]> git.proxmox.com Git - mirror_ubuntu-jammy-kernel.git/blob - kernel/sysctl.c
projects / mirror_ubuntu-jammy-kernel.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Merge tag 'apparmor-pr-2018-06-13' of git://git.kernel.org/pub/scm/linux/kernel/git...
[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 {
372 .procname = "sched_time_avg_ms",
373 .data = &sysctl_sched_time_avg,
374 .maxlen = sizeof(unsigned int),
375 .mode = 0644,
376 .proc_handler = proc_dointvec_minmax,
377 .extra1 = &one,
378 },
379 #ifdef CONFIG_SCHEDSTATS
380 {
381 .procname = "sched_schedstats",
382 .data = NULL,
383 .maxlen = sizeof(unsigned int),
384 .mode = 0644,
385 .proc_handler = sysctl_schedstats,
386 .extra1 = &zero,
387 .extra2 = &one,
388 },
389 #endif /* CONFIG_SCHEDSTATS */
390 #endif /* CONFIG_SMP */
391 #ifdef CONFIG_NUMA_BALANCING
392 {
393 .procname = "numa_balancing_scan_delay_ms",
394 .data = &sysctl_numa_balancing_scan_delay,
395 .maxlen = sizeof(unsigned int),
396 .mode = 0644,
397 .proc_handler = proc_dointvec,
398 },
399 {
400 .procname = "numa_balancing_scan_period_min_ms",
401 .data = &sysctl_numa_balancing_scan_period_min,
402 .maxlen = sizeof(unsigned int),
403 .mode = 0644,
404 .proc_handler = proc_dointvec,
405 },
406 {
407 .procname = "numa_balancing_scan_period_max_ms",
408 .data = &sysctl_numa_balancing_scan_period_max,
409 .maxlen = sizeof(unsigned int),
410 .mode = 0644,
411 .proc_handler = proc_dointvec,
412 },
413 {
414 .procname = "numa_balancing_scan_size_mb",
415 .data = &sysctl_numa_balancing_scan_size,
416 .maxlen = sizeof(unsigned int),
417 .mode = 0644,
418 .proc_handler = proc_dointvec_minmax,
419 .extra1 = &one,
420 },
421 {
422 .procname = "numa_balancing",
423 .data = NULL, /* filled in by handler */
424 .maxlen = sizeof(unsigned int),
425 .mode = 0644,
426 .proc_handler = sysctl_numa_balancing,
427 .extra1 = &zero,
428 .extra2 = &one,
429 },
430 #endif /* CONFIG_NUMA_BALANCING */
431 #endif /* CONFIG_SCHED_DEBUG */
432 {
433 .procname = "sched_rt_period_us",
434 .data = &sysctl_sched_rt_period,
435 .maxlen = sizeof(unsigned int),
436 .mode = 0644,
437 .proc_handler = sched_rt_handler,
438 },
439 {
440 .procname = "sched_rt_runtime_us",
441 .data = &sysctl_sched_rt_runtime,
442 .maxlen = sizeof(int),
443 .mode = 0644,
444 .proc_handler = sched_rt_handler,
445 },
446 {
447 .procname = "sched_rr_timeslice_ms",
448 .data = &sysctl_sched_rr_timeslice,
449 .maxlen = sizeof(int),
450 .mode = 0644,
451 .proc_handler = sched_rr_handler,
452 },
453 #ifdef CONFIG_SCHED_AUTOGROUP
454 {
455 .procname = "sched_autogroup_enabled",
456 .data = &sysctl_sched_autogroup_enabled,
457 .maxlen = sizeof(unsigned int),
458 .mode = 0644,
459 .proc_handler = proc_dointvec_minmax,
460 .extra1 = &zero,
461 .extra2 = &one,
462 },
463 #endif
464 #ifdef CONFIG_CFS_BANDWIDTH
465 {
466 .procname = "sched_cfs_bandwidth_slice_us",
467 .data = &sysctl_sched_cfs_bandwidth_slice,
468 .maxlen = sizeof(unsigned int),
469 .mode = 0644,
470 .proc_handler = proc_dointvec_minmax,
471 .extra1 = &one,
472 },
473 #endif
474 #ifdef CONFIG_PROVE_LOCKING
475 {
476 .procname = "prove_locking",
477 .data = &prove_locking,
478 .maxlen = sizeof(int),
479 .mode = 0644,
480 .proc_handler = proc_dointvec,
481 },
482 #endif
483 #ifdef CONFIG_LOCK_STAT
484 {
485 .procname = "lock_stat",
486 .data = &lock_stat,
487 .maxlen = sizeof(int),
488 .mode = 0644,
489 .proc_handler = proc_dointvec,
490 },
491 #endif
492 {
493 .procname = "panic",
494 .data = &panic_timeout,
495 .maxlen = sizeof(int),
496 .mode = 0644,
497 .proc_handler = proc_dointvec,
498 },
499 #ifdef CONFIG_COREDUMP
500 {
501 .procname = "core_uses_pid",
502 .data = &core_uses_pid,
503 .maxlen = sizeof(int),
504 .mode = 0644,
505 .proc_handler = proc_dointvec,
506 },
507 {
508 .procname = "core_pattern",
509 .data = core_pattern,
510 .maxlen = CORENAME_MAX_SIZE,
511 .mode = 0644,
512 .proc_handler = proc_dostring_coredump,
513 },
514 {
515 .procname = "core_pipe_limit",
516 .data = &core_pipe_limit,
517 .maxlen = sizeof(unsigned int),
518 .mode = 0644,
519 .proc_handler = proc_dointvec,
520 },
521 #endif
522 #ifdef CONFIG_PROC_SYSCTL
523 {
524 .procname = "tainted",
525 .maxlen = sizeof(long),
526 .mode = 0644,
527 .proc_handler = proc_taint,
528 },
529 {
530 .procname = "sysctl_writes_strict",
531 .data = &sysctl_writes_strict,
532 .maxlen = sizeof(int),
533 .mode = 0644,
534 .proc_handler = proc_dointvec_minmax,
535 .extra1 = &neg_one,
536 .extra2 = &one,
537 },
538 #endif
539 #ifdef CONFIG_LATENCYTOP
540 {
541 .procname = "latencytop",
542 .data = &latencytop_enabled,
543 .maxlen = sizeof(int),
544 .mode = 0644,
545 .proc_handler = sysctl_latencytop,
546 },
547 #endif
548 #ifdef CONFIG_BLK_DEV_INITRD
549 {
550 .procname = "real-root-dev",
551 .data = &real_root_dev,
552 .maxlen = sizeof(int),
553 .mode = 0644,
554 .proc_handler = proc_dointvec,
555 },
556 #endif
557 {
558 .procname = "print-fatal-signals",
559 .data = &print_fatal_signals,
560 .maxlen = sizeof(int),
561 .mode = 0644,
562 .proc_handler = proc_dointvec,
563 },
564 #ifdef CONFIG_SPARC
565 {
566 .procname = "reboot-cmd",
567 .data = reboot_command,
568 .maxlen = 256,
569 .mode = 0644,
570 .proc_handler = proc_dostring,
571 },
572 {
573 .procname = "stop-a",
574 .data = &stop_a_enabled,
575 .maxlen = sizeof (int),
576 .mode = 0644,
577 .proc_handler = proc_dointvec,
578 },
579 {
580 .procname = "scons-poweroff",
581 .data = &scons_pwroff,
582 .maxlen = sizeof (int),
583 .mode = 0644,
584 .proc_handler = proc_dointvec,
585 },
586 #endif
587 #ifdef CONFIG_SPARC64
588 {
589 .procname = "tsb-ratio",
590 .data = &sysctl_tsb_ratio,
591 .maxlen = sizeof (int),
592 .mode = 0644,
593 .proc_handler = proc_dointvec,
594 },
595 #endif
596 #ifdef __hppa__
597 {
598 .procname = "soft-power",
599 .data = &pwrsw_enabled,
600 .maxlen = sizeof (int),
601 .mode = 0644,
602 .proc_handler = proc_dointvec,
603 },
604 #endif
605 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
606 {
607 .procname = "unaligned-trap",
608 .data = &unaligned_enabled,
609 .maxlen = sizeof (int),
610 .mode = 0644,
611 .proc_handler = proc_dointvec,
612 },
613 #endif
614 {
615 .procname = "ctrl-alt-del",
616 .data = &C_A_D,
617 .maxlen = sizeof(int),
618 .mode = 0644,
619 .proc_handler = proc_dointvec,
620 },
621 #ifdef CONFIG_FUNCTION_TRACER
622 {
623 .procname = "ftrace_enabled",
624 .data = &ftrace_enabled,
625 .maxlen = sizeof(int),
626 .mode = 0644,
627 .proc_handler = ftrace_enable_sysctl,
628 },
629 #endif
630 #ifdef CONFIG_STACK_TRACER
631 {
632 .procname = "stack_tracer_enabled",
633 .data = &stack_tracer_enabled,
634 .maxlen = sizeof(int),
635 .mode = 0644,
636 .proc_handler = stack_trace_sysctl,
637 },
638 #endif
639 #ifdef CONFIG_TRACING
640 {
641 .procname = "ftrace_dump_on_oops",
642 .data = &ftrace_dump_on_oops,
643 .maxlen = sizeof(int),
644 .mode = 0644,
645 .proc_handler = proc_dointvec,
646 },
647 {
648 .procname = "traceoff_on_warning",
649 .data = &__disable_trace_on_warning,
650 .maxlen = sizeof(__disable_trace_on_warning),
651 .mode = 0644,
652 .proc_handler = proc_dointvec,
653 },
654 {
655 .procname = "tracepoint_printk",
656 .data = &tracepoint_printk,
657 .maxlen = sizeof(tracepoint_printk),
658 .mode = 0644,
659 .proc_handler = tracepoint_printk_sysctl,
660 },
661 #endif
662 #ifdef CONFIG_KEXEC_CORE
663 {
664 .procname = "kexec_load_disabled",
665 .data = &kexec_load_disabled,
666 .maxlen = sizeof(int),
667 .mode = 0644,
668 /* only handle a transition from default "0" to "1" */
669 .proc_handler = proc_dointvec_minmax,
670 .extra1 = &one,
671 .extra2 = &one,
672 },
673 #endif
674 #ifdef CONFIG_MODULES
675 {
676 .procname = "modprobe",
677 .data = &modprobe_path,
678 .maxlen = KMOD_PATH_LEN,
679 .mode = 0644,
680 .proc_handler = proc_dostring,
681 },
682 {
683 .procname = "modules_disabled",
684 .data = &modules_disabled,
685 .maxlen = sizeof(int),
686 .mode = 0644,
687 /* only handle a transition from default "0" to "1" */
688 .proc_handler = proc_dointvec_minmax,
689 .extra1 = &one,
690 .extra2 = &one,
691 },
692 #endif
693 #ifdef CONFIG_UEVENT_HELPER
694 {
695 .procname = "hotplug",
696 .data = &uevent_helper,
697 .maxlen = UEVENT_HELPER_PATH_LEN,
698 .mode = 0644,
699 .proc_handler = proc_dostring,
700 },
701 #endif
702 #ifdef CONFIG_CHR_DEV_SG
703 {
704 .procname = "sg-big-buff",
705 .data = &sg_big_buff,
706 .maxlen = sizeof (int),
707 .mode = 0444,
708 .proc_handler = proc_dointvec,
709 },
710 #endif
711 #ifdef CONFIG_BSD_PROCESS_ACCT
712 {
713 .procname = "acct",
714 .data = &acct_parm,
715 .maxlen = 3*sizeof(int),
716 .mode = 0644,
717 .proc_handler = proc_dointvec,
718 },
719 #endif
720 #ifdef CONFIG_MAGIC_SYSRQ
721 {
722 .procname = "sysrq",
723 .data = &__sysrq_enabled,
724 .maxlen = sizeof (int),
725 .mode = 0644,
726 .proc_handler = sysrq_sysctl_handler,
727 },
728 #endif
729 #ifdef CONFIG_PROC_SYSCTL
730 {
731 .procname = "cad_pid",
732 .data = NULL,
733 .maxlen = sizeof (int),
734 .mode = 0600,
735 .proc_handler = proc_do_cad_pid,
736 },
737 #endif
738 {
739 .procname = "threads-max",
740 .data = NULL,
741 .maxlen = sizeof(int),
742 .mode = 0644,
743 .proc_handler = sysctl_max_threads,
744 },
745 {
746 .procname = "random",
747 .mode = 0555,
748 .child = random_table,
749 },
750 {
751 .procname = "usermodehelper",
752 .mode = 0555,
