]> git.ipfire.org Git - thirdparty/linux.git/blob - kernel/sysctl.c
projects / thirdparty / linux.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Merge tag '5.2-rc-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6
[thirdparty/linux.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/userfaultfd_k.h>
70
71 #include "../lib/kstrtox.h"
72
73 #include <linux/uaccess.h>
74 #include <asm/processor.h>
75
76 #ifdef CONFIG_X86
77 #include <asm/nmi.h>
78 #include <asm/stacktrace.h>
79 #include <asm/io.h>
80 #endif
81 #ifdef CONFIG_SPARC
82 #include <asm/setup.h>
83 #endif
84 #ifdef CONFIG_BSD_PROCESS_ACCT
85 #include <linux/acct.h>
86 #endif
87 #ifdef CONFIG_RT_MUTEXES
88 #include <linux/rtmutex.h>
89 #endif
90 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
91 #include <linux/lockdep.h>
92 #endif
93 #ifdef CONFIG_CHR_DEV_SG
94 #include <scsi/sg.h>
95 #endif
96 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
97 #include <linux/stackleak.h>
98 #endif
99 #ifdef CONFIG_LOCKUP_DETECTOR
100 #include <linux/nmi.h>
101 #endif
102
103 #if defined(CONFIG_SYSCTL)
104
105 /* External variables not in a header file. */
106 extern int suid_dumpable;
107 #ifdef CONFIG_COREDUMP
108 extern int core_uses_pid;
109 extern char core_pattern[];
110 extern unsigned int core_pipe_limit;
111 #endif
112 extern int pid_max;
113 extern int pid_max_min, pid_max_max;
114 extern int percpu_pagelist_fraction;
115 extern int latencytop_enabled;
116 extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
117 #ifndef CONFIG_MMU
118 extern int sysctl_nr_trim_pages;
119 #endif
120
121 /* Constants used for minimum and maximum */
122 #ifdef CONFIG_LOCKUP_DETECTOR
123 static int sixty = 60;
124 #endif
125
126 static int __maybe_unused neg_one = -1;
127
128 static int zero;
129 static int __maybe_unused one = 1;
130 static int __maybe_unused two = 2;
131 static int __maybe_unused four = 4;
132 static unsigned long zero_ul;
133 static unsigned long one_ul = 1;
134 static unsigned long long_max = LONG_MAX;
135 static int one_hundred = 100;
136 static int one_thousand = 1000;
137 #ifdef CONFIG_PRINTK
138 static int ten_thousand = 10000;
139 #endif
140 #ifdef CONFIG_PERF_EVENTS
141 static int six_hundred_forty_kb = 640 * 1024;
142 #endif
143
144 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
145 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
146
147 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
148 static int maxolduid = 65535;
149 static int minolduid;
150
151 static int ngroups_max = NGROUPS_MAX;
152 static const int cap_last_cap = CAP_LAST_CAP;
153
154 /*
155 * This is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs
156 * and hung_task_check_interval_secs
157 */
158 #ifdef CONFIG_DETECT_HUNG_TASK
159 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
160 #endif
161
162 #ifdef CONFIG_INOTIFY_USER
163 #include <linux/inotify.h>
164 #endif
165 #ifdef CONFIG_SPARC
166 #endif
167
168 #ifdef __hppa__
169 extern int pwrsw_enabled;
170 #endif
171
172 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
173 extern int unaligned_enabled;
174 #endif
175
176 #ifdef CONFIG_IA64
177 extern int unaligned_dump_stack;
178 #endif
179
180 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
181 extern int no_unaligned_warning;
182 #endif
183
184 #ifdef CONFIG_PROC_SYSCTL
185
186 /**
187 * enum sysctl_writes_mode - supported sysctl write modes
188 *
189 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
190 * to be written, and multiple writes on the same sysctl file descriptor
191 * will rewrite the sysctl value, regardless of file position. No warning
192 * is issued when the initial position is not 0.
193 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
194 * not 0.
195 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
196 * file position 0 and the value must be fully contained in the buffer
197 * sent to the write syscall. If dealing with strings respect the file
198 * position, but restrict this to the max length of the buffer, anything
199 * passed the max lenght will be ignored. Multiple writes will append
200 * to the buffer.
201 *
202 * These write modes control how current file position affects the behavior of
203 * updating sysctl values through the proc interface on each write.
