
456 #ifdef CONFIG_CFS_BANDWIDTH
457 {
458 .procname = "sched_cfs_bandwidth_slice_us",
459 .data = &sysctl_sched_cfs_bandwidth_slice,
460 .maxlen = sizeof(unsigned int),
461 .mode = 0644,
462 .proc_handler = proc_dointvec_minmax,
463 .extra1 = &one,
464 },
465 #endif
466 #ifdef CONFIG_PROVE_LOCKING
467 {
468 .procname = "prove_locking",
469 .data = &prove_locking,
470 .maxlen = sizeof(int),
471 .mode = 0644,
472 .proc_handler = proc_dointvec,
473 },
474 #endif
475 #ifdef CONFIG_LOCK_STAT
476 {
477 .procname = "lock_stat",
478 .data = &lock_stat,
479 .maxlen = sizeof(int),
480 .mode = 0644,
481 .proc_handler = proc_dointvec,
482 },
483 #endif
484 {
485 .procname = "panic",
486 .data = &panic_timeout,
487 .maxlen = sizeof(int),
488 .mode = 0644,
489 .proc_handler = proc_dointvec,
490 },
491 #ifdef CONFIG_COREDUMP
492 {
493 .procname = "core_uses_pid",
494 .data = &core_uses_pid,
495 .maxlen = sizeof(int),
496 .mode = 0644,
497 .proc_handler = proc_dointvec,
498 },
499 {
500 .procname = "core_pattern",
501 .data = core_pattern,
502 .maxlen = CORENAME_MAX_SIZE,
503 .mode = 0644,
504 .proc_handler = proc_dostring_coredump,
505 },
506 {
507 .procname = "core_pipe_limit",
508 .data = &core_pipe_limit,
509 .maxlen = sizeof(unsigned int),
510 .mode = 0644,
511 .proc_handler = proc_dointvec,
512 },
513 #endif
514 #ifdef CONFIG_PROC_SYSCTL
515 {
516 .procname = "tainted",
517 .maxlen = sizeof(long),
518 .mode = 0644,
519 .proc_handler = proc_taint,
520 },
521 {
522 .procname = "sysctl_writes_strict",
523 .data = &sysctl_writes_strict,
524 .maxlen = sizeof(int),
525 .mode = 0644,
526 .proc_handler = proc_dointvec_minmax,
527 .extra1 = &neg_one,
528 .extra2 = &one,
529 },
530 #endif
531 #ifdef CONFIG_LATENCYTOP
532 {
533 .procname = "latencytop",
534 .data = &latencytop_enabled,
535 .maxlen = sizeof(int),
536 .mode = 0644,
537 .proc_handler = sysctl_latencytop,
538 },
539 #endif
540 #ifdef CONFIG_BLK_DEV_INITRD
541 {
542 .procname = "real-root-dev",
543 .data = &real_root_dev,
544 .maxlen = sizeof(int),
545 .mode = 0644,
546 .proc_handler = proc_dointvec,
547 },
548 #endif
549 {
550 .procname = "print-fatal-signals",
551 .data = &print_fatal_signals,
552 .maxlen = sizeof(int),
553 .mode = 0644,
554 .proc_handler = proc_dointvec,
555 },
556 #ifdef CONFIG_SPARC
557 {
558 .procname = "reboot-cmd",
559 .data = reboot_command,
560 .maxlen = 256,
561 .mode = 0644,
562 .proc_handler = proc_dostring,
563 },
564 {
565 .procname = "stop-a",
566 .data = &stop_a_enabled,
567 .maxlen = sizeof (int),
568 .mode = 0644,
569 .proc_handler = proc_dointvec,
570 },
571 {
572 .procname = "scons-poweroff",
573 .data = &scons_pwroff,
574 .maxlen = sizeof (int),
575 .mode = 0644,
576 .proc_handler = proc_dointvec,
577 },
578 #endif
579 #ifdef CONFIG_SPARC64
580 {
581 .procname = "tsb-ratio",
582 .data = &sysctl_tsb_ratio,
583 .maxlen = sizeof (int),
584 .mode = 0644,
585 .proc_handler = proc_dointvec,
586 },
587 #endif
588 #ifdef __hppa__
589 {
590 .procname = "soft-power",
591 .data = &pwrsw_enabled,
592 .maxlen = sizeof (int),
593 .mode = 0644,
594 .proc_handler = proc_dointvec,
595 },
596 #endif
597 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
598 {
599 .procname = "unaligned-trap",
600 .data = &unaligned_enabled,
601 .maxlen = sizeof (int),
602 .mode = 0644,
603 .proc_handler = proc_dointvec,
604 },
605 #endif
606 {
607 .procname = "ctrl-alt-del",
608 .data = &C_A_D,
609 .maxlen = sizeof(int),
610 .mode = 0644,
611 .proc_handler = proc_dointvec,
612 },
613 #ifdef CONFIG_FUNCTION_TRACER
614 {
615 .procname = "ftrace_enabled",
616 .data = &ftrace_enabled,
617 .maxlen = sizeof(int),
618 .mode = 0644,
619 .proc_handler = ftrace_enable_sysctl,
620 },
621 #endif
622 #ifdef CONFIG_STACK_TRACER
623 {
624 .procname = "stack_tracer_enabled",
625 .data = &stack_tracer_enabled,
626 .maxlen = sizeof(int),
627 .mode = 0644,
628 .proc_handler = stack_trace_sysctl,
629 },
630 #endif
631 #ifdef CONFIG_TRACING
632 {
633 .procname = "ftrace_dump_on_oops",
634 .data = &ftrace_dump_on_oops,
635 .maxlen = sizeof(int),
636 .mode = 0644,
637 .proc_handler = proc_dointvec,
638 },
639 {
640 .procname = "traceoff_on_warning",
641 .data = &__disable_trace_on_warning,
642 .maxlen = sizeof(__disable_trace_on_warning),
643 .mode = 0644,
644 .proc_handler = proc_dointvec,
645 },
646 {
647 .procname = "tracepoint_printk",
648 .data = &tracepoint_printk,
649 .maxlen = sizeof(tracepoint_printk),
650 .mode = 0644,
651 .proc_handler = tracepoint_printk_sysctl,
652 },
653 #endif
654 #ifdef CONFIG_KEXEC_CORE
655 {
656 .procname = "kexec_load_disabled",
657 .data = &kexec_load_disabled,
658 .maxlen = sizeof(int),
659 .mode = 0644,
660 /* only handle a transition from default "0" to "1" */
661 .proc_handler = proc_dointvec_minmax,
662 .extra1 = &one,
663 .extra2 = &one,
664 },
665 #endif
666 #ifdef CONFIG_MODULES
667 {
668 .procname = "modprobe",
669 .data = &modprobe_path,
670 .maxlen = KMOD_PATH_LEN,
671 .mode = 0644,
672 .proc_handler = proc_dostring,
673 },
674 {
675 .procname = "modules_disabled",
676 .data = &modules_disabled,
677 .maxlen = sizeof(int),
678 .mode = 0644,
679 /* only handle a transition from default "0" to "1" */
680 .proc_handler = proc_dointvec_minmax,
681 .extra1 = &one,
682 .extra2 = &one,
683 },
684 #endif
685 #ifdef CONFIG_UEVENT_HELPER
686 {
687 .procname = "hotplug",
688 .data = &uevent_helper,
689 .maxlen = UEVENT_HELPER_PATH_LEN,
690 .mode = 0644,
691 .proc_handler = proc_dostring,
692 },
693 #endif
694 #ifdef CONFIG_CHR_DEV_SG
695 {
696 .procname = "sg-big-buff",
697 .data = &sg_big_buff,
698 .maxlen = sizeof (int),
699 .mode = 0444,
700 .proc_handler = proc_dointvec,
701 },
702 #endif
703 #ifdef CONFIG_BSD_PROCESS_ACCT
704 {
705 .procname = "acct",
706 .data = &acct_parm,
707 .maxlen = 3*sizeof(int),
708 .mode = 0644,
709 .proc_handler = proc_dointvec,
710 },
711 #endif
712 #ifdef CONFIG_MAGIC_SYSRQ
713 {
714 .procname = "sysrq",
715 .data = &__sysrq_enabled,
716 .maxlen = sizeof (int),
717 .mode = 0644,
718 .proc_handler = sysrq_sysctl_handler,
719 },
720 #endif
721 #ifdef CONFIG_PROC_SYSCTL
722 {
723 .procname = "cad_pid",
724 .data = NULL,
725 .maxlen = sizeof (int),
726 .mode = 0600,
727 .proc_handler = proc_do_cad_pid,
728 },
729 #endif
730 {
731 .procname = "threads-max",
732 .data = NULL,
733 .maxlen = sizeof(int),
734 .mode = 0644,
735 .proc_handler = sysctl_max_threads,
736 },
737 {
738 .procname = "random",
739 .mode = 0555,
740 .child = random_table,
741 },
742 {
743 .procname = "usermodehelper",
744 .mode = 0555,
745 .child = usermodehelper_table,
746 },
747 #ifdef CONFIG_FW_LOADER_USER_HELPER
748 {
749 .procname = "firmware_config",
750 .mode = 0555,
751 .child = firmware_config_table,
752 },
753 #endif
754 {
755 .procname = "overflowuid",
756 .data = &overflowuid,
757 .maxlen = sizeof(int),
758 .mode = 0644,
759 .proc_handler = proc_dointvec_minmax,
760 .extra1 = &minolduid,
761 .extra2 = &maxolduid,
762 },
763 {
764 .procname = "overflowgid",
765 .data = &overflowgid,
766 .maxlen = sizeof(int),
767 .mode = 0644,
768 .proc_handler = proc_dointvec_minmax,
769 .extra1 = &minolduid,
770 .extra2 = &maxolduid,
771 },
772 #ifdef CONFIG_S390
773 #ifdef CONFIG_MATHEMU
774 {
775 .procname = "ieee_emulation_warnings",
776 .data = &sysctl_ieee_emulation_warnings,
777 .maxlen = sizeof(int),
778 .mode = 0644,
779 .proc_handler = proc_dointvec,
780 },
781 #endif
782 {
783 .procname = "userprocess_debug",
784 .data = &show_unhandled_signals,
785 .maxlen = sizeof(int),
786 .mode = 0644,
787 .proc_handler = proc_dointvec,
788 },
789 #endif
790 {
791 .procname = "pid_max",
792 .data = &pid_max,
793 .maxlen = sizeof (int),
794 .mode = 0644,
795 .proc_handler = proc_dointvec_minmax,
796 .extra1 = &pid_max_min,
797 .extra2 = &pid_max_max,
798 },
799 {
800 .procname = "panic_on_oops",
801 .data = &panic_on_oops,
802 .maxlen = sizeof(int),
803 .mode = 0644,
804 .proc_handler = proc_dointvec,
805 },
806 #if defined CONFIG_PRINTK
807 {