753 .child = usermodehelper_table,
754 },
755 #ifdef CONFIG_FW_LOADER_USER_HELPER
756 {
757 .procname = "firmware_config",
758 .mode = 0555,
759 .child = firmware_config_table,
760 },
761 #endif
762 {
763 .procname = "overflowuid",
764 .data = &overflowuid,
765 .maxlen = sizeof(int),
766 .mode = 0644,
767 .proc_handler = proc_dointvec_minmax,
768 .extra1 = &minolduid,
769 .extra2 = &maxolduid,
770 },
771 {
772 .procname = "overflowgid",
773 .data = &overflowgid,
774 .maxlen = sizeof(int),
775 .mode = 0644,
776 .proc_handler = proc_dointvec_minmax,
777 .extra1 = &minolduid,
778 .extra2 = &maxolduid,
779 },
780 #ifdef CONFIG_S390
781 #ifdef CONFIG_MATHEMU
782 {
783 .procname = "ieee_emulation_warnings",
784 .data = &sysctl_ieee_emulation_warnings,
785 .maxlen = sizeof(int),
786 .mode = 0644,
787 .proc_handler = proc_dointvec,
788 },
789 #endif
790 {
791 .procname = "userprocess_debug",
792 .data = &show_unhandled_signals,
793 .maxlen = sizeof(int),
794 .mode = 0644,
795 .proc_handler = proc_dointvec,
796 },
797 #endif
798 {
799 .procname = "pid_max",
800 .data = &pid_max,
801 .maxlen = sizeof (int),
802 .mode = 0644,
803 .proc_handler = proc_dointvec_minmax,
804 .extra1 = &pid_max_min,
805 .extra2 = &pid_max_max,
806 },
807 {
808 .procname = "panic_on_oops",
809 .data = &panic_on_oops,
810 .maxlen = sizeof(int),
811 .mode = 0644,
812 .proc_handler = proc_dointvec,
813 },
814 #if defined CONFIG_PRINTK
815 {
816 .procname = "printk",
817 .data = &console_loglevel,
818 .maxlen = 4*sizeof(int),
819 .mode = 0644,
820 .proc_handler = proc_dointvec,
821 },
822 {
823 .procname = "printk_ratelimit",
824 .data = &printk_ratelimit_state.interval,
825 .maxlen = sizeof(int),
826 .mode = 0644,
827 .proc_handler = proc_dointvec_jiffies,
828 },
829 {
830 .procname = "printk_ratelimit_burst",
831 .data = &printk_ratelimit_state.burst,
832 .maxlen = sizeof(int),
833 .mode = 0644,
834 .proc_handler = proc_dointvec,
835 },
836 {
837 .procname = "printk_delay",
838 .data = &printk_delay_msec,
839 .maxlen = sizeof(int),
840 .mode = 0644,
841 .proc_handler = proc_dointvec_minmax,
842 .extra1 = &zero,
843 .extra2 = &ten_thousand,
844 },
845 {
846 .procname = "printk_devkmsg",
847 .data = devkmsg_log_str,
848 .maxlen = DEVKMSG_STR_MAX_SIZE,
849 .mode = 0644,
850 .proc_handler = devkmsg_sysctl_set_loglvl,
851 },
852 {
853 .procname = "dmesg_restrict",
854 .data = &dmesg_restrict,
855 .maxlen = sizeof(int),
856 .mode = 0644,
857 .proc_handler = proc_dointvec_minmax_sysadmin,
858 .extra1 = &zero,
859 .extra2 = &one,
860 },
861 {
862 .procname = "kptr_restrict",
863 .data = &kptr_restrict,
864 .maxlen = sizeof(int),
865 .mode = 0644,
866 .proc_handler = proc_dointvec_minmax_sysadmin,
867 .extra1 = &zero,
868 .extra2 = &two,
869 },
870 #endif
871 {
872 .procname = "ngroups_max",
873 .data = &ngroups_max,
874 .maxlen = sizeof (int),
875 .mode = 0444,
876 .proc_handler = proc_dointvec,
877 },
878 {
879 .procname = "cap_last_cap",
880 .data = (void *)&cap_last_cap,
881 .maxlen = sizeof(int),
882 .mode = 0444,
883 .proc_handler = proc_dointvec,
884 },
885 #if defined(CONFIG_LOCKUP_DETECTOR)
886 {
887 .procname = "watchdog",
888 .data = &watchdog_user_enabled,
889 .maxlen = sizeof(int),
890 .mode = 0644,
891 .proc_handler = proc_watchdog,
892 .extra1 = &zero,
893 .extra2 = &one,
894 },
895 {
896 .procname = "watchdog_thresh",
897 .data = &watchdog_thresh,
898 .maxlen = sizeof(int),
899 .mode = 0644,
900 .proc_handler = proc_watchdog_thresh,
901 .extra1 = &zero,
902 .extra2 = &sixty,
903 },
904 {
905 .procname = "nmi_watchdog",
906 .data = &nmi_watchdog_user_enabled,
907 .maxlen = sizeof(int),
908 .mode = NMI_WATCHDOG_SYSCTL_PERM,
909 .proc_handler = proc_nmi_watchdog,
910 .extra1 = &zero,
911 .extra2 = &one,
912 },
913 {
914 .procname = "watchdog_cpumask",
915 .data = &watchdog_cpumask_bits,
916 .maxlen = NR_CPUS,
917 .mode = 0644,
918 .proc_handler = proc_watchdog_cpumask,
919 },
920 #ifdef CONFIG_SOFTLOCKUP_DETECTOR
921 {
922 .procname = "soft_watchdog",
923 .data = &soft_watchdog_user_enabled,
924 .maxlen = sizeof(int),
925 .mode = 0644,
926 .proc_handler = proc_soft_watchdog,
927 .extra1 = &zero,
928 .extra2 = &one,
929 },
930 {
931 .procname = "softlockup_panic",
932 .data = &softlockup_panic,
933 .maxlen = sizeof(int),
934 .mode = 0644,
935 .proc_handler = proc_dointvec_minmax,
936 .extra1 = &zero,
937 .extra2 = &one,
938 },
939 #ifdef CONFIG_SMP
940 {
941 .procname = "softlockup_all_cpu_backtrace",
942 .data = &sysctl_softlockup_all_cpu_backtrace,
943 .maxlen = sizeof(int),
944 .mode = 0644,
945 .proc_handler = proc_dointvec_minmax,
946 .extra1 = &zero,
947 .extra2 = &one,
948 },
949 #endif /* CONFIG_SMP */
950 #endif
951 #ifdef CONFIG_HARDLOCKUP_DETECTOR
952 {
953 .procname = "hardlockup_panic",
954 .data = &hardlockup_panic,
955 .maxlen = sizeof(int),
956 .mode = 0644,
957 .proc_handler = proc_dointvec_minmax,
958 .extra1 = &zero,
959 .extra2 = &one,
960 },
961 #ifdef CONFIG_SMP
962 {
963 .procname = "hardlockup_all_cpu_backtrace",
964 .data = &sysctl_hardlockup_all_cpu_backtrace,
965 .maxlen = sizeof(int),
966 .mode = 0644,
967 .proc_handler = proc_dointvec_minmax,
968 .extra1 = &zero,
969 .extra2 = &one,
970 },
971 #endif /* CONFIG_SMP */
972 #endif
973 #endif
974
975 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
976 {
977 .procname = "unknown_nmi_panic",
978 .data = &unknown_nmi_panic,
979 .maxlen = sizeof (int),
980 .mode = 0644,
981 .proc_handler = proc_dointvec,
982 },
983 #endif
984 #if defined(CONFIG_X86)
985 {
986 .procname = "panic_on_unrecovered_nmi",
987 .data = &panic_on_unrecovered_nmi,
988 .maxlen = sizeof(int),
989 .mode = 0644,
990 .proc_handler = proc_dointvec,
991 },
992 {
993 .procname = "panic_on_io_nmi",
994 .data = &panic_on_io_nmi,
995 .maxlen = sizeof(int),
996 .mode = 0644,
997 .proc_handler = proc_dointvec,
998 },
999 #ifdef CONFIG_DEBUG_STACKOVERFLOW
1000 {
1001 .procname = "panic_on_stackoverflow",
1002 .data = &sysctl_panic_on_stackoverflow,
1003 .maxlen = sizeof(int),
1004 .mode = 0644,
1005 .proc_handler = proc_dointvec,
1006 },
1007 #endif
1008 {
1009 .procname = "bootloader_type",
1010 .data = &bootloader_type,
1011 .maxlen = sizeof (int),
1012 .mode = 0444,
1013 .proc_handler = proc_dointvec,
1014 },
1015 {
1016 .procname = "bootloader_version",
1017 .data = &bootloader_version,
1018 .maxlen = sizeof (int),
1019 .mode = 0444,
1020 .proc_handler = proc_dointvec,
1021 },
1022 {
1023 .procname = "io_delay_type",
1024 .data = &io_delay_type,
1025 .maxlen = sizeof(int),
1026 .mode = 0644,
1027 .proc_handler = proc_dointvec,
1028 },
1029 #endif
1030 #if defined(CONFIG_MMU)
1031 {
1032 .procname = "randomize_va_space",
1033 .data = &randomize_va_space,
1034 .maxlen = sizeof(int),
1035 .mode = 0644,
1036 .proc_handler = proc_dointvec,
1037 },
1038 #endif
1039 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1040 {
1041 .procname = "spin_retry",
1042 .data = &spin_retry,
1043 .maxlen = sizeof (int),
1044 .mode = 0644,
1045 .proc_handler = proc_dointvec,
1046 },
1047 #endif
1048 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1049 {
1050 .procname = "acpi_video_flags",
1051 .data = &acpi_realmode_flags,
1052 .maxlen = sizeof (unsigned long),
1053 .mode = 0644,
1054 .proc_handler = proc_doulongvec_minmax,
1055 },
1056 #endif
1057 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1058 {
1059 .procname = "ignore-unaligned-usertrap",
1060 .data = &no_unaligned_warning,
1061 .maxlen = sizeof (int),
1062 .mode = 0644,
1063 .proc_handler = proc_dointvec,
1064 },
1065 #endif
1066 #ifdef CONFIG_IA64
1067 {
1068 .procname = "unaligned-dump-stack",
1069 .data = &unaligned_dump_stack,
1070 .maxlen = sizeof (int),
1071 .mode = 0644,
1072 .proc_handler = proc_dointvec,
1073 },
1074 #endif
1075 #ifdef CONFIG_DETECT_HUNG_TASK
1076 {
1077 .procname = "hung_task_panic",
1078 .data = &sysctl_hung_task_panic,
1079 .maxlen = sizeof(int),
1080 .mode = 0644,
1081 .proc_handler = proc_dointvec_minmax,
1082 .extra1 = &zero,
1083 .extra2 = &one,
1084 },
1085 {
1086 .procname = "hung_task_check_count",
1087 .data = &sysctl_hung_task_check_count,
1088 .maxlen = sizeof(int),
1089 .mode = 0644,
1090 .proc_handler = proc_dointvec_minmax,
1091 .extra1 = &zero,
1092 },
1093 {
1094 .procname = "hung_task_timeout_secs",
1095 .data = &sysctl_hung_task_timeout_secs,
1096 .maxlen = sizeof(unsigned long),
1097 .mode = 0644,
1098 .proc_handler = proc_dohung_task_timeout_secs,
1099 .extra2 = &hung_task_timeout_max,
1100 },
1101 {
1102 .procname = "hung_task_warnings",
1103 .data = &sysctl_hung_task_warnings,
1104 .maxlen = sizeof(int),
1105 .mode = 0644,
1106 .proc_handler = proc_dointvec_minmax,
1107 .extra1 = &neg_one,
1108 },
1109 #endif
1110 #ifdef CONFIG_RT_MUTEXES
1111 {
1112 .procname = "max_lock_depth",
1113 .data = &max_lock_depth,
1114 .maxlen = sizeof(int),
1115 .mode = 0644,
1116 .proc_handler = proc_dointvec,
1117 },
1118 #endif
1119 {
1120 .procname = "poweroff_cmd",
1121 .data = &poweroff_cmd,
1122 .maxlen = POWEROFF_CMD_PATH_LEN,
1123 .mode = 0644,
1124 .proc_handler = proc_dostring,
1125 },
1126 #ifdef CONFIG_KEYS
1127 {
1128 .procname = "keys",
1129 .mode = 0555,
1130 .child = key_sysctls,
1131 },
1132 #endif
1133 #ifdef CONFIG_PERF_EVENTS
1134 /*
1135 * User-space scripts rely on the existence of this file
1136 * as a feature check for perf_events being enabled.