204 */
205 enum sysctl_writes_mode {
206 SYSCTL_WRITES_LEGACY = -1,
207 SYSCTL_WRITES_WARN = 0,
208 SYSCTL_WRITES_STRICT = 1,
209 };
210
211 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
212
213 static int proc_do_cad_pid(struct ctl_table *table, int write,
214 void __user *buffer, size_t *lenp, loff_t *ppos);
215 static int proc_taint(struct ctl_table *table, int write,
216 void __user *buffer, size_t *lenp, loff_t *ppos);
217 #endif
218
219 #ifdef CONFIG_PRINTK
220 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
221 void __user *buffer, size_t *lenp, loff_t *ppos);
222 #endif
223
224 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
225 void __user *buffer, size_t *lenp, loff_t *ppos);
226 #ifdef CONFIG_COREDUMP
227 static int proc_dostring_coredump(struct ctl_table *table, int write,
228 void __user *buffer, size_t *lenp, loff_t *ppos);
229 #endif
230 static int proc_dopipe_max_size(struct ctl_table *table, int write,
231 void __user *buffer, size_t *lenp, loff_t *ppos);
232 #ifdef CONFIG_BPF_SYSCALL
233 static int proc_dointvec_minmax_bpf_stats(struct ctl_table *table, int write,
234 void __user *buffer, size_t *lenp,
235 loff_t *ppos);
236 #endif
237
238 #ifdef CONFIG_MAGIC_SYSRQ
239 /* Note: sysrq code uses its own private copy */
240 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
241
242 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
243 void __user *buffer, size_t *lenp,
244 loff_t *ppos)
245 {
246 int error;
247
248 error = proc_dointvec(table, write, buffer, lenp, ppos);
249 if (error)
250 return error;
251
252 if (write)
253 sysrq_toggle_support(__sysrq_enabled);
254
255 return 0;
256 }
257
258 #endif
259
260 static struct ctl_table kern_table[];
261 static struct ctl_table vm_table[];
262 static struct ctl_table fs_table[];
263 static struct ctl_table debug_table[];
264 static struct ctl_table dev_table[];
265 extern struct ctl_table random_table[];
266 #ifdef CONFIG_EPOLL
267 extern struct ctl_table epoll_table[];
268 #endif
269
270 #ifdef CONFIG_FW_LOADER_USER_HELPER
271 extern struct ctl_table firmware_config_table[];
272 #endif
273
274 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
275 int sysctl_legacy_va_layout;
276 #endif
277
278 /* The default sysctl tables: */
279
280 static struct ctl_table sysctl_base_table[] = {
281 {
282 .procname = "kernel",
283 .mode = 0555,
284 .child = kern_table,
285 },
286 {
287 .procname = "vm",
288 .mode = 0555,
289 .child = vm_table,
290 },
291 {
292 .procname = "fs",
293 .mode = 0555,
294 .child = fs_table,
295 },
296 {
297 .procname = "debug",
298 .mode = 0555,
299 .child = debug_table,
300 },
301 {
302 .procname = "dev",
303 .mode = 0555,
304 .child = dev_table,
305 },
306 { }
307 };
308
309 #ifdef CONFIG_SCHED_DEBUG
310 static int min_sched_granularity_ns = 100000; /* 100 usecs */
311 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
312 static int min_wakeup_granularity_ns; /* 0 usecs */
313 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
314 #ifdef CONFIG_SMP
315 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
316 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
317 #endif /* CONFIG_SMP */
318 #endif /* CONFIG_SCHED_DEBUG */
319
320 #ifdef CONFIG_COMPACTION
321 static int min_extfrag_threshold;
322 static int max_extfrag_threshold = 1000;
323 #endif
324
325 static struct ctl_table kern_table[] = {
326 {
327 .procname = "sched_child_runs_first",
328 .data = &sysctl_sched_child_runs_first,
329 .maxlen = sizeof(unsigned int),
330 .mode = 0644,
331 .proc_handler = proc_dointvec,
332 },
333 #ifdef CONFIG_SCHED_DEBUG
334 {
335 .procname = "sched_min_granularity_ns",
336 .data = &sysctl_sched_min_granularity,
337 .maxlen = sizeof(unsigned int),
338 .mode = 0644,
339 .proc_handler = sched_proc_update_handler,
340 .extra1 = &min_sched_granularity_ns,
341 .extra2 = &max_sched_granularity_ns,
342 },
343 {
344 .procname = "sched_latency_ns",
345 .data = &sysctl_sched_latency,
346 .maxlen = sizeof(unsigned int),
347 .mode = 0644,
348 .proc_handler = sched_proc_update_handler,
349 .extra1 = &min_sched_granularity_ns,
350 .extra2 = &max_sched_granularity_ns,
351 },
352 {
353 .procname = "sched_wakeup_granularity_ns",
354 .data = &sysctl_sched_wakeup_granularity,
355 .maxlen = sizeof(unsigned int),
