]> git.proxmox.com Git - mirror_ubuntu-eoan-kernel.git/blob - kernel/sysctl.c
Remove 'type' argument from access_ok() function
[mirror_ubuntu-eoan-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
70 #include <linux/uaccess.h>
71 #include <asm/processor.h>
72
73 #ifdef CONFIG_X86
74 #include <asm/nmi.h>
75 #include <asm/stacktrace.h>
76 #include <asm/io.h>
77 #endif
78 #ifdef CONFIG_SPARC
79 #include <asm/setup.h>
80 #endif
81 #ifdef CONFIG_BSD_PROCESS_ACCT
82 #include <linux/acct.h>
83 #endif
84 #ifdef CONFIG_RT_MUTEXES
85 #include <linux/rtmutex.h>
86 #endif
87 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
88 #include <linux/lockdep.h>
89 #endif
90 #ifdef CONFIG_CHR_DEV_SG
91 #include <scsi/sg.h>
92 #endif
93 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
94 #include <linux/stackleak.h>
95 #endif
96 #ifdef CONFIG_LOCKUP_DETECTOR
97 #include <linux/nmi.h>
98 #endif
99
100 #if defined(CONFIG_SYSCTL)
101
102 /* External variables not in a header file. */
103 extern int suid_dumpable;
104 #ifdef CONFIG_COREDUMP
105 extern int core_uses_pid;
106 extern char core_pattern[];
107 extern unsigned int core_pipe_limit;
108 #endif
109 extern int pid_max;
110 extern int pid_max_min, pid_max_max;
111 extern int percpu_pagelist_fraction;
112 extern int latencytop_enabled;
113 extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
114 #ifndef CONFIG_MMU
115 extern int sysctl_nr_trim_pages;
116 #endif
117
118 /* Constants used for minimum and maximum */
119 #ifdef CONFIG_LOCKUP_DETECTOR
120 static int sixty = 60;
121 #endif
122
123 static int __maybe_unused neg_one = -1;
124
125 static int zero;
126 static int __maybe_unused one = 1;
127 static int __maybe_unused two = 2;
128 static int __maybe_unused four = 4;
129 static unsigned long one_ul = 1;
130 static int one_hundred = 100;
131 static int one_thousand = 1000;
132 #ifdef CONFIG_PRINTK
133 static int ten_thousand = 10000;
134 #endif
135 #ifdef CONFIG_PERF_EVENTS
136 static int six_hundred_forty_kb = 640 * 1024;
137 #endif
138
139 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
140 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
141
142 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
143 static int maxolduid = 65535;
144 static int minolduid;
145
146 static int ngroups_max = NGROUPS_MAX;
147 static const int cap_last_cap = CAP_LAST_CAP;
148
149 /*
150 * This is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs
151 * and hung_task_check_interval_secs
152 */
153 #ifdef CONFIG_DETECT_HUNG_TASK
154 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
155 #endif
156
157 #ifdef CONFIG_INOTIFY_USER
158 #include <linux/inotify.h>
159 #endif
160 #ifdef CONFIG_SPARC
161 #endif
162
163 #ifdef __hppa__
164 extern int pwrsw_enabled;
165 #endif
166
167 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
168 extern int unaligned_enabled;
169 #endif
170
171 #ifdef CONFIG_IA64
172 extern int unaligned_dump_stack;
173 #endif
174
175 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
176 extern int no_unaligned_warning;
177 #endif
178
179 #ifdef CONFIG_PROC_SYSCTL
180
181 /**
182 * enum sysctl_writes_mode - supported sysctl write modes
183 *
184 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
185 * to be written, and multiple writes on the same sysctl file descriptor
186 * will rewrite the sysctl value, regardless of file position. No warning
187 * is issued when the initial position is not 0.
188 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
189 * not 0.
190 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
191 * file position 0 and the value must be fully contained in the buffer
192 * sent to the write syscall. If dealing with strings respect the file
193 * position, but restrict this to the max length of the buffer, anything
194 * passed the max lenght will be ignored. Multiple writes will append
195 * to the buffer.
196 *
197 * These write modes control how current file position affects the behavior of
198 * updating sysctl values through the proc interface on each write.
199 */
200 enum sysctl_writes_mode {
201 SYSCTL_WRITES_LEGACY = -1,
202 SYSCTL_WRITES_WARN = 0,
203 SYSCTL_WRITES_STRICT = 1,
204 };
205
206 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
207
208 static int proc_do_cad_pid(struct ctl_table *table, int write,
209 void __user *buffer, size_t *lenp, loff_t *ppos);
210 static int proc_taint(struct ctl_table *table, int write,
211 void __user *buffer, size_t *lenp, loff_t *ppos);
212 #endif
213
214 #ifdef CONFIG_PRINTK
215 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
216 void __user *buffer, size_t *lenp, loff_t *ppos);
217 #endif
218
219 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
220 void __user *buffer, size_t *lenp, loff_t *ppos);
221 #ifdef CONFIG_COREDUMP
222 static int proc_dostring_coredump(struct ctl_table *table, int write,
223 void __user *buffer, size_t *lenp, loff_t *ppos);
224 #endif
225 static int proc_dopipe_max_size(struct ctl_table *table, int write,
226 void __user *buffer, size_t *lenp, loff_t *ppos);
227
228 #ifdef CONFIG_MAGIC_SYSRQ
229 /* Note: sysrq code uses its own private copy */
230 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
231
232 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
233 void __user *buffer, size_t *lenp,
234 loff_t *ppos)
235 {
236 int error;
237
238 error = proc_dointvec(table, write, buffer, lenp, ppos);
239 if (error)
240 return error;
241
242 if (write)
243 sysrq_toggle_support(__sysrq_enabled);
244
245 return 0;
246 }
247
248 #endif
249
250 static struct ctl_table kern_table[];
251 static struct ctl_table vm_table[];
252 static struct ctl_table fs_table[];
253 static struct ctl_table debug_table[];
254 static struct ctl_table dev_table[];
255 extern struct ctl_table random_table[];
256 #ifdef CONFIG_EPOLL
257 extern struct ctl_table epoll_table[];
258 #endif
259
260 #ifdef CONFIG_FW_LOADER_USER_HELPER
261 extern struct ctl_table firmware_config_table[];
262 #endif
263
264 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
265 int sysctl_legacy_va_layout;
266 #endif
267
268 /* The default sysctl tables: */
269
270 static struct ctl_table sysctl_base_table[] = {
271 {
272 .procname = "kernel",
273 .mode = 0555,
274 .child = kern_table,
275 },
276 {
277 .procname = "vm",
278 .mode = 0555,
279 .child = vm_table,
280 },
281 {
282 .procname = "fs",
283 .mode = 0555,
284 .child = fs_table,
285 },
286 {
287 .procname = "debug",
288 .mode = 0555,
289 .child = debug_table,
290 },
291 {
292 .procname = "dev",
293 .mode = 0555,
294 .child = dev_table,
295 },
296 { }
297 };
298
299 #ifdef CONFIG_SCHED_DEBUG
300 static int min_sched_granularity_ns = 100000; /* 100 usecs */
301 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
302 static int min_wakeup_granularity_ns; /* 0 usecs */
303 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
304 #ifdef CONFIG_SMP
305 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
306 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
307 #endif /* CONFIG_SMP */
308 #endif /* CONFIG_SCHED_DEBUG */
309
310 #ifdef CONFIG_COMPACTION
311 static int min_extfrag_threshold;
312 static int max_extfrag_threshold = 1000;
313 #endif
314
315 static struct ctl_table kern_table[] = {
316 {
317 .procname = "sched_child_runs_first",
318 .data = &sysctl_sched_child_runs_first,
319 .maxlen = sizeof(unsigned int),
320 .mode = 0644,
321 .proc_handler = proc_dointvec,
322 },
323 #ifdef CONFIG_SCHED_DEBUG
324 {
325 .procname = "sched_min_granularity_ns",
326 .data = &sysctl_sched_min_granularity,
327 .maxlen = sizeof(unsigned int),
328 .mode = 0644,
329 .proc_handler = sched_proc_update_handler,
330 .extra1 = &min_sched_granularity_ns,
331 .extra2 = &max_sched_granularity_ns,
332 },
333 {
334 .procname = "sched_latency_ns",
335 .data = &sysctl_sched_latency,
336 .maxlen = sizeof(unsigned int),
337 .mode = 0644,
338 .proc_handler = sched_proc_update_handler,
339 .extra1 = &min_sched_granularity_ns,
340 .extra2 = &max_sched_granularity_ns,
341 },
342 {
343 .procname = "sched_wakeup_granularity_ns",
344 .data = &sysctl_sched_wakeup_granularity,
345 .maxlen = sizeof(unsigned int),
346 .mode = 0644,
347 .proc_handler = sched_proc_update_handler,
348 .extra1 = &min_wakeup_granularity_ns,
349 .extra2 = &max_wakeup_granularity_ns,
350 },
351 #ifdef CONFIG_SMP
352 {
353 .procname = "sched_tunable_scaling",
354 .data = &sysctl_sched_tunable_scaling,
355 .maxlen = sizeof(enum sched_tunable_scaling),
356 .mode = 0644,
357 .proc_handler = sched_proc_update_handler,
358 .extra1 = &min_sched_tunable_scaling,
359 .extra2 = &max_sched_tunable_scaling,
360 },
361 {
362 .procname = "sched_migration_cost_ns",
363 .data = &sysctl_sched_migration_cost,
364 .maxlen = sizeof(unsigned int),
365 .mode = 0644,
366 .proc_handler = proc_dointvec,
367 },
368 {
369 .procname = "sched_nr_migrate",
370 .data = &sysctl_sched_nr_migrate,
371 .maxlen = sizeof(unsigned int),
372 .mode = 0644,
373 .proc_handler = proc_dointvec,
374 },
375 #ifdef CONFIG_SCHEDSTATS
376 {
377 .procname = "sched_schedstats",
378 .data = NULL,
379 .maxlen = sizeof(unsigned int),
380 .mode = 0644,
381 .proc_handler = sysctl_schedstats,