808 .procname = "printk",
809 .data = &console_loglevel,
810 .maxlen = 4*sizeof(int),
811 .mode = 0644,
812 .proc_handler = proc_dointvec,
813 },
814 {
815 .procname = "printk_ratelimit",
816 .data = &printk_ratelimit_state.interval,
817 .maxlen = sizeof(int),
818 .mode = 0644,
819 .proc_handler = proc_dointvec_jiffies,
820 },
821 {
822 .procname = "printk_ratelimit_burst",
823 .data = &printk_ratelimit_state.burst,
824 .maxlen = sizeof(int),
825 .mode = 0644,
826 .proc_handler = proc_dointvec,
827 },
828 {
829 .procname = "printk_delay",
830 .data = &printk_delay_msec,
831 .maxlen = sizeof(int),
832 .mode = 0644,
833 .proc_handler = proc_dointvec_minmax,
834 .extra1 = &zero,
835 .extra2 = &ten_thousand,
836 },
837 {
838 .procname = "printk_devkmsg",
839 .data = devkmsg_log_str,
840 .maxlen = DEVKMSG_STR_MAX_SIZE,
841 .mode = 0644,
842 .proc_handler = devkmsg_sysctl_set_loglvl,
843 },
844 {
845 .procname = "dmesg_restrict",
846 .data = &dmesg_restrict,
847 .maxlen = sizeof(int),
848 .mode = 0644,
849 .proc_handler = proc_dointvec_minmax_sysadmin,
850 .extra1 = &zero,
851 .extra2 = &one,
852 },
853 {
854 .procname = "kptr_restrict",
855 .data = &kptr_restrict,
856 .maxlen = sizeof(int),
857 .mode = 0644,
858 .proc_handler = proc_dointvec_minmax_sysadmin,
859 .extra1 = &zero,
860 .extra2 = &two,
861 },
862 #endif
863 {
864 .procname = "ngroups_max",
865 .data = &ngroups_max,
866 .maxlen = sizeof (int),
867 .mode = 0444,
868 .proc_handler = proc_dointvec,
869 },
870 {
871 .procname = "cap_last_cap",
872 .data = (void *)&cap_last_cap,
873 .maxlen = sizeof(int),
874 .mode = 0444,
875 .proc_handler = proc_dointvec,
876 },
877 #if defined(CONFIG_LOCKUP_DETECTOR)
878 {
879 .procname = "watchdog",
880 .data = &watchdog_user_enabled,
881 .maxlen = sizeof(int),
882 .mode = 0644,
883 .proc_handler = proc_watchdog,
884 .extra1 = &zero,
885 .extra2 = &one,
886 },
887 {
888 .procname = "watchdog_thresh",
889 .data = &watchdog_thresh,
890 .maxlen = sizeof(int),
891 .mode = 0644,
892 .proc_handler = proc_watchdog_thresh,
893 .extra1 = &zero,
894 .extra2 = &sixty,
895 },
896 {
897 .procname = "nmi_watchdog",
898 .data = &nmi_watchdog_user_enabled,
899 .maxlen = sizeof(int),
900 .mode = NMI_WATCHDOG_SYSCTL_PERM,
901 .proc_handler = proc_nmi_watchdog,
902 .extra1 = &zero,
903 .extra2 = &one,
904 },
905 {
906 .procname = "watchdog_cpumask",
907 .data = &watchdog_cpumask_bits,
908 .maxlen = NR_CPUS,
909 .mode = 0644,
910 .proc_handler = proc_watchdog_cpumask,
911 },
912 #ifdef CONFIG_SOFTLOCKUP_DETECTOR
913 {
914 .procname = "soft_watchdog",
915 .data = &soft_watchdog_user_enabled,
916 .maxlen = sizeof(int),
917 .mode = 0644,
918 .proc_handler = proc_soft_watchdog,
919 .extra1 = &zero,
920 .extra2 = &one,
921 },
922 {
923 .procname = "softlockup_panic",
924 .data = &softlockup_panic,
925 .maxlen = sizeof(int),
926 .mode = 0644,
927 .proc_handler = proc_dointvec_minmax,
928 .extra1 = &zero,
929 .extra2 = &one,
930 },
931 #ifdef CONFIG_SMP
932 {
933 .procname = "softlockup_all_cpu_backtrace",
934 .data = &sysctl_softlockup_all_cpu_backtrace,
935 .maxlen = sizeof(int),
936 .mode = 0644,
937 .proc_handler = proc_dointvec_minmax,
938 .extra1 = &zero,
939 .extra2 = &one,
940 },
941 #endif /* CONFIG_SMP */
942 #endif
943 #ifdef CONFIG_HARDLOCKUP_DETECTOR
944 {
945 .procname = "hardlockup_panic",
946 .data = &hardlockup_panic,
947 .maxlen = sizeof(int),
948 .mode = 0644,
949 .proc_handler = proc_dointvec_minmax,
950 .extra1 = &zero,
951 .extra2 = &one,
952 },
953 #ifdef CONFIG_SMP
954 {
955 .procname = "hardlockup_all_cpu_backtrace",
956 .data = &sysctl_hardlockup_all_cpu_backtrace,
957 .maxlen = sizeof(int),
958 .mode = 0644,
959 .proc_handler = proc_dointvec_minmax,
960 .extra1 = &zero,
961 .extra2 = &one,
962 },
963 #endif /* CONFIG_SMP */
964 #endif
965 #endif
966
967 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
968 {
969 .procname = "unknown_nmi_panic",
970 .data = &unknown_nmi_panic,
971 .maxlen = sizeof (int),
972 .mode = 0644,
973 .proc_handler = proc_dointvec,
974 },
975 #endif
976 #if defined(CONFIG_X86)
977 {
978 .procname = "panic_on_unrecovered_nmi",
979 .data = &panic_on_unrecovered_nmi,
980 .maxlen = sizeof(int),
981 .mode = 0644,
982 .proc_handler = proc_dointvec,
983 },
984 {
985 .procname = "panic_on_io_nmi",
986 .data = &panic_on_io_nmi,
987 .maxlen = sizeof(int),
988 .mode = 0644,
989 .proc_handler = proc_dointvec,
990 },
991 #ifdef CONFIG_DEBUG_STACKOVERFLOW
992 {
993 .procname = "panic_on_stackoverflow",
994 .data = &sysctl_panic_on_stackoverflow,
995 .maxlen = sizeof(int),
996 .mode = 0644,
997 .proc_handler = proc_dointvec,
998 },
999 #endif
1000 {
1001 .procname = "bootloader_type",
1002 .data = &bootloader_type,
1003 .maxlen = sizeof (int),
1004 .mode = 0444,
1005 .proc_handler = proc_dointvec,
1006 },
1007 {
1008 .procname = "bootloader_version",
1009 .data = &bootloader_version,
1010 .maxlen = sizeof (int),
1011 .mode = 0444,
1012 .proc_handler = proc_dointvec,
1013 },
1014 {
1015 .procname = "io_delay_type",
1016 .data = &io_delay_type,
1017 .maxlen = sizeof(int),
1018 .mode = 0644,
1019 .proc_handler = proc_dointvec,
1020 },
1021 #endif
1022 #if defined(CONFIG_MMU)
1023 {
1024 .procname = "randomize_va_space",
1025 .data = &randomize_va_space,
1026 .maxlen = sizeof(int),
1027 .mode = 0644,
1028 .proc_handler = proc_dointvec,
1029 },
1030 #endif
1031 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1032 {
1033 .procname = "spin_retry",
1034 .data = &spin_retry,
1035 .maxlen = sizeof (int),
1036 .mode = 0644,
1037 .proc_handler = proc_dointvec,
1038 },
1039 #endif
1040 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1041 {
1042 .procname = "acpi_video_flags",
1043 .data = &acpi_realmode_flags,
1044 .maxlen = sizeof (unsigned long),
1045 .mode = 0644,
1046 .proc_handler = proc_doulongvec_minmax,
1047 },
1048 #endif
1049 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1050 {
1051 .procname = "ignore-unaligned-usertrap",
1052 .data = &no_unaligned_warning,
1053 .maxlen = sizeof (int),
1054 .mode = 0644,
1055 .proc_handler = proc_dointvec,
1056 },
1057 #endif
1058 #ifdef CONFIG_IA64
1059 {
1060 .procname = "unaligned-dump-stack",
1061 .data = &unaligned_dump_stack,
1062 .maxlen = sizeof (int),
1063 .mode = 0644,
1064 .proc_handler = proc_dointvec,
1065 },
1066 #endif
1067 #ifdef CONFIG_DETECT_HUNG_TASK
1068 {
1069 .procname = "hung_task_panic",
1070 .data = &sysctl_hung_task_panic,
1071 .maxlen = sizeof(int),
1072 .mode = 0644,
1073 .proc_handler = proc_dointvec_minmax,
1074 .extra1 = &zero,
1075 .extra2 = &one,
1076 },
1077 {
1078 .procname = "hung_task_check_count",
1079 .data = &sysctl_hung_task_check_count,
1080 .maxlen = sizeof(int),
1081 .mode = 0644,
1082 .proc_handler = proc_dointvec_minmax,
1083 .extra1 = &zero,
1084 },
1085 {
1086 .procname = "hung_task_timeout_secs",
1087 .data = &sysctl_hung_task_timeout_secs,
1088 .maxlen = sizeof(unsigned long),
1089 .mode = 0644,
1090 .proc_handler = proc_dohung_task_timeout_secs,
1091 .extra2 = &hung_task_timeout_max,
1092 },
1093 {
1094 .procname = "hung_task_warnings",
1095 .data = &sysctl_hung_task_warnings,
1096 .maxlen = sizeof(int),
1097 .mode = 0644,
1098 .proc_handler = proc_dointvec_minmax,
1099 .extra1 = &neg_one,
1100 },
1101 #endif
1102 #ifdef CONFIG_RT_MUTEXES
1103 {
1104 .procname = "max_lock_depth",
1105 .data = &max_lock_depth,
1106 .maxlen = sizeof(int),
1107 .mode = 0644,
1108 .proc_handler = proc_dointvec,
1109 },
1110 #endif
1111 {
1112 .procname = "poweroff_cmd",
1113 .data = &poweroff_cmd,
1114 .maxlen = POWEROFF_CMD_PATH_LEN,
1115 .mode = 0644,
1116 .proc_handler = proc_dostring,
1117 },
1118 #ifdef CONFIG_KEYS
1119 {
1120 .procname = "keys",
1121 .mode = 0555,
1122 .child = key_sysctls,
1123 },
1124 #endif
1125 #ifdef CONFIG_PERF_EVENTS
1126 /*
1127 * User-space scripts rely on the existence of this file
1128 * as a feature check for perf_events being enabled.
1129 *
1130 * So it's an ABI, do not remove!