356 .mode = 0644,
357 .proc_handler = sched_proc_update_handler,
358 .extra1 = &min_wakeup_granularity_ns,
359 .extra2 = &max_wakeup_granularity_ns,
360 },
361 #ifdef CONFIG_SMP
362 {
363 .procname = "sched_tunable_scaling",
364 .data = &sysctl_sched_tunable_scaling,
365 .maxlen = sizeof(enum sched_tunable_scaling),
366 .mode = 0644,
367 .proc_handler = sched_proc_update_handler,
368 .extra1 = &min_sched_tunable_scaling,
369 .extra2 = &max_sched_tunable_scaling,
370 },
371 {
372 .procname = "sched_migration_cost_ns",
373 .data = &sysctl_sched_migration_cost,
374 .maxlen = sizeof(unsigned int),
375 .mode = 0644,
376 .proc_handler = proc_dointvec,
377 },
378 {
379 .procname = "sched_nr_migrate",
380 .data = &sysctl_sched_nr_migrate,
381 .maxlen = sizeof(unsigned int),
382 .mode = 0644,
383 .proc_handler = proc_dointvec,
384 },
385 #ifdef CONFIG_SCHEDSTATS
386 {
387 .procname = "sched_schedstats",
388 .data = NULL,
389 .maxlen = sizeof(unsigned int),
390 .mode = 0644,
391 .proc_handler = sysctl_schedstats,
392 .extra1 = &zero,
393 .extra2 = &one,
394 },
395 #endif /* CONFIG_SCHEDSTATS */
396 #endif /* CONFIG_SMP */
397 #ifdef CONFIG_NUMA_BALANCING
398 {
399 .procname = "numa_balancing_scan_delay_ms",
400 .data = &sysctl_numa_balancing_scan_delay,
401 .maxlen = sizeof(unsigned int),
402 .mode = 0644,
403 .proc_handler = proc_dointvec,
404 },
405 {
406 .procname = "numa_balancing_scan_period_min_ms",
407 .data = &sysctl_numa_balancing_scan_period_min,
408 .maxlen = sizeof(unsigned int),
409 .mode = 0644,
410 .proc_handler = proc_dointvec,
411 },
412 {
413 .procname = "numa_balancing_scan_period_max_ms",
414 .data = &sysctl_numa_balancing_scan_period_max,
415 .maxlen = sizeof(unsigned int),
416 .mode = 0644,
417 .proc_handler = proc_dointvec,
418 },
419 {
420 .procname = "numa_balancing_scan_size_mb",
421 .data = &sysctl_numa_balancing_scan_size,
422 .maxlen = sizeof(unsigned int),
423 .mode = 0644,
424 .proc_handler = proc_dointvec_minmax,
425 .extra1 = &one,
426 },
427 {
428 .procname = "numa_balancing",
429 .data = NULL, /* filled in by handler */
430 .maxlen = sizeof(unsigned int),
431 .mode = 0644,
432 .proc_handler = sysctl_numa_balancing,
433 .extra1 = &zero,
434 .extra2 = &one,
435 },
436 #endif /* CONFIG_NUMA_BALANCING */
437 #endif /* CONFIG_SCHED_DEBUG */
438 {
439 .procname = "sched_rt_period_us",
440 .data = &sysctl_sched_rt_period,
441 .maxlen = sizeof(unsigned int),
442 .mode = 0644,
443 .proc_handler = sched_rt_handler,
444 },
445 {
446 .procname = "sched_rt_runtime_us",
447 .data = &sysctl_sched_rt_runtime,
448 .maxlen = sizeof(int),
449 .mode = 0644,
450 .proc_handler = sched_rt_handler,
451 },
452 {
453 .procname = "sched_rr_timeslice_ms",
454 .data = &sysctl_sched_rr_timeslice,
455 .maxlen = sizeof(int),
456 .mode = 0644,
457 .proc_handler = sched_rr_handler,
458 },
459 #ifdef CONFIG_SCHED_AUTOGROUP
460 {
461 .procname = "sched_autogroup_enabled",
462 .data = &sysctl_sched_autogroup_enabled,
463 .maxlen = sizeof(unsigned int),
464 .mode = 0644,
465 .proc_handler = proc_dointvec_minmax,
466 .extra1 = &zero,
467 .extra2 = &one,
468 },
469 #endif
470 #ifdef CONFIG_CFS_BANDWIDTH
471 {
472 .procname = "sched_cfs_bandwidth_slice_us",
473 .data = &sysctl_sched_cfs_bandwidth_slice,
474 .maxlen = sizeof(unsigned int),
475 .mode = 0644,
476 .proc_handler = proc_dointvec_minmax,
477 .extra1 = &one,
478 },
479 #endif
480 #if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_CPU_FREQ_GOV_SCHEDUTIL)
481 {
482 .procname = "sched_energy_aware",
483 .data = &sysctl_sched_energy_aware,
484 .maxlen = sizeof(unsigned int),
485 .mode = 0644,
486 .proc_handler = sched_energy_aware_handler,
487 .extra1 = &zero,
488 .extra2 = &one,
489 },
490 #endif
491 #ifdef CONFIG_PROVE_LOCKING
492 {
493 .procname = "prove_locking",
494 .data = &prove_locking,
495 .maxlen = sizeof(int),
496 .mode = 0644,
497 .proc_handler = proc_dointvec,
498 },
499 #endif
500 #ifdef CONFIG_LOCK_STAT
501 {
502 .procname = "lock_stat",
503 .data = &lock_stat,
504 .maxlen = sizeof(int),
505 .mode = 0644,
506 .proc_handler = proc_dointvec,
507 },
508 #endif
509 {
510 .procname = "panic",
511 .data = &panic_timeout,
512 .maxlen = sizeof(int),
513 .mode = 0644,
514 .proc_handler = proc_dointvec,
515 },