382 .extra1 = &zero,
383 .extra2 = &one,
384 },
385 #endif /* CONFIG_SCHEDSTATS */
386 #endif /* CONFIG_SMP */
387 #ifdef CONFIG_NUMA_BALANCING
388 {
389 .procname = "numa_balancing_scan_delay_ms",
390 .data = &sysctl_numa_balancing_scan_delay,
391 .maxlen = sizeof(unsigned int),
392 .mode = 0644,
393 .proc_handler = proc_dointvec,
394 },
395 {
396 .procname = "numa_balancing_scan_period_min_ms",
397 .data = &sysctl_numa_balancing_scan_period_min,
398 .maxlen = sizeof(unsigned int),
399 .mode = 0644,
400 .proc_handler = proc_dointvec,
401 },
402 {
403 .procname = "numa_balancing_scan_period_max_ms",
404 .data = &sysctl_numa_balancing_scan_period_max,
405 .maxlen = sizeof(unsigned int),
406 .mode = 0644,
407 .proc_handler = proc_dointvec,
408 },
409 {
410 .procname = "numa_balancing_scan_size_mb",
411 .data = &sysctl_numa_balancing_scan_size,
412 .maxlen = sizeof(unsigned int),
413 .mode = 0644,
414 .proc_handler = proc_dointvec_minmax,
415 .extra1 = &one,
416 },
417 {
418 .procname = "numa_balancing",
419 .data = NULL, /* filled in by handler */
420 .maxlen = sizeof(unsigned int),
421 .mode = 0644,
422 .proc_handler = sysctl_numa_balancing,
423 .extra1 = &zero,
424 .extra2 = &one,
425 },
426 #endif /* CONFIG_NUMA_BALANCING */
427 #endif /* CONFIG_SCHED_DEBUG */
428 {
429 .procname = "sched_rt_period_us",
430 .data = &sysctl_sched_rt_period,
431 .maxlen = sizeof(unsigned int),
432 .mode = 0644,
433 .proc_handler = sched_rt_handler,
434 },
435 {
436 .procname = "sched_rt_runtime_us",
437 .data = &sysctl_sched_rt_runtime,
438 .maxlen = sizeof(int),
439 .mode = 0644,
440 .proc_handler = sched_rt_handler,
441 },
442 {
443 .procname = "sched_rr_timeslice_ms",
444 .data = &sysctl_sched_rr_timeslice,
445 .maxlen = sizeof(int),
446 .mode = 0644,
447 .proc_handler = sched_rr_handler,
448 },
449 #ifdef CONFIG_SCHED_AUTOGROUP
450 {
451 .procname = "sched_autogroup_enabled",
452 .data = &sysctl_sched_autogroup_enabled,
453 .maxlen = sizeof(unsigned int),
454 .mode = 0644,
455 .proc_handler = proc_dointvec_minmax,
456 .extra1 = &zero,
457 .extra2 = &one,
458 },
459 #endif
460 #ifdef CONFIG_CFS_BANDWIDTH
461 {
462 .procname = "sched_cfs_bandwidth_slice_us",
463 .data = &sysctl_sched_cfs_bandwidth_slice,
464 .maxlen = sizeof(unsigned int),
465 .mode = 0644,
466 .proc_handler = proc_dointvec_minmax,
467 .extra1 = &one,
468 },
469 #endif
470 #ifdef CONFIG_PROVE_LOCKING
471 {
472 .procname = "prove_locking",
473 .data = &prove_locking,
474 .maxlen = sizeof(int),
475 .mode = 0644,
476 .proc_handler = proc_dointvec,
477 },
478 #endif
479 #ifdef CONFIG_LOCK_STAT
480 {
481 .procname = "lock_stat",
482 .data = &lock_stat,
483 .maxlen = sizeof(int),
484 .mode = 0644,
485 .proc_handler = proc_dointvec,
486 },
487 #endif
488 {
489 .procname = "panic",
490 .data = &panic_timeout,
491 .maxlen = sizeof(int),
492 .mode = 0644,
493 .proc_handler = proc_dointvec,
494 },
495 #ifdef CONFIG_COREDUMP
496 {
497 .procname = "core_uses_pid",
498 .data = &core_uses_pid,
499 .maxlen = sizeof(int),
500 .mode = 0644,
501 .proc_handler = proc_dointvec,
502 },
503 {
504 .procname = "core_pattern",
505 .data = core_pattern,
506 .maxlen = CORENAME_MAX_SIZE,
507 .mode = 0644,
508 .proc_handler = proc_dostring_coredump,
509 },
510 {
511 .procname = "core_pipe_limit",
512 .data = &core_pipe_limit,
513 .maxlen = sizeof(unsigned int),
514 .mode = 0644,
515 .proc_handler = proc_dointvec,
516 },
517 #endif
518 #ifdef CONFIG_PROC_SYSCTL
519 {
520 .procname = "tainted",
521 .maxlen = sizeof(long),
522 .mode = 0644,
523 .proc_handler = proc_taint,
524 },
525 {
526 .procname = "sysctl_writes_strict",
527 .data = &sysctl_writes_strict,
528 .maxlen = sizeof(int),
529 .mode = 0644,
530 .proc_handler = proc_dointvec_minmax,
531 .extra1 = &neg_one,
532 .extra2 = &one,
533 },
534 #endif
535 #ifdef CONFIG_LATENCYTOP
536 {
537 .procname = "latencytop",
538 .data = &latencytop_enabled,
539 .maxlen = sizeof(int),
540 .mode = 0644,
541 .proc_handler = sysctl_latencytop,
542 },
543 #endif
544 #ifdef CONFIG_BLK_DEV_INITRD
545 {
546 .procname = "real-root-dev",
547 .data = &real_root_dev,
548 .maxlen = sizeof(int),
549 .mode = 0644,
550 .proc_handler = proc_dointvec,
551 },
552 #endif
553 {
554 .procname = "print-fatal-signals",
555 .data = &print_fatal_signals,
556 .maxlen = sizeof(int),
557 .mode = 0644,
558 .proc_handler = proc_dointvec,
559 },
560 #ifdef CONFIG_SPARC
561 {
562 .procname = "reboot-cmd",
563 .data = reboot_command,
564 .maxlen = 256,
565 .mode = 0644,
566 .proc_handler = proc_dostring,
567 },
568 {
569 .procname = "stop-a",
570 .data = &stop_a_enabled,
571 .maxlen = sizeof (int),
572 .mode = 0644,
573 .proc_handler = proc_dointvec,
574 },
575 {
576 .procname = "scons-poweroff",
577 .data = &scons_pwroff,
578 .maxlen = sizeof (int),
579 .mode = 0644,
580 .proc_handler = proc_dointvec,
581 },
582 #endif
583 #ifdef CONFIG_SPARC64
584 {
585 .procname = "tsb-ratio",
586 .data = &sysctl_tsb_ratio,
587 .maxlen = sizeof (int),
588 .mode = 0644,
589 .proc_handler = proc_dointvec,
590 },
591 #endif
592 #ifdef __hppa__
593 {
594 .procname = "soft-power",
595 .data = &pwrsw_enabled,
596 .maxlen = sizeof (int),
597 .mode = 0644,
598 .proc_handler = proc_dointvec,
599 },
600 #endif
601 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
602 {
603 .procname = "unaligned-trap",
604 .data = &unaligned_enabled,
605 .maxlen = sizeof (int),
606 .mode = 0644,
607 .proc_handler = proc_dointvec,
608 },
609 #endif
610 {
611 .procname = "ctrl-alt-del",
612 .data = &C_A_D,
613 .maxlen = sizeof(int),
614 .mode = 0644,
615 .proc_handler = proc_dointvec,
616 },
617 #ifdef CONFIG_FUNCTION_TRACER
618 {
619 .procname = "ftrace_enabled",
620 .data = &ftrace_enabled,
621 .maxlen = sizeof(int),
622 .mode = 0644,
623 .proc_handler = ftrace_enable_sysctl,
624 },
625 #endif
626 #ifdef CONFIG_STACK_TRACER
627 {
628 .procname = "stack_tracer_enabled",
629 .data = &stack_tracer_enabled,
630 .maxlen = sizeof(int),
631 .mode = 0644,
632 .proc_handler = stack_trace_sysctl,
633 },
634 #endif
635 #ifdef CONFIG_TRACING
636 {
637 .procname = "ftrace_dump_on_oops",
638 .data = &ftrace_dump_on_oops,
639 .maxlen = sizeof(int),
640 .mode = 0644,
641 .proc_handler = proc_dointvec,
642 },
643 {
644 .procname = "traceoff_on_warning",
645 .data = &__disable_trace_on_warning,
646 .maxlen = sizeof(__disable_trace_on_warning),
647 .mode = 0644,
648 .proc_handler = proc_dointvec,
649 },
650 {
651 .procname = "tracepoint_printk",
652 .data = &tracepoint_printk,
653 .maxlen = sizeof(tracepoint_printk),
654 .mode = 0644,
655 .proc_handler = tracepoint_printk_sysctl,
656 },
657 #endif
658 #ifdef CONFIG_KEXEC_CORE
659 {
660 .procname = "kexec_load_disabled",
661 .data = &kexec_load_disabled,
662 .maxlen = sizeof(int),
663 .mode = 0644,
664 /* only handle a transition from default "0" to "1" */
665 .proc_handler = proc_dointvec_minmax,
666 .extra1 = &one,
667 .extra2 = &one,
668 },
669 #endif
670 #ifdef CONFIG_MODULES
671 {
672 .procname = "modprobe",
673 .data = &modprobe_path,
674 .maxlen = KMOD_PATH_LEN,
675 .mode = 0644,
676 .proc_handler = proc_dostring,
677 },
678 {
679 .procname = "modules_disabled",
680 .data = &modules_disabled,
681 .maxlen = sizeof(int),
682 .mode = 0644,
683 /* only handle a transition from default "0" to "1" */
684 .proc_handler = proc_dointvec_minmax,
685 .extra1 = &one,
686 .extra2 = &one,
687 },
688 #endif
689 #ifdef CONFIG_UEVENT_HELPER
690 {
691 .procname = "hotplug",
692 .data = &uevent_helper,
693 .maxlen = UEVENT_HELPER_PATH_LEN,
694 .mode = 0644,
695 .proc_handler = proc_dostring,
696 },
697 #endif
698 #ifdef CONFIG_CHR_DEV_SG
699 {
700 .procname = "sg-big-buff",
701 .data = &sg_big_buff,
702 .maxlen = sizeof (int),
703 .mode = 0444,
704 .proc_handler = proc_dointvec,
705 },
706 #endif
707 #ifdef CONFIG_BSD_PROCESS_ACCT
708 {
709 .procname = "acct",
710 .data = &acct_parm,
711 .maxlen = 3*sizeof(int),
712 .mode = 0644,
713 .proc_handler = proc_dointvec,
714 },
715 #endif
716 #ifdef CONFIG_MAGIC_SYSRQ
717 {
718 .procname = "sysrq",
719 .data = &__sysrq_enabled,
720 .maxlen = sizeof (int),
721 .mode = 0644,
722 .proc_handler = sysrq_sysctl_handler,
723 },
724 #endif
725 #ifdef CONFIG_PROC_SYSCTL
726 {
727 .procname = "cad_pid",
728 .data = NULL,
729 .maxlen = sizeof (int),
730 .mode = 0600,
731 .proc_handler = proc_do_cad_pid,
732 },
733 #endif
734 {
735 .procname = "threads-max",
736 .data = NULL,
737 .maxlen = sizeof(int),
738 .mode = 0644,
739 .proc_handler = sysctl_max_threads,
740 },
741 {
742 .procname = "random",
743 .mode = 0555,
744 .child = random_table,
745 },
746 {
747 .procname = "usermodehelper",
748 .mode = 0555,
749 .child = usermodehelper_table,
750 },
751 #ifdef CONFIG_FW_LOADER_USER_HELPER
752 {
753 .procname = "firmware_config",
754 .mode = 0555,
755 .child = firmware_config_table,
756 },
757 #endif
758 {
759 .procname = "overflowuid",
760 .data = &overflowuid,
761 .maxlen = sizeof(int),
762 .mode = 0644,
763 .proc_handler = proc_dointvec_minmax,
764 .extra1 = &minolduid,
765 .extra2 = &maxolduid,
766 },
767 {
768 .procname = "overflowgid",
769 .data = &overflowgid,
770 .maxlen = sizeof(int),
771 .mode = 0644,