1131 */
1132 {
1133 .procname = "perf_event_paranoid",
1134 .data = &sysctl_perf_event_paranoid,
1135 .maxlen = sizeof(sysctl_perf_event_paranoid),
1136 .mode = 0644,
1137 .proc_handler = proc_dointvec,
1138 },
1139 {
1140 .procname = "perf_event_mlock_kb",
1141 .data = &sysctl_perf_event_mlock,
1142 .maxlen = sizeof(sysctl_perf_event_mlock),
1143 .mode = 0644,
1144 .proc_handler = proc_dointvec,
1145 },
1146 {
1147 .procname = "perf_event_max_sample_rate",
1148 .data = &sysctl_perf_event_sample_rate,
1149 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1150 .mode = 0644,
1151 .proc_handler = perf_proc_update_handler,
1152 .extra1 = &one,
1153 },
1154 {
1155 .procname = "perf_cpu_time_max_percent",
1156 .data = &sysctl_perf_cpu_time_max_percent,
1157 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1158 .mode = 0644,
1159 .proc_handler = perf_cpu_time_max_percent_handler,
1160 .extra1 = &zero,
1161 .extra2 = &one_hundred,
1162 },
1163 {
1164 .procname = "perf_event_max_stack",
1165 .data = &sysctl_perf_event_max_stack,
1166 .maxlen = sizeof(sysctl_perf_event_max_stack),
1167 .mode = 0644,
1168 .proc_handler = perf_event_max_stack_handler,
1169 .extra1 = &zero,
1170 .extra2 = &six_hundred_forty_kb,
1171 },
1172 {
1173 .procname = "perf_event_max_contexts_per_stack",
1174 .data = &sysctl_perf_event_max_contexts_per_stack,
1175 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
1176 .mode = 0644,
1177 .proc_handler = perf_event_max_stack_handler,
1178 .extra1 = &zero,
1179 .extra2 = &one_thousand,
1180 },
1181 #endif
1182 {
1183 .procname = "panic_on_warn",
1184 .data = &panic_on_warn,
1185 .maxlen = sizeof(int),
1186 .mode = 0644,
1187 .proc_handler = proc_dointvec_minmax,
1188 .extra1 = &zero,
1189 .extra2 = &one,
1190 },
1191 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1192 {
1193 .procname = "timer_migration",
1194 .data = &sysctl_timer_migration,
1195 .maxlen = sizeof(unsigned int),
1196 .mode = 0644,
1197 .proc_handler = timer_migration_handler,
1198 .extra1 = &zero,
1199 .extra2 = &one,
1200 },
1201 #endif
1202 #ifdef CONFIG_BPF_SYSCALL
1203 {
1204 .procname = "unprivileged_bpf_disabled",
1205 .data = &sysctl_unprivileged_bpf_disabled,
1206 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
1207 .mode = 0644,
1208 /* only handle a transition from default "0" to "1" */
1209 .proc_handler = proc_dointvec_minmax,
1210 .extra1 = &one,
1211 .extra2 = &one,
1212 },
1213 #endif
1214 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1215 {
1216 .procname = "panic_on_rcu_stall",
1217 .data = &sysctl_panic_on_rcu_stall,
1218 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
1219 .mode = 0644,
1220 .proc_handler = proc_dointvec_minmax,
1221 .extra1 = &zero,
1222 .extra2 = &one,
1223 },
1224 #endif
1225 { }
1226 };
1227
1228 static struct ctl_table vm_table[] = {
1229 {
1230 .procname = "overcommit_memory",
1231 .data = &sysctl_overcommit_memory,
1232 .maxlen = sizeof(sysctl_overcommit_memory),
1233 .mode = 0644,
1234 .proc_handler = proc_dointvec_minmax,
1235 .extra1 = &zero,
1236 .extra2 = &two,
1237 },
1238 {
1239 .procname = "panic_on_oom",
1240 .data = &sysctl_panic_on_oom,
1241 .maxlen = sizeof(sysctl_panic_on_oom),
1242 .mode = 0644,
1243 .proc_handler = proc_dointvec_minmax,
1244 .extra1 = &zero,
1245 .extra2 = &two,
1246 },
1247 {
1248 .procname = "oom_kill_allocating_task",
1249 .data = &sysctl_oom_kill_allocating_task,
1250 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1251 .mode = 0644,
1252 .proc_handler = proc_dointvec,
1253 },
1254 {
1255 .procname = "oom_dump_tasks",
1256 .data = &sysctl_oom_dump_tasks,
1257 .maxlen = sizeof(sysctl_oom_dump_tasks),
1258 .mode = 0644,
1259 .proc_handler = proc_dointvec,
1260 },
1261 {
1262 .procname = "overcommit_ratio",
1263 .data = &sysctl_overcommit_ratio,
1264 .maxlen = sizeof(sysctl_overcommit_ratio),
1265 .mode = 0644,
1266 .proc_handler = overcommit_ratio_handler,
1267 },
1268 {
1269 .procname = "overcommit_kbytes",
1270 .data = &sysctl_overcommit_kbytes,
1271 .maxlen = sizeof(sysctl_overcommit_kbytes),
1272 .mode = 0644,
1273 .proc_handler = overcommit_kbytes_handler,
1274 },
1275 {
1276 .procname = "page-cluster",
1277 .data = &page_cluster,
1278 .maxlen = sizeof(int),
1279 .mode = 0644,
1280 .proc_handler = proc_dointvec_minmax,
1281 .extra1 = &zero,
1282 },
1283 {
1284 .procname = "dirty_background_ratio",
1285 .data = &dirty_background_ratio,
1286 .maxlen = sizeof(dirty_background_ratio),
1287 .mode = 0644,
1288 .proc_handler = dirty_background_ratio_handler,
1289 .extra1 = &zero,
1290 .extra2 = &one_hundred,
1291 },
1292 {
1293 .procname = "dirty_background_bytes",
1294 .data = &dirty_background_bytes,
1295 .maxlen = sizeof(dirty_background_bytes),
1296 .mode = 0644,
1297 .proc_handler = dirty_background_bytes_handler,
1298 .extra1 = &one_ul,
1299 },
1300 {
1301 .procname = "dirty_ratio",
1302 .data = &vm_dirty_ratio,
1303 .maxlen = sizeof(vm_dirty_ratio),
1304 .mode = 0644,
1305 .proc_handler = dirty_ratio_handler,
1306 .extra1 = &zero,
1307 .extra2 = &one_hundred,
1308 },
1309 {
1310 .procname = "dirty_bytes",
1311 .data = &vm_dirty_bytes,
1312 .maxlen = sizeof(vm_dirty_bytes),
1313 .mode = 0644,
1314 .proc_handler = dirty_bytes_handler,
1315 .extra1 = &dirty_bytes_min,
1316 },
1317 {
1318 .procname = "dirty_writeback_centisecs",
1319 .data = &dirty_writeback_interval,
1320 .maxlen = sizeof(dirty_writeback_interval),
1321 .mode = 0644,
1322 .proc_handler = dirty_writeback_centisecs_handler,
1323 },
1324 {
1325 .procname = "dirty_expire_centisecs",
1326 .data = &dirty_expire_interval,
1327 .maxlen = sizeof(dirty_expire_interval),
1328 .mode = 0644,
1329 .proc_handler = proc_dointvec_minmax,
1330 .extra1 = &zero,
1331 },
1332 {
1333 .procname = "dirtytime_expire_seconds",
1334 .data = &dirtytime_expire_interval,
1335 .maxlen = sizeof(dirtytime_expire_interval),
1336 .mode = 0644,
1337 .proc_handler = dirtytime_interval_handler,
1338 .extra1 = &zero,
1339 },
1340 {
1341 .procname = "swappiness",
1342 .data = &vm_swappiness,
1343 .maxlen = sizeof(vm_swappiness),
1344 .mode = 0644,
1345 .proc_handler = proc_dointvec_minmax,
1346 .extra1 = &zero,
1347 .extra2 = &one_hundred,
1348 },
1349 #ifdef CONFIG_HUGETLB_PAGE
1350 {
1351 .procname = "nr_hugepages",
1352 .data = NULL,
1353 .maxlen = sizeof(unsigned long),
1354 .mode = 0644,
1355 .proc_handler = hugetlb_sysctl_handler,
1356 },
1357 #ifdef CONFIG_NUMA
1358 {
1359 .procname = "nr_hugepages_mempolicy",
1360 .data = NULL,
1361 .maxlen = sizeof(unsigned long),
1362 .mode = 0644,
1363 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1364 },
1365 {
1366 .procname = "numa_stat",
1367 .data = &sysctl_vm_numa_stat,
1368 .maxlen = sizeof(int),
1369 .mode = 0644,
1370 .proc_handler = sysctl_vm_numa_stat_handler,
1371 .extra1 = &zero,
1372 .extra2 = &one,
1373 },
1374 #endif
1375 {
1376 .procname = "hugetlb_shm_group",
1377 .data = &sysctl_hugetlb_shm_group,
1378 .maxlen = sizeof(gid_t),
1379 .mode = 0644,
1380 .proc_handler = proc_dointvec,
1381 },
1382 {
1383 .procname = "nr_overcommit_hugepages",
1384 .data = NULL,
1385 .maxlen = sizeof(unsigned long),
1386 .mode = 0644,
1387 .proc_handler = hugetlb_overcommit_handler,
1388 },
1389 #endif
1390 {
1391 .procname = "lowmem_reserve_ratio",
1392 .data = &sysctl_lowmem_reserve_ratio,
1393 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1394 .mode = 0644,
1395 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1396 },
1397 {
1398 .procname = "drop_caches",
1399 .data = &sysctl_drop_caches,
1400 .maxlen = sizeof(int),
1401 .mode = 0644,
1402 .proc_handler = drop_caches_sysctl_handler,
1403 .extra1 = &one,
1404 .extra2 = &four,
1405 },
1406 #ifdef CONFIG_COMPACTION
1407 {
1408 .procname = "compact_memory",
1409 .data = &sysctl_compact_memory,
1410 .maxlen = sizeof(int),
1411 .mode = 0200,
1412 .proc_handler = sysctl_compaction_handler,
1413 },
1414 {
1415 .procname = "extfrag_threshold",
1416 .data = &sysctl_extfrag_threshold,
1417 .maxlen = sizeof(int),
1418 .mode = 0644,
1419 .proc_handler = sysctl_extfrag_handler,
1420 .extra1 = &min_extfrag_threshold,
1421 .extra2 = &max_extfrag_threshold,
1422 },
1423 {
1424 .procname = "compact_unevictable_allowed",
1425 .data = &sysctl_compact_unevictable_allowed,
1426 .maxlen = sizeof(int),
1427 .mode = 0644,
1428 .proc_handler = proc_dointvec,
1429 .extra1 = &zero,
1430 .extra2 = &one,
1431 },
1432
1433 #endif /* CONFIG_COMPACTION */
1434 {
1435 .procname = "min_free_kbytes",
1436 .data = &min_free_kbytes,
1437 .maxlen = sizeof(min_free_kbytes),
1438 .mode = 0644,
1439 .proc_handler = min_free_kbytes_sysctl_handler,
1440 .extra1 = &zero,
1441 },
1442 {
1443 .procname = "watermark_scale_factor",