516 #ifdef CONFIG_COREDUMP
517 {
518 .procname = "core_uses_pid",
519 .data = &core_uses_pid,
520 .maxlen = sizeof(int),
521 .mode = 0644,
522 .proc_handler = proc_dointvec,
523 },
524 {
525 .procname = "core_pattern",
526 .data = core_pattern,
527 .maxlen = CORENAME_MAX_SIZE,
528 .mode = 0644,
529 .proc_handler = proc_dostring_coredump,
530 },
531 {
532 .procname = "core_pipe_limit",
533 .data = &core_pipe_limit,
534 .maxlen = sizeof(unsigned int),
535 .mode = 0644,
536 .proc_handler = proc_dointvec,
537 },
538 #endif
539 #ifdef CONFIG_PROC_SYSCTL
540 {
541 .procname = "tainted",
542 .maxlen = sizeof(long),
543 .mode = 0644,
544 .proc_handler = proc_taint,
545 },
546 {
547 .procname = "sysctl_writes_strict",
548 .data = &sysctl_writes_strict,
549 .maxlen = sizeof(int),
550 .mode = 0644,
551 .proc_handler = proc_dointvec_minmax,
552 .extra1 = &neg_one,
553 .extra2 = &one,
554 },
555 #endif
556 #ifdef CONFIG_LATENCYTOP
557 {
558 .procname = "latencytop",
559 .data = &latencytop_enabled,
560 .maxlen = sizeof(int),
561 .mode = 0644,
562 .proc_handler = sysctl_latencytop,
563 },
564 #endif
565 #ifdef CONFIG_BLK_DEV_INITRD
566 {
567 .procname = "real-root-dev",
568 .data = &real_root_dev,
569 .maxlen = sizeof(int),
570 .mode = 0644,
571 .proc_handler = proc_dointvec,
572 },
573 #endif
574 {
575 .procname = "print-fatal-signals",
576 .data = &print_fatal_signals,
577 .maxlen = sizeof(int),
578 .mode = 0644,
579 .proc_handler = proc_dointvec,
580 },
581 #ifdef CONFIG_SPARC
582 {
583 .procname = "reboot-cmd",
584 .data = reboot_command,
585 .maxlen = 256,
586 .mode = 0644,
587 .proc_handler = proc_dostring,
588 },
589 {
590 .procname = "stop-a",
591 .data = &stop_a_enabled,
592 .maxlen = sizeof (int),
593 .mode = 0644,
594 .proc_handler = proc_dointvec,
595 },
596 {
597 .procname = "scons-poweroff",
598 .data = &scons_pwroff,
599 .maxlen = sizeof (int),
600 .mode = 0644,
601 .proc_handler = proc_dointvec,
602 },
603 #endif
604 #ifdef CONFIG_SPARC64
605 {
606 .procname = "tsb-ratio",
607 .data = &sysctl_tsb_ratio,
608 .maxlen = sizeof (int),
609 .mode = 0644,
610 .proc_handler = proc_dointvec,
611 },
612 #endif
613 #ifdef __hppa__
614 {
615 .procname = "soft-power",
616 .data = &pwrsw_enabled,
617 .maxlen = sizeof (int),
618 .mode = 0644,
619 .proc_handler = proc_dointvec,
620 },
621 #endif
622 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
623 {
624 .procname = "unaligned-trap",
625 .data = &unaligned_enabled,
626 .maxlen = sizeof (int),
627 .mode = 0644,
628 .proc_handler = proc_dointvec,
629 },
630 #endif
631 {
632 .procname = "ctrl-alt-del",
633 .data = &C_A_D,
634 .maxlen = sizeof(int),
635 .mode = 0644,
636 .proc_handler = proc_dointvec,
637 },
638 #ifdef CONFIG_FUNCTION_TRACER
639 {
640 .procname = "ftrace_enabled",
641 .data = &ftrace_enabled,
642 .maxlen = sizeof(int),
643 .mode = 0644,
644 .proc_handler = ftrace_enable_sysctl,
645 },
646 #endif
647 #ifdef CONFIG_STACK_TRACER
648 {
649 .procname = "stack_tracer_enabled",
650 .data = &stack_tracer_enabled,
651 .maxlen = sizeof(int),
652 .mode = 0644,
653 .proc_handler = stack_trace_sysctl,
654 },
655 #endif
656 #ifdef CONFIG_TRACING
657 {
658 .procname = "ftrace_dump_on_oops",
659 .data = &ftrace_dump_on_oops,
660 .maxlen = sizeof(int),
661 .mode = 0644,
662 .proc_handler = proc_dointvec,
663 },
664 {
665 .procname = "traceoff_on_warning",
666 .data = &__disable_trace_on_warning,
667 .maxlen = sizeof(__disable_trace_on_warning),
668 .mode = 0644,
669 .proc_handler = proc_dointvec,
670 },
671 {
672 .procname = "tracepoint_printk",
673 .data = &tracepoint_printk,
674 .maxlen = sizeof(tracepoint_printk),
675 .mode = 0644,
676 .proc_handler = tracepoint_printk_sysctl,
677 },
678 #endif
679 #ifdef CONFIG_KEXEC_CORE
680 {
681 .procname = "kexec_load_disabled",
682 .data = &kexec_load_disabled,
683 .maxlen = sizeof(int),
684 .mode = 0644,
685 /* only handle a transition from default "0" to "1" */
686 .proc_handler = proc_dointvec_minmax,
687 .extra1 = &one,
688 .extra2 = &one,
689 },
690 #endif
691 #ifdef CONFIG_MODULES
692 {
693 .procname = "modprobe",
694 .data = &modprobe_path,
695 .maxlen = KMOD_PATH_LEN,
696 .mode = 0644,
697 .proc_handler = proc_dostring,
698 },
699 {
700 .procname = "modules_disabled",
701 .data = &modules_disabled,