772 .proc_handler = proc_dointvec_minmax,
773 .extra1 = &minolduid,
774 .extra2 = &maxolduid,
775 },
776 #ifdef CONFIG_S390
777 #ifdef CONFIG_MATHEMU
778 {
779 .procname = "ieee_emulation_warnings",
780 .data = &sysctl_ieee_emulation_warnings,
781 .maxlen = sizeof(int),
782 .mode = 0644,
783 .proc_handler = proc_dointvec,
784 },
785 #endif
786 {
787 .procname = "userprocess_debug",
788 .data = &show_unhandled_signals,
789 .maxlen = sizeof(int),
790 .mode = 0644,
791 .proc_handler = proc_dointvec,
792 },
793 #endif
794 {
795 .procname = "pid_max",
796 .data = &pid_max,
797 .maxlen = sizeof (int),
798 .mode = 0644,
799 .proc_handler = proc_dointvec_minmax,
800 .extra1 = &pid_max_min,
801 .extra2 = &pid_max_max,
802 },
803 {
804 .procname = "panic_on_oops",
805 .data = &panic_on_oops,
806 .maxlen = sizeof(int),
807 .mode = 0644,
808 .proc_handler = proc_dointvec,
809 },
810 #if defined CONFIG_PRINTK
811 {
812 .procname = "printk",
813 .data = &console_loglevel,
814 .maxlen = 4*sizeof(int),
815 .mode = 0644,
816 .proc_handler = proc_dointvec,
817 },
818 {
819 .procname = "printk_ratelimit",
820 .data = &printk_ratelimit_state.interval,
821 .maxlen = sizeof(int),
822 .mode = 0644,
823 .proc_handler = proc_dointvec_jiffies,
824 },
825 {
826 .procname = "printk_ratelimit_burst",
827 .data = &printk_ratelimit_state.burst,
828 .maxlen = sizeof(int),
829 .mode = 0644,
830 .proc_handler = proc_dointvec,
831 },
832 {
833 .procname = "printk_delay",
834 .data = &printk_delay_msec,
835 .maxlen = sizeof(int),
836 .mode = 0644,
837 .proc_handler = proc_dointvec_minmax,
838 .extra1 = &zero,
839 .extra2 = &ten_thousand,
840 },
841 {
842 .procname = "printk_devkmsg",
843 .data = devkmsg_log_str,
844 .maxlen = DEVKMSG_STR_MAX_SIZE,
845 .mode = 0644,
846 .proc_handler = devkmsg_sysctl_set_loglvl,
847 },
848 {
849 .procname = "dmesg_restrict",
850 .data = &dmesg_restrict,
851 .maxlen = sizeof(int),
852 .mode = 0644,
853 .proc_handler = proc_dointvec_minmax_sysadmin,
854 .extra1 = &zero,
855 .extra2 = &one,
856 },
857 {
858 .procname = "kptr_restrict",
859 .data = &kptr_restrict,
860 .maxlen = sizeof(int),
861 .mode = 0644,
862 .proc_handler = proc_dointvec_minmax_sysadmin,
863 .extra1 = &zero,
864 .extra2 = &two,
865 },
866 #endif
867 {
868 .procname = "ngroups_max",
869 .data = &ngroups_max,
870 .maxlen = sizeof (int),
871 .mode = 0444,
872 .proc_handler = proc_dointvec,
873 },
874 {
875 .procname = "cap_last_cap",
876 .data = (void *)&cap_last_cap,
877 .maxlen = sizeof(int),
878 .mode = 0444,
879 .proc_handler = proc_dointvec,
880 },
881 #if defined(CONFIG_LOCKUP_DETECTOR)
882 {
883 .procname = "watchdog",
884 .data = &watchdog_user_enabled,
885 .maxlen = sizeof(int),
886 .mode = 0644,
887 .proc_handler = proc_watchdog,
888 .extra1 = &zero,
889 .extra2 = &one,
890 },
891 {
892 .procname = "watchdog_thresh",
893 .data = &watchdog_thresh,
894 .maxlen = sizeof(int),
895 .mode = 0644,
896 .proc_handler = proc_watchdog_thresh,
897 .extra1 = &zero,
898 .extra2 = &sixty,
899 },
900 {
901 .procname = "nmi_watchdog",
902 .data = &nmi_watchdog_user_enabled,
903 .maxlen = sizeof(int),
904 .mode = NMI_WATCHDOG_SYSCTL_PERM,
905 .proc_handler = proc_nmi_watchdog,
906 .extra1 = &zero,
907 .extra2 = &one,
908 },
909 {
910 .procname = "watchdog_cpumask",
911 .data = &watchdog_cpumask_bits,
912 .maxlen = NR_CPUS,
913 .mode = 0644,
914 .proc_handler = proc_watchdog_cpumask,
915 },
916 #ifdef CONFIG_SOFTLOCKUP_DETECTOR
917 {
918 .procname = "soft_watchdog",
919 .data = &soft_watchdog_user_enabled,
920 .maxlen = sizeof(int),
921 .mode = 0644,
922 .proc_handler = proc_soft_watchdog,
923 .extra1 = &zero,
924 .extra2 = &one,
925 },
926 {
927 .procname = "softlockup_panic",
928 .data = &softlockup_panic,
929 .maxlen = sizeof(int),
930 .mode = 0644,
931 .proc_handler = proc_dointvec_minmax,
932 .extra1 = &zero,
933 .extra2 = &one,
934 },
935 #ifdef CONFIG_SMP
936 {
937 .procname = "softlockup_all_cpu_backtrace",
938 .data = &sysctl_softlockup_all_cpu_backtrace,
939 .maxlen = sizeof(int),
940 .mode = 0644,
941 .proc_handler = proc_dointvec_minmax,
942 .extra1 = &zero,
943 .extra2 = &one,
944 },
945 #endif /* CONFIG_SMP */
946 #endif
947 #ifdef CONFIG_HARDLOCKUP_DETECTOR
948 {
949 .procname = "hardlockup_panic",
950 .data = &hardlockup_panic,
951 .maxlen = sizeof(int),
952 .mode = 0644,
953 .proc_handler = proc_dointvec_minmax,
954 .extra1 = &zero,
955 .extra2 = &one,
956 },
957 #ifdef CONFIG_SMP
958 {
959 .procname = "hardlockup_all_cpu_backtrace",
960 .data = &sysctl_hardlockup_all_cpu_backtrace,
961 .maxlen = sizeof(int),
962 .mode = 0644,
963 .proc_handler = proc_dointvec_minmax,
964 .extra1 = &zero,
965 .extra2 = &one,
966 },
967 #endif /* CONFIG_SMP */
968 #endif
969 #endif
970
971 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
972 {
973 .procname = "unknown_nmi_panic",
974 .data = &unknown_nmi_panic,
975 .maxlen = sizeof (int),
976 .mode = 0644,
977 .proc_handler = proc_dointvec,
978 },
979 #endif
980 #if defined(CONFIG_X86)
981 {
982 .procname = "panic_on_unrecovered_nmi",
983 .data = &panic_on_unrecovered_nmi,
984 .maxlen = sizeof(int),
985 .mode = 0644,
986 .proc_handler = proc_dointvec,
987 },
988 {
989 .procname = "panic_on_io_nmi",
990 .data = &panic_on_io_nmi,
991 .maxlen = sizeof(int),
992 .mode = 0644,
993 .proc_handler = proc_dointvec,
994 },
995 #ifdef CONFIG_DEBUG_STACKOVERFLOW
996 {
997 .procname = "panic_on_stackoverflow",
998 .data = &sysctl_panic_on_stackoverflow,
999 .maxlen = sizeof(int),
1000 .mode = 0644,
1001 .proc_handler = proc_dointvec,
1002 },
1003 #endif
1004 {
1005 .procname = "bootloader_type",
1006 .data = &bootloader_type,
1007 .maxlen = sizeof (int),
1008 .mode = 0444,
1009 .proc_handler = proc_dointvec,
1010 },
1011 {
1012 .procname = "bootloader_version",
1013 .data = &bootloader_version,
1014 .maxlen = sizeof (int),
1015 .mode = 0444,
1016 .proc_handler = proc_dointvec,
1017 },
1018 {
1019 .procname = "io_delay_type",
1020 .data = &io_delay_type,
1021 .maxlen = sizeof(int),
1022 .mode = 0644,
1023 .proc_handler = proc_dointvec,
1024 },
1025 #endif
1026 #if defined(CONFIG_MMU)
1027 {
1028 .procname = "randomize_va_space",
1029 .data = &randomize_va_space,
1030 .maxlen = sizeof(int),
1031 .mode = 0644,
1032 .proc_handler = proc_dointvec,
1033 },
1034 #endif
1035 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1036 {
1037 .procname = "spin_retry",
1038 .data = &spin_retry,
1039 .maxlen = sizeof (int),
1040 .mode = 0644,
1041 .proc_handler = proc_dointvec,
1042 },
1043 #endif
1044 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1045 {
1046 .procname = "acpi_video_flags",
1047 .data = &acpi_realmode_flags,
1048 .maxlen = sizeof (unsigned long),
1049 .mode = 0644,
1050 .proc_handler = proc_doulongvec_minmax,
1051 },
1052 #endif
1053 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1054 {
1055 .procname = "ignore-unaligned-usertrap",
1056 .data = &no_unaligned_warning,
1057 .maxlen = sizeof (int),
1058 .mode = 0644,
1059 .proc_handler = proc_dointvec,
1060 },
1061 #endif
1062 #ifdef CONFIG_IA64
1063 {
1064 .procname = "unaligned-dump-stack",
1065 .data = &unaligned_dump_stack,
1066 .maxlen = sizeof (int),
1067 .mode = 0644,
1068 .proc_handler = proc_dointvec,
1069 },
1070 #endif
1071 #ifdef CONFIG_DETECT_HUNG_TASK
1072 {
1073 .procname = "hung_task_panic",
1074 .data = &sysctl_hung_task_panic,
1075 .maxlen = sizeof(int),
1076 .mode = 0644,
1077 .proc_handler = proc_dointvec_minmax,
1078 .extra1 = &zero,
1079 .extra2 = &one,
1080 },
1081 {
1082 .procname = "hung_task_check_count",
1083 .data = &sysctl_hung_task_check_count,
1084 .maxlen = sizeof(int),
1085 .mode = 0644,
1086 .proc_handler = proc_dointvec_minmax,
1087 .extra1 = &zero,
1088 },
1089 {
1090 .procname = "hung_task_timeout_secs",
1091 .data = &sysctl_hung_task_timeout_secs,
1092 .maxlen = sizeof(unsigned long),
1093 .mode = 0644,
1094 .proc_handler = proc_dohung_task_timeout_secs,
1095 .extra2 = &hung_task_timeout_max,
1096 },
1097 {
1098 .procname = "hung_task_check_interval_secs",
1099 .data = &sysctl_hung_task_check_interval_secs,
1100 .maxlen = sizeof(unsigned long),
1101 .mode = 0644,
1102 .proc_handler = proc_dohung_task_timeout_secs,
1103 .extra2 = &hung_task_timeout_max,
1104 },
1105 {
1106 .procname = "hung_task_warnings",
1107 .data = &sysctl_hung_task_warnings,
1108 .maxlen = sizeof(int),
1109 .mode = 0644,
1110 .proc_handler = proc_dointvec_minmax,
1111 .extra1 = &neg_one,
1112 },
1113 #endif
1114 #ifdef CONFIG_RT_MUTEXES
1115 {
1116 .procname = "max_lock_depth",
1117 .data = &max_lock_depth,
1118 .maxlen = sizeof(int),
1119 .mode = 0644,
1120 .proc_handler = proc_dointvec,
1121 },
1122 #endif
1123 {
1124 .procname = "poweroff_cmd",
1125 .data = &poweroff_cmd,
1126 .maxlen = POWEROFF_CMD_PATH_LEN,
1127 .mode = 0644,
1128 .proc_handler = proc_dostring,
1129 },
1130 #ifdef CONFIG_KEYS
1131 {
1132 .procname = "keys",
1133 .mode = 0555,
1134 .child = key_sysctls,
1135 },
1136 #endif
1137 #ifdef CONFIG_PERF_EVENTS
1138 /*
1139 * User-space scripts rely on the existence of this file
1140 * as a feature check for perf_events being enabled.
1141 *
1142 * So it's an ABI, do not remove!