1444 .data = &watermark_scale_factor,
1445 .maxlen = sizeof(watermark_scale_factor),
1446 .mode = 0644,
1447 .proc_handler = watermark_scale_factor_sysctl_handler,
1448 .extra1 = &one,
1449 .extra2 = &one_thousand,
1450 },
1451 {
1452 .procname = "percpu_pagelist_fraction",
1453 .data = &percpu_pagelist_fraction,
1454 .maxlen = sizeof(percpu_pagelist_fraction),
1455 .mode = 0644,
1456 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1457 .extra1 = &zero,
1458 },
1459 #ifdef CONFIG_MMU
1460 {
1461 .procname = "max_map_count",
1462 .data = &sysctl_max_map_count,
1463 .maxlen = sizeof(sysctl_max_map_count),
1464 .mode = 0644,
1465 .proc_handler = proc_dointvec_minmax,
1466 .extra1 = &zero,
1467 },
1468 #else
1469 {
1470 .procname = "nr_trim_pages",
1471 .data = &sysctl_nr_trim_pages,
1472 .maxlen = sizeof(sysctl_nr_trim_pages),
1473 .mode = 0644,
1474 .proc_handler = proc_dointvec_minmax,
1475 .extra1 = &zero,
1476 },
1477 #endif
1478 {
1479 .procname = "laptop_mode",
1480 .data = &laptop_mode,
1481 .maxlen = sizeof(laptop_mode),
1482 .mode = 0644,
1483 .proc_handler = proc_dointvec_jiffies,
1484 },
1485 {
1486 .procname = "block_dump",
1487 .data = &block_dump,
1488 .maxlen = sizeof(block_dump),
1489 .mode = 0644,
1490 .proc_handler = proc_dointvec,
1491 .extra1 = &zero,
1492 },
1493 {
1494 .procname = "vfs_cache_pressure",
1495 .data = &sysctl_vfs_cache_pressure,
1496 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1497 .mode = 0644,
1498 .proc_handler = proc_dointvec,
1499 .extra1 = &zero,
1500 },
1501 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1502 {
1503 .procname = "legacy_va_layout",
1504 .data = &sysctl_legacy_va_layout,
1505 .maxlen = sizeof(sysctl_legacy_va_layout),
1506 .mode = 0644,
1507 .proc_handler = proc_dointvec,
1508 .extra1 = &zero,
1509 },
1510 #endif
1511 #ifdef CONFIG_NUMA
1512 {
1513 .procname = "zone_reclaim_mode",
1514 .data = &node_reclaim_mode,
1515 .maxlen = sizeof(node_reclaim_mode),
1516 .mode = 0644,
1517 .proc_handler = proc_dointvec,
1518 .extra1 = &zero,
1519 },
1520 {
1521 .procname = "min_unmapped_ratio",
1522 .data = &sysctl_min_unmapped_ratio,
1523 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1524 .mode = 0644,
1525 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1526 .extra1 = &zero,
1527 .extra2 = &one_hundred,
1528 },
1529 {
1530 .procname = "min_slab_ratio",
1531 .data = &sysctl_min_slab_ratio,
1532 .maxlen = sizeof(sysctl_min_slab_ratio),
1533 .mode = 0644,
1534 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1535 .extra1 = &zero,
1536 .extra2 = &one_hundred,
1537 },
1538 #endif
1539 #ifdef CONFIG_SMP
1540 {
1541 .procname = "stat_interval",
1542 .data = &sysctl_stat_interval,
1543 .maxlen = sizeof(sysctl_stat_interval),
1544 .mode = 0644,
1545 .proc_handler = proc_dointvec_jiffies,
1546 },
1547 {
1548 .procname = "stat_refresh",
1549 .data = NULL,
1550 .maxlen = 0,
1551 .mode = 0600,
1552 .proc_handler = vmstat_refresh,
1553 },
1554 #endif
1555 #ifdef CONFIG_MMU
1556 {
1557 .procname = "mmap_min_addr",
1558 .data = &dac_mmap_min_addr,
1559 .maxlen = sizeof(unsigned long),
1560 .mode = 0644,
1561 .proc_handler = mmap_min_addr_handler,
1562 },
1563 #endif
1564 #ifdef CONFIG_NUMA
1565 {
1566 .procname = "numa_zonelist_order",
1567 .data = &numa_zonelist_order,
1568 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1569 .mode = 0644,
1570 .proc_handler = numa_zonelist_order_handler,
1571 },
1572 #endif
1573 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1574 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1575 {
1576 .procname = "vdso_enabled",
1577 #ifdef CONFIG_X86_32
1578 .data = &vdso32_enabled,
1579 .maxlen = sizeof(vdso32_enabled),
1580 #else
1581 .data = &vdso_enabled,
1582 .maxlen = sizeof(vdso_enabled),
1583 #endif
1584 .mode = 0644,
1585 .proc_handler = proc_dointvec,
1586 .extra1 = &zero,
1587 },
1588 #endif
1589 #ifdef CONFIG_HIGHMEM
1590 {
1591 .procname = "highmem_is_dirtyable",
1592 .data = &vm_highmem_is_dirtyable,
1593 .maxlen = sizeof(vm_highmem_is_dirtyable),
1594 .mode = 0644,
1595 .proc_handler = proc_dointvec_minmax,
1596 .extra1 = &zero,
1597 .extra2 = &one,
1598 },
1599 #endif
1600 #ifdef CONFIG_MEMORY_FAILURE
1601 {
1602 .procname = "memory_failure_early_kill",
1603 .data = &sysctl_memory_failure_early_kill,
1604 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1605 .mode = 0644,
1606 .proc_handler = proc_dointvec_minmax,
1607 .extra1 = &zero,
1608 .extra2 = &one,
1609 },
1610 {
1611 .procname = "memory_failure_recovery",
1612 .data = &sysctl_memory_failure_recovery,
1613 .maxlen = sizeof(sysctl_memory_failure_recovery),
1614 .mode = 0644,
1615 .proc_handler = proc_dointvec_minmax,
1616 .extra1 = &zero,
1617 .extra2 = &one,
1618 },
1619 #endif
1620 {
1621 .procname = "user_reserve_kbytes",
1622 .data = &sysctl_user_reserve_kbytes,
1623 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1624 .mode = 0644,
1625 .proc_handler = proc_doulongvec_minmax,
1626 },
1627 {
1628 .procname = "admin_reserve_kbytes",
1629 .data = &sysctl_admin_reserve_kbytes,
1630 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1631 .mode = 0644,
1632 .proc_handler = proc_doulongvec_minmax,
1633 },
1634 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1635 {
1636 .procname = "mmap_rnd_bits",
1637 .data = &mmap_rnd_bits,
1638 .maxlen = sizeof(mmap_rnd_bits),
1639 .mode = 0600,
1640 .proc_handler = proc_dointvec_minmax,
1641 .extra1 = (void *)&mmap_rnd_bits_min,
1642 .extra2 = (void *)&mmap_rnd_bits_max,
1643 },
1644 #endif
1645 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1646 {
1647 .procname = "mmap_rnd_compat_bits",
1648 .data = &mmap_rnd_compat_bits,
1649 .maxlen = sizeof(mmap_rnd_compat_bits),
1650 .mode = 0600,
1651 .proc_handler = proc_dointvec_minmax,
1652 .extra1 = (void *)&mmap_rnd_compat_bits_min,
1653 .extra2 = (void *)&mmap_rnd_compat_bits_max,
1654 },
1655 #endif
1656 { }
1657 };
1658
1659 static struct ctl_table fs_table[] = {
1660 {
1661 .procname = "inode-nr",
1662 .data = &inodes_stat,
1663 .maxlen = 2*sizeof(long),
1664 .mode = 0444,
1665 .proc_handler = proc_nr_inodes,
1666 },
1667 {
1668 .procname = "inode-state",
1669 .data = &inodes_stat,
1670 .maxlen = 7*sizeof(long),
1671 .mode = 0444,
1672 .proc_handler = proc_nr_inodes,
1673 },
1674 {
1675 .procname = "file-nr",
1676 .data = &files_stat,
1677 .maxlen = sizeof(files_stat),
1678 .mode = 0444,
1679 .proc_handler = proc_nr_files,
1680 },
1681 {
1682 .procname = "file-max",
1683 .data = &files_stat.max_files,
1684 .maxlen = sizeof(files_stat.max_files),
1685 .mode = 0644,
1686 .proc_handler = proc_doulongvec_minmax,
1687 },
1688 {
1689 .procname = "nr_open",
1690 .data = &sysctl_nr_open,
1691 .maxlen = sizeof(unsigned int),
1692 .mode = 0644,
1693 .proc_handler = proc_dointvec_minmax,
1694 .extra1 = &sysctl_nr_open_min,
1695 .extra2 = &sysctl_nr_open_max,
1696 },
1697 {
1698 .procname = "dentry-state",
1699 .data = &dentry_stat,
1700 .maxlen = 6*sizeof(long),
1701 .mode = 0444,
1702 .proc_handler = proc_nr_dentry,
1703 },
1704 {
1705 .procname = "overflowuid",
1706 .data = &fs_overflowuid,
1707 .maxlen = sizeof(int),
1708 .mode = 0644,
1709 .proc_handler = proc_dointvec_minmax,
1710 .extra1 = &minolduid,
1711 .extra2 = &maxolduid,
1712 },
1713 {
1714 .procname = "overflowgid",
1715 .data = &fs_overflowgid,
1716 .maxlen = sizeof(int),
1717 .mode = 0644,
1718 .proc_handler = proc_dointvec_minmax,
1719 .extra1 = &minolduid,
1720 .extra2 = &maxolduid,
1721 },
1722 #ifdef CONFIG_FILE_LOCKING
1723 {
1724 .procname = "leases-enable",
1725 .data = &leases_enable,
1726 .maxlen = sizeof(int),
1727 .mode = 0644,
1728 .proc_handler = proc_dointvec,
1729 },
1730 #endif
1731 #ifdef CONFIG_DNOTIFY
1732 {
1733 .procname = "dir-notify-enable",
1734 .data = &dir_notify_enable,
1735 .maxlen = sizeof(int),
1736 .mode = 0644,
1737 .proc_handler = proc_dointvec,
1738 },
1739 #endif
1740 #ifdef CONFIG_MMU
1741 #ifdef CONFIG_FILE_LOCKING
1742 {
1743 .procname = "lease-break-time",
1744 .data = &lease_break_time,
1745 .maxlen = sizeof(int),
1746 .mode = 0644,
1747 .proc_handler = proc_dointvec,
1748 },
1749 #endif
1750 #ifdef CONFIG_AIO
1751 {
1752 .procname = "aio-nr",
1753 .data = &aio_nr,
1754 .maxlen = sizeof(aio_nr),
1755 .mode = 0444,
1756 .proc_handler = proc_doulongvec_minmax,
1757 },
1758 {
1759 .procname = "aio-max-nr",
1760 .data = &aio_max_nr,
1761 .maxlen = sizeof(aio_max_nr),
1762 .mode = 0644,
1763 .proc_handler = proc_doulongvec_minmax,
1764 },
1765 #endif /* CONFIG_AIO */
1766 #ifdef CONFIG_INOTIFY_USER
1767 {
1768 .procname = "inotify",
1769 .mode = 0555,
1770 .child = inotify_table,
1771 },
1772 #endif