702 .maxlen = sizeof(int),
703 .mode = 0644,
704 /* only handle a transition from default "0" to "1" */
705 .proc_handler = proc_dointvec_minmax,
706 .extra1 = &one,
707 .extra2 = &one,
708 },
709 #endif
710 #ifdef CONFIG_UEVENT_HELPER
711 {
712 .procname = "hotplug",
713 .data = &uevent_helper,
714 .maxlen = UEVENT_HELPER_PATH_LEN,
715 .mode = 0644,
716 .proc_handler = proc_dostring,
717 },
718 #endif
719 #ifdef CONFIG_CHR_DEV_SG
720 {
721 .procname = "sg-big-buff",
722 .data = &sg_big_buff,
723 .maxlen = sizeof (int),
724 .mode = 0444,
725 .proc_handler = proc_dointvec,
726 },
727 #endif
728 #ifdef CONFIG_BSD_PROCESS_ACCT
729 {
730 .procname = "acct",
731 .data = &acct_parm,
732 .maxlen = 3*sizeof(int),
733 .mode = 0644,
734 .proc_handler = proc_dointvec,
735 },
736 #endif
737 #ifdef CONFIG_MAGIC_SYSRQ
738 {
739 .procname = "sysrq",
740 .data = &__sysrq_enabled,
741 .maxlen = sizeof (int),
742 .mode = 0644,
743 .proc_handler = sysrq_sysctl_handler,
744 },
745 #endif
746 #ifdef CONFIG_PROC_SYSCTL
747 {
748 .procname = "cad_pid",
749 .data = NULL,
750 .maxlen = sizeof (int),
751 .mode = 0600,
752 .proc_handler = proc_do_cad_pid,
753 },
754 #endif
755 {
756 .procname = "threads-max",
757 .data = NULL,
758 .maxlen = sizeof(int),
759 .mode = 0644,
760 .proc_handler = sysctl_max_threads,
761 },
762 {
763 .procname = "random",
764 .mode = 0555,
765 .child = random_table,
766 },
767 {
768 .procname = "usermodehelper",
769 .mode = 0555,
770 .child = usermodehelper_table,
771 },
772 #ifdef CONFIG_FW_LOADER_USER_HELPER
773 {
774 .procname = "firmware_config",
775 .mode = 0555,
776 .child = firmware_config_table,
777 },
778 #endif
779 {
780 .procname = "overflowuid",
781 .data = &overflowuid,
782 .maxlen = sizeof(int),
783 .mode = 0644,
784 .proc_handler = proc_dointvec_minmax,
785 .extra1 = &minolduid,
786 .extra2 = &maxolduid,
787 },
788 {
789 .procname = "overflowgid",
790 .data = &overflowgid,
791 .maxlen = sizeof(int),
792 .mode = 0644,
793 .proc_handler = proc_dointvec_minmax,
794 .extra1 = &minolduid,
795 .extra2 = &maxolduid,
796 },
797 #ifdef CONFIG_S390
798 #ifdef CONFIG_MATHEMU
799 {
800 .procname = "ieee_emulation_warnings",
801 .data = &sysctl_ieee_emulation_warnings,
802 .maxlen = sizeof(int),
803 .mode = 0644,
804 .proc_handler = proc_dointvec,
805 },
806 #endif
807 {
808 .procname = "userprocess_debug",
809 .data = &show_unhandled_signals,
810 .maxlen = sizeof(int),
811 .mode = 0644,
812 .proc_handler = proc_dointvec,
813 },
814 #endif
815 {
816 .procname = "pid_max",
817 .data = &pid_max,
818 .maxlen = sizeof (int),
819 .mode = 0644,
820 .proc_handler = proc_dointvec_minmax,
821 .extra1 = &pid_max_min,
822 .extra2 = &pid_max_max,
823 },
824 {
825 .procname = "panic_on_oops",
826 .data = &panic_on_oops,
827 .maxlen = sizeof(int),
828 .mode = 0644,
829 .proc_handler = proc_dointvec,
830 },
831 {
832 .procname = "panic_print",
833 .data = &panic_print,
834 .maxlen = sizeof(unsigned long),
835 .mode = 0644,
836 .proc_handler = proc_doulongvec_minmax,
837 },
838 #if defined CONFIG_PRINTK
839 {
840 .procname = "printk",
841 .data = &console_loglevel,
842 .maxlen = 4*sizeof(int),
843 .mode = 0644,
844 .proc_handler = proc_dointvec,
845 },
846 {
847 .procname = "printk_ratelimit",
848 .data = &printk_ratelimit_state.interval,
849 .maxlen = sizeof(int),
850 .mode = 0644,
851 .proc_handler = proc_dointvec_jiffies,
852 },
853 {
854 .procname = "printk_ratelimit_burst",
855 .data = &printk_ratelimit_state.burst,
856 .maxlen = sizeof(int),
857 .mode = 0644,
858 .proc_handler = proc_dointvec,
859 },
860 {
861 .procname = "printk_delay",
862 .data = &printk_delay_msec,
863 .maxlen = sizeof(int),
864 .mode = 0644,
865 .proc_handler = proc_dointvec_minmax,
866 .extra1 = &zero,
867 .extra2 = &ten_thousand,
868 },
869 {
870 .procname = "printk_devkmsg",
871 .data = devkmsg_log_str,
872 .maxlen = DEVKMSG_STR_MAX_SIZE,
873 .mode = 0644,
874 .proc_handler = devkmsg_sysctl_set_loglvl,
875 },
876 {
877 .procname = "dmesg_restrict",
878 .data = &dmesg_restrict,
879 .maxlen = sizeof(int),
880 .mode = 0644,
881 .proc_handler = proc_dointvec_minmax_sysadmin,
882 .extra1 = &zero,
883 .extra2 = &one,
884 },
885 {
886 .procname = "kptr_restrict",
887 .data = &kptr_restrict,
888 .maxlen = sizeof(int),
889 .mode = 0644,