1143 */
1144 {
1145 .procname = "perf_event_paranoid",
1146 .data = &sysctl_perf_event_paranoid,
1147 .maxlen = sizeof(sysctl_perf_event_paranoid),
1148 .mode = 0644,
1149 .proc_handler = proc_dointvec,
1150 },
1151 {
1152 .procname = "perf_event_mlock_kb",
1153 .data = &sysctl_perf_event_mlock,
1154 .maxlen = sizeof(sysctl_perf_event_mlock),
1155 .mode = 0644,
1156 .proc_handler = proc_dointvec,
1157 },
1158 {
1159 .procname = "perf_event_max_sample_rate",
1160 .data = &sysctl_perf_event_sample_rate,
1161 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1162 .mode = 0644,
1163 .proc_handler = perf_proc_update_handler,
1164 .extra1 = &one,
1165 },
1166 {
1167 .procname = "perf_cpu_time_max_percent",
1168 .data = &sysctl_perf_cpu_time_max_percent,
1169 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1170 .mode = 0644,
1171 .proc_handler = perf_cpu_time_max_percent_handler,
1172 .extra1 = &zero,
1173 .extra2 = &one_hundred,
1174 },
1175 {
1176 .procname = "perf_event_max_stack",
1177 .data = &sysctl_perf_event_max_stack,
1178 .maxlen = sizeof(sysctl_perf_event_max_stack),
1179 .mode = 0644,
1180 .proc_handler = perf_event_max_stack_handler,
1181 .extra1 = &zero,
1182 .extra2 = &six_hundred_forty_kb,
1183 },
1184 {
1185 .procname = "perf_event_max_contexts_per_stack",
1186 .data = &sysctl_perf_event_max_contexts_per_stack,
1187 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
1188 .mode = 0644,
1189 .proc_handler = perf_event_max_stack_handler,
1190 .extra1 = &zero,
1191 .extra2 = &one_thousand,
1192 },
1193 #endif
1194 {
1195 .procname = "panic_on_warn",
1196 .data = &panic_on_warn,
1197 .maxlen = sizeof(int),
1198 .mode = 0644,
1199 .proc_handler = proc_dointvec_minmax,
1200 .extra1 = &zero,
1201 .extra2 = &one,
1202 },
1203 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1204 {
1205 .procname = "timer_migration",
1206 .data = &sysctl_timer_migration,
1207 .maxlen = sizeof(unsigned int),
1208 .mode = 0644,
1209 .proc_handler = timer_migration_handler,
1210 .extra1 = &zero,
1211 .extra2 = &one,
1212 },
1213 #endif
1214 #ifdef CONFIG_BPF_SYSCALL
1215 {
1216 .procname = "unprivileged_bpf_disabled",
1217 .data = &sysctl_unprivileged_bpf_disabled,
1218 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
1219 .mode = 0644,
1220 /* only handle a transition from default "0" to "1" */
1221 .proc_handler = proc_dointvec_minmax,
1222 .extra1 = &one,
1223 .extra2 = &one,
1224 },
1225 #endif
1226 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1227 {
1228 .procname = "panic_on_rcu_stall",
1229 .data = &sysctl_panic_on_rcu_stall,
1230 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
1231 .mode = 0644,
1232 .proc_handler = proc_dointvec_minmax,
1233 .extra1 = &zero,
1234 .extra2 = &one,
1235 },
1236 #endif
1237 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
1238 {
1239 .procname = "stack_erasing",
1240 .data = NULL,
1241 .maxlen = sizeof(int),
1242 .mode = 0600,
1243 .proc_handler = stack_erasing_sysctl,
1244 .extra1 = &zero,
1245 .extra2 = &one,
1246 },
1247 #endif
1248 { }
1249 };
1250
1251 static struct ctl_table vm_table[] = {
1252 {
1253 .procname = "overcommit_memory",
1254 .data = &sysctl_overcommit_memory,
1255 .maxlen = sizeof(sysctl_overcommit_memory),
1256 .mode = 0644,
1257 .proc_handler = proc_dointvec_minmax,
1258 .extra1 = &zero,
1259 .extra2 = &two,
1260 },
1261 {
1262 .procname = "panic_on_oom",
1263 .data = &sysctl_panic_on_oom,
1264 .maxlen = sizeof(sysctl_panic_on_oom),
1265 .mode = 0644,
1266 .proc_handler = proc_dointvec_minmax,
1267 .extra1 = &zero,
1268 .extra2 = &two,
1269 },
1270 {
1271 .procname = "oom_kill_allocating_task",
1272 .data = &sysctl_oom_kill_allocating_task,
1273 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1274 .mode = 0644,
1275 .proc_handler = proc_dointvec,
1276 },
1277 {
1278 .procname = "oom_dump_tasks",
1279 .data = &sysctl_oom_dump_tasks,
1280 .maxlen = sizeof(sysctl_oom_dump_tasks),
1281 .mode = 0644,
1282 .proc_handler = proc_dointvec,
1283 },
1284 {
1285 .procname = "overcommit_ratio",
1286 .data = &sysctl_overcommit_ratio,
1287 .maxlen = sizeof(sysctl_overcommit_ratio),
1288 .mode = 0644,
1289 .proc_handler = overcommit_ratio_handler,
1290 },
1291 {
1292 .procname = "overcommit_kbytes",
1293 .data = &sysctl_overcommit_kbytes,
1294 .maxlen = sizeof(sysctl_overcommit_kbytes),
1295 .mode = 0644,
1296 .proc_handler = overcommit_kbytes_handler,
1297 },
1298 {
1299 .procname = "page-cluster",
1300 .data = &page_cluster,
1301 .maxlen = sizeof(int),
1302 .mode = 0644,
1303 .proc_handler = proc_dointvec_minmax,
1304 .extra1 = &zero,
1305 },
1306 {
1307 .procname = "dirty_background_ratio",
1308 .data = &dirty_background_ratio,
1309 .maxlen = sizeof(dirty_background_ratio),
1310 .mode = 0644,
1311 .proc_handler = dirty_background_ratio_handler,
1312 .extra1 = &zero,
1313 .extra2 = &one_hundred,
1314 },
1315 {
1316 .procname = "dirty_background_bytes",
1317 .data = &dirty_background_bytes,
1318 .maxlen = sizeof(dirty_background_bytes),
1319 .mode = 0644,
1320 .proc_handler = dirty_background_bytes_handler,
1321 .extra1 = &one_ul,
1322 },
1323 {
1324 .procname = "dirty_ratio",
1325 .data = &vm_dirty_ratio,
1326 .maxlen = sizeof(vm_dirty_ratio),
1327 .mode = 0644,
1328 .proc_handler = dirty_ratio_handler,
1329 .extra1 = &zero,
1330 .extra2 = &one_hundred,
1331 },
1332 {
1333 .procname = "dirty_bytes",
1334 .data = &vm_dirty_bytes,
1335 .maxlen = sizeof(vm_dirty_bytes),
1336 .mode = 0644,
1337 .proc_handler = dirty_bytes_handler,
1338 .extra1 = &dirty_bytes_min,
1339 },
1340 {
1341 .procname = "dirty_writeback_centisecs",
1342 .data = &dirty_writeback_interval,
1343 .maxlen = sizeof(dirty_writeback_interval),
1344 .mode = 0644,
1345 .proc_handler = dirty_writeback_centisecs_handler,
1346 },
1347 {
1348 .procname = "dirty_expire_centisecs",
1349 .data = &dirty_expire_interval,
1350 .maxlen = sizeof(dirty_expire_interval),
1351 .mode = 0644,
1352 .proc_handler = proc_dointvec_minmax,
1353 .extra1 = &zero,
1354 },
1355 {
1356 .procname = "dirtytime_expire_seconds",
1357 .data = &dirtytime_expire_interval,
1358 .maxlen = sizeof(dirtytime_expire_interval),
1359 .mode = 0644,
1360 .proc_handler = dirtytime_interval_handler,
1361 .extra1 = &zero,
1362 },
1363 {
1364 .procname = "swappiness",
1365 .data = &vm_swappiness,
1366 .maxlen = sizeof(vm_swappiness),
1367 .mode = 0644,
1368 .proc_handler = proc_dointvec_minmax,
1369 .extra1 = &zero,
1370 .extra2 = &one_hundred,
1371 },
1372 #ifdef CONFIG_HUGETLB_PAGE
1373 {
1374 .procname = "nr_hugepages",
1375 .data = NULL,
1376 .maxlen = sizeof(unsigned long),
1377 .mode = 0644,
1378 .proc_handler = hugetlb_sysctl_handler,
1379 },
1380 #ifdef CONFIG_NUMA
1381 {
1382 .procname = "nr_hugepages_mempolicy",
1383 .data = NULL,
1384 .maxlen = sizeof(unsigned long),
1385 .mode = 0644,
1386 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1387 },
1388 {
1389 .procname = "numa_stat",
1390 .data = &sysctl_vm_numa_stat,
1391 .maxlen = sizeof(int),
1392 .mode = 0644,
1393 .proc_handler = sysctl_vm_numa_stat_handler,
1394 .extra1 = &zero,
1395 .extra2 = &one,
1396 },
1397 #endif
1398 {
1399 .procname = "hugetlb_shm_group",
1400 .data = &sysctl_hugetlb_shm_group,
1401 .maxlen = sizeof(gid_t),
1402 .mode = 0644,
1403 .proc_handler = proc_dointvec,
1404 },
1405 {
1406 .procname = "nr_overcommit_hugepages",
1407 .data = NULL,
1408 .maxlen = sizeof(unsigned long),
1409 .mode = 0644,
1410 .proc_handler = hugetlb_overcommit_handler,
1411 },
1412 #endif
1413 {
1414 .procname = "lowmem_reserve_ratio",
1415 .data = &sysctl_lowmem_reserve_ratio,
1416 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1417 .mode = 0644,
1418 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1419 },
1420 {
1421 .procname = "drop_caches",
1422 .data = &sysctl_drop_caches,
1423 .maxlen = sizeof(int),
1424 .mode = 0644,
1425 .proc_handler = drop_caches_sysctl_handler,
1426 .extra1 = &one,
1427 .extra2 = &four,
1428 },
1429 #ifdef CONFIG_COMPACTION
1430 {
1431 .procname = "compact_memory",
1432 .data = &sysctl_compact_memory,
1433 .maxlen = sizeof(int),
1434 .mode = 0200,
1435 .proc_handler = sysctl_compaction_handler,
1436 },
1437 {
1438 .procname = "extfrag_threshold",
1439 .data = &sysctl_extfrag_threshold,
1440 .maxlen = sizeof(int),
1441 .mode = 0644,
1442 .proc_handler = sysctl_extfrag_handler,
1443 .extra1 = &min_extfrag_threshold,
1444 .extra2 = &max_extfrag_threshold,
1445 },
1446 {
1447 .procname = "compact_unevictable_allowed",
1448 .data = &sysctl_compact_unevictable_allowed,
1449 .maxlen = sizeof(int),
1450 .mode = 0644,
1451 .proc_handler = proc_dointvec,
1452 .extra1 = &zero,
1453 .extra2 = &one,
1454 },
1455
1456 #endif /* CONFIG_COMPACTION */
1457 {
1458 .procname = "min_free_kbytes",
1459 .data = &min_free_kbytes,
1460 .maxlen = sizeof(min_free_kbytes),
1461 .mode = 0644,
1462 .proc_handler = min_free_kbytes_sysctl_handler,
1463 .extra1 = &zero,
1464 },
1465 {
1466 .procname = "watermark_boost_factor",
1467 .data = &watermark_boost_factor,
1468 .maxlen = sizeof(watermark_boost_factor),
1469 .mode = 0644,
1470 .proc_handler = watermark_boost_factor_sysctl_handler,
1471 .extra1 = &zero,
1472 },
1473 {
1474 .procname = "watermark_scale_factor",
1475 .data = &watermark_scale_factor,
1476 .maxlen = sizeof(watermark_scale_factor),
1477 .mode = 0644,
1478 .proc_handler = watermark_scale_factor_sysctl_handler,
1479 .extra1 = &one,
1480 .extra2 = &one_thousand,
1481 },
1482 {
1483 .procname = "percpu_pagelist_fraction",
1484 .data = &percpu_pagelist_fraction,
1485 .maxlen = sizeof(percpu_pagelist_fraction),
1486 .mode = 0644,
1487 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1488 .extra1 = &zero,
1489 },
1490 #ifdef CONFIG_MMU
1491 {
1492 .procname = "max_map_count",
1493 .data = &sysctl_max_map_count,
1494 .maxlen = sizeof(sysctl_max_map_count),
1495 .mode = 0644,
1496 .proc_handler = proc_dointvec_minmax,
1497 .extra1 = &zero,
1498 },
1499 #else
1500 {
1501 .procname = "nr_trim_pages",
1502 .data = &sysctl_nr_trim_pages,
1503 .maxlen = sizeof(sysctl_nr_trim_pages),
1504 .mode = 0644,
1505 .proc_handler = proc_dointvec_minmax,
1506 .extra1 = &zero,
1507 },
1508 #endif
1509 {
1510 .procname = "laptop_mode",
1511 .data = &laptop_mode,
1512 .maxlen = sizeof(laptop_mode),
1513 .mode = 0644,
1514 .proc_handler = proc_dointvec_jiffies,
1515 },
1516 {
1517 .procname = "block_dump",
1518 .data = &block_dump,
1519 .maxlen = sizeof(block_dump),
1520 .mode = 0644,
1521 .proc_handler = proc_dointvec,
1522 .extra1 = &zero,
1523 },
1524 {
1525 .procname = "vfs_cache_pressure",
1526 .data = &sysctl_vfs_cache_pressure,
1527 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1528 .mode = 0644,
1529 .proc_handler = proc_dointvec,
1530 .extra1 = &zero,
1531 },
1532 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1533 {
1534 .procname = "legacy_va_layout",
1535 .data = &sysctl_legacy_va_layout,
1536 .maxlen = sizeof(sysctl_legacy_va_layout),
1537 .mode = 0644,
1538 .proc_handler = proc_dointvec,
1539 .extra1 = &zero,
1540 },
1541 #endif
1542 #ifdef CONFIG_NUMA
1543 {
1544 .procname = "zone_reclaim_mode",
1545 .data = &node_reclaim_mode,
1546 .maxlen = sizeof(node_reclaim_mode),
1547 .mode = 0644,
1548 .proc_handler = proc_dointvec,
1549 .extra1 = &zero,
1550 },
1551 {
1552 .procname = "min_unmapped_ratio",
1553 .data = &sysctl_min_unmapped_ratio,
1554 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1555 .mode = 0644,
1556 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1557 .extra1 = &zero,
1558 .extra2 = &one_hundred,
1559 },