1137 *
1138 * So it's an ABI, do not remove!
1139 */
1140 {
1141 .procname = "perf_event_paranoid",
1142 .data = &sysctl_perf_event_paranoid,
1143 .maxlen = sizeof(sysctl_perf_event_paranoid),
1144 .mode = 0644,
1145 .proc_handler = proc_dointvec,
1146 },
1147 {
1148 .procname = "perf_event_mlock_kb",
1149 .data = &sysctl_perf_event_mlock,
1150 .maxlen = sizeof(sysctl_perf_event_mlock),
1151 .mode = 0644,
1152 .proc_handler = proc_dointvec,
1153 },
1154 {
1155 .procname = "perf_event_max_sample_rate",
1156 .data = &sysctl_perf_event_sample_rate,
1157 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1158 .mode = 0644,
1159 .proc_handler = perf_proc_update_handler,
1160 .extra1 = &one,
1161 },
1162 {
1163 .procname = "perf_cpu_time_max_percent",
1164 .data = &sysctl_perf_cpu_time_max_percent,
1165 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1166 .mode = 0644,
1167 .proc_handler = perf_cpu_time_max_percent_handler,
1168 .extra1 = &zero,
1169 .extra2 = &one_hundred,
1170 },
1171 {
1172 .procname = "perf_event_max_stack",
1173 .data = &sysctl_perf_event_max_stack,
1174 .maxlen = sizeof(sysctl_perf_event_max_stack),
1175 .mode = 0644,
1176 .proc_handler = perf_event_max_stack_handler,
1177 .extra1 = &zero,
1178 .extra2 = &six_hundred_forty_kb,
1179 },
1180 {
1181 .procname = "perf_event_max_contexts_per_stack",
1182 .data = &sysctl_perf_event_max_contexts_per_stack,
1183 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
1184 .mode = 0644,
1185 .proc_handler = perf_event_max_stack_handler,
1186 .extra1 = &zero,
1187 .extra2 = &one_thousand,
1188 },
1189 #endif
1190 {
1191 .procname = "panic_on_warn",
1192 .data = &panic_on_warn,
1193 .maxlen = sizeof(int),
1194 .mode = 0644,
1195 .proc_handler = proc_dointvec_minmax,
1196 .extra1 = &zero,
1197 .extra2 = &one,
1198 },
1199 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1200 {
1201 .procname = "timer_migration",
1202 .data = &sysctl_timer_migration,
1203 .maxlen = sizeof(unsigned int),
1204 .mode = 0644,
1205 .proc_handler = timer_migration_handler,
1206 .extra1 = &zero,
1207 .extra2 = &one,
1208 },
1209 #endif
1210 #ifdef CONFIG_BPF_SYSCALL
1211 {
1212 .procname = "unprivileged_bpf_disabled",
1213 .data = &sysctl_unprivileged_bpf_disabled,
1214 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
1215 .mode = 0644,
1216 /* only handle a transition from default "0" to "1" */
1217 .proc_handler = proc_dointvec_minmax,
1218 .extra1 = &one,
1219 .extra2 = &one,
1220 },
1221 #endif
1222 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1223 {
1224 .procname = "panic_on_rcu_stall",
1225 .data = &sysctl_panic_on_rcu_stall,
1226 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
1227 .mode = 0644,
1228 .proc_handler = proc_dointvec_minmax,
1229 .extra1 = &zero,
1230 .extra2 = &one,
1231 },
1232 #endif
1233 { }
1234 };
1235
1236 static struct ctl_table vm_table[] = {
1237 {
1238 .procname = "overcommit_memory",
1239 .data = &sysctl_overcommit_memory,
1240 .maxlen = sizeof(sysctl_overcommit_memory),
1241 .mode = 0644,
1242 .proc_handler = proc_dointvec_minmax,
1243 .extra1 = &zero,
1244 .extra2 = &two,
1245 },
1246 {
1247 .procname = "panic_on_oom",
1248 .data = &sysctl_panic_on_oom,
1249 .maxlen = sizeof(sysctl_panic_on_oom),
1250 .mode = 0644,
1251 .proc_handler = proc_dointvec_minmax,
1252 .extra1 = &zero,
1253 .extra2 = &two,
1254 },
1255 {
1256 .procname = "oom_kill_allocating_task",
1257 .data = &sysctl_oom_kill_allocating_task,
1258 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1259 .mode = 0644,
1260 .proc_handler = proc_dointvec,
1261 },
1262 {
1263 .procname = "oom_dump_tasks",
1264 .data = &sysctl_oom_dump_tasks,
1265 .maxlen = sizeof(sysctl_oom_dump_tasks),
1266 .mode = 0644,
1267 .proc_handler = proc_dointvec,
1268 },
1269 {
1270 .procname = "overcommit_ratio",
1271 .data = &sysctl_overcommit_ratio,
1272 .maxlen = sizeof(sysctl_overcommit_ratio),
1273 .mode = 0644,
1274 .proc_handler = overcommit_ratio_handler,
1275 },
1276 {
1277 .procname = "overcommit_kbytes",
1278 .data = &sysctl_overcommit_kbytes,
1279 .maxlen = sizeof(sysctl_overcommit_kbytes),
1280 .mode = 0644,
1281 .proc_handler = overcommit_kbytes_handler,
1282 },
1283 {
1284 .procname = "page-cluster",
1285 .data = &page_cluster,
1286 .maxlen = sizeof(int),
1287 .mode = 0644,
1288 .proc_handler = proc_dointvec_minmax,
1289 .extra1 = &zero,
1290 },
1291 {
1292 .procname = "dirty_background_ratio",
1293 .data = &dirty_background_ratio,
1294 .maxlen = sizeof(dirty_background_ratio),
1295 .mode = 0644,
1296 .proc_handler = dirty_background_ratio_handler,
1297 .extra1 = &zero,
1298 .extra2 = &one_hundred,
1299 },
1300 {
1301 .procname = "dirty_background_bytes",
1302 .data = &dirty_background_bytes,
1303 .maxlen = sizeof(dirty_background_bytes),
1304 .mode = 0644,
1305 .proc_handler = dirty_background_bytes_handler,
1306 .extra1 = &one_ul,
1307 },
1308 {
1309 .procname = "dirty_ratio",
1310 .data = &vm_dirty_ratio,
1311 .maxlen = sizeof(vm_dirty_ratio),
1312 .mode = 0644,
1313 .proc_handler = dirty_ratio_handler,
1314 .extra1 = &zero,
1315 .extra2 = &one_hundred,
1316 },
1317 {
1318 .procname = "dirty_bytes",
1319 .data = &vm_dirty_bytes,
1320 .maxlen = sizeof(vm_dirty_bytes),
1321 .mode = 0644,
1322 .proc_handler = dirty_bytes_handler,
1323 .extra1 = &dirty_bytes_min,
1324 },
1325 {
1326 .procname = "dirty_writeback_centisecs",
1327 .data = &dirty_writeback_interval,
1328 .maxlen = sizeof(dirty_writeback_interval),
1329 .mode = 0644,
1330 .proc_handler = dirty_writeback_centisecs_handler,
1331 },
1332 {
1333 .procname = "dirty_expire_centisecs",
1334 .data = &dirty_expire_interval,
1335 .maxlen = sizeof(dirty_expire_interval),
1336 .mode = 0644,
1337 .proc_handler = proc_dointvec_minmax,
1338 .extra1 = &zero,
1339 },
1340 {
1341 .procname = "dirtytime_expire_seconds",
1342 .data = &dirtytime_expire_interval,
1343 .maxlen = sizeof(dirtytime_expire_interval),
1344 .mode = 0644,
1345 .proc_handler = dirtytime_interval_handler,
1346 .extra1 = &zero,
1347 },
1348 {
1349 .procname = "swappiness",
1350 .data = &vm_swappiness,
1351 .maxlen = sizeof(vm_swappiness),
1352 .mode = 0644,
1353 .proc_handler = proc_dointvec_minmax,
1354 .extra1 = &zero,
1355 .extra2 = &one_hundred,
1356 },
1357 #ifdef CONFIG_HUGETLB_PAGE
1358 {
1359 .procname = "nr_hugepages",
1360 .data = NULL,
1361 .maxlen = sizeof(unsigned long),
1362 .mode = 0644,
1363 .proc_handler = hugetlb_sysctl_handler,
1364 },
1365 #ifdef CONFIG_NUMA
1366 {
1367 .procname = "nr_hugepages_mempolicy",
1368 .data = NULL,
1369 .maxlen = sizeof(unsigned long),
1370 .mode = 0644,
1371 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1372 },
1373 {
1374 .procname = "numa_stat",
1375 .data = &sysctl_vm_numa_stat,
1376 .maxlen = sizeof(int),
1377 .mode = 0644,
1378 .proc_handler = sysctl_vm_numa_stat_handler,
1379 .extra1 = &zero,
1380 .extra2 = &one,
1381 },
1382 #endif
1383 {
1384 .procname = "hugetlb_shm_group",
1385 .data = &sysctl_hugetlb_shm_group,
1386 .maxlen = sizeof(gid_t),
1387 .mode = 0644,
1388 .proc_handler = proc_dointvec,
1389 },
1390 {
1391 .procname = "nr_overcommit_hugepages",
1392 .data = NULL,
1393 .maxlen = sizeof(unsigned long),
1394 .mode = 0644,
1395 .proc_handler = hugetlb_overcommit_handler,
1396 },
1397 #endif
1398 {
1399 .procname = "lowmem_reserve_ratio",
1400 .data = &sysctl_lowmem_reserve_ratio,
1401 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1402 .mode = 0644,
1403 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1404 },
1405 {
1406 .procname = "drop_caches",
1407 .data = &sysctl_drop_caches,
1408 .maxlen = sizeof(int),
1409 .mode = 0644,
1410 .proc_handler = drop_caches_sysctl_handler,
1411 .extra1 = &one,
1412 .extra2 = &four,
1413 },
1414 #ifdef CONFIG_COMPACTION
1415 {
1416 .procname = "compact_memory",
1417 .data = &sysctl_compact_memory,
1418 .maxlen = sizeof(int),
1419 .mode = 0200,
1420 .proc_handler = sysctl_compaction_handler,
1421 },
1422 {
1423 .procname = "extfrag_threshold",
1424 .data = &sysctl_extfrag_threshold,
1425 .maxlen = sizeof(int),
1426 .mode = 0644,
1427 .proc_handler = sysctl_extfrag_handler,
1428 .extra1 = &min_extfrag_threshold,
1429 .extra2 = &max_extfrag_threshold,
1430 },
1431 {