890 .proc_handler = proc_dointvec_minmax_sysadmin,
891 .extra1 = &zero,
892 .extra2 = &two,
893 },
894 #endif
895 {
896 .procname = "ngroups_max",
897 .data = &ngroups_max,
898 .maxlen = sizeof (int),
899 .mode = 0444,
900 .proc_handler = proc_dointvec,
901 },
902 {
903 .procname = "cap_last_cap",
904 .data = (void *)&cap_last_cap,
905 .maxlen = sizeof(int),
906 .mode = 0444,
907 .proc_handler = proc_dointvec,
908 },
909 #if defined(CONFIG_LOCKUP_DETECTOR)
910 {
911 .procname = "watchdog",
912 .data = &watchdog_user_enabled,
913 .maxlen = sizeof(int),
914 .mode = 0644,
915 .proc_handler = proc_watchdog,
916 .extra1 = &zero,
917 .extra2 = &one,
918 },
919 {
920 .procname = "watchdog_thresh",
921 .data = &watchdog_thresh,
922 .maxlen = sizeof(int),
923 .mode = 0644,
924 .proc_handler = proc_watchdog_thresh,
925 .extra1 = &zero,
926 .extra2 = &sixty,
927 },
928 {
929 .procname = "nmi_watchdog",
930 .data = &nmi_watchdog_user_enabled,
931 .maxlen = sizeof(int),
932 .mode = NMI_WATCHDOG_SYSCTL_PERM,
933 .proc_handler = proc_nmi_watchdog,
934 .extra1 = &zero,
935 .extra2 = &one,
936 },
937 {
938 .procname = "watchdog_cpumask",
939 .data = &watchdog_cpumask_bits,
940 .maxlen = NR_CPUS,
941 .mode = 0644,
942 .proc_handler = proc_watchdog_cpumask,
943 },
944 #ifdef CONFIG_SOFTLOCKUP_DETECTOR
945 {
946 .procname = "soft_watchdog",
947 .data = &soft_watchdog_user_enabled,
948 .maxlen = sizeof(int),
949 .mode = 0644,
950 .proc_handler = proc_soft_watchdog,
951 .extra1 = &zero,
952 .extra2 = &one,
953 },
954 {
955 .procname = "softlockup_panic",
956 .data = &softlockup_panic,
957 .maxlen = sizeof(int),
958 .mode = 0644,
959 .proc_handler = proc_dointvec_minmax,
960 .extra1 = &zero,
961 .extra2 = &one,
962 },
963 #ifdef CONFIG_SMP
964 {
965 .procname = "softlockup_all_cpu_backtrace",
966 .data = &sysctl_softlockup_all_cpu_backtrace,
967 .maxlen = sizeof(int),
968 .mode = 0644,
969 .proc_handler = proc_dointvec_minmax,
970 .extra1 = &zero,
971 .extra2 = &one,
972 },
973 #endif /* CONFIG_SMP */
974 #endif
975 #ifdef CONFIG_HARDLOCKUP_DETECTOR
976 {
977 .procname = "hardlockup_panic",
978 .data = &hardlockup_panic,
979 .maxlen = sizeof(int),
980 .mode = 0644,
981 .proc_handler = proc_dointvec_minmax,
982 .extra1 = &zero,
983 .extra2 = &one,
984 },
985 #ifdef CONFIG_SMP
986 {
987 .procname = "hardlockup_all_cpu_backtrace",
988 .data = &sysctl_hardlockup_all_cpu_backtrace,
989 .maxlen = sizeof(int),
990 .mode = 0644,
991 .proc_handler = proc_dointvec_minmax,
992 .extra1 = &zero,
993 .extra2 = &one,
994 },
995 #endif /* CONFIG_SMP */
996 #endif
997 #endif
998
999 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
1000 {
1001 .procname = "unknown_nmi_panic",
1002 .data = &unknown_nmi_panic,
1003 .maxlen = sizeof (int),
1004 .mode = 0644,
1005 .proc_handler = proc_dointvec,
1006 },
1007 #endif
1008 #if defined(CONFIG_X86)
1009 {
1010 .procname = "panic_on_unrecovered_nmi",
1011 .data = &panic_on_unrecovered_nmi,
1012 .maxlen = sizeof(int),
1013 .mode = 0644,
1014 .proc_handler = proc_dointvec,
1015 },
1016 {
1017 .procname = "panic_on_io_nmi",
1018 .data = &panic_on_io_nmi,
1019 .maxlen = sizeof(int),
1020 .mode = 0644,
1021 .proc_handler = proc_dointvec,
1022 },
1023 #ifdef CONFIG_DEBUG_STACKOVERFLOW
1024 {
1025 .procname = "panic_on_stackoverflow",
1026 .data = &sysctl_panic_on_stackoverflow,
1027 .maxlen = sizeof(int),
1028 .mode = 0644,
1029 .proc_handler = proc_dointvec,
1030 },
1031 #endif
1032 {
1033 .procname = "bootloader_type",
1034 .data = &bootloader_type,
1035 .maxlen = sizeof (int),
1036 .mode = 0444,
1037 .proc_handler = proc_dointvec,
1038 },
1039 {
1040 .procname = "bootloader_version",
1041 .data = &bootloader_version,
1042 .maxlen = sizeof (int),
1043 .mode = 0444,
1044 .proc_handler = proc_dointvec,
1045 },
1046 {
1047 .procname = "io_delay_type",
1048 .data = &io_delay_type,
1049 .maxlen = sizeof(int),
1050 .mode = 0644,
1051 .proc_handler = proc_dointvec,
1052 },
1053 #endif
1054 #if defined(CONFIG_MMU)
1055 {
1056 .procname = "randomize_va_space",
1057 .data = &randomize_va_space,
1058 .maxlen = sizeof(int),
1059 .mode = 0644,
1060 .proc_handler = proc_dointvec,
1061 },
1062 #endif
1063 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1064 {
1065 .procname = "spin_retry",