1560 {
1561 .procname = "min_slab_ratio",
1562 .data = &sysctl_min_slab_ratio,
1563 .maxlen = sizeof(sysctl_min_slab_ratio),
1564 .mode = 0644,
1565 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1566 .extra1 = &zero,
1567 .extra2 = &one_hundred,
1568 },
1569 #endif
1570 #ifdef CONFIG_SMP
1571 {
1572 .procname = "stat_interval",
1573 .data = &sysctl_stat_interval,
1574 .maxlen = sizeof(sysctl_stat_interval),
1575 .mode = 0644,
1576 .proc_handler = proc_dointvec_jiffies,
1577 },
1578 {
1579 .procname = "stat_refresh",
1580 .data = NULL,
1581 .maxlen = 0,
1582 .mode = 0600,
1583 .proc_handler = vmstat_refresh,
1584 },
1585 #endif
1586 #ifdef CONFIG_MMU
1587 {
1588 .procname = "mmap_min_addr",
1589 .data = &dac_mmap_min_addr,
1590 .maxlen = sizeof(unsigned long),
1591 .mode = 0644,
1592 .proc_handler = mmap_min_addr_handler,
1593 },
1594 #endif
1595 #ifdef CONFIG_NUMA
1596 {
1597 .procname = "numa_zonelist_order",
1598 .data = &numa_zonelist_order,
1599 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1600 .mode = 0644,
1601 .proc_handler = numa_zonelist_order_handler,
1602 },
1603 #endif
1604 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1605 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1606 {
1607 .procname = "vdso_enabled",
1608 #ifdef CONFIG_X86_32
1609 .data = &vdso32_enabled,
1610 .maxlen = sizeof(vdso32_enabled),
1611 #else
1612 .data = &vdso_enabled,
1613 .maxlen = sizeof(vdso_enabled),
1614 #endif
1615 .mode = 0644,
1616 .proc_handler = proc_dointvec,
1617 .extra1 = &zero,
1618 },
1619 #endif
1620 #ifdef CONFIG_HIGHMEM
1621 {
1622 .procname = "highmem_is_dirtyable",
1623 .data = &vm_highmem_is_dirtyable,
1624 .maxlen = sizeof(vm_highmem_is_dirtyable),
1625 .mode = 0644,
1626 .proc_handler = proc_dointvec_minmax,
1627 .extra1 = &zero,
1628 .extra2 = &one,
1629 },
1630 #endif
1631 #ifdef CONFIG_MEMORY_FAILURE
1632 {
1633 .procname = "memory_failure_early_kill",
1634 .data = &sysctl_memory_failure_early_kill,
1635 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1636 .mode = 0644,
1637 .proc_handler = proc_dointvec_minmax,
1638 .extra1 = &zero,
1639 .extra2 = &one,
1640 },
1641 {
1642 .procname = "memory_failure_recovery",
1643 .data = &sysctl_memory_failure_recovery,
1644 .maxlen = sizeof(sysctl_memory_failure_recovery),
1645 .mode = 0644,
1646 .proc_handler = proc_dointvec_minmax,
1647 .extra1 = &zero,
1648 .extra2 = &one,
1649 },
1650 #endif
1651 {
1652 .procname = "user_reserve_kbytes",
1653 .data = &sysctl_user_reserve_kbytes,
1654 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1655 .mode = 0644,
1656 .proc_handler = proc_doulongvec_minmax,
1657 },
1658 {
1659 .procname = "admin_reserve_kbytes",
1660 .data = &sysctl_admin_reserve_kbytes,
1661 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1662 .mode = 0644,
1663 .proc_handler = proc_doulongvec_minmax,
1664 },
1665 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1666 {
1667 .procname = "mmap_rnd_bits",
1668 .data = &mmap_rnd_bits,
1669 .maxlen = sizeof(mmap_rnd_bits),
1670 .mode = 0600,
1671 .proc_handler = proc_dointvec_minmax,
1672 .extra1 = (void *)&mmap_rnd_bits_min,
1673 .extra2 = (void *)&mmap_rnd_bits_max,
1674 },
1675 #endif
1676 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1677 {
1678 .procname = "mmap_rnd_compat_bits",
1679 .data = &mmap_rnd_compat_bits,
1680 .maxlen = sizeof(mmap_rnd_compat_bits),
1681 .mode = 0600,
1682 .proc_handler = proc_dointvec_minmax,
1683 .extra1 = (void *)&mmap_rnd_compat_bits_min,
1684 .extra2 = (void *)&mmap_rnd_compat_bits_max,
1685 },
1686 #endif
1687 { }
1688 };
1689
1690 static struct ctl_table fs_table[] = {
1691 {
1692 .procname = "inode-nr",
1693 .data = &inodes_stat,
1694 .maxlen = 2*sizeof(long),
1695 .mode = 0444,
1696 .proc_handler = proc_nr_inodes,
1697 },
1698 {
1699 .procname = "inode-state",
1700 .data = &inodes_stat,
1701 .maxlen = 7*sizeof(long),
1702 .mode = 0444,
1703 .proc_handler = proc_nr_inodes,
1704 },
1705 {
1706 .procname = "file-nr",
1707 .data = &files_stat,
1708 .maxlen = sizeof(files_stat),
1709 .mode = 0444,
1710 .proc_handler = proc_nr_files,
1711 },
1712 {
1713 .procname = "file-max",
1714 .data = &files_stat.max_files,
1715 .maxlen = sizeof(files_stat.max_files),
1716 .mode = 0644,
1717 .proc_handler = proc_doulongvec_minmax,
1718 },
1719 {
1720 .procname = "nr_open",
1721 .data = &sysctl_nr_open,
1722 .maxlen = sizeof(unsigned int),
1723 .mode = 0644,
1724 .proc_handler = proc_dointvec_minmax,
1725 .extra1 = &sysctl_nr_open_min,
1726 .extra2 = &sysctl_nr_open_max,
1727 },
1728 {
1729 .procname = "dentry-state",
1730 .data = &dentry_stat,
1731 .maxlen = 6*sizeof(long),
1732 .mode = 0444,
1733 .proc_handler = proc_nr_dentry,
1734 },
1735 {
1736 .procname = "overflowuid",
1737 .data = &fs_overflowuid,
1738 .maxlen = sizeof(int),
1739 .mode = 0644,
1740 .proc_handler = proc_dointvec_minmax,
1741 .extra1 = &minolduid,
1742 .extra2 = &maxolduid,
1743 },
1744 {
1745 .procname = "overflowgid",
1746 .data = &fs_overflowgid,
1747 .maxlen = sizeof(int),
1748 .mode = 0644,
1749 .proc_handler = proc_dointvec_minmax,
1750 .extra1 = &minolduid,
1751 .extra2 = &maxolduid,
1752 },
1753 #ifdef CONFIG_FILE_LOCKING
1754 {
1755 .procname = "leases-enable",
1756 .data = &leases_enable,
1757 .maxlen = sizeof(int),
1758 .mode = 0644,
1759 .proc_handler = proc_dointvec,
1760 },
1761 #endif
1762 #ifdef CONFIG_DNOTIFY
1763 {
1764 .procname = "dir-notify-enable",
1765 .data = &dir_notify_enable,
1766 .maxlen = sizeof(int),
1767 .mode = 0644,
1768 .proc_handler = proc_dointvec,
1769 },
1770 #endif
1771 #ifdef CONFIG_MMU
1772 #ifdef CONFIG_FILE_LOCKING
1773 {
1774 .procname = "lease-break-time",
1775 .data = &lease_break_time,
1776 .maxlen = sizeof(int),
1777 .mode = 0644,
1778 .proc_handler = proc_dointvec,
1779 },
1780 #endif
1781 #ifdef CONFIG_AIO
1782 {
1783 .procname = "aio-nr",
1784 .data = &aio_nr,
1785 .maxlen = sizeof(aio_nr),
1786 .mode = 0444,
1787 .proc_handler = proc_doulongvec_minmax,
1788 },
1789 {
1790 .procname = "aio-max-nr",
1791 .data = &aio_max_nr,
1792 .maxlen = sizeof(aio_max_nr),
1793 .mode = 0644,
1794 .proc_handler = proc_doulongvec_minmax,
1795 },
1796 #endif /* CONFIG_AIO */
1797 #ifdef CONFIG_INOTIFY_USER
1798 {
1799 .procname = "inotify",
1800 .mode = 0555,
1801 .child = inotify_table,
1802 },
1803 #endif
1804 #ifdef CONFIG_EPOLL
1805 {
1806 .procname = "epoll",
1807 .mode = 0555,
1808 .child = epoll_table,
1809 },
1810 #endif
1811 #endif
1812 {
1813 .procname = "protected_symlinks",
1814 .data = &sysctl_protected_symlinks,
1815 .maxlen = sizeof(int),
1816 .mode = 0600,
1817 .proc_handler = proc_dointvec_minmax,
1818 .extra1 = &zero,
1819 .extra2 = &one,
1820 },
1821 {
1822 .procname = "protected_hardlinks",
1823 .data = &sysctl_protected_hardlinks,
1824 .maxlen = sizeof(int),
1825 .mode = 0600,
1826 .proc_handler = proc_dointvec_minmax,
1827 .extra1 = &zero,
1828 .extra2 = &one,
1829 },
1830 {
1831 .procname = "protected_fifos",
1832 .data = &sysctl_protected_fifos,
1833 .maxlen = sizeof(int),
1834 .mode = 0600,
1835 .proc_handler = proc_dointvec_minmax,
1836 .extra1 = &zero,
1837 .extra2 = &two,
1838 },
1839 {
1840 .procname = "protected_regular",
1841 .data = &sysctl_protected_regular,
1842 .maxlen = sizeof(int),
1843 .mode = 0600,
1844 .proc_handler = proc_dointvec_minmax,
1845 .extra1 = &zero,
1846 .extra2 = &two,
1847 },
1848 {
1849 .procname = "suid_dumpable",
1850 .data = &suid_dumpable,
1851 .maxlen = sizeof(int),
1852 .mode = 0644,
1853 .proc_handler = proc_dointvec_minmax_coredump,
1854 .extra1 = &zero,
1855 .extra2 = &two,
1856 },
1857 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1858 {
1859 .procname = "binfmt_misc",
1860 .mode = 0555,
1861 .child = sysctl_mount_point,
1862 },
1863 #endif
1864 {
1865 .procname = "pipe-max-size",
1866 .data = &pipe_max_size,
1867 .maxlen = sizeof(pipe_max_size),
1868 .mode = 0644,
1869 .proc_handler = proc_dopipe_max_size,
1870 },
1871 {
1872 .procname = "pipe-user-pages-hard",
1873 .data = &pipe_user_pages_hard,
1874 .maxlen = sizeof(pipe_user_pages_hard),
1875 .mode = 0644,
1876 .proc_handler = proc_doulongvec_minmax,
1877 },
1878 {
1879 .procname = "pipe-user-pages-soft",
1880 .data = &pipe_user_pages_soft,
1881 .maxlen = sizeof(pipe_user_pages_soft),
1882 .mode = 0644,
1883 .proc_handler = proc_doulongvec_minmax,
1884 },
1885 {
1886 .procname = "mount-max",
1887 .data = &sysctl_mount_max,
1888 .maxlen = sizeof(unsigned int),
1889 .mode = 0644,
1890 .proc_handler = proc_dointvec_minmax,
1891 .extra1 = &one,
1892 },
1893 { }
1894 };
1895
1896 static struct ctl_table debug_table[] = {
1897 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1898 {
1899 .procname = "exception-trace",
1900 .data = &show_unhandled_signals,
1901 .maxlen = sizeof(int),
1902 .mode = 0644,
1903 .proc_handler = proc_dointvec
1904 },
1905 #endif
1906 #if defined(CONFIG_OPTPROBES)
1907 {
1908 .procname = "kprobes-optimization",
1909 .data = &sysctl_kprobes_optimization,
1910 .maxlen = sizeof(int),
1911 .mode = 0644,
1912 .proc_handler = proc_kprobes_optimization_handler,
1913 .extra1 = &zero,
1914 .extra2 = &one,
1915 },
1916 #endif
1917 { }
1918 };
1919
1920 static struct ctl_table dev_table[] = {
1921 { }
1922 };
1923
1924 int __init sysctl_init(void)
1925 {
1926 struct ctl_table_header *hdr;
1927
1928 hdr = register_sysctl_table(sysctl_base_table);
1929 kmemleak_not_leak(hdr);
1930 return 0;
1931 }
1932
1933 #endif /* CONFIG_SYSCTL */
1934
1935 /*
1936 * /proc/sys support
1937 */
1938
1939 #ifdef CONFIG_PROC_SYSCTL
1940
1941 static int _proc_do_string(char *data, int maxlen, int write,
1942 char __user *buffer,
1943 size_t *lenp, loff_t *ppos)
1944 {
1945 size_t len;
1946 char __user *p;
1947 char c;
1948
1949 if (!data || !maxlen || !*lenp) {
1950 *lenp = 0;
1951 return 0;
1952 }
1953
1954 if (write) {
1955 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1956 /* Only continue writes not past the end of buffer. */
1957 len = strlen(data);
1958 if (len > maxlen - 1)
1959 len = maxlen - 1;
1960
1961 if (*ppos > len)
1962 return 0;
1963 len = *ppos;
1964 } else {
1965 /* Start writing from beginning of buffer. */
1966 len = 0;
1967 }
1968
1969 *ppos += *lenp;
1970 p = buffer;
1971 while ((p - buffer) < *lenp && len < maxlen - 1) {
1972 if (get_user(c, p++))
1973 return -EFAULT;
1974 if (c == 0 || c == '\n')
1975 break;
1976 data[len++] = c;
1977 }
1978 data[len] = 0;
1979 } else {
1980 len = strlen(data);
1981 if (len > maxlen)
1982 len = maxlen;
1983
1984 if (*ppos > len) {
1985 *lenp = 0;
1986 return 0;
1987 }
1988
1989 data += *ppos;
1990 len -= *ppos;
1991
1992 if (len > *lenp)
1993 len = *lenp;
1994 if (len)
1995 if (copy_to_user(buffer, data, len))
1996 return -EFAULT;
1997 if (len < *lenp) {
1998 if (put_user('\n', buffer + len))
1999 return -EFAULT;
2000 len++;
2001 }
2002 *lenp = len;
2003 *ppos += len;
2004 }
2005 return 0;
2006 }
2007
2008 static void warn_sysctl_write(struct ctl_table *table)
2009 {
2010 pr_warn_once("%s wrote to %s when file position was not 0!\n"
2011 "This will not be supported in the future. To silence this\n"
2012 "warning, set kernel.sysctl_writes_strict = -1\n",
2013 current->comm, table->procname);
2014 }
2015
2016 /**
2017 * proc_first_pos_non_zero_ignore - check if first position is allowed
2018 * @ppos: file position
2019 * @table: the sysctl table
2020 *
2021 * Returns true if the first position is non-zero and the sysctl_writes_strict
2022 * mode indicates this is not allowed for numeric input types. String proc
2023 * handlers can ignore the return value.