1432 .procname = "compact_unevictable_allowed",
1433 .data = &sysctl_compact_unevictable_allowed,
1434 .maxlen = sizeof(int),
1435 .mode = 0644,
1436 .proc_handler = proc_dointvec,
1437 .extra1 = &zero,
1438 .extra2 = &one,
1439 },
1440
1441 #endif /* CONFIG_COMPACTION */
1442 {
1443 .procname = "min_free_kbytes",
1444 .data = &min_free_kbytes,
1445 .maxlen = sizeof(min_free_kbytes),
1446 .mode = 0644,
1447 .proc_handler = min_free_kbytes_sysctl_handler,
1448 .extra1 = &zero,
1449 },
1450 {
1451 .procname = "watermark_scale_factor",
1452 .data = &watermark_scale_factor,
1453 .maxlen = sizeof(watermark_scale_factor),
1454 .mode = 0644,
1455 .proc_handler = watermark_scale_factor_sysctl_handler,
1456 .extra1 = &one,
1457 .extra2 = &one_thousand,
1458 },
1459 {
1460 .procname = "percpu_pagelist_fraction",
1461 .data = &percpu_pagelist_fraction,
1462 .maxlen = sizeof(percpu_pagelist_fraction),
1463 .mode = 0644,
1464 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1465 .extra1 = &zero,
1466 },
1467 #ifdef CONFIG_MMU
1468 {
1469 .procname = "max_map_count",
1470 .data = &sysctl_max_map_count,
1471 .maxlen = sizeof(sysctl_max_map_count),
1472 .mode = 0644,
1473 .proc_handler = proc_dointvec_minmax,
1474 .extra1 = &zero,
1475 },
1476 #else
1477 {
1478 .procname = "nr_trim_pages",
1479 .data = &sysctl_nr_trim_pages,
1480 .maxlen = sizeof(sysctl_nr_trim_pages),
1481 .mode = 0644,
1482 .proc_handler = proc_dointvec_minmax,
1483 .extra1 = &zero,
1484 },
1485 #endif
1486 {
1487 .procname = "laptop_mode",
1488 .data = &laptop_mode,
1489 .maxlen = sizeof(laptop_mode),
1490 .mode = 0644,
1491 .proc_handler = proc_dointvec_jiffies,
1492 },
1493 {
1494 .procname = "block_dump",
1495 .data = &block_dump,
1496 .maxlen = sizeof(block_dump),
1497 .mode = 0644,
1498 .proc_handler = proc_dointvec,
1499 .extra1 = &zero,
1500 },
1501 {
1502 .procname = "vfs_cache_pressure",
1503 .data = &sysctl_vfs_cache_pressure,
1504 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1505 .mode = 0644,
1506 .proc_handler = proc_dointvec,
1507 .extra1 = &zero,
1508 },
1509 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1510 {
1511 .procname = "legacy_va_layout",
1512 .data = &sysctl_legacy_va_layout,
1513 .maxlen = sizeof(sysctl_legacy_va_layout),
1514 .mode = 0644,
1515 .proc_handler = proc_dointvec,
1516 .extra1 = &zero,
1517 },
1518 #endif
1519 #ifdef CONFIG_NUMA
1520 {
1521 .procname = "zone_reclaim_mode",
1522 .data = &node_reclaim_mode,
1523 .maxlen = sizeof(node_reclaim_mode),
1524 .mode = 0644,
1525 .proc_handler = proc_dointvec,
1526 .extra1 = &zero,
1527 },
1528 {
1529 .procname = "min_unmapped_ratio",
1530 .data = &sysctl_min_unmapped_ratio,
1531 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1532 .mode = 0644,
1533 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1534 .extra1 = &zero,
1535 .extra2 = &one_hundred,
1536 },
1537 {
1538 .procname = "min_slab_ratio",
1539 .data = &sysctl_min_slab_ratio,
1540 .maxlen = sizeof(sysctl_min_slab_ratio),
1541 .mode = 0644,
1542 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1543 .extra1 = &zero,
1544 .extra2 = &one_hundred,
1545 },
1546 #endif
1547 #ifdef CONFIG_SMP
1548 {
1549 .procname = "stat_interval",
1550 .data = &sysctl_stat_interval,
1551 .maxlen = sizeof(sysctl_stat_interval),
1552 .mode = 0644,
1553 .proc_handler = proc_dointvec_jiffies,
1554 },
1555 {
1556 .procname = "stat_refresh",
1557 .data = NULL,
1558 .maxlen = 0,
1559 .mode = 0600,
1560 .proc_handler = vmstat_refresh,
1561 },
1562 #endif
1563 #ifdef CONFIG_MMU
1564 {
1565 .procname = "mmap_min_addr",
1566 .data = &dac_mmap_min_addr,
1567 .maxlen = sizeof(unsigned long),
1568 .mode = 0644,
1569 .proc_handler = mmap_min_addr_handler,
1570 },
1571 #endif
1572 #ifdef CONFIG_NUMA
1573 {
1574 .procname = "numa_zonelist_order",
1575 .data = &numa_zonelist_order,
1576 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1577 .mode = 0644,
1578 .proc_handler = numa_zonelist_order_handler,
1579 },
1580 #endif
1581 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1582 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1583 {
1584 .procname = "vdso_enabled",
1585 #ifdef CONFIG_X86_32
1586 .data = &vdso32_enabled,
1587 .maxlen = sizeof(vdso32_enabled),
1588 #else
1589 .data = &vdso_enabled,
1590 .maxlen = sizeof(vdso_enabled),
1591 #endif
1592 .mode = 0644,
1593 .proc_handler = proc_dointvec,
1594 .extra1 = &zero,
1595 },
1596 #endif
1597 #ifdef CONFIG_HIGHMEM
1598 {
1599 .procname = "highmem_is_dirtyable",
1600 .data = &vm_highmem_is_dirtyable,
1601 .maxlen = sizeof(vm_highmem_is_dirtyable),
1602 .mode = 0644,
1603 .proc_handler = proc_dointvec_minmax,
1604 .extra1 = &zero,
1605 .extra2 = &one,
1606 },
1607 #endif
1608 #ifdef CONFIG_MEMORY_FAILURE
1609 {
1610 .procname = "memory_failure_early_kill",
1611 .data = &sysctl_memory_failure_early_kill,
1612 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1613 .mode = 0644,
1614 .proc_handler = proc_dointvec_minmax,
1615 .extra1 = &zero,
1616 .extra2 = &one,
1617 },
1618 {
1619 .procname = "memory_failure_recovery",
1620 .data = &sysctl_memory_failure_recovery,
1621 .maxlen = sizeof(sysctl_memory_failure_recovery),
1622 .mode = 0644,
1623 .proc_handler = proc_dointvec_minmax,
1624 .extra1 = &zero,
1625 .extra2 = &one,
1626 },
1627 #endif
1628 {
1629 .procname = "user_reserve_kbytes",
1630 .data = &sysctl_user_reserve_kbytes,
1631 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1632 .mode = 0644,
1633 .proc_handler = proc_doulongvec_minmax,
1634 },
1635 {
1636 .procname = "admin_reserve_kbytes",
1637 .data = &sysctl_admin_reserve_kbytes,
1638 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1639 .mode = 0644,
1640 .proc_handler = proc_doulongvec_minmax,
1641 },
1642 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1643 {
1644 .procname = "mmap_rnd_bits",
1645 .data = &mmap_rnd_bits,
1646 .maxlen = sizeof(mmap_rnd_bits),
1647 .mode = 0600,
1648 .proc_handler = proc_dointvec_minmax,
1649 .extra1 = (void *)&mmap_rnd_bits_min,
1650 .extra2 = (void *)&mmap_rnd_bits_max,
1651 },
1652 #endif
1653 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1654 {
1655 .procname = "mmap_rnd_compat_bits",
1656 .data = &mmap_rnd_compat_bits,
1657 .maxlen = sizeof(mmap_rnd_compat_bits),
1658 .mode = 0600,
1659 .proc_handler = proc_dointvec_minmax,
1660 .extra1 = (void *)&mmap_rnd_compat_bits_min,
1661 .extra2 = (void *)&mmap_rnd_compat_bits_max,
1662 },
1663 #endif
1664 { }
1665 };
1666
1667 static struct ctl_table fs_table[] = {
1668 {
1669 .procname = "inode-nr",
1670 .data = &inodes_stat,
1671 .maxlen = 2*sizeof(long),
1672 .mode = 0444,
1673 .proc_handler = proc_nr_inodes,
1674 },
1675 {
1676 .procname = "inode-state",
1677 .data = &inodes_stat,
1678 .maxlen = 7*sizeof(long),
1679 .mode = 0444,
1680 .proc_handler = proc_nr_inodes,
1681 },
1682 {
1683 .procname = "file-nr",
1684 .data = &files_stat,
1685 .maxlen = sizeof(files_stat),
1686 .mode = 0444,
1687 .proc_handler = proc_nr_files,
1688 },
1689 {
1690 .procname = "file-max",
1691 .data = &files_stat.max_files,
1692 .maxlen = sizeof(files_stat.max_files),
1693 .mode = 0644,
1694 .proc_handler = proc_doulongvec_minmax,
1695 },
1696 {
1697 .procname = "nr_open",
1698 .data = &sysctl_nr_open,
1699 .maxlen = sizeof(unsigned int),
1700 .mode = 0644,
1701 .proc_handler = proc_dointvec_minmax,
1702 .extra1 = &sysctl_nr_open_min,
1703 .extra2 = &sysctl_nr_open_max,
1704 },
1705 {
1706 .procname = "dentry-state",
1707 .data = &dentry_stat,
1708 .maxlen = 6*sizeof(long),
1709 .mode = 0444,
1710 .proc_handler = proc_nr_dentry,
1711 },
1712 {
1713 .procname = "overflowuid",
1714 .data = &fs_overflowuid,
1715 .maxlen = sizeof(int),
1716 .mode = 0644,
1717 .proc_handler = proc_dointvec_minmax,
1718 .extra1 = &minolduid,
1719 .extra2 = &maxolduid,
1720 },
1721 {
1722 .procname = "overflowgid",
1723 .data = &fs_overflowgid,
1724 .maxlen = sizeof(int),
1725 .mode = 0644,
1726 .proc_handler = proc_dointvec_minmax,
1727 .extra1 = &minolduid,
1728 .extra2 = &maxolduid,
1729 },
1730 #ifdef CONFIG_FILE_LOCKING
1731 {
1732 .procname = "leases-enable",
1733 .data = &leases_enable,
1734 .maxlen = sizeof(int),
1735 .mode = 0644,
1736 .proc_handler = proc_dointvec,