1066 .data = &spin_retry,
1067 .maxlen = sizeof (int),
1068 .mode = 0644,
1069 .proc_handler = proc_dointvec,
1070 },
1071 #endif
1072 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1073 {
1074 .procname = "acpi_video_flags",
1075 .data = &acpi_realmode_flags,
1076 .maxlen = sizeof (unsigned long),
1077 .mode = 0644,
1078 .proc_handler = proc_doulongvec_minmax,
1079 },
1080 #endif
1081 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1082 {
1083 .procname = "ignore-unaligned-usertrap",
1084 .data = &no_unaligned_warning,
1085 .maxlen = sizeof (int),
1086 .mode = 0644,
1087 .proc_handler = proc_dointvec,
1088 },
1089 #endif
1090 #ifdef CONFIG_IA64
1091 {
1092 .procname = "unaligned-dump-stack",
1093 .data = &unaligned_dump_stack,
1094 .maxlen = sizeof (int),
1095 .mode = 0644,
1096 .proc_handler = proc_dointvec,
1097 },
1098 #endif
1099 #ifdef CONFIG_DETECT_HUNG_TASK
1100 {
1101 .procname = "hung_task_panic",
1102 .data = &sysctl_hung_task_panic,
1103 .maxlen = sizeof(int),
1104 .mode = 0644,
1105 .proc_handler = proc_dointvec_minmax,
1106 .extra1 = &zero,
1107 .extra2 = &one,
1108 },
1109 {
1110 .procname = "hung_task_check_count",
1111 .data = &sysctl_hung_task_check_count,
1112 .maxlen = sizeof(int),
1113 .mode = 0644,
1114 .proc_handler = proc_dointvec_minmax,
1115 .extra1 = &zero,
1116 },
1117 {
1118 .procname = "hung_task_timeout_secs",
1119 .data = &sysctl_hung_task_timeout_secs,
1120 .maxlen = sizeof(unsigned long),
1121 .mode = 0644,
1122 .proc_handler = proc_dohung_task_timeout_secs,
1123 .extra2 = &hung_task_timeout_max,
1124 },
1125 {
1126 .procname = "hung_task_check_interval_secs",
1127 .data = &sysctl_hung_task_check_interval_secs,
1128 .maxlen = sizeof(unsigned long),
1129 .mode = 0644,
1130 .proc_handler = proc_dohung_task_timeout_secs,
1131 .extra2 = &hung_task_timeout_max,
1132 },
1133 {
1134 .procname = "hung_task_warnings",
1135 .data = &sysctl_hung_task_warnings,
1136 .maxlen = sizeof(int),
1137 .mode = 0644,
1138 .proc_handler = proc_dointvec_minmax,
1139 .extra1 = &neg_one,
1140 },
1141 #endif
1142 #ifdef CONFIG_RT_MUTEXES
1143 {
1144 .procname = "max_lock_depth",
1145 .data = &max_lock_depth,
1146 .maxlen = sizeof(int),
1147 .mode = 0644,
1148 .proc_handler = proc_dointvec,
1149 },
1150 #endif
1151 {
1152 .procname = "poweroff_cmd",
1153 .data = &poweroff_cmd,
1154 .maxlen = POWEROFF_CMD_PATH_LEN,
1155 .mode = 0644,
1156 .proc_handler = proc_dostring,
1157 },
1158 #ifdef CONFIG_KEYS
1159 {
1160 .procname = "keys",
1161 .mode = 0555,
1162 .child = key_sysctls,
1163 },
1164 #endif
1165 #ifdef CONFIG_PERF_EVENTS
1166 /*
1167 * User-space scripts rely on the existence of this file
1168 * as a feature check for perf_events being enabled.
1169 *
1170 * So it's an ABI, do not remove!
1171 */
1172 {
1173 .procname = "perf_event_paranoid",
1174 .data = &sysctl_perf_event_paranoid,
1175 .maxlen = sizeof(sysctl_perf_event_paranoid),
1176 .mode = 0644,
1177 .proc_handler = proc_dointvec,
1178 },
1179 {
1180 .procname = "perf_event_mlock_kb",
1181 .data = &sysctl_perf_event_mlock,
1182 .maxlen = sizeof(sysctl_perf_event_mlock),
1183 .mode = 0644,
1184 .proc_handler = proc_dointvec,
1185 },
1186 {
1187 .procname = "perf_event_max_sample_rate",
1188 .data = &sysctl_perf_event_sample_rate,
1189 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1190 .mode = 0644,
1191 .proc_handler = perf_proc_update_handler,
1192 .extra1 = &one,
1193 },
1194 {
1195 .procname = "perf_cpu_time_max_percent",
1196 .data = &sysctl_perf_cpu_time_max_percent,
1197 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1198 .mode = 0644,
1199 .proc_handler = perf_cpu_time_max_percent_handler,
1200 .extra1 = &zero,
1201 .extra2 = &one_hundred,
1202 },
1203 {
1204 .procname = "perf_event_max_stack",
1205 .data = &sysctl_perf_event_max_stack,
1206 .maxlen = sizeof(sysctl_perf_event_max_stack),
1207 .mode = 0644,
1208 .proc_handler = perf_event_max_stack_handler,
1209 .extra1 = &zero,
1210 .extra2 = &six_hundred_forty_kb,
1211 },
1212 {
1213 .procname = "perf_event_max_contexts_per_stack",
1214 .data = &sysctl_perf_event_max_contexts_per_stack,
1215 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
1216 .mode = 0644,
1217 .proc_handler = perf_event_max_stack_handler,
1218 .extra1 = &zero,