2024 */
2025 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
2026 struct ctl_table *table)
2027 {
2028 if (!*ppos)
2029 return false;
2030
2031 switch (sysctl_writes_strict) {
2032 case SYSCTL_WRITES_STRICT:
2033 return true;
2034 case SYSCTL_WRITES_WARN:
2035 warn_sysctl_write(table);
2036 return false;
2037 default:
2038 return false;
2039 }
2040 }
2041
2042 /**
2043 * proc_dostring - read a string sysctl
2044 * @table: the sysctl table
2045 * @write: %TRUE if this is a write to the sysctl file
2046 * @buffer: the user buffer
2047 * @lenp: the size of the user buffer
2048 * @ppos: file position
2049 *
2050 * Reads/writes a string from/to the user buffer. If the kernel
2051 * buffer provided is not large enough to hold the string, the
2052 * string is truncated. The copied string is %NULL-terminated.
2053 * If the string is being read by the user process, it is copied
2054 * and a newline '\n' is added. It is truncated if the buffer is
2055 * not large enough.
2056 *
2057 * Returns 0 on success.
2058 */
2059 int proc_dostring(struct ctl_table *table, int write,
2060 void __user *buffer, size_t *lenp, loff_t *ppos)
2061 {
2062 if (write)
2063 proc_first_pos_non_zero_ignore(ppos, table);
2064
2065 return _proc_do_string((char *)(table->data), table->maxlen, write,
2066 (char __user *)buffer, lenp, ppos);
2067 }
2068
2069 static size_t proc_skip_spaces(char **buf)
2070 {
2071 size_t ret;
2072 char *tmp = skip_spaces(*buf);
2073 ret = tmp - *buf;
2074 *buf = tmp;
2075 return ret;
2076 }
2077
2078 static void proc_skip_char(char **buf, size_t *size, const char v)
2079 {
2080 while (*size) {
2081 if (**buf != v)
2082 break;
2083 (*size)--;
2084 (*buf)++;
2085 }
2086 }
2087
2088 #define TMPBUFLEN 22
2089 /**
2090 * proc_get_long - reads an ASCII formatted integer from a user buffer
2091 *
2092 * @buf: a kernel buffer
2093 * @size: size of the kernel buffer
2094 * @val: this is where the number will be stored
2095 * @neg: set to %TRUE if number is negative
2096 * @perm_tr: a vector which contains the allowed trailers
2097 * @perm_tr_len: size of the perm_tr vector
2098 * @tr: pointer to store the trailer character
2099 *
2100 * In case of success %0 is returned and @buf and @size are updated with
2101 * the amount of bytes read. If @tr is non-NULL and a trailing
2102 * character exists (size is non-zero after returning from this
2103 * function), @tr is updated with the trailing character.
2104 */
2105 static int proc_get_long(char **buf, size_t *size,
2106 unsigned long *val, bool *neg,
2107 const char *perm_tr, unsigned perm_tr_len, char *tr)
2108 {
2109 int len;
2110 char *p, tmp[TMPBUFLEN];
2111
2112 if (!*size)
2113 return -EINVAL;
2114
2115 len = *size;
2116 if (len > TMPBUFLEN - 1)
2117 len = TMPBUFLEN - 1;
2118
2119 memcpy(tmp, *buf, len);
2120
2121 tmp[len] = 0;
2122 p = tmp;
2123 if (*p == '-' && *size > 1) {
2124 *neg = true;
2125 p++;
2126 } else
2127 *neg = false;
2128 if (!isdigit(*p))
2129 return -EINVAL;
2130
2131 *val = simple_strtoul(p, &p, 0);
2132
2133 len = p - tmp;
2134
2135 /* We don't know if the next char is whitespace thus we may accept
2136 * invalid integers (e.g. 1234...a) or two integers instead of one
2137 * (e.g. 123...1). So lets not allow such large numbers. */
2138 if (len == TMPBUFLEN - 1)
2139 return -EINVAL;
2140
2141 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2142 return -EINVAL;
2143
2144 if (tr && (len < *size))
2145 *tr = *p;
2146
2147 *buf += len;
2148 *size -= len;
2149
2150 return 0;
2151 }
2152
2153 /**
2154 * proc_put_long - converts an integer to a decimal ASCII formatted string
2155 *
2156 * @buf: the user buffer
2157 * @size: the size of the user buffer
2158 * @val: the integer to be converted
2159 * @neg: sign of the number, %TRUE for negative
2160 *
2161 * In case of success %0 is returned and @buf and @size are updated with
2162 * the amount of bytes written.
2163 */
2164 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2165 bool neg)
2166 {
2167 int len;
2168 char tmp[TMPBUFLEN], *p = tmp;
2169
2170 sprintf(p, "%s%lu", neg ? "-" : "", val);
2171 len = strlen(tmp);
2172 if (len > *size)
2173 len = *size;
2174 if (copy_to_user(*buf, tmp, len))
2175 return -EFAULT;
2176 *size -= len;
2177 *buf += len;
2178 return 0;
2179 }
2180 #undef TMPBUFLEN
2181
2182 static int proc_put_char(void __user **buf, size_t *size, char c)
2183 {
2184 if (*size) {
2185 char __user **buffer = (char __user **)buf;
2186 if (put_user(c, *buffer))
2187 return -EFAULT;
2188 (*size)--, (*buffer)++;
2189 *buf = *buffer;
2190 }
2191 return 0;
2192 }
2193
2194 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2195 int *valp,
2196 int write, void *data)
2197 {
2198 if (write) {
2199 if (*negp) {
2200 if (*lvalp > (unsigned long) INT_MAX + 1)
2201 return -EINVAL;
2202 *valp = -*lvalp;
2203 } else {
2204 if (*lvalp > (unsigned long) INT_MAX)
2205 return -EINVAL;
2206 *valp = *lvalp;
2207 }
2208 } else {
2209 int val = *valp;
2210 if (val < 0) {
2211 *negp = true;
2212 *lvalp = -(unsigned long)val;
2213 } else {
2214 *negp = false;
2215 *lvalp = (unsigned long)val;
2216 }
2217 }
2218 return 0;
2219 }
2220
2221 static int do_proc_douintvec_conv(unsigned long *lvalp,
2222 unsigned int *valp,
2223 int write, void *data)
2224 {
2225 if (write) {
2226 if (*lvalp > UINT_MAX)
2227 return -EINVAL;
2228 *valp = *lvalp;
2229 } else {
2230 unsigned int val = *valp;
2231 *lvalp = (unsigned long)val;
2232 }
2233 return 0;
2234 }
2235
2236 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2237
2238 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2239 int write, void __user *buffer,
2240 size_t *lenp, loff_t *ppos,
2241 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2242 int write, void *data),
2243 void *data)
2244 {
2245 int *i, vleft, first = 1, err = 0;
2246 size_t left;
2247 char *kbuf = NULL, *p;
2248
2249 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2250 *lenp = 0;
2251 return 0;
2252 }
2253
2254 i = (int *) tbl_data;
2255 vleft = table->maxlen / sizeof(*i);
2256 left = *lenp;
2257
2258 if (!conv)
2259 conv = do_proc_dointvec_conv;
2260
2261 if (write) {
2262 if (proc_first_pos_non_zero_ignore(ppos, table))
2263 goto out;
2264
2265 if (left > PAGE_SIZE - 1)
2266 left = PAGE_SIZE - 1;
2267 p = kbuf = memdup_user_nul(buffer, left);
2268 if (IS_ERR(kbuf))
2269 return PTR_ERR(kbuf);
2270 }
2271
2272 for (; left && vleft--; i++, first=0) {
2273 unsigned long lval;
2274 bool neg;
2275
2276 if (write) {
2277 left -= proc_skip_spaces(&p);
2278
2279 if (!left)
2280 break;
2281 err = proc_get_long(&p, &left, &lval, &neg,
2282 proc_wspace_sep,
2283 sizeof(proc_wspace_sep), NULL);
2284 if (err)
2285 break;
2286 if (conv(&neg, &lval, i, 1, data)) {
2287 err = -EINVAL;
2288 break;
2289 }
2290 } else {
2291 if (conv(&neg, &lval, i, 0, data)) {
2292 err = -EINVAL;
2293 break;
2294 }
2295 if (!first)
2296 err = proc_put_char(&buffer, &left, '\t');
2297 if (err)
2298 break;
2299 err = proc_put_long(&buffer, &left, lval, neg);
2300 if (err)
2301 break;
2302 }
2303 }
2304
2305 if (!write && !first && left && !err)
2306 err = proc_put_char(&buffer, &left, '\n');
2307 if (write && !err && left)
2308 left -= proc_skip_spaces(&p);
2309 if (write) {
2310 kfree(kbuf);
2311 if (first)
2312 return err ? : -EINVAL;
2313 }
2314 *lenp -= left;
2315 out:
2316 *ppos += *lenp;
2317 return err;
2318 }
2319
2320 static int do_proc_dointvec(struct ctl_table *table, int write,
2321 void __user *buffer, size_t *lenp, loff_t *ppos,
2322 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2323 int write, void *data),
2324 void *data)
2325 {
2326 return __do_proc_dointvec(table->data, table, write,
2327 buffer, lenp, ppos, conv, data);
2328 }
2329
2330 static int do_proc_douintvec_w(unsigned int *tbl_data,
2331 struct ctl_table *table,
2332 void __user *buffer,
2333 size_t *lenp, loff_t *ppos,
2334 int (*conv)(unsigned long *lvalp,
2335 unsigned int *valp,
2336 int write, void *data),
2337 void *data)
2338 {
2339 unsigned long lval;
2340 int err = 0;
2341 size_t left;
2342 bool neg;
2343 char *kbuf = NULL, *p;
2344
2345 left = *lenp;
2346
2347 if (proc_first_pos_non_zero_ignore(ppos, table))
2348 goto bail_early;
2349
2350 if (left > PAGE_SIZE - 1)
2351 left = PAGE_SIZE - 1;
2352
2353 p = kbuf = memdup_user_nul(buffer, left);
2354 if (IS_ERR(kbuf))
2355 return -EINVAL;
2356
2357 left -= proc_skip_spaces(&p);
2358 if (!left) {
2359 err = -EINVAL;
2360 goto out_free;
2361 }
2362
2363 err = proc_get_long(&p, &left, &lval, &neg,
2364 proc_wspace_sep,
2365 sizeof(proc_wspace_sep), NULL);
2366 if (err || neg) {
2367 err = -EINVAL;
2368 goto out_free;
2369 }
2370
2371 if (conv(&lval, tbl_data, 1, data)) {
2372 err = -EINVAL;
2373 goto out_free;
2374 }
2375
2376 if (!err && left)
2377 left -= proc_skip_spaces(&p);
2378
2379 out_free:
2380 kfree(kbuf);
2381 if (err)
2382 return -EINVAL;
2383
2384 return 0;
2385
2386 /* This is in keeping with old __do_proc_dointvec() */
2387 bail_early:
2388 *ppos += *lenp;
2389 return err;
2390 }
2391
2392 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2393 size_t *lenp, loff_t *ppos,
2394 int (*conv)(unsigned long *lvalp,
2395 unsigned int *valp,
2396 int write, void *data),
2397 void *data)
2398 {
2399 unsigned long lval;
2400 int err = 0;
2401 size_t left;
2402
2403 left = *lenp;
2404
2405 if (conv(&lval, tbl_data, 0, data)) {
2406 err = -EINVAL;
2407 goto out;
2408 }
2409
2410 err = proc_put_long(&buffer, &left, lval, false);
2411 if (err || !left)
2412 goto out;
2413
2414 err = proc_put_char(&buffer, &left, '\n');
2415
2416 out:
2417 *lenp -= left;
2418 *ppos += *lenp;
2419
2420 return err;
2421 }
2422
2423 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2424 int write, void __user *buffer,
2425 size_t *lenp, loff_t *ppos,
2426 int (*conv)(unsigned long *lvalp,
2427 unsigned int *valp,
2428 int write, void *data),
2429 void *data)
2430 {
2431 unsigned int *i, vleft;
2432
2433 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2434 *lenp = 0;
2435 return 0;
2436 }
2437
2438 i = (unsigned int *) tbl_data;
2439 vleft = table->maxlen / sizeof(*i);
2440
2441 /*
2442 * Arrays are not supported, keep this simple. *Do not* add
2443 * support for them.