1737 },
1738 #endif
1739 #ifdef CONFIG_DNOTIFY
1740 {
1741 .procname = "dir-notify-enable",
1742 .data = &dir_notify_enable,
1743 .maxlen = sizeof(int),
1744 .mode = 0644,
1745 .proc_handler = proc_dointvec,
1746 },
1747 #endif
1748 #ifdef CONFIG_MMU
1749 #ifdef CONFIG_FILE_LOCKING
1750 {
1751 .procname = "lease-break-time",
1752 .data = &lease_break_time,
1753 .maxlen = sizeof(int),
1754 .mode = 0644,
1755 .proc_handler = proc_dointvec,
1756 },
1757 #endif
1758 #ifdef CONFIG_AIO
1759 {
1760 .procname = "aio-nr",
1761 .data = &aio_nr,
1762 .maxlen = sizeof(aio_nr),
1763 .mode = 0444,
1764 .proc_handler = proc_doulongvec_minmax,
1765 },
1766 {
1767 .procname = "aio-max-nr",
1768 .data = &aio_max_nr,
1769 .maxlen = sizeof(aio_max_nr),
1770 .mode = 0644,
1771 .proc_handler = proc_doulongvec_minmax,
1772 },
1773 #endif /* CONFIG_AIO */
1774 #ifdef CONFIG_INOTIFY_USER
1775 {
1776 .procname = "inotify",
1777 .mode = 0555,
1778 .child = inotify_table,
1779 },
1780 #endif
1781 #ifdef CONFIG_EPOLL
1782 {
1783 .procname = "epoll",
1784 .mode = 0555,
1785 .child = epoll_table,
1786 },
1787 #endif
1788 #endif
1789 {
1790 .procname = "protected_symlinks",
1791 .data = &sysctl_protected_symlinks,
1792 .maxlen = sizeof(int),
1793 .mode = 0600,
1794 .proc_handler = proc_dointvec_minmax,
1795 .extra1 = &zero,
1796 .extra2 = &one,
1797 },
1798 {
1799 .procname = "protected_hardlinks",
1800 .data = &sysctl_protected_hardlinks,
1801 .maxlen = sizeof(int),
1802 .mode = 0600,
1803 .proc_handler = proc_dointvec_minmax,
1804 .extra1 = &zero,
1805 .extra2 = &one,
1806 },
1807 {
1808 .procname = "suid_dumpable",
1809 .data = &suid_dumpable,
1810 .maxlen = sizeof(int),
1811 .mode = 0644,
1812 .proc_handler = proc_dointvec_minmax_coredump,
1813 .extra1 = &zero,
1814 .extra2 = &two,
1815 },
1816 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1817 {
1818 .procname = "binfmt_misc",
1819 .mode = 0555,
1820 .child = sysctl_mount_point,
1821 },
1822 #endif
1823 {
1824 .procname = "pipe-max-size",
1825 .data = &pipe_max_size,
1826 .maxlen = sizeof(pipe_max_size),
1827 .mode = 0644,
1828 .proc_handler = proc_dopipe_max_size,
1829 },
1830 {
1831 .procname = "pipe-user-pages-hard",
1832 .data = &pipe_user_pages_hard,
1833 .maxlen = sizeof(pipe_user_pages_hard),
1834 .mode = 0644,
1835 .proc_handler = proc_doulongvec_minmax,
1836 },
1837 {
1838 .procname = "pipe-user-pages-soft",
1839 .data = &pipe_user_pages_soft,
1840 .maxlen = sizeof(pipe_user_pages_soft),
1841 .mode = 0644,
1842 .proc_handler = proc_doulongvec_minmax,
1843 },
1844 {
1845 .procname = "mount-max",
1846 .data = &sysctl_mount_max,
1847 .maxlen = sizeof(unsigned int),
1848 .mode = 0644,
1849 .proc_handler = proc_dointvec_minmax,
1850 .extra1 = &one,
1851 },
1852 { }
1853 };
1854
1855 static struct ctl_table debug_table[] = {
1856 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1857 {
1858 .procname = "exception-trace",
1859 .data = &show_unhandled_signals,
1860 .maxlen = sizeof(int),
1861 .mode = 0644,
1862 .proc_handler = proc_dointvec
1863 },
1864 #endif
1865 #if defined(CONFIG_OPTPROBES)
1866 {
1867 .procname = "kprobes-optimization",
1868 .data = &sysctl_kprobes_optimization,
1869 .maxlen = sizeof(int),
1870 .mode = 0644,
1871 .proc_handler = proc_kprobes_optimization_handler,
1872 .extra1 = &zero,
1873 .extra2 = &one,
1874 },
1875 #endif
1876 { }
1877 };
1878
1879 static struct ctl_table dev_table[] = {
1880 { }
1881 };
1882
1883 int __init sysctl_init(void)
1884 {
1885 struct ctl_table_header *hdr;
1886
1887 hdr = register_sysctl_table(sysctl_base_table);
1888 kmemleak_not_leak(hdr);
1889 return 0;
1890 }
1891
1892 #endif /* CONFIG_SYSCTL */
1893
1894 /*
1895 * /proc/sys support
1896 */
1897
1898 #ifdef CONFIG_PROC_SYSCTL
1899
1900 static int _proc_do_string(char *data, int maxlen, int write,
1901 char __user *buffer,
1902 size_t *lenp, loff_t *ppos)
1903 {
1904 size_t len;
1905 char __user *p;
1906 char c;
1907
1908 if (!data || !maxlen || !*lenp) {
1909 *lenp = 0;
1910 return 0;
1911 }
1912
1913 if (write) {
1914 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1915 /* Only continue writes not past the end of buffer. */
1916 len = strlen(data);
1917 if (len > maxlen - 1)
1918 len = maxlen - 1;
1919
1920 if (*ppos > len)
1921 return 0;
1922 len = *ppos;
1923 } else {
1924 /* Start writing from beginning of buffer. */
1925 len = 0;
1926 }
1927
1928 *ppos += *lenp;
1929 p = buffer;
1930 while ((p - buffer) < *lenp && len < maxlen - 1) {
1931 if (get_user(c, p++))
1932 return -EFAULT;
1933 if (c == 0 || c == '\n')
1934 break;
1935 data[len++] = c;
1936 }
1937 data[len] = 0;
1938 } else {
1939 len = strlen(data);
1940 if (len > maxlen)
1941 len = maxlen;
1942
1943 if (*ppos > len) {
1944 *lenp = 0;
1945 return 0;
1946 }
1947
1948 data += *ppos;
1949 len -= *ppos;
1950
1951 if (len > *lenp)
1952 len = *lenp;
1953 if (len)
1954 if (copy_to_user(buffer, data, len))
1955 return -EFAULT;
1956 if (len < *lenp) {
1957 if (put_user('\n', buffer + len))
1958 return -EFAULT;
1959 len++;
1960 }
1961 *lenp = len;
1962 *ppos += len;
1963 }
1964 return 0;
1965 }
1966
1967 static void warn_sysctl_write(struct ctl_table *table)
1968 {
1969 pr_warn_once("%s wrote to %s when file position was not 0!\n"
1970 "This will not be supported in the future. To silence this\n"
1971 "warning, set kernel.sysctl_writes_strict = -1\n",
1972 current->comm, table->procname);
1973 }
1974
1975 /**
1976 * proc_first_pos_non_zero_ignore - check if firs position is allowed
1977 * @ppos: file position
1978 * @table: the sysctl table
1979 *
1980 * Returns true if the first position is non-zero and the sysctl_writes_strict
1981 * mode indicates this is not allowed for numeric input types. String proc
1982 * hadlers can ignore the return value.
1983 */
1984 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
1985 struct ctl_table *table)
1986 {
1987 if (!*ppos)
1988 return false;
1989
1990 switch (sysctl_writes_strict) {
1991 case SYSCTL_WRITES_STRICT:
1992 return true;
1993 case SYSCTL_WRITES_WARN:
1994 warn_sysctl_write(table);
1995 return false;
1996 default:
1997 return false;
1998 }
1999 }
2000
2001 /**
2002 * proc_dostring - read a string sysctl
2003 * @table: the sysctl table
2004 * @write: %TRUE if this is a write to the sysctl file
2005 * @buffer: the user buffer
2006 * @lenp: the size of the user buffer
2007 * @ppos: file position
2008 *
2009 * Reads/writes a string from/to the user buffer. If the kernel
2010 * buffer provided is not large enough to hold the string, the
2011 * string is truncated. The copied string is %NULL-terminated.
2012 * If the string is being read by the user process, it is copied
2013 * and a newline '\n' is added. It is truncated if the buffer is
2014 * not large enough.
2015 *
2016 * Returns 0 on success.
2017 */
2018 int proc_dostring(struct ctl_table *table, int write,
2019 void __user *buffer, size_t *lenp, loff_t *ppos)
2020 {
2021 if (write)
2022 proc_first_pos_non_zero_ignore(ppos, table);
2023
2024 return _proc_do_string((char *)(table->data), table->maxlen, write,
2025 (char __user *)buffer, lenp, ppos);
2026 }
2027
2028 static size_t proc_skip_spaces(char **buf)
2029 {
2030 size_t ret;
2031 char *tmp = skip_spaces(*buf);
2032 ret = tmp - *buf;
2033 *buf = tmp;
2034 return ret;
2035 }
2036
2037 static void proc_skip_char(char **buf, size_t *size, const char v)
2038 {
2039 while (*size) {
2040 if (**buf != v)
2041 break;
2042 (*size)--;
2043 (*buf)++;
2044 }
2045 }
2046
2047 #define TMPBUFLEN 22
2048 /**
2049 * proc_get_long - reads an ASCII formatted integer from a user buffer
2050 *
2051 * @buf: a kernel buffer
2052 * @size: size of the kernel buffer
2053 * @val: this is where the number will be stored
2054 * @neg: set to %TRUE if number is negative
2055 * @perm_tr: a vector which contains the allowed trailers
2056 * @perm_tr_len: size of the perm_tr vector
2057 * @tr: pointer to store the trailer character
2058 *
2059 * In case of success %0 is returned and @buf and @size are updated with
2060 * the amount of bytes read. If @tr is non-NULL and a trailing
2061 * character exists (size is non-zero after returning from this
2062 * function), @tr is updated with the trailing character.