1773 #ifdef CONFIG_EPOLL
1774 {
1775 .procname = "epoll",
1776 .mode = 0555,
1777 .child = epoll_table,
1778 },
1779 #endif
1780 #endif
1781 {
1782 .procname = "protected_symlinks",
1783 .data = &sysctl_protected_symlinks,
1784 .maxlen = sizeof(int),
1785 .mode = 0600,
1786 .proc_handler = proc_dointvec_minmax,
1787 .extra1 = &zero,
1788 .extra2 = &one,
1789 },
1790 {
1791 .procname = "protected_hardlinks",
1792 .data = &sysctl_protected_hardlinks,
1793 .maxlen = sizeof(int),
1794 .mode = 0600,
1795 .proc_handler = proc_dointvec_minmax,
1796 .extra1 = &zero,
1797 .extra2 = &one,
1798 },
1799 {
1800 .procname = "suid_dumpable",
1801 .data = &suid_dumpable,
1802 .maxlen = sizeof(int),
1803 .mode = 0644,
1804 .proc_handler = proc_dointvec_minmax_coredump,
1805 .extra1 = &zero,
1806 .extra2 = &two,
1807 },
1808 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1809 {
1810 .procname = "binfmt_misc",
1811 .mode = 0555,
1812 .child = sysctl_mount_point,
1813 },
1814 #endif
1815 {
1816 .procname = "pipe-max-size",
1817 .data = &pipe_max_size,
1818 .maxlen = sizeof(pipe_max_size),
1819 .mode = 0644,
1820 .proc_handler = proc_dopipe_max_size,
1821 },
1822 {
1823 .procname = "pipe-user-pages-hard",
1824 .data = &pipe_user_pages_hard,
1825 .maxlen = sizeof(pipe_user_pages_hard),
1826 .mode = 0644,
1827 .proc_handler = proc_doulongvec_minmax,
1828 },
1829 {
1830 .procname = "pipe-user-pages-soft",
1831 .data = &pipe_user_pages_soft,
1832 .maxlen = sizeof(pipe_user_pages_soft),
1833 .mode = 0644,
1834 .proc_handler = proc_doulongvec_minmax,
1835 },
1836 {
1837 .procname = "mount-max",
1838 .data = &sysctl_mount_max,
1839 .maxlen = sizeof(unsigned int),
1840 .mode = 0644,
1841 .proc_handler = proc_dointvec_minmax,
1842 .extra1 = &one,
1843 },
1844 { }
1845 };
1846
1847 static struct ctl_table debug_table[] = {
1848 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1849 {
1850 .procname = "exception-trace",
1851 .data = &show_unhandled_signals,
1852 .maxlen = sizeof(int),
1853 .mode = 0644,
1854 .proc_handler = proc_dointvec
1855 },
1856 #endif
1857 #if defined(CONFIG_OPTPROBES)
1858 {
1859 .procname = "kprobes-optimization",
1860 .data = &sysctl_kprobes_optimization,
1861 .maxlen = sizeof(int),
1862 .mode = 0644,
1863 .proc_handler = proc_kprobes_optimization_handler,
1864 .extra1 = &zero,
1865 .extra2 = &one,
1866 },
1867 #endif
1868 { }
1869 };
1870
1871 static struct ctl_table dev_table[] = {
1872 { }
1873 };
1874
1875 int __init sysctl_init(void)
1876 {
1877 struct ctl_table_header *hdr;
1878
1879 hdr = register_sysctl_table(sysctl_base_table);
1880 kmemleak_not_leak(hdr);
1881 return 0;
1882 }
1883
1884 #endif /* CONFIG_SYSCTL */
1885
1886 /*
1887 * /proc/sys support
1888 */
1889
1890 #ifdef CONFIG_PROC_SYSCTL
1891
1892 static int _proc_do_string(char *data, int maxlen, int write,
1893 char __user *buffer,
1894 size_t *lenp, loff_t *ppos)
1895 {
1896 size_t len;
1897 char __user *p;
1898 char c;
1899
1900 if (!data || !maxlen || !*lenp) {
1901 *lenp = 0;
1902 return 0;
1903 }
1904
1905 if (write) {
1906 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1907 /* Only continue writes not past the end of buffer. */
1908 len = strlen(data);
1909 if (len > maxlen - 1)
1910 len = maxlen - 1;
1911
1912 if (*ppos > len)
1913 return 0;
1914 len = *ppos;
1915 } else {
1916 /* Start writing from beginning of buffer. */
1917 len = 0;
1918 }
1919
1920 *ppos += *lenp;
1921 p = buffer;
1922 while ((p - buffer) < *lenp && len < maxlen - 1) {
1923 if (get_user(c, p++))
1924 return -EFAULT;
1925 if (c == 0 || c == '\n')
1926 break;
1927 data[len++] = c;
1928 }
1929 data[len] = 0;
1930 } else {
1931 len = strlen(data);
1932 if (len > maxlen)
1933 len = maxlen;
1934
1935 if (*ppos > len) {
1936 *lenp = 0;
1937 return 0;
1938 }
1939
1940 data += *ppos;
1941 len -= *ppos;
1942
1943 if (len > *lenp)
1944 len = *lenp;
1945 if (len)
1946 if (copy_to_user(buffer, data, len))
1947 return -EFAULT;
1948 if (len < *lenp) {
1949 if (put_user('\n', buffer + len))
1950 return -EFAULT;
1951 len++;
1952 }
1953 *lenp = len;
1954 *ppos += len;
1955 }
1956 return 0;
1957 }
1958
1959 static void warn_sysctl_write(struct ctl_table *table)
1960 {
1961 pr_warn_once("%s wrote to %s when file position was not 0!\n"
1962 "This will not be supported in the future. To silence this\n"
1963 "warning, set kernel.sysctl_writes_strict = -1\n",
1964 current->comm, table->procname);
1965 }
1966
1967 /**
1968 * proc_first_pos_non_zero_ignore - check if firs position is allowed
1969 * @ppos: file position
1970 * @table: the sysctl table
1971 *
1972 * Returns true if the first position is non-zero and the sysctl_writes_strict
1973 * mode indicates this is not allowed for numeric input types. String proc
1974 * hadlers can ignore the return value.
1975 */
1976 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
1977 struct ctl_table *table)
1978 {
1979 if (!*ppos)
1980 return false;
1981
1982 switch (sysctl_writes_strict) {
1983 case SYSCTL_WRITES_STRICT:
1984 return true;
1985 case SYSCTL_WRITES_WARN:
1986 warn_sysctl_write(table);
1987 return false;
1988 default:
1989 return false;
1990 }
1991 }
1992
1993 /**
1994 * proc_dostring - read a string sysctl
1995 * @table: the sysctl table
1996 * @write: %TRUE if this is a write to the sysctl file
1997 * @buffer: the user buffer
1998 * @lenp: the size of the user buffer
1999 * @ppos: file position
2000 *
2001 * Reads/writes a string from/to the user buffer. If the kernel
2002 * buffer provided is not large enough to hold the string, the
2003 * string is truncated. The copied string is %NULL-terminated.
2004 * If the string is being read by the user process, it is copied
2005 * and a newline '\n' is added. It is truncated if the buffer is
2006 * not large enough.
2007 *
2008 * Returns 0 on success.
2009 */
2010 int proc_dostring(struct ctl_table *table, int write,
2011 void __user *buffer, size_t *lenp, loff_t *ppos)
2012 {
2013 if (write)
2014 proc_first_pos_non_zero_ignore(ppos, table);
2015
2016 return _proc_do_string((char *)(table->data), table->maxlen, write,
2017 (char __user *)buffer, lenp, ppos);
2018 }
2019
2020 static size_t proc_skip_spaces(char **buf)
2021 {
2022 size_t ret;
2023 char *tmp = skip_spaces(*buf);
2024 ret = tmp - *buf;
2025 *buf = tmp;
2026 return ret;
2027 }
2028
2029 static void proc_skip_char(char **buf, size_t *size, const char v)
2030 {
2031 while (*size) {
2032 if (**buf != v)
2033 break;
2034 (*size)--;
2035 (*buf)++;
2036 }
2037 }
2038
2039 #define TMPBUFLEN 22
2040 /**
2041 * proc_get_long - reads an ASCII formatted integer from a user buffer
2042 *
2043 * @buf: a kernel buffer
2044 * @size: size of the kernel buffer
2045 * @val: this is where the number will be stored
2046 * @neg: set to %TRUE if number is negative
2047 * @perm_tr: a vector which contains the allowed trailers
2048 * @perm_tr_len: size of the perm_tr vector
2049 * @tr: pointer to store the trailer character
2050 *
2051 * In case of success %0 is returned and @buf and @size are updated with
2052 * the amount of bytes read. If @tr is non-NULL and a trailing
2053 * character exists (size is non-zero after returning from this
2054 * function), @tr is updated with the trailing character.
2055 */
2056 static int proc_get_long(char **buf, size_t *size,
2057 unsigned long *val, bool *neg,
2058 const char *perm_tr, unsigned perm_tr_len, char *tr)
2059 {
2060 int len;
2061 char *p, tmp[TMPBUFLEN];
2062
2063 if (!*size)
2064 return -EINVAL;
2065
2066 len = *size;
2067 if (len > TMPBUFLEN - 1)
2068 len = TMPBUFLEN - 1;
2069
2070 memcpy(tmp, *buf, len);
2071
2072 tmp[len] = 0;
2073 p = tmp;
2074 if (*p == '-' && *size > 1) {
2075 *neg = true;
2076 p++;
2077 } else
2078 *neg = false;
2079 if (!isdigit(*p))
2080 return -EINVAL;
2081
2082 *val = simple_strtoul(p, &p, 0);
2083
2084 len = p - tmp;
2085
2086 /* We don't know if the next char is whitespace thus we may accept
2087 * invalid integers (e.g. 1234...a) or two integers instead of one
2088 * (e.g. 123...1). So lets not allow such large numbers. */
2089 if (len == TMPBUFLEN - 1)
2090 return -EINVAL;
2091
2092 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2093 return -EINVAL;
2094
2095 if (tr && (len < *size))
2096 *tr = *p;
2097
2098 *buf += len;
2099 *size -= len;
2100
2101 return 0;
2102 }
2103
2104 /**
2105 * proc_put_long - converts an integer to a decimal ASCII formatted string
2106 *
2107 * @buf: the user buffer
2108 * @size: the size of the user buffer
2109 * @val: the integer to be converted
2110 * @neg: sign of the number, %TRUE for negative
2111 *
2112 * In case of success %0 is returned and @buf and @size are updated with
2113 * the amount of bytes written.