2444 */
2445 if (vleft != 1) {
2446 *lenp = 0;
2447 return -EINVAL;
2448 }
2449
2450 if (!conv)
2451 conv = do_proc_douintvec_conv;
2452
2453 if (write)
2454 return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2455 conv, data);
2456 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2457 }
2458
2459 static int do_proc_douintvec(struct ctl_table *table, int write,
2460 void __user *buffer, size_t *lenp, loff_t *ppos,
2461 int (*conv)(unsigned long *lvalp,
2462 unsigned int *valp,
2463 int write, void *data),
2464 void *data)
2465 {
2466 return __do_proc_douintvec(table->data, table, write,
2467 buffer, lenp, ppos, conv, data);
2468 }
2469
2470 /**
2471 * proc_dointvec - read a vector of integers
2472 * @table: the sysctl table
2473 * @write: %TRUE if this is a write to the sysctl file
2474 * @buffer: the user buffer
2475 * @lenp: the size of the user buffer
2476 * @ppos: file position
2477 *
2478 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2479 * values from/to the user buffer, treated as an ASCII string.
2480 *
2481 * Returns 0 on success.
2482 */
2483 int proc_dointvec(struct ctl_table *table, int write,
2484 void __user *buffer, size_t *lenp, loff_t *ppos)
2485 {
2486 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2487 }
2488
2489 /**
2490 * proc_douintvec - read a vector of unsigned integers
2491 * @table: the sysctl table
2492 * @write: %TRUE if this is a write to the sysctl file
2493 * @buffer: the user buffer
2494 * @lenp: the size of the user buffer
2495 * @ppos: file position
2496 *
2497 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2498 * values from/to the user buffer, treated as an ASCII string.
2499 *
2500 * Returns 0 on success.
2501 */
2502 int proc_douintvec(struct ctl_table *table, int write,
2503 void __user *buffer, size_t *lenp, loff_t *ppos)
2504 {
2505 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2506 do_proc_douintvec_conv, NULL);
2507 }
2508
2509 /*
2510 * Taint values can only be increased
2511 * This means we can safely use a temporary.
2512 */
2513 static int proc_taint(struct ctl_table *table, int write,
2514 void __user *buffer, size_t *lenp, loff_t *ppos)
2515 {
2516 struct ctl_table t;
2517 unsigned long tmptaint = get_taint();
2518 int err;
2519
2520 if (write && !capable(CAP_SYS_ADMIN))
2521 return -EPERM;
2522
2523 t = *table;
2524 t.data = &tmptaint;
2525 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2526 if (err < 0)
2527 return err;
2528
2529 if (write) {
2530 /*
2531 * Poor man's atomic or. Not worth adding a primitive
2532 * to everyone's atomic.h for this
2533 */
2534 int i;
2535 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2536 if ((tmptaint >> i) & 1)
2537 add_taint(i, LOCKDEP_STILL_OK);
2538 }
2539 }
2540
2541 return err;
2542 }
2543
2544 #ifdef CONFIG_PRINTK
2545 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2546 void __user *buffer, size_t *lenp, loff_t *ppos)
2547 {
2548 if (write && !capable(CAP_SYS_ADMIN))
2549 return -EPERM;
2550
2551 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2552 }
2553 #endif
2554
2555 /**
2556 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
2557 * @min: pointer to minimum allowable value
2558 * @max: pointer to maximum allowable value
2559 *
2560 * The do_proc_dointvec_minmax_conv_param structure provides the
2561 * minimum and maximum values for doing range checking for those sysctl
2562 * parameters that use the proc_dointvec_minmax() handler.
2563 */
2564 struct do_proc_dointvec_minmax_conv_param {
2565 int *min;
2566 int *max;
2567 };
2568
2569 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2570 int *valp,
2571 int write, void *data)
2572 {
2573 struct do_proc_dointvec_minmax_conv_param *param = data;
2574 if (write) {
2575 int val = *negp ? -*lvalp : *lvalp;
2576 if ((param->min && *param->min > val) ||
2577 (param->max && *param->max < val))
2578 return -EINVAL;
2579 *valp = val;
2580 } else {
2581 int val = *valp;
2582 if (val < 0) {
2583 *negp = true;
2584 *lvalp = -(unsigned long)val;
2585 } else {
2586 *negp = false;
2587 *lvalp = (unsigned long)val;
2588 }
2589 }
2590 return 0;
2591 }
2592
2593 /**
2594 * proc_dointvec_minmax - read a vector of integers with min/max values
2595 * @table: the sysctl table
2596 * @write: %TRUE if this is a write to the sysctl file
2597 * @buffer: the user buffer
2598 * @lenp: the size of the user buffer
2599 * @ppos: file position
2600 *
2601 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2602 * values from/to the user buffer, treated as an ASCII string.
2603 *
2604 * This routine will ensure the values are within the range specified by
2605 * table->extra1 (min) and table->extra2 (max).
2606 *
2607 * Returns 0 on success or -EINVAL on write when the range check fails.
2608 */
2609 int proc_dointvec_minmax(struct ctl_table *table, int write,
2610 void __user *buffer, size_t *lenp, loff_t *ppos)
2611 {
2612 struct do_proc_dointvec_minmax_conv_param param = {
2613 .min = (int *) table->extra1,
2614 .max = (int *) table->extra2,
2615 };
2616 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2617 do_proc_dointvec_minmax_conv, &param);
2618 }
2619
2620 /**
2621 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
2622 * @min: pointer to minimum allowable value
2623 * @max: pointer to maximum allowable value
2624 *
2625 * The do_proc_douintvec_minmax_conv_param structure provides the
2626 * minimum and maximum values for doing range checking for those sysctl
2627 * parameters that use the proc_douintvec_minmax() handler.
2628 */
2629 struct do_proc_douintvec_minmax_conv_param {
2630 unsigned int *min;
2631 unsigned int *max;
2632 };
2633
2634 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2635 unsigned int *valp,
2636 int write, void *data)
2637 {
2638 struct do_proc_douintvec_minmax_conv_param *param = data;
2639
2640 if (write) {
2641 unsigned int val = *lvalp;
2642
2643 if (*lvalp > UINT_MAX)
2644 return -EINVAL;
2645
2646 if ((param->min && *param->min > val) ||
2647 (param->max && *param->max < val))
2648 return -ERANGE;
2649
2650 *valp = val;
2651 } else {
2652 unsigned int val = *valp;
2653 *lvalp = (unsigned long) val;
2654 }
2655
2656 return 0;
2657 }
2658
2659 /**
2660 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2661 * @table: the sysctl table
2662 * @write: %TRUE if this is a write to the sysctl file
2663 * @buffer: the user buffer
2664 * @lenp: the size of the user buffer
2665 * @ppos: file position
2666 *
2667 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2668 * values from/to the user buffer, treated as an ASCII string. Negative
2669 * strings are not allowed.
2670 *
2671 * This routine will ensure the values are within the range specified by
2672 * table->extra1 (min) and table->extra2 (max). There is a final sanity
2673 * check for UINT_MAX to avoid having to support wrap around uses from
2674 * userspace.
2675 *
2676 * Returns 0 on success or -ERANGE on write when the range check fails.
2677 */
2678 int proc_douintvec_minmax(struct ctl_table *table, int write,
2679 void __user *buffer, size_t *lenp, loff_t *ppos)
2680 {
2681 struct do_proc_douintvec_minmax_conv_param param = {
2682 .min = (unsigned int *) table->extra1,
2683 .max = (unsigned int *) table->extra2,
2684 };
2685 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2686 do_proc_douintvec_minmax_conv, &param);
2687 }
2688
2689 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
2690 unsigned int *valp,
2691 int write, void *data)
2692 {
2693 if (write) {
2694 unsigned int val;
2695
2696 val = round_pipe_size(*lvalp);
2697 if (val == 0)
2698 return -EINVAL;
2699
2700 *valp = val;
2701 } else {
2702 unsigned int val = *valp;
2703 *lvalp = (unsigned long) val;
2704 }
2705
2706 return 0;
2707 }
2708
2709 static int proc_dopipe_max_size(struct ctl_table *table, int write,
2710 void __user *buffer, size_t *lenp, loff_t *ppos)
2711 {
2712 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2713 do_proc_dopipe_max_size_conv, NULL);
2714 }
2715
2716 static void validate_coredump_safety(void)
2717 {
2718 #ifdef CONFIG_COREDUMP
2719 if (suid_dumpable == SUID_DUMP_ROOT &&
2720 core_pattern[0] != '/' && core_pattern[0] != '|') {
2721 printk(KERN_WARNING
2722 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2723 "Pipe handler or fully qualified core dump path required.\n"
2724 "Set kernel.core_pattern before fs.suid_dumpable.\n"
2725 );
2726 }
2727 #endif
2728 }
2729
2730 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2731 void __user *buffer, size_t *lenp, loff_t *ppos)
2732 {
2733 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2734 if (!error)
2735 validate_coredump_safety();
2736 return error;
2737 }
2738
2739 #ifdef CONFIG_COREDUMP
2740 static int proc_dostring_coredump(struct ctl_table *table, int write,
2741 void __user *buffer, size_t *lenp, loff_t *ppos)
2742 {
2743 int error = proc_dostring(table, write, buffer, lenp, ppos);
2744 if (!error)
2745 validate_coredump_safety();
2746 return error;
2747 }
2748 #endif
2749
2750 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2751 void __user *buffer,
2752 size_t *lenp, loff_t *ppos,
2753 unsigned long convmul,
2754 unsigned long convdiv)
2755 {
2756 unsigned long *i, *min, *max;
2757 int vleft, first = 1, err = 0;
2758 size_t left;
2759 char *kbuf = NULL, *p;
2760
2761 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2762 *lenp = 0;
2763 return 0;
2764 }
2765
2766 i = (unsigned long *) data;
2767 min = (unsigned long *) table->extra1;
2768 max = (unsigned long *) table->extra2;
2769 vleft = table->maxlen / sizeof(unsigned long);
2770 left = *lenp;
2771
2772 if (write) {
2773 if (proc_first_pos_non_zero_ignore(ppos, table))
2774 goto out;
2775
2776 if (left > PAGE_SIZE - 1)
2777 left = PAGE_SIZE - 1;
2778 p = kbuf = memdup_user_nul(buffer, left);
2779 if (IS_ERR(kbuf))
2780 return PTR_ERR(kbuf);
2781 }
2782
2783 for (; left && vleft--; i++, first = 0) {
2784 unsigned long val;
2785
2786 if (write) {
2787 bool neg;
2788
2789 left -= proc_skip_spaces(&p);
2790
2791 err = proc_get_long(&p, &left, &val, &neg,
2792 proc_wspace_sep,
2793 sizeof(proc_wspace_sep), NULL);
2794 if (err)
2795 break;
2796 if (neg)
2797 continue;
2798 val = convmul * val / convdiv;
2799 if ((min && val < *min) || (max && val > *max))
2800 continue;
2801 *i = val;
2802 } else {
2803 val = convdiv * (*i) / convmul;
2804 if (!first) {
2805 err = proc_put_char(&buffer, &left, '\t');
2806 if (err)
2807 break;
2808 }
2809 err = proc_put_long(&buffer, &left, val, false);
2810 if (err)
2811 break;
2812 }
2813 }
2814
2815 if (!write && !first && left && !err)
2816 err = proc_put_char(&buffer, &left, '\n');
2817 if (write && !err)
2818 left -= proc_skip_spaces(&p);
2819 if (write) {
2820 kfree(kbuf);
2821 if (first)
2822 return err ? : -EINVAL;
2823 }
2824 *lenp -= left;
2825 out:
2826 *ppos += *lenp;
2827 return err;
2828 }
2829
2830 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2831 void __user *buffer,
2832 size_t *lenp, loff_t *ppos,
2833 unsigned long convmul,
2834 unsigned long convdiv)
2835 {
2836 return __do_proc_doulongvec_minmax(table->data, table, write,
2837 buffer, lenp, ppos, convmul, convdiv);
2838 }
2839
2840 /**
2841 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2842 * @table: the sysctl table
2843 * @write: %TRUE if this is a write to the sysctl file
2844 * @buffer: the user buffer
2845 * @lenp: the size of the user buffer
2846 * @ppos: file position
2847 *
2848 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2849 * values from/to the user buffer, treated as an ASCII string.