2063 */
2064 static int proc_get_long(char **buf, size_t *size,
2065 unsigned long *val, bool *neg,
2066 const char *perm_tr, unsigned perm_tr_len, char *tr)
2067 {
2068 int len;
2069 char *p, tmp[TMPBUFLEN];
2070
2071 if (!*size)
2072 return -EINVAL;
2073
2074 len = *size;
2075 if (len > TMPBUFLEN - 1)
2076 len = TMPBUFLEN - 1;
2077
2078 memcpy(tmp, *buf, len);
2079
2080 tmp[len] = 0;
2081 p = tmp;
2082 if (*p == '-' && *size > 1) {
2083 *neg = true;
2084 p++;
2085 } else
2086 *neg = false;
2087 if (!isdigit(*p))
2088 return -EINVAL;
2089
2090 *val = simple_strtoul(p, &p, 0);
2091
2092 len = p - tmp;
2093
2094 /* We don't know if the next char is whitespace thus we may accept
2095 * invalid integers (e.g. 1234...a) or two integers instead of one
2096 * (e.g. 123...1). So lets not allow such large numbers. */
2097 if (len == TMPBUFLEN - 1)
2098 return -EINVAL;
2099
2100 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2101 return -EINVAL;
2102
2103 if (tr && (len < *size))
2104 *tr = *p;
2105
2106 *buf += len;
2107 *size -= len;
2108
2109 return 0;
2110 }
2111
2112 /**
2113 * proc_put_long - converts an integer to a decimal ASCII formatted string
2114 *
2115 * @buf: the user buffer
2116 * @size: the size of the user buffer
2117 * @val: the integer to be converted
2118 * @neg: sign of the number, %TRUE for negative
2119 *
2120 * In case of success %0 is returned and @buf and @size are updated with
2121 * the amount of bytes written.
2122 */
2123 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2124 bool neg)
2125 {
2126 int len;
2127 char tmp[TMPBUFLEN], *p = tmp;
2128
2129 sprintf(p, "%s%lu", neg ? "-" : "", val);
2130 len = strlen(tmp);
2131 if (len > *size)
2132 len = *size;
2133 if (copy_to_user(*buf, tmp, len))
2134 return -EFAULT;
2135 *size -= len;
2136 *buf += len;
2137 return 0;
2138 }
2139 #undef TMPBUFLEN
2140
2141 static int proc_put_char(void __user **buf, size_t *size, char c)
2142 {
2143 if (*size) {
2144 char __user **buffer = (char __user **)buf;
2145 if (put_user(c, *buffer))
2146 return -EFAULT;
2147 (*size)--, (*buffer)++;
2148 *buf = *buffer;
2149 }
2150 return 0;
2151 }
2152
2153 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2154 int *valp,
2155 int write, void *data)
2156 {
2157 if (write) {
2158 if (*negp) {
2159 if (*lvalp > (unsigned long) INT_MAX + 1)
2160 return -EINVAL;
2161 *valp = -*lvalp;
2162 } else {
2163 if (*lvalp > (unsigned long) INT_MAX)
2164 return -EINVAL;
2165 *valp = *lvalp;
2166 }
2167 } else {
2168 int val = *valp;
2169 if (val < 0) {
2170 *negp = true;
2171 *lvalp = -(unsigned long)val;
2172 } else {
2173 *negp = false;
2174 *lvalp = (unsigned long)val;
2175 }
2176 }
2177 return 0;
2178 }
2179
2180 static int do_proc_douintvec_conv(unsigned long *lvalp,
2181 unsigned int *valp,
2182 int write, void *data)
2183 {
2184 if (write) {
2185 if (*lvalp > UINT_MAX)
2186 return -EINVAL;
2187 *valp = *lvalp;
2188 } else {
2189 unsigned int val = *valp;
2190 *lvalp = (unsigned long)val;
2191 }
2192 return 0;
2193 }
2194
2195 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2196
2197 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2198 int write, void __user *buffer,
2199 size_t *lenp, loff_t *ppos,
2200 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2201 int write, void *data),
2202 void *data)
2203 {
2204 int *i, vleft, first = 1, err = 0;
2205 size_t left;
2206 char *kbuf = NULL, *p;
2207
2208 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2209 *lenp = 0;
2210 return 0;
2211 }
2212
2213 i = (int *) tbl_data;
2214 vleft = table->maxlen / sizeof(*i);
2215 left = *lenp;
2216
2217 if (!conv)
2218 conv = do_proc_dointvec_conv;
2219
2220 if (write) {
2221 if (proc_first_pos_non_zero_ignore(ppos, table))
2222 goto out;
2223
2224 if (left > PAGE_SIZE - 1)
2225 left = PAGE_SIZE - 1;
2226 p = kbuf = memdup_user_nul(buffer, left);
2227 if (IS_ERR(kbuf))
2228 return PTR_ERR(kbuf);
2229 }
2230
2231 for (; left && vleft--; i++, first=0) {
2232 unsigned long lval;
2233 bool neg;
2234
2235 if (write) {
2236 left -= proc_skip_spaces(&p);
2237
2238 if (!left)
2239 break;
2240 err = proc_get_long(&p, &left, &lval, &neg,
2241 proc_wspace_sep,
2242 sizeof(proc_wspace_sep), NULL);
2243 if (err)
2244 break;
2245 if (conv(&neg, &lval, i, 1, data)) {
2246 err = -EINVAL;
2247 break;
2248 }
2249 } else {
2250 if (conv(&neg, &lval, i, 0, data)) {
2251 err = -EINVAL;
2252 break;
2253 }
2254 if (!first)
2255 err = proc_put_char(&buffer, &left, '\t');
2256 if (err)
2257 break;
2258 err = proc_put_long(&buffer, &left, lval, neg);
2259 if (err)
2260 break;
2261 }
2262 }
2263
2264 if (!write && !first && left && !err)
2265 err = proc_put_char(&buffer, &left, '\n');
2266 if (write && !err && left)
2267 left -= proc_skip_spaces(&p);
2268 if (write) {
2269 kfree(kbuf);
2270 if (first)
2271 return err ? : -EINVAL;
2272 }
2273 *lenp -= left;
2274 out:
2275 *ppos += *lenp;
2276 return err;
2277 }
2278
2279 static int do_proc_dointvec(struct ctl_table *table, int write,
2280 void __user *buffer, size_t *lenp, loff_t *ppos,
2281 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2282 int write, void *data),
2283 void *data)
2284 {
2285 return __do_proc_dointvec(table->data, table, write,
2286 buffer, lenp, ppos, conv, data);
2287 }
2288
2289 static int do_proc_douintvec_w(unsigned int *tbl_data,
2290 struct ctl_table *table,
2291 void __user *buffer,
2292 size_t *lenp, loff_t *ppos,
2293 int (*conv)(unsigned long *lvalp,
2294 unsigned int *valp,
2295 int write, void *data),
2296 void *data)
2297 {
2298 unsigned long lval;
2299 int err = 0;
2300 size_t left;
2301 bool neg;
2302 char *kbuf = NULL, *p;
2303
2304 left = *lenp;
2305
2306 if (proc_first_pos_non_zero_ignore(ppos, table))
2307 goto bail_early;
2308
2309 if (left > PAGE_SIZE - 1)
2310 left = PAGE_SIZE - 1;
2311
2312 p = kbuf = memdup_user_nul(buffer, left);
2313 if (IS_ERR(kbuf))
2314 return -EINVAL;
2315
2316 left -= proc_skip_spaces(&p);
2317 if (!left) {
2318 err = -EINVAL;
2319 goto out_free;
2320 }
2321
2322 err = proc_get_long(&p, &left, &lval, &neg,
2323 proc_wspace_sep,
2324 sizeof(proc_wspace_sep), NULL);
2325 if (err || neg) {
2326 err = -EINVAL;
2327 goto out_free;
2328 }
2329
2330 if (conv(&lval, tbl_data, 1, data)) {
2331 err = -EINVAL;
2332 goto out_free;
2333 }
2334
2335 if (!err && left)
2336 left -= proc_skip_spaces(&p);
2337
2338 out_free:
2339 kfree(kbuf);
2340 if (err)
2341 return -EINVAL;
2342
2343 return 0;
2344
2345 /* This is in keeping with old __do_proc_dointvec() */
2346 bail_early:
2347 *ppos += *lenp;
2348 return err;
2349 }
2350
2351 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2352 size_t *lenp, loff_t *ppos,
2353 int (*conv)(unsigned long *lvalp,
2354 unsigned int *valp,
2355 int write, void *data),
2356 void *data)
2357 {
2358 unsigned long lval;
2359 int err = 0;
2360 size_t left;
2361
2362 left = *lenp;
2363
2364 if (conv(&lval, tbl_data, 0, data)) {
2365 err = -EINVAL;
2366 goto out;
2367 }
2368
2369 err = proc_put_long(&buffer, &left, lval, false);
2370 if (err || !left)
2371 goto out;
2372
2373 err = proc_put_char(&buffer, &left, '\n');
2374
2375 out:
2376 *lenp -= left;
2377 *ppos += *lenp;
2378
2379 return err;
2380 }
2381
2382 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2383 int write, void __user *buffer,
2384 size_t *lenp, loff_t *ppos,
2385 int (*conv)(unsigned long *lvalp,
2386 unsigned int *valp,
2387 int write, void *data),
2388 void *data)
2389 {
2390 unsigned int *i, vleft;
2391
2392 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2393 *lenp = 0;
2394 return 0;
2395 }
2396
2397 i = (unsigned int *) tbl_data;
2398 vleft = table->maxlen / sizeof(*i);
2399
2400 /*
2401 * Arrays are not supported, keep this simple. *Do not* add
2402 * support for them.
2403 */
2404 if (vleft != 1) {
2405 *lenp = 0;
2406 return -EINVAL;
2407 }
2408
2409 if (!conv)
2410 conv = do_proc_douintvec_conv;
2411
2412 if (write)
2413 return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2414 conv, data);
2415 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2416 }
2417
2418 static int do_proc_douintvec(struct ctl_table *table, int write,
2419 void __user *buffer, size_t *lenp, loff_t *ppos,
2420 int (*conv)(unsigned long *lvalp,
2421 unsigned int *valp,
2422 int write, void *data),
2423 void *data)
2424 {
2425 return __do_proc_douintvec(table->data, table, write,
2426 buffer, lenp, ppos, conv, data);
2427 }
2428
2429 /**
2430 * proc_dointvec - read a vector of integers
2431 * @table: the sysctl table
2432 * @write: %TRUE if this is a write to the sysctl file
2433 * @buffer: the user buffer
2434 * @lenp: the size of the user buffer
2435 * @ppos: file position
2436 *
2437 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2438 * values from/to the user buffer, treated as an ASCII string.
2439 *
2440 * Returns 0 on success.
2441 */
2442 int proc_dointvec(struct ctl_table *table, int write,
2443 void __user *buffer, size_t *lenp, loff_t *ppos)
2444 {
2445 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2446 }
2447
2448 /**
2449 * proc_douintvec - read a vector of unsigned integers
2450 * @table: the sysctl table
2451 * @write: %TRUE if this is a write to the sysctl file
2452 * @buffer: the user buffer
2453 * @lenp: the size of the user buffer
2454 * @ppos: file position
2455 *
2456 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2457 * values from/to the user buffer, treated as an ASCII string.
2458 *
2459 * Returns 0 on success.