2114 */
2115 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2116 bool neg)
2117 {
2118 int len;
2119 char tmp[TMPBUFLEN], *p = tmp;
2120
2121 sprintf(p, "%s%lu", neg ? "-" : "", val);
2122 len = strlen(tmp);
2123 if (len > *size)
2124 len = *size;
2125 if (copy_to_user(*buf, tmp, len))
2126 return -EFAULT;
2127 *size -= len;
2128 *buf += len;
2129 return 0;
2130 }
2131 #undef TMPBUFLEN
2132
2133 static int proc_put_char(void __user **buf, size_t *size, char c)
2134 {
2135 if (*size) {
2136 char __user **buffer = (char __user **)buf;
2137 if (put_user(c, *buffer))
2138 return -EFAULT;
2139 (*size)--, (*buffer)++;
2140 *buf = *buffer;
2141 }
2142 return 0;
2143 }
2144
2145 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2146 int *valp,
2147 int write, void *data)
2148 {
2149 if (write) {
2150 if (*negp) {
2151 if (*lvalp > (unsigned long) INT_MAX + 1)
2152 return -EINVAL;
2153 *valp = -*lvalp;
2154 } else {
2155 if (*lvalp > (unsigned long) INT_MAX)
2156 return -EINVAL;
2157 *valp = *lvalp;
2158 }
2159 } else {
2160 int val = *valp;
2161 if (val < 0) {
2162 *negp = true;
2163 *lvalp = -(unsigned long)val;
2164 } else {
2165 *negp = false;
2166 *lvalp = (unsigned long)val;
2167 }
2168 }
2169 return 0;
2170 }
2171
2172 static int do_proc_douintvec_conv(unsigned long *lvalp,
2173 unsigned int *valp,
2174 int write, void *data)
2175 {
2176 if (write) {
2177 if (*lvalp > UINT_MAX)
2178 return -EINVAL;
2179 *valp = *lvalp;
2180 } else {
2181 unsigned int val = *valp;
2182 *lvalp = (unsigned long)val;
2183 }
2184 return 0;
2185 }
2186
2187 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2188
2189 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2190 int write, void __user *buffer,
2191 size_t *lenp, loff_t *ppos,
2192 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2193 int write, void *data),
2194 void *data)
2195 {
2196 int *i, vleft, first = 1, err = 0;
2197 size_t left;
2198 char *kbuf = NULL, *p;
2199
2200 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2201 *lenp = 0;
2202 return 0;
2203 }
2204
2205 i = (int *) tbl_data;
2206 vleft = table->maxlen / sizeof(*i);
2207 left = *lenp;
2208
2209 if (!conv)
2210 conv = do_proc_dointvec_conv;
2211
2212 if (write) {
2213 if (proc_first_pos_non_zero_ignore(ppos, table))
2214 goto out;
2215
2216 if (left > PAGE_SIZE - 1)
2217 left = PAGE_SIZE - 1;
2218 p = kbuf = memdup_user_nul(buffer, left);
2219 if (IS_ERR(kbuf))
2220 return PTR_ERR(kbuf);
2221 }
2222
2223 for (; left && vleft--; i++, first=0) {
2224 unsigned long lval;
2225 bool neg;
2226
2227 if (write) {
2228 left -= proc_skip_spaces(&p);
2229
2230 if (!left)
2231 break;
2232 err = proc_get_long(&p, &left, &lval, &neg,
2233 proc_wspace_sep,
2234 sizeof(proc_wspace_sep), NULL);
2235 if (err)
2236 break;
2237 if (conv(&neg, &lval, i, 1, data)) {
2238 err = -EINVAL;
2239 break;
2240 }
2241 } else {
2242 if (conv(&neg, &lval, i, 0, data)) {
2243 err = -EINVAL;
2244 break;
2245 }
2246 if (!first)
2247 err = proc_put_char(&buffer, &left, '\t');
2248 if (err)
2249 break;
2250 err = proc_put_long(&buffer, &left, lval, neg);
2251 if (err)
2252 break;
2253 }
2254 }
2255
2256 if (!write && !first && left && !err)
2257 err = proc_put_char(&buffer, &left, '\n');
2258 if (write && !err && left)
2259 left -= proc_skip_spaces(&p);
2260 if (write) {
2261 kfree(kbuf);
2262 if (first)
2263 return err ? : -EINVAL;
2264 }
2265 *lenp -= left;
2266 out:
2267 *ppos += *lenp;
2268 return err;
2269 }
2270
2271 static int do_proc_dointvec(struct ctl_table *table, int write,
2272 void __user *buffer, size_t *lenp, loff_t *ppos,
2273 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2274 int write, void *data),
2275 void *data)
2276 {
2277 return __do_proc_dointvec(table->data, table, write,
2278 buffer, lenp, ppos, conv, data);
2279 }
2280
2281 static int do_proc_douintvec_w(unsigned int *tbl_data,
2282 struct ctl_table *table,
2283 void __user *buffer,
2284 size_t *lenp, loff_t *ppos,
2285 int (*conv)(unsigned long *lvalp,
2286 unsigned int *valp,
2287 int write, void *data),
2288 void *data)
2289 {
2290 unsigned long lval;
2291 int err = 0;
2292 size_t left;
2293 bool neg;
2294 char *kbuf = NULL, *p;
2295
2296 left = *lenp;
2297
2298 if (proc_first_pos_non_zero_ignore(ppos, table))
2299 goto bail_early;
2300
2301 if (left > PAGE_SIZE - 1)
2302 left = PAGE_SIZE - 1;
2303
2304 p = kbuf = memdup_user_nul(buffer, left);
2305 if (IS_ERR(kbuf))
2306 return -EINVAL;
2307
2308 left -= proc_skip_spaces(&p);
2309 if (!left) {
2310 err = -EINVAL;
2311 goto out_free;
2312 }
2313
2314 err = proc_get_long(&p, &left, &lval, &neg,
2315 proc_wspace_sep,
2316 sizeof(proc_wspace_sep), NULL);
2317 if (err || neg) {
2318 err = -EINVAL;
2319 goto out_free;
2320 }
2321
2322 if (conv(&lval, tbl_data, 1, data)) {
2323 err = -EINVAL;
2324 goto out_free;
2325 }
2326
2327 if (!err && left)
2328 left -= proc_skip_spaces(&p);
2329
2330 out_free:
2331 kfree(kbuf);
2332 if (err)
2333 return -EINVAL;
2334
2335 return 0;
2336
2337 /* This is in keeping with old __do_proc_dointvec() */
2338 bail_early:
2339 *ppos += *lenp;
2340 return err;
2341 }
2342
2343 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2344 size_t *lenp, loff_t *ppos,
2345 int (*conv)(unsigned long *lvalp,
2346 unsigned int *valp,
2347 int write, void *data),
2348 void *data)
2349 {
2350 unsigned long lval;
2351 int err = 0;
2352 size_t left;
2353
2354 left = *lenp;
2355
2356 if (conv(&lval, tbl_data, 0, data)) {
2357 err = -EINVAL;
2358 goto out;
2359 }
2360
2361 err = proc_put_long(&buffer, &left, lval, false);
2362 if (err || !left)
2363 goto out;
2364
2365 err = proc_put_char(&buffer, &left, '\n');
2366
2367 out:
2368 *lenp -= left;
2369 *ppos += *lenp;
2370
2371 return err;
2372 }
2373
2374 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2375 int write, void __user *buffer,
2376 size_t *lenp, loff_t *ppos,
2377 int (*conv)(unsigned long *lvalp,
2378 unsigned int *valp,
2379 int write, void *data),
2380 void *data)
2381 {
2382 unsigned int *i, vleft;
2383
2384 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2385 *lenp = 0;
2386 return 0;
2387 }
2388
2389 i = (unsigned int *) tbl_data;
2390 vleft = table->maxlen / sizeof(*i);
2391
2392 /*
2393 * Arrays are not supported, keep this simple. *Do not* add
2394 * support for them.
2395 */
2396 if (vleft != 1) {
2397 *lenp = 0;
2398 return -EINVAL;
2399 }
2400
2401 if (!conv)
2402 conv = do_proc_douintvec_conv;
2403
2404 if (write)
2405 return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2406 conv, data);
2407 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2408 }
2409
2410 static int do_proc_douintvec(struct ctl_table *table, int write,
2411 void __user *buffer, size_t *lenp, loff_t *ppos,
2412 int (*conv)(unsigned long *lvalp,
2413 unsigned int *valp,
2414 int write, void *data),
2415 void *data)
2416 {
2417 return __do_proc_douintvec(table->data, table, write,
2418 buffer, lenp, ppos, conv, data);
2419 }
2420
2421 /**
2422 * proc_dointvec - read a vector of integers
2423 * @table: the sysctl table
2424 * @write: %TRUE if this is a write to the sysctl file
2425 * @buffer: the user buffer
2426 * @lenp: the size of the user buffer
2427 * @ppos: file position
2428 *
2429 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2430 * values from/to the user buffer, treated as an ASCII string.
2431 *
2432 * Returns 0 on success.
2433 */
2434 int proc_dointvec(struct ctl_table *table, int write,
2435 void __user *buffer, size_t *lenp, loff_t *ppos)
2436 {
2437 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2438 }
2439
2440 /**
2441 * proc_douintvec - read a vector of unsigned integers
2442 * @table: the sysctl table
2443 * @write: %TRUE if this is a write to the sysctl file
2444 * @buffer: the user buffer
2445 * @lenp: the size of the user buffer
2446 * @ppos: file position
2447 *
2448 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2449 * values from/to the user buffer, treated as an ASCII string.
2450 *
2451 * Returns 0 on success.
2452 */
2453 int proc_douintvec(struct ctl_table *table, int write,
2454 void __user *buffer, size_t *lenp, loff_t *ppos)
2455 {
2456 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2457 do_proc_douintvec_conv, NULL);
2458 }
2459
2460 /*
2461 * Taint values can only be increased
2462 * This means we can safely use a temporary.