2850 *
2851 * This routine will ensure the values are within the range specified by
2852 * table->extra1 (min) and table->extra2 (max).
2853 *
2854 * Returns 0 on success.
2855 */
2856 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2857 void __user *buffer, size_t *lenp, loff_t *ppos)
2858 {
2859 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2860 }
2861
2862 /**
2863 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2864 * @table: the sysctl table
2865 * @write: %TRUE if this is a write to the sysctl file
2866 * @buffer: the user buffer
2867 * @lenp: the size of the user buffer
2868 * @ppos: file position
2869 *
2870 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2871 * values from/to the user buffer, treated as an ASCII string. The values
2872 * are treated as milliseconds, and converted to jiffies when they are stored.
2873 *
2874 * This routine will ensure the values are within the range specified by
2875 * table->extra1 (min) and table->extra2 (max).
2876 *
2877 * Returns 0 on success.
2878 */
2879 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2880 void __user *buffer,
2881 size_t *lenp, loff_t *ppos)
2882 {
2883 return do_proc_doulongvec_minmax(table, write, buffer,
2884 lenp, ppos, HZ, 1000l);
2885 }
2886
2887
2888 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2889 int *valp,
2890 int write, void *data)
2891 {
2892 if (write) {
2893 if (*lvalp > INT_MAX / HZ)
2894 return 1;
2895 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2896 } else {
2897 int val = *valp;
2898 unsigned long lval;
2899 if (val < 0) {
2900 *negp = true;
2901 lval = -(unsigned long)val;
2902 } else {
2903 *negp = false;
2904 lval = (unsigned long)val;
2905 }
2906 *lvalp = lval / HZ;
2907 }
2908 return 0;
2909 }
2910
2911 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2912 int *valp,
2913 int write, void *data)
2914 {
2915 if (write) {
2916 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2917 return 1;
2918 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2919 } else {
2920 int val = *valp;
2921 unsigned long lval;
2922 if (val < 0) {
2923 *negp = true;
2924 lval = -(unsigned long)val;
2925 } else {
2926 *negp = false;
2927 lval = (unsigned long)val;
2928 }
2929 *lvalp = jiffies_to_clock_t(lval);
2930 }
2931 return 0;
2932 }
2933
2934 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2935 int *valp,
2936 int write, void *data)
2937 {
2938 if (write) {
2939 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2940
2941 if (jif > INT_MAX)
2942 return 1;
2943 *valp = (int)jif;
2944 } else {
2945 int val = *valp;
2946 unsigned long lval;
2947 if (val < 0) {
2948 *negp = true;
2949 lval = -(unsigned long)val;
2950 } else {
2951 *negp = false;
2952 lval = (unsigned long)val;
2953 }
2954 *lvalp = jiffies_to_msecs(lval);
2955 }
2956 return 0;
2957 }
2958
2959 /**
2960 * proc_dointvec_jiffies - read a vector of integers as seconds
2961 * @table: the sysctl table
2962 * @write: %TRUE if this is a write to the sysctl file
2963 * @buffer: the user buffer
2964 * @lenp: the size of the user buffer
2965 * @ppos: file position
2966 *
2967 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2968 * values from/to the user buffer, treated as an ASCII string.
2969 * The values read are assumed to be in seconds, and are converted into
2970 * jiffies.
2971 *
2972 * Returns 0 on success.
2973 */
2974 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2975 void __user *buffer, size_t *lenp, loff_t *ppos)
2976 {
2977 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2978 do_proc_dointvec_jiffies_conv,NULL);
2979 }
2980
2981 /**
2982 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2983 * @table: the sysctl table
2984 * @write: %TRUE if this is a write to the sysctl file
2985 * @buffer: the user buffer
2986 * @lenp: the size of the user buffer
2987 * @ppos: pointer to the file position
2988 *
2989 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2990 * values from/to the user buffer, treated as an ASCII string.
2991 * The values read are assumed to be in 1/USER_HZ seconds, and
2992 * are converted into jiffies.
2993 *
2994 * Returns 0 on success.
2995 */
2996 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2997 void __user *buffer, size_t *lenp, loff_t *ppos)
2998 {
2999 return do_proc_dointvec(table,write,buffer,lenp,ppos,
3000 do_proc_dointvec_userhz_jiffies_conv,NULL);
3001 }
3002
3003 /**
3004 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
3005 * @table: the sysctl table
3006 * @write: %TRUE if this is a write to the sysctl file
3007 * @buffer: the user buffer
3008 * @lenp: the size of the user buffer
3009 * @ppos: file position
3010 * @ppos: the current position in the file
3011 *
3012 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3013 * values from/to the user buffer, treated as an ASCII string.
3014 * The values read are assumed to be in 1/1000 seconds, and
3015 * are converted into jiffies.
3016 *
3017 * Returns 0 on success.
3018 */
3019 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3020 void __user *buffer, size_t *lenp, loff_t *ppos)
3021 {
3022 return do_proc_dointvec(table, write, buffer, lenp, ppos,
3023 do_proc_dointvec_ms_jiffies_conv, NULL);
3024 }
3025
3026 static int proc_do_cad_pid(struct ctl_table *table, int write,
3027 void __user *buffer, size_t *lenp, loff_t *ppos)
3028 {
3029 struct pid *new_pid;
3030 pid_t tmp;
3031 int r;
3032
3033 tmp = pid_vnr(cad_pid);
3034
3035 r = __do_proc_dointvec(&tmp, table, write, buffer,
3036 lenp, ppos, NULL, NULL);
3037 if (r || !write)
3038 return r;
3039
3040 new_pid = find_get_pid(tmp);
3041 if (!new_pid)
3042 return -ESRCH;
3043
3044 put_pid(xchg(&cad_pid, new_pid));
3045 return 0;
3046 }
3047
3048 /**
3049 * proc_do_large_bitmap - read/write from/to a large bitmap
3050 * @table: the sysctl table
3051 * @write: %TRUE if this is a write to the sysctl file
3052 * @buffer: the user buffer
3053 * @lenp: the size of the user buffer
3054 * @ppos: file position
3055 *
3056 * The bitmap is stored at table->data and the bitmap length (in bits)
3057 * in table->maxlen.
3058 *
3059 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
3060 * large bitmaps may be represented in a compact manner. Writing into
3061 * the file will clear the bitmap then update it with the given input.
3062 *
3063 * Returns 0 on success.
3064 */
3065 int proc_do_large_bitmap(struct ctl_table *table, int write,
3066 void __user *buffer, size_t *lenp, loff_t *ppos)
3067 {
3068 int err = 0;
3069 bool first = 1;
3070 size_t left = *lenp;
3071 unsigned long bitmap_len = table->maxlen;
3072 unsigned long *bitmap = *(unsigned long **) table->data;
3073 unsigned long *tmp_bitmap = NULL;
3074 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
3075
3076 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
3077 *lenp = 0;
3078 return 0;
3079 }
3080
3081 if (write) {
3082 char *kbuf, *p;
3083
3084 if (left > PAGE_SIZE - 1)
3085 left = PAGE_SIZE - 1;
3086
3087 p = kbuf = memdup_user_nul(buffer, left);
3088 if (IS_ERR(kbuf))
3089 return PTR_ERR(kbuf);
3090
3091 tmp_bitmap = kcalloc(BITS_TO_LONGS(bitmap_len),
3092 sizeof(unsigned long),
3093 GFP_KERNEL);
3094 if (!tmp_bitmap) {
3095 kfree(kbuf);
3096 return -ENOMEM;
3097 }
3098 proc_skip_char(&p, &left, '\n');
3099 while (!err && left) {
3100 unsigned long val_a, val_b;
3101 bool neg;
3102
3103 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3104 sizeof(tr_a), &c);
3105 if (err)
3106 break;
3107 if (val_a >= bitmap_len || neg) {
3108 err = -EINVAL;
3109 break;
3110 }
3111
3112 val_b = val_a;
3113 if (left) {
3114 p++;
3115 left--;
3116 }
3117
3118 if (c == '-') {
3119 err = proc_get_long(&p, &left, &val_b,
3120 &neg, tr_b, sizeof(tr_b),
3121 &c);
3122 if (err)
3123 break;
3124 if (val_b >= bitmap_len || neg ||
3125 val_a > val_b) {
3126 err = -EINVAL;
3127 break;
3128 }
3129 if (left) {
3130 p++;
3131 left--;
3132 }
3133 }
3134
3135 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3136 first = 0;
3137 proc_skip_char(&p, &left, '\n');
3138 }
3139 kfree(kbuf);
3140 } else {
3141 unsigned long bit_a, bit_b = 0;
3142
3143 while (left) {
3144 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3145 if (bit_a >= bitmap_len)
3146 break;
3147 bit_b = find_next_zero_bit(bitmap, bitmap_len,
3148 bit_a + 1) - 1;
3149
3150 if (!first) {
3151 err = proc_put_char(&buffer, &left, ',');
3152 if (err)
3153 break;
3154 }
3155 err = proc_put_long(&buffer, &left, bit_a, false);
3156 if (err)
3157 break;
3158 if (bit_a != bit_b) {
3159 err = proc_put_char(&buffer, &left, '-');
3160 if (err)
3161 break;
3162 err = proc_put_long(&buffer, &left, bit_b, false);
3163 if (err)
3164 break;
3165 }
3166
3167 first = 0; bit_b++;
3168 }
3169 if (!err)
3170 err = proc_put_char(&buffer, &left, '\n');
3171 }
3172
3173 if (!err) {
3174 if (write) {
3175 if (*ppos)
3176 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3177 else
3178 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3179 }
3180 *lenp -= left;
3181 *ppos += *lenp;
3182 }
3183
3184 kfree(tmp_bitmap);
3185 return err;
3186 }
3187
3188 #else /* CONFIG_PROC_SYSCTL */
3189
3190 int proc_dostring(struct ctl_table *table, int write,
3191 void __user *buffer, size_t *lenp, loff_t *ppos)
3192 {
3193 return -ENOSYS;
3194 }
3195
3196 int proc_dointvec(struct ctl_table *table, int write,
3197 void __user *buffer, size_t *lenp, loff_t *ppos)
3198 {
3199 return -ENOSYS;
3200 }
3201
3202 int proc_douintvec(struct ctl_table *table, int write,
3203 void __user *buffer, size_t *lenp, loff_t *ppos)
3204 {
3205 return -ENOSYS;
3206 }
3207
3208 int proc_dointvec_minmax(struct ctl_table *table, int write,
3209 void __user *buffer, size_t *lenp, loff_t *ppos)
3210 {
3211 return -ENOSYS;
3212 }
3213
3214 int proc_douintvec_minmax(struct ctl_table *table, int write,
3215 void __user *buffer, size_t *lenp, loff_t *ppos)
3216 {
3217 return -ENOSYS;
3218 }
3219
3220 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3221 void __user *buffer, size_t *lenp, loff_t *ppos)
3222 {
3223 return -ENOSYS;
3224 }
3225
3226 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3227 void __user *buffer, size_t *lenp, loff_t *ppos)
3228 {
3229 return -ENOSYS;
3230 }
3231
3232 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3233 void __user *buffer, size_t *lenp, loff_t *ppos)
3234 {
3235 return -ENOSYS;
3236 }
3237
3238 int proc_doulongvec_minmax(struct ctl_table *table, int write,
3239 void __user *buffer, size_t *lenp, loff_t *ppos)
3240 {
3241 return -ENOSYS;
3242 }
3243
3244 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3245 void __user *buffer,
3246 size_t *lenp, loff_t *ppos)
3247 {
3248 return -ENOSYS;
3249 }
3250
3251
3252 #endif /* CONFIG_PROC_SYSCTL */
3253
3254 /*
3255 * No sense putting this after each symbol definition, twice,
3256 * exception granted :-)
3257 */
3258 EXPORT_SYMBOL(proc_dointvec);
3259 EXPORT_SYMBOL(proc_douintvec);
3260 EXPORT_SYMBOL(proc_dointvec_jiffies);
3261 EXPORT_SYMBOL(proc_dointvec_minmax);
3262 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3263 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3264 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3265 EXPORT_SYMBOL(proc_dostring);
3266 EXPORT_SYMBOL(proc_doulongvec_minmax);
3267 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);