2460 */
2461 int proc_douintvec(struct ctl_table *table, int write,
2462 void __user *buffer, size_t *lenp, loff_t *ppos)
2463 {
2464 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2465 do_proc_douintvec_conv, NULL);
2466 }
2467
2468 /*
2469 * Taint values can only be increased
2470 * This means we can safely use a temporary.
2471 */
2472 static int proc_taint(struct ctl_table *table, int write,
2473 void __user *buffer, size_t *lenp, loff_t *ppos)
2474 {
2475 struct ctl_table t;
2476 unsigned long tmptaint = get_taint();
2477 int err;
2478
2479 if (write && !capable(CAP_SYS_ADMIN))
2480 return -EPERM;
2481
2482 t = *table;
2483 t.data = &tmptaint;
2484 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2485 if (err < 0)
2486 return err;
2487
2488 if (write) {
2489 /*
2490 * Poor man's atomic or. Not worth adding a primitive
2491 * to everyone's atomic.h for this
2492 */
2493 int i;
2494 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2495 if ((tmptaint >> i) & 1)
2496 add_taint(i, LOCKDEP_STILL_OK);
2497 }
2498 }
2499
2500 return err;
2501 }
2502
2503 #ifdef CONFIG_PRINTK
2504 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2505 void __user *buffer, size_t *lenp, loff_t *ppos)
2506 {
2507 if (write && !capable(CAP_SYS_ADMIN))
2508 return -EPERM;
2509
2510 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2511 }
2512 #endif
2513
2514 /**
2515 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
2516 * @min: pointer to minimum allowable value
2517 * @max: pointer to maximum allowable value
2518 *
2519 * The do_proc_dointvec_minmax_conv_param structure provides the
2520 * minimum and maximum values for doing range checking for those sysctl
2521 * parameters that use the proc_dointvec_minmax() handler.
2522 */
2523 struct do_proc_dointvec_minmax_conv_param {
2524 int *min;
2525 int *max;
2526 };
2527
2528 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2529 int *valp,
2530 int write, void *data)
2531 {
2532 struct do_proc_dointvec_minmax_conv_param *param = data;
2533 if (write) {
2534 int val = *negp ? -*lvalp : *lvalp;
2535 if ((param->min && *param->min > val) ||
2536 (param->max && *param->max < val))
2537 return -EINVAL;
2538 *valp = val;
2539 } else {
2540 int val = *valp;
2541 if (val < 0) {
2542 *negp = true;
2543 *lvalp = -(unsigned long)val;
2544 } else {
2545 *negp = false;
2546 *lvalp = (unsigned long)val;
2547 }
2548 }
2549 return 0;
2550 }
2551
2552 /**
2553 * proc_dointvec_minmax - read a vector of integers with min/max values
2554 * @table: the sysctl table
2555 * @write: %TRUE if this is a write to the sysctl file
2556 * @buffer: the user buffer
2557 * @lenp: the size of the user buffer
2558 * @ppos: file position
2559 *
2560 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2561 * values from/to the user buffer, treated as an ASCII string.
2562 *
2563 * This routine will ensure the values are within the range specified by
2564 * table->extra1 (min) and table->extra2 (max).
2565 *
2566 * Returns 0 on success or -EINVAL on write when the range check fails.
2567 */
2568 int proc_dointvec_minmax(struct ctl_table *table, int write,
2569 void __user *buffer, size_t *lenp, loff_t *ppos)
2570 {
2571 struct do_proc_dointvec_minmax_conv_param param = {
2572 .min = (int *) table->extra1,
2573 .max = (int *) table->extra2,
2574 };
2575 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2576 do_proc_dointvec_minmax_conv, &param);
2577 }
2578
2579 /**
2580 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
2581 * @min: pointer to minimum allowable value
2582 * @max: pointer to maximum allowable value
2583 *
2584 * The do_proc_douintvec_minmax_conv_param structure provides the
2585 * minimum and maximum values for doing range checking for those sysctl
2586 * parameters that use the proc_douintvec_minmax() handler.
2587 */
2588 struct do_proc_douintvec_minmax_conv_param {
2589 unsigned int *min;
2590 unsigned int *max;
2591 };
2592
2593 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2594 unsigned int *valp,
2595 int write, void *data)
2596 {
2597 struct do_proc_douintvec_minmax_conv_param *param = data;
2598
2599 if (write) {
2600 unsigned int val = *lvalp;
2601
2602 if (*lvalp > UINT_MAX)
2603 return -EINVAL;
2604
2605 if ((param->min && *param->min > val) ||
2606 (param->max && *param->max < val))
2607 return -ERANGE;
2608
2609 *valp = val;
2610 } else {
2611 unsigned int val = *valp;
2612 *lvalp = (unsigned long) val;
2613 }
2614
2615 return 0;
2616 }
2617
2618 /**
2619 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2620 * @table: the sysctl table
2621 * @write: %TRUE if this is a write to the sysctl file
2622 * @buffer: the user buffer
2623 * @lenp: the size of the user buffer
2624 * @ppos: file position
2625 *
2626 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2627 * values from/to the user buffer, treated as an ASCII string. Negative
2628 * strings are not allowed.
2629 *
2630 * This routine will ensure the values are within the range specified by
2631 * table->extra1 (min) and table->extra2 (max). There is a final sanity
2632 * check for UINT_MAX to avoid having to support wrap around uses from
2633 * userspace.
2634 *
2635 * Returns 0 on success or -ERANGE on write when the range check fails.
2636 */
2637 int proc_douintvec_minmax(struct ctl_table *table, int write,
2638 void __user *buffer, size_t *lenp, loff_t *ppos)
2639 {
2640 struct do_proc_douintvec_minmax_conv_param param = {
2641 .min = (unsigned int *) table->extra1,
2642 .max = (unsigned int *) table->extra2,
2643 };
2644 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2645 do_proc_douintvec_minmax_conv, &param);
2646 }
2647
2648 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
2649 unsigned int *valp,
2650 int write, void *data)
2651 {
2652 if (write) {
2653 unsigned int val;
2654
2655 val = round_pipe_size(*lvalp);
2656 if (val == 0)
2657 return -EINVAL;
2658
2659 *valp = val;
2660 } else {
2661 unsigned int val = *valp;
2662 *lvalp = (unsigned long) val;
2663 }
2664
2665 return 0;
2666 }
2667
2668 static int proc_dopipe_max_size(struct ctl_table *table, int write,
2669 void __user *buffer, size_t *lenp, loff_t *ppos)
2670 {
2671 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2672 do_proc_dopipe_max_size_conv, NULL);
2673 }
2674
2675 static void validate_coredump_safety(void)
2676 {
2677 #ifdef CONFIG_COREDUMP
2678 if (suid_dumpable == SUID_DUMP_ROOT &&
2679 core_pattern[0] != '/' && core_pattern[0] != '|') {
2680 printk(KERN_WARNING
2681 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2682 "Pipe handler or fully qualified core dump path required.\n"
2683 "Set kernel.core_pattern before fs.suid_dumpable.\n"
2684 );
2685 }
2686 #endif
2687 }
2688
2689 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2690 void __user *buffer, size_t *lenp, loff_t *ppos)
2691 {
2692 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2693 if (!error)
2694 validate_coredump_safety();
2695 return error;
2696 }
2697
2698 #ifdef CONFIG_COREDUMP
2699 static int proc_dostring_coredump(struct ctl_table *table, int write,
2700 void __user *buffer, size_t *lenp, loff_t *ppos)
2701 {
2702 int error = proc_dostring(table, write, buffer, lenp, ppos);
2703 if (!error)
2704 validate_coredump_safety();
2705 return error;
2706 }
2707 #endif
2708
2709 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2710 void __user *buffer,
2711 size_t *lenp, loff_t *ppos,
2712 unsigned long convmul,
2713 unsigned long convdiv)
2714 {
2715 unsigned long *i, *min, *max;
2716 int vleft, first = 1, err = 0;
2717 size_t left;
2718 char *kbuf = NULL, *p;
2719
2720 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2721 *lenp = 0;
2722 return 0;
2723 }
2724
2725 i = (unsigned long *) data;
2726 min = (unsigned long *) table->extra1;
2727 max = (unsigned long *) table->extra2;
2728 vleft = table->maxlen / sizeof(unsigned long);
2729 left = *lenp;
2730
2731 if (write) {
2732 if (proc_first_pos_non_zero_ignore(ppos, table))
2733 goto out;
2734
2735 if (left > PAGE_SIZE - 1)
2736 left = PAGE_SIZE - 1;
2737 p = kbuf = memdup_user_nul(buffer, left);
2738 if (IS_ERR(kbuf))
2739 return PTR_ERR(kbuf);
2740 }
2741
2742 for (; left && vleft--; i++, first = 0) {
2743 unsigned long val;
2744
2745 if (write) {
2746 bool neg;
2747
2748 left -= proc_skip_spaces(&p);
2749
2750 err = proc_get_long(&p, &left, &val, &neg,
2751 proc_wspace_sep,
2752 sizeof(proc_wspace_sep), NULL);
2753 if (err)
2754 break;
2755 if (neg)
2756 continue;
2757 val = convmul * val / convdiv;
2758 if ((min && val < *min) || (max && val > *max))
2759 continue;
2760 *i = val;
2761 } else {
2762 val = convdiv * (*i) / convmul;
2763 if (!first) {
2764 err = proc_put_char(&buffer, &left, '\t');
2765 if (err)
2766 break;
2767 }
2768 err = proc_put_long(&buffer, &left, val, false);
2769 if (err)
2770 break;
2771 }
2772 }
2773
2774 if (!write && !first && left && !err)
2775 err = proc_put_char(&buffer, &left, '\n');
2776 if (write && !err)
2777 left -= proc_skip_spaces(&p);
2778 if (write) {
2779 kfree(kbuf);
2780 if (first)
2781 return err ? : -EINVAL;
2782 }
2783 *lenp -= left;
2784 out:
2785 *ppos += *lenp;
2786 return err;
2787 }
2788
2789 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2790 void __user *buffer,
2791 size_t *lenp, loff_t *ppos,
2792 unsigned long convmul,
2793 unsigned long convdiv)
2794 {
2795 return __do_proc_doulongvec_minmax(table->data, table, write,
2796 buffer, lenp, ppos, convmul, convdiv);
2797 }
2798
2799 /**
2800 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2801 * @table: the sysctl table
2802 * @write: %TRUE if this is a write to the sysctl file
2803 * @buffer: the user buffer
2804 * @lenp: the size of the user buffer
2805 * @ppos: file position
2806 *
2807 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2808 * values from/to the user buffer, treated as an ASCII string.
2809 *
2810 * This routine will ensure the values are within the range specified by
2811 * table->extra1 (min) and table->extra2 (max).
2812 *
2813 * Returns 0 on success.
2814 */
2815 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2816 void __user *buffer, size_t *lenp, loff_t *ppos)
2817 {
2818 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2819 }
2820
2821 /**
2822 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2823 * @table: the sysctl table
2824 * @write: %TRUE if this is a write to the sysctl file
2825 * @buffer: the user buffer
2826 * @lenp: the size of the user buffer
2827 * @ppos: file position
2828 *
2829 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2830 * values from/to the user buffer, treated as an ASCII string. The values
2831 * are treated as milliseconds, and converted to jiffies when they are stored.
2832 *
2833 * This routine will ensure the values are within the range specified by
2834 * table->extra1 (min) and table->extra2 (max).