2463 */
2464 static int proc_taint(struct ctl_table *table, int write,
2465 void __user *buffer, size_t *lenp, loff_t *ppos)
2466 {
2467 struct ctl_table t;
2468 unsigned long tmptaint = get_taint();
2469 int err;
2470
2471 if (write && !capable(CAP_SYS_ADMIN))
2472 return -EPERM;
2473
2474 t = *table;
2475 t.data = &tmptaint;
2476 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2477 if (err < 0)
2478 return err;
2479
2480 if (write) {
2481 /*
2482 * Poor man's atomic or. Not worth adding a primitive
2483 * to everyone's atomic.h for this
2484 */
2485 int i;
2486 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2487 if ((tmptaint >> i) & 1)
2488 add_taint(i, LOCKDEP_STILL_OK);
2489 }
2490 }
2491
2492 return err;
2493 }
2494
2495 #ifdef CONFIG_PRINTK
2496 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2497 void __user *buffer, size_t *lenp, loff_t *ppos)
2498 {
2499 if (write && !capable(CAP_SYS_ADMIN))
2500 return -EPERM;
2501
2502 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2503 }
2504 #endif
2505
2506 /**
2507 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
2508 * @min: pointer to minimum allowable value
2509 * @max: pointer to maximum allowable value
2510 *
2511 * The do_proc_dointvec_minmax_conv_param structure provides the
2512 * minimum and maximum values for doing range checking for those sysctl
2513 * parameters that use the proc_dointvec_minmax() handler.
2514 */
2515 struct do_proc_dointvec_minmax_conv_param {
2516 int *min;
2517 int *max;
2518 };
2519
2520 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2521 int *valp,
2522 int write, void *data)
2523 {
2524 struct do_proc_dointvec_minmax_conv_param *param = data;
2525 if (write) {
2526 int val = *negp ? -*lvalp : *lvalp;
2527 if ((param->min && *param->min > val) ||
2528 (param->max && *param->max < val))
2529 return -EINVAL;
2530 *valp = val;
2531 } else {
2532 int val = *valp;
2533 if (val < 0) {
2534 *negp = true;
2535 *lvalp = -(unsigned long)val;
2536 } else {
2537 *negp = false;
2538 *lvalp = (unsigned long)val;
2539 }
2540 }
2541 return 0;
2542 }
2543
2544 /**
2545 * proc_dointvec_minmax - read a vector of integers with min/max values
2546 * @table: the sysctl table
2547 * @write: %TRUE if this is a write to the sysctl file
2548 * @buffer: the user buffer
2549 * @lenp: the size of the user buffer
2550 * @ppos: file position
2551 *
2552 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2553 * values from/to the user buffer, treated as an ASCII string.
2554 *
2555 * This routine will ensure the values are within the range specified by
2556 * table->extra1 (min) and table->extra2 (max).
2557 *
2558 * Returns 0 on success or -EINVAL on write when the range check fails.
2559 */
2560 int proc_dointvec_minmax(struct ctl_table *table, int write,
2561 void __user *buffer, size_t *lenp, loff_t *ppos)
2562 {
2563 struct do_proc_dointvec_minmax_conv_param param = {
2564 .min = (int *) table->extra1,
2565 .max = (int *) table->extra2,
2566 };
2567 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2568 do_proc_dointvec_minmax_conv, &param);
2569 }
2570
2571 /**
2572 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
2573 * @min: pointer to minimum allowable value
2574 * @max: pointer to maximum allowable value
2575 *
2576 * The do_proc_douintvec_minmax_conv_param structure provides the
2577 * minimum and maximum values for doing range checking for those sysctl
2578 * parameters that use the proc_douintvec_minmax() handler.
2579 */
2580 struct do_proc_douintvec_minmax_conv_param {
2581 unsigned int *min;
2582 unsigned int *max;
2583 };
2584
2585 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2586 unsigned int *valp,
2587 int write, void *data)
2588 {
2589 struct do_proc_douintvec_minmax_conv_param *param = data;
2590
2591 if (write) {
2592 unsigned int val = *lvalp;
2593
2594 if (*lvalp > UINT_MAX)
2595 return -EINVAL;
2596
2597 if ((param->min && *param->min > val) ||
2598 (param->max && *param->max < val))
2599 return -ERANGE;
2600
2601 *valp = val;
2602 } else {
2603 unsigned int val = *valp;
2604 *lvalp = (unsigned long) val;
2605 }
2606
2607 return 0;
2608 }
2609
2610 /**
2611 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2612 * @table: the sysctl table
2613 * @write: %TRUE if this is a write to the sysctl file
2614 * @buffer: the user buffer
2615 * @lenp: the size of the user buffer
2616 * @ppos: file position
2617 *
2618 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2619 * values from/to the user buffer, treated as an ASCII string. Negative
2620 * strings are not allowed.
2621 *
2622 * This routine will ensure the values are within the range specified by
2623 * table->extra1 (min) and table->extra2 (max). There is a final sanity
2624 * check for UINT_MAX to avoid having to support wrap around uses from
2625 * userspace.
2626 *
2627 * Returns 0 on success or -ERANGE on write when the range check fails.
2628 */
2629 int proc_douintvec_minmax(struct ctl_table *table, int write,
2630 void __user *buffer, size_t *lenp, loff_t *ppos)
2631 {
2632 struct do_proc_douintvec_minmax_conv_param param = {
2633 .min = (unsigned int *) table->extra1,
2634 .max = (unsigned int *) table->extra2,
2635 };
2636 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2637 do_proc_douintvec_minmax_conv, &param);
2638 }
2639
2640 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
2641 unsigned int *valp,
2642 int write, void *data)
2643 {
2644 if (write) {
2645 unsigned int val;
2646
2647 val = round_pipe_size(*lvalp);
2648 if (val == 0)
2649 return -EINVAL;
2650
2651 *valp = val;
2652 } else {
2653 unsigned int val = *valp;
2654 *lvalp = (unsigned long) val;
2655 }
2656
2657 return 0;
2658 }
2659
2660 static int proc_dopipe_max_size(struct ctl_table *table, int write,
2661 void __user *buffer, size_t *lenp, loff_t *ppos)
2662 {
2663 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2664 do_proc_dopipe_max_size_conv, NULL);
2665 }
2666
2667 static void validate_coredump_safety(void)
2668 {
2669 #ifdef CONFIG_COREDUMP
2670 if (suid_dumpable == SUID_DUMP_ROOT &&
2671 core_pattern[0] != '/' && core_pattern[0] != '|') {
2672 printk(KERN_WARNING
2673 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2674 "Pipe handler or fully qualified core dump path required.\n"
2675 "Set kernel.core_pattern before fs.suid_dumpable.\n"
2676 );
2677 }
2678 #endif
2679 }
2680
2681 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2682 void __user *buffer, size_t *lenp, loff_t *ppos)
2683 {
2684 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2685 if (!error)
2686 validate_coredump_safety();
2687 return error;
2688 }
2689
2690 #ifdef CONFIG_COREDUMP
2691 static int proc_dostring_coredump(struct ctl_table *table, int write,
2692 void __user *buffer, size_t *lenp, loff_t *ppos)
2693 {
2694 int error = proc_dostring(table, write, buffer, lenp, ppos);
2695 if (!error)
2696 validate_coredump_safety();
2697 return error;
2698 }
2699 #endif
2700
2701 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2702 void __user *buffer,
2703 size_t *lenp, loff_t *ppos,
2704 unsigned long convmul,
2705 unsigned long convdiv)
2706 {
2707 unsigned long *i, *min, *max;
2708 int vleft, first = 1, err = 0;
2709 size_t left;
2710 char *kbuf = NULL, *p;
2711
2712 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2713 *lenp = 0;
2714 return 0;
2715 }
2716
2717 i = (unsigned long *) data;
2718 min = (unsigned long *) table->extra1;
2719 max = (unsigned long *) table->extra2;
2720 vleft = table->maxlen / sizeof(unsigned long);
2721 left = *lenp;
2722
2723 if (write) {
2724 if (proc_first_pos_non_zero_ignore(ppos, table))
2725 goto out;
2726
2727 if (left > PAGE_SIZE - 1)
2728 left = PAGE_SIZE - 1;
2729 p = kbuf = memdup_user_nul(buffer, left);
2730 if (IS_ERR(kbuf))
2731 return PTR_ERR(kbuf);
2732 }
2733
2734 for (; left && vleft--; i++, first = 0) {
2735 unsigned long val;
2736
2737 if (write) {
2738 bool neg;
2739
2740 left -= proc_skip_spaces(&p);
2741
2742 err = proc_get_long(&p, &left, &val, &neg,
2743 proc_wspace_sep,
2744 sizeof(proc_wspace_sep), NULL);
2745 if (err)
2746 break;
2747 if (neg)
2748 continue;
2749 val = convmul * val / convdiv;
2750 if ((min && val < *min) || (max && val > *max))
2751 continue;
2752 *i = val;
2753 } else {
2754 val = convdiv * (*i) / convmul;
2755 if (!first) {
2756 err = proc_put_char(&buffer, &left, '\t');
2757 if (err)
2758 break;
2759 }
2760 err = proc_put_long(&buffer, &left, val, false);
2761 if (err)
2762 break;
2763 }
2764 }
2765
2766 if (!write && !first && left && !err)
2767 err = proc_put_char(&buffer, &left, '\n');
2768 if (write && !err)
2769 left -= proc_skip_spaces(&p);
2770 if (write) {
2771 kfree(kbuf);
2772 if (first)
2773 return err ? : -EINVAL;
2774 }
2775 *lenp -= left;
2776 out:
2777 *ppos += *lenp;
2778 return err;
2779 }
2780
2781 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2782 void __user *buffer,
2783 size_t *lenp, loff_t *ppos,
2784 unsigned long convmul,
2785 unsigned long convdiv)
2786 {
2787 return __do_proc_doulongvec_minmax(table->data, table, write,
2788 buffer, lenp, ppos, convmul, convdiv);
2789 }
2790
2791 /**
2792 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2793 * @table: the sysctl table
2794 * @write: %TRUE if this is a write to the sysctl file
2795 * @buffer: the user buffer
2796 * @lenp: the size of the user buffer
2797 * @ppos: file position
2798 *
2799 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2800 * values from/to the user buffer, treated as an ASCII string.
2801 *
2802 * This routine will ensure the values are within the range specified by
2803 * table->extra1 (min) and table->extra2 (max).
2804 *
2805 * Returns 0 on success.
2806 */
2807 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2808 void __user *buffer, size_t *lenp, loff_t *ppos)
2809 {
2810 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2811 }
2812
2813 /**
2814 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2815 * @table: the sysctl table
2816 * @write: %TRUE if this is a write to the sysctl file
2817 * @buffer: the user buffer
2818 * @lenp: the size of the user buffer
2819 * @ppos: file position
2820 *
2821 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2822 * values from/to the user buffer, treated as an ASCII string. The values
2823 * are treated as milliseconds, and converted to jiffies when they are stored.
2824 *
2825 * This routine will ensure the values are within the range specified by
2826 * table->extra1 (min) and table->extra2 (max).
2827 *
2828 * Returns 0 on success.