2835 *
2836 * Returns 0 on success.
2837 */
2838 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2839 void __user *buffer,
2840 size_t *lenp, loff_t *ppos)
2841 {
2842 return do_proc_doulongvec_minmax(table, write, buffer,
2843 lenp, ppos, HZ, 1000l);
2844 }
2845
2846
2847 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2848 int *valp,
2849 int write, void *data)
2850 {
2851 if (write) {
2852 if (*lvalp > INT_MAX / HZ)
2853 return 1;
2854 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2855 } else {
2856 int val = *valp;
2857 unsigned long lval;
2858 if (val < 0) {
2859 *negp = true;
2860 lval = -(unsigned long)val;
2861 } else {
2862 *negp = false;
2863 lval = (unsigned long)val;
2864 }
2865 *lvalp = lval / HZ;
2866 }
2867 return 0;
2868 }
2869
2870 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2871 int *valp,
2872 int write, void *data)
2873 {
2874 if (write) {
2875 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2876 return 1;
2877 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2878 } else {
2879 int val = *valp;
2880 unsigned long lval;
2881 if (val < 0) {
2882 *negp = true;
2883 lval = -(unsigned long)val;
2884 } else {
2885 *negp = false;
2886 lval = (unsigned long)val;
2887 }
2888 *lvalp = jiffies_to_clock_t(lval);
2889 }
2890 return 0;
2891 }
2892
2893 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2894 int *valp,
2895 int write, void *data)
2896 {
2897 if (write) {
2898 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2899
2900 if (jif > INT_MAX)
2901 return 1;
2902 *valp = (int)jif;
2903 } else {
2904 int val = *valp;
2905 unsigned long lval;
2906 if (val < 0) {
2907 *negp = true;
2908 lval = -(unsigned long)val;
2909 } else {
2910 *negp = false;
2911 lval = (unsigned long)val;
2912 }
2913 *lvalp = jiffies_to_msecs(lval);
2914 }
2915 return 0;
2916 }
2917
2918 /**
2919 * proc_dointvec_jiffies - read a vector of integers as seconds
2920 * @table: the sysctl table
2921 * @write: %TRUE if this is a write to the sysctl file
2922 * @buffer: the user buffer
2923 * @lenp: the size of the user buffer
2924 * @ppos: file position
2925 *
2926 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2927 * values from/to the user buffer, treated as an ASCII string.
2928 * The values read are assumed to be in seconds, and are converted into
2929 * jiffies.
2930 *
2931 * Returns 0 on success.
2932 */
2933 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2934 void __user *buffer, size_t *lenp, loff_t *ppos)
2935 {
2936 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2937 do_proc_dointvec_jiffies_conv,NULL);
2938 }
2939
2940 /**
2941 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2942 * @table: the sysctl table
2943 * @write: %TRUE if this is a write to the sysctl file
2944 * @buffer: the user buffer
2945 * @lenp: the size of the user buffer
2946 * @ppos: pointer to the file position
2947 *
2948 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2949 * values from/to the user buffer, treated as an ASCII string.
2950 * The values read are assumed to be in 1/USER_HZ seconds, and
2951 * are converted into jiffies.
2952 *
2953 * Returns 0 on success.
2954 */
2955 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2956 void __user *buffer, size_t *lenp, loff_t *ppos)
2957 {
2958 return do_proc_dointvec(table,write,buffer,lenp,ppos,
2959 do_proc_dointvec_userhz_jiffies_conv,NULL);
2960 }
2961
2962 /**
2963 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2964 * @table: the sysctl table
2965 * @write: %TRUE if this is a write to the sysctl file
2966 * @buffer: the user buffer
2967 * @lenp: the size of the user buffer
2968 * @ppos: file position
2969 * @ppos: the current position in the file
2970 *
2971 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2972 * values from/to the user buffer, treated as an ASCII string.
2973 * The values read are assumed to be in 1/1000 seconds, and
2974 * are converted into jiffies.
2975 *
2976 * Returns 0 on success.
2977 */
2978 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2979 void __user *buffer, size_t *lenp, loff_t *ppos)
2980 {
2981 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2982 do_proc_dointvec_ms_jiffies_conv, NULL);
2983 }
2984
2985 static int proc_do_cad_pid(struct ctl_table *table, int write,
2986 void __user *buffer, size_t *lenp, loff_t *ppos)
2987 {
2988 struct pid *new_pid;
2989 pid_t tmp;
2990 int r;
2991
2992 tmp = pid_vnr(cad_pid);
2993
2994 r = __do_proc_dointvec(&tmp, table, write, buffer,
2995 lenp, ppos, NULL, NULL);
2996 if (r || !write)
2997 return r;
2998
2999 new_pid = find_get_pid(tmp);
3000 if (!new_pid)
3001 return -ESRCH;
3002
3003 put_pid(xchg(&cad_pid, new_pid));
3004 return 0;
3005 }
3006
3007 /**
3008 * proc_do_large_bitmap - read/write from/to a large bitmap
3009 * @table: the sysctl table
3010 * @write: %TRUE if this is a write to the sysctl file
3011 * @buffer: the user buffer
3012 * @lenp: the size of the user buffer
3013 * @ppos: file position
3014 *
3015 * The bitmap is stored at table->data and the bitmap length (in bits)
3016 * in table->maxlen.
3017 *
3018 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
3019 * large bitmaps may be represented in a compact manner. Writing into
3020 * the file will clear the bitmap then update it with the given input.
3021 *
3022 * Returns 0 on success.
3023 */
3024 int proc_do_large_bitmap(struct ctl_table *table, int write,
3025 void __user *buffer, size_t *lenp, loff_t *ppos)
3026 {
3027 int err = 0;
3028 bool first = 1;
3029 size_t left = *lenp;
3030 unsigned long bitmap_len = table->maxlen;
3031 unsigned long *bitmap = *(unsigned long **) table->data;
3032 unsigned long *tmp_bitmap = NULL;
3033 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
3034
3035 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
3036 *lenp = 0;
3037 return 0;
3038 }
3039
3040 if (write) {
3041 char *kbuf, *p;
3042
3043 if (left > PAGE_SIZE - 1)
3044 left = PAGE_SIZE - 1;
3045
3046 p = kbuf = memdup_user_nul(buffer, left);
3047 if (IS_ERR(kbuf))
3048 return PTR_ERR(kbuf);
3049
3050 tmp_bitmap = kcalloc(BITS_TO_LONGS(bitmap_len),
3051 sizeof(unsigned long),
3052 GFP_KERNEL);
3053 if (!tmp_bitmap) {
3054 kfree(kbuf);
3055 return -ENOMEM;
3056 }
3057 proc_skip_char(&p, &left, '\n');
3058 while (!err && left) {
3059 unsigned long val_a, val_b;
3060 bool neg;
3061
3062 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3063 sizeof(tr_a), &c);
3064 if (err)
3065 break;
3066 if (val_a >= bitmap_len || neg) {
3067 err = -EINVAL;
3068 break;
3069 }
3070
3071 val_b = val_a;
3072 if (left) {
3073 p++;
3074 left--;
3075 }
3076
3077 if (c == '-') {
3078 err = proc_get_long(&p, &left, &val_b,
3079 &neg, tr_b, sizeof(tr_b),
3080 &c);
3081 if (err)
3082 break;
3083 if (val_b >= bitmap_len || neg ||
3084 val_a > val_b) {
3085 err = -EINVAL;
3086 break;
3087 }
3088 if (left) {
3089 p++;
3090 left--;
3091 }
3092 }
3093
3094 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3095 first = 0;
3096 proc_skip_char(&p, &left, '\n');
3097 }
3098 kfree(kbuf);
3099 } else {
3100 unsigned long bit_a, bit_b = 0;
3101
3102 while (left) {
3103 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3104 if (bit_a >= bitmap_len)
3105 break;
3106 bit_b = find_next_zero_bit(bitmap, bitmap_len,
3107 bit_a + 1) - 1;
3108
3109 if (!first) {
3110 err = proc_put_char(&buffer, &left, ',');
3111 if (err)
3112 break;
3113 }
3114 err = proc_put_long(&buffer, &left, bit_a, false);
3115 if (err)
3116 break;
3117 if (bit_a != bit_b) {
3118 err = proc_put_char(&buffer, &left, '-');
3119 if (err)
3120 break;
3121 err = proc_put_long(&buffer, &left, bit_b, false);
3122 if (err)
3123 break;
3124 }
3125
3126 first = 0; bit_b++;
3127 }
3128 if (!err)
3129 err = proc_put_char(&buffer, &left, '\n');
3130 }
3131
3132 if (!err) {
3133 if (write) {
3134 if (*ppos)
3135 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3136 else
3137 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3138 }
3139 *lenp -= left;
3140 *ppos += *lenp;
3141 }
3142
3143 kfree(tmp_bitmap);
3144 return err;
3145 }
3146
3147 #else /* CONFIG_PROC_SYSCTL */
3148
3149 int proc_dostring(struct ctl_table *table, int write,
3150 void __user *buffer, size_t *lenp, loff_t *ppos)
3151 {
3152 return -ENOSYS;
3153 }
3154
3155 int proc_dointvec(struct ctl_table *table, int write,
3156 void __user *buffer, size_t *lenp, loff_t *ppos)
3157 {
3158 return -ENOSYS;
3159 }
3160
3161 int proc_douintvec(struct ctl_table *table, int write,
3162 void __user *buffer, size_t *lenp, loff_t *ppos)
3163 {
3164 return -ENOSYS;
3165 }
3166
3167 int proc_dointvec_minmax(struct ctl_table *table, int write,
3168 void __user *buffer, size_t *lenp, loff_t *ppos)
3169 {
3170 return -ENOSYS;
3171 }
3172
3173 int proc_douintvec_minmax(struct ctl_table *table, int write,
3174 void __user *buffer, size_t *lenp, loff_t *ppos)
3175 {
3176 return -ENOSYS;
3177 }
3178
3179 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3180 void __user *buffer, size_t *lenp, loff_t *ppos)
3181 {
3182 return -ENOSYS;
3183 }
3184
3185 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3186 void __user *buffer, size_t *lenp, loff_t *ppos)
3187 {
3188 return -ENOSYS;
3189 }
3190
3191 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3192 void __user *buffer, size_t *lenp, loff_t *ppos)
3193 {
3194 return -ENOSYS;
3195 }
3196
3197 int proc_doulongvec_minmax(struct ctl_table *table, int write,
3198 void __user *buffer, size_t *lenp, loff_t *ppos)
3199 {
3200 return -ENOSYS;
3201 }
3202
3203 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3204 void __user *buffer,
3205 size_t *lenp, loff_t *ppos)
3206 {
3207 return -ENOSYS;
3208 }
3209
3210
3211 #endif /* CONFIG_PROC_SYSCTL */
3212
3213 /*
3214 * No sense putting this after each symbol definition, twice,
3215 * exception granted :-)
3216 */
3217 EXPORT_SYMBOL(proc_dointvec);
3218 EXPORT_SYMBOL(proc_douintvec);
3219 EXPORT_SYMBOL(proc_dointvec_jiffies);
3220 EXPORT_SYMBOL(proc_dointvec_minmax);
3221 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3222 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3223 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3224 EXPORT_SYMBOL(proc_dostring);
3225 EXPORT_SYMBOL(proc_doulongvec_minmax);
3226 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
Ubuntu Jammy Linux kernel mirror