2829 */
2830 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2831 void __user *buffer,
2832 size_t *lenp, loff_t *ppos)
2833 {
2834 return do_proc_doulongvec_minmax(table, write, buffer,
2835 lenp, ppos, HZ, 1000l);
2836 }
2837
2838
2839 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2840 int *valp,
2841 int write, void *data)
2842 {
2843 if (write) {
2844 if (*lvalp > INT_MAX / HZ)
2845 return 1;
2846 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2847 } else {
2848 int val = *valp;
2849 unsigned long lval;
2850 if (val < 0) {
2851 *negp = true;
2852 lval = -(unsigned long)val;
2853 } else {
2854 *negp = false;
2855 lval = (unsigned long)val;
2856 }
2857 *lvalp = lval / HZ;
2858 }
2859 return 0;
2860 }
2861
2862 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2863 int *valp,
2864 int write, void *data)
2865 {
2866 if (write) {
2867 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2868 return 1;
2869 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2870 } else {
2871 int val = *valp;
2872 unsigned long lval;
2873 if (val < 0) {
2874 *negp = true;
2875 lval = -(unsigned long)val;
2876 } else {
2877 *negp = false;
2878 lval = (unsigned long)val;
2879 }
2880 *lvalp = jiffies_to_clock_t(lval);
2881 }
2882 return 0;
2883 }
2884
2885 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2886 int *valp,
2887 int write, void *data)
2888 {
2889 if (write) {
2890 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2891
2892 if (jif > INT_MAX)
2893 return 1;
2894 *valp = (int)jif;
2895 } else {
2896 int val = *valp;
2897 unsigned long lval;
2898 if (val < 0) {
2899 *negp = true;
2900 lval = -(unsigned long)val;
2901 } else {
2902 *negp = false;
2903 lval = (unsigned long)val;
2904 }
2905 *lvalp = jiffies_to_msecs(lval);
2906 }
2907 return 0;
2908 }
2909
2910 /**
2911 * proc_dointvec_jiffies - read a vector of integers as seconds
2912 * @table: the sysctl table
2913 * @write: %TRUE if this is a write to the sysctl file
2914 * @buffer: the user buffer
2915 * @lenp: the size of the user buffer
2916 * @ppos: file position
2917 *
2918 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2919 * values from/to the user buffer, treated as an ASCII string.
2920 * The values read are assumed to be in seconds, and are converted into
2921 * jiffies.
2922 *
2923 * Returns 0 on success.
2924 */
2925 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2926 void __user *buffer, size_t *lenp, loff_t *ppos)
2927 {
2928 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2929 do_proc_dointvec_jiffies_conv,NULL);
2930 }
2931
2932 /**
2933 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2934 * @table: the sysctl table
2935 * @write: %TRUE if this is a write to the sysctl file
2936 * @buffer: the user buffer
2937 * @lenp: the size of the user buffer
2938 * @ppos: pointer to the file position
2939 *
2940 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2941 * values from/to the user buffer, treated as an ASCII string.
2942 * The values read are assumed to be in 1/USER_HZ seconds, and
2943 * are converted into jiffies.
2944 *
2945 * Returns 0 on success.
2946 */
2947 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2948 void __user *buffer, size_t *lenp, loff_t *ppos)
2949 {
2950 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2951 do_proc_dointvec_userhz_jiffies_conv,NULL);
2952 }
2953
2954 /**
2955 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2956 * @table: the sysctl table
2957 * @write: %TRUE if this is a write to the sysctl file
2958 * @buffer: the user buffer
2959 * @lenp: the size of the user buffer
2960 * @ppos: file position
2961 * @ppos: the current position in the file
2962 *
2963 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2964 * values from/to the user buffer, treated as an ASCII string.
2965 * The values read are assumed to be in 1/1000 seconds, and
2966 * are converted into jiffies.
2967 *
2968 * Returns 0 on success.
2969 */
2970 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2971 void __user *buffer, size_t *lenp, loff_t *ppos)
2972 {
2973 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2974 do_proc_dointvec_ms_jiffies_conv, NULL);
2975 }
2976
2977 static int proc_do_cad_pid(struct ctl_table *table, int write,
2978 void __user *buffer, size_t *lenp, loff_t *ppos)
2979 {
2980 struct pid *new_pid;
2981 pid_t tmp;
2982 int r;
2983
2984 tmp = pid_vnr(cad_pid);
2985
2986 r = __do_proc_dointvec(&tmp, table, write, buffer,
2987 lenp, ppos, NULL, NULL);
2988 if (r || !write)
2989 return r;
2990
2991 new_pid = find_get_pid(tmp);
2992 if (!new_pid)
2993 return -ESRCH;
2994
2995 put_pid(xchg(&cad_pid, new_pid));
2996 return 0;
2997 }
2998
2999 /**
3000 * proc_do_large_bitmap - read/write from/to a large bitmap
3001 * @table: the sysctl table
3002 * @write: %TRUE if this is a write to the sysctl file
3003 * @buffer: the user buffer
3004 * @lenp: the size of the user buffer
3005 * @ppos: file position
3006 *
3007 * The bitmap is stored at table->data and the bitmap length (in bits)
3008 * in table->maxlen.
3009 *
3010 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
3011 * large bitmaps may be represented in a compact manner. Writing into
3012 * the file will clear the bitmap then update it with the given input.
3013 *
3014 * Returns 0 on success.
3015 */
3016 int proc_do_large_bitmap(struct ctl_table *table, int write,
3017 void __user *buffer, size_t *lenp, loff_t *ppos)
3018 {
3019 int err = 0;
3020 bool first = 1;
3021 size_t left = *lenp;
3022 unsigned long bitmap_len = table->maxlen;
3023 unsigned long *bitmap = *(unsigned long **) table->data;
3024 unsigned long *tmp_bitmap = NULL;
3025 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
3026
3027 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
3028 *lenp = 0;
3029 return 0;
3030 }
3031
3032 if (write) {
3033 char *kbuf, *p;
3034
3035 if (left > PAGE_SIZE - 1)
3036 left = PAGE_SIZE - 1;
3037
3038 p = kbuf = memdup_user_nul(buffer, left);
3039 if (IS_ERR(kbuf))
3040 return PTR_ERR(kbuf);
3041
3042 tmp_bitmap = kcalloc(BITS_TO_LONGS(bitmap_len),
3043 sizeof(unsigned long),
3044 GFP_KERNEL);
3045 if (!tmp_bitmap) {
3046 kfree(kbuf);
3047 return -ENOMEM;
3048 }
3049 proc_skip_char(&p, &left, '\n');
3050 while (!err && left) {
3051 unsigned long val_a, val_b;
3052 bool neg;
3053
3054 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3055 sizeof(tr_a), &c);
3056 if (err)
3057 break;
3058 if (val_a >= bitmap_len || neg) {
3059 err = -EINVAL;
3060 break;
3061 }
3062
3063 val_b = val_a;
3064 if (left) {
3065 p++;
3066 left--;
3067 }
3068
3069 if (c == '-') {
3070 err = proc_get_long(&p, &left, &val_b,
3071 &neg, tr_b, sizeof(tr_b),
3072 &c);
3073 if (err)
3074 break;
3075 if (val_b >= bitmap_len || neg ||
3076 val_a > val_b) {
3077 err = -EINVAL;
3078 break;
3079 }
3080 if (left) {
3081 p++;
3082 left--;
3083 }
3084 }
3085
3086 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3087 first = 0;
3088 proc_skip_char(&p, &left, '\n');
3089 }
3090 kfree(kbuf);
3091 } else {
3092 unsigned long bit_a, bit_b = 0;
3093
3094 while (left) {
3095 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3096 if (bit_a >= bitmap_len)
3097 break;
3098 bit_b = find_next_zero_bit(bitmap, bitmap_len,
3099 bit_a + 1) - 1;
3100
3101 if (!first) {
3102 err = proc_put_char(&buffer, &left, ',');
3103 if (err)
3104 break;
3105 }
3106 err = proc_put_long(&buffer, &left, bit_a, false);
3107 if (err)
3108 break;
3109 if (bit_a != bit_b) {
3110 err = proc_put_char(&buffer, &left, '-');
3111 if (err)
3112 break;
3113 err = proc_put_long(&buffer, &left, bit_b, false);
3114 if (err)
3115 break;
3116 }
3117
3118 first = 0; bit_b++;
3119 }
3120 if (!err)
3121 err = proc_put_char(&buffer, &left, '\n');
3122 }
3123
3124 if (!err) {
3125 if (write) {
3126 if (*ppos)
3127 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3128 else
3129 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3130 }
3131 *lenp -= left;
3132 *ppos += *lenp;
3133 }
3134
3135 kfree(tmp_bitmap);
3136 return err;
3137 }
3138
3139 #else /* CONFIG_PROC_SYSCTL */
3140
3141 int proc_dostring(struct ctl_table *table, int write,
3142 void __user *buffer, size_t *lenp, loff_t *ppos)
3143 {
3144 return -ENOSYS;
3145 }
3146
3147 int proc_dointvec(struct ctl_table *table, int write,
3148 void __user *buffer, size_t *lenp, loff_t *ppos)
3149 {
3150 return -ENOSYS;
3151 }
3152
3153 int proc_douintvec(struct ctl_table *table, int write,
3154 void __user *buffer, size_t *lenp, loff_t *ppos)
3155 {
3156 return -ENOSYS;
3157 }
3158
3159 int proc_dointvec_minmax(struct ctl_table *table, int write,
3160 void __user *buffer, size_t *lenp, loff_t *ppos)
3161 {
3162 return -ENOSYS;
3163 }
3164
3165 int proc_douintvec_minmax(struct ctl_table *table, int write,
3166 void __user *buffer, size_t *lenp, loff_t *ppos)
3167 {
3168 return -ENOSYS;
3169 }
3170
3171 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3172 void __user *buffer, size_t *lenp, loff_t *ppos)
3173 {
3174 return -ENOSYS;
3175 }
3176
3177 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3178 void __user *buffer, size_t *lenp, loff_t *ppos)
3179 {
3180 return -ENOSYS;
3181 }
3182
3183 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3184 void __user *buffer, size_t *lenp, loff_t *ppos)
3185 {
3186 return -ENOSYS;
3187 }
3188
3189 int proc_doulongvec_minmax(struct ctl_table *table, int write,
3190 void __user *buffer, size_t *lenp, loff_t *ppos)
3191 {
3192 return -ENOSYS;
3193 }
3194
3195 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3196 void __user *buffer,
3197 size_t *lenp, loff_t *ppos)
3198 {
3199 return -ENOSYS;
3200 }
3201
3202
3203 #endif /* CONFIG_PROC_SYSCTL */
3204
3205 /*
3206 * No sense putting this after each symbol definition, twice,
3207 * exception granted :-)
3208 */
3209 EXPORT_SYMBOL(proc_dointvec);
3210 EXPORT_SYMBOL(proc_douintvec);
3211 EXPORT_SYMBOL(proc_dointvec_jiffies);
3212 EXPORT_SYMBOL(proc_dointvec_minmax);
3213 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3214 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3215 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3216 EXPORT_SYMBOL(proc_dostring);
3217 EXPORT_SYMBOL(proc_doulongvec_minmax);
3218 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);