1219 .extra2 = &one_thousand,
1220 },
1221 #endif
1222 {
1223 .procname = "panic_on_warn",
1224 .data = &panic_on_warn,
1225 .maxlen = sizeof(int),
1226 .mode = 0644,
1227 .proc_handler = proc_dointvec_minmax,
1228 .extra1 = &zero,
1229 .extra2 = &one,
1230 },
1231 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1232 {
1233 .procname = "timer_migration",
1234 .data = &sysctl_timer_migration,
1235 .maxlen = sizeof(unsigned int),
1236 .mode = 0644,
1237 .proc_handler = timer_migration_handler,
1238 .extra1 = &zero,
1239 .extra2 = &one,
1240 },
1241 #endif
1242 #ifdef CONFIG_BPF_SYSCALL
1243 {
1244 .procname = "unprivileged_bpf_disabled",
1245 .data = &sysctl_unprivileged_bpf_disabled,
1246 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
1247 .mode = 0644,
1248 /* only handle a transition from default "0" to "1" */
1249 .proc_handler = proc_dointvec_minmax,
1250 .extra1 = &one,
1251 .extra2 = &one,
1252 },
1253 {
1254 .procname = "bpf_stats_enabled",
1255 .data = &sysctl_bpf_stats_enabled,
1256 .maxlen = sizeof(sysctl_bpf_stats_enabled),
1257 .mode = 0644,
1258 .proc_handler = proc_dointvec_minmax_bpf_stats,
1259 .extra1 = &zero,
1260 .extra2 = &one,
1261 },
1262 #endif
1263 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1264 {
1265 .procname = "panic_on_rcu_stall",
1266 .data = &sysctl_panic_on_rcu_stall,
1267 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
1268 .mode = 0644,
1269 .proc_handler = proc_dointvec_minmax,
1270 .extra1 = &zero,
1271 .extra2 = &one,
1272 },
1273 #endif
1274 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
1275 {
1276 .procname = "stack_erasing",
1277 .data = NULL,
1278 .maxlen = sizeof(int),
1279 .mode = 0600,
1280 .proc_handler = stack_erasing_sysctl,
1281 .extra1 = &zero,
1282 .extra2 = &one,
1283 },
1284 #endif
1285 { }
1286 };
1287
1288 static struct ctl_table vm_table[] = {
1289 {
1290 .procname = "overcommit_memory",
1291 .data = &sysctl_overcommit_memory,
1292 .maxlen = sizeof(sysctl_overcommit_memory),
1293 .mode = 0644,
1294 .proc_handler = proc_dointvec_minmax,
1295 .extra1 = &zero,
1296 .extra2 = &two,
1297 },
1298 {
1299 .procname = "panic_on_oom",
1300 .data = &sysctl_panic_on_oom,
1301 .maxlen = sizeof(sysctl_panic_on_oom),
1302 .mode = 0644,
1303 .proc_handler = proc_dointvec_minmax,
1304 .extra1 = &zero,
1305 .extra2 = &two,
1306 },
1307 {
1308 .procname = "oom_kill_allocating_task",
1309 .data = &sysctl_oom_kill_allocating_task,
1310 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1311 .mode = 0644,
1312 .proc_handler = proc_dointvec,
1313 },
1314 {
1315 .procname = "oom_dump_tasks",
1316 .data = &sysctl_oom_dump_tasks,
1317 .maxlen = sizeof(sysctl_oom_dump_tasks),
1318 .mode = 0644,
1319 .proc_handler = proc_dointvec,
1320 },
1321 {
1322 .procname = "overcommit_ratio",
1323 .data = &sysctl_overcommit_ratio,
1324 .maxlen = sizeof(sysctl_overcommit_ratio),
1325 .mode = 0644,
1326 .proc_handler = overcommit_ratio_handler,
1327 },
1328 {
1329 .procname = "overcommit_kbytes",
1330 .data = &sysctl_overcommit_kbytes,
1331 .maxlen = sizeof(sysctl_overcommit_kbytes),
1332 .mode = 0644,
1333 .proc_handler = overcommit_kbytes_handler,
1334 },
1335 {
1336 .procname = "page-cluster",
1337 .data = &page_cluster,
1338 .maxlen = sizeof(int),
1339 .mode = 0644,
1340 .proc_handler = proc_dointvec_minmax,
1341 .extra1 = &zero,
1342 },
1343 {
1344 .procname = "dirty_background_ratio",
1345 .data = &dirty_background_ratio,
1346 .maxlen = sizeof(dirty_background_ratio),
1347 .mode = 0644,
1348 .proc_handler = dirty_background_ratio_handler,
1349 .extra1 = &zero,
1350 .extra2 = &one_hundred,
1351 },
1352 {
1353 .procname = "dirty_background_bytes",
1354 .data = &dirty_background_bytes,
1355 .maxlen = sizeof(dirty_background_bytes),
1356 .mode = 0644,
1357 .proc_handler = dirty_background_bytes_handler,
1358 .extra1 = &one_ul,
1359 },
1360 {
1361 .procname = "dirty_ratio",
1362 .data = &vm_dirty_ratio,
1363 .maxlen = sizeof(vm_dirty_ratio),
1364 .mode = 0644,
1365 .proc_handler = dirty_ratio_handler,
1366 .extra1 = &zero,
1367 .extra2 = &one_hundred,
1368 },
1369 {
1370 .procname = "dirty_bytes",
1371 .data = &vm_dirty_bytes,
1372 .maxlen = sizeof(vm_dirty_bytes),
1373 .mode = 0644,
1374 .proc_handler = dirty_bytes_handler,
1375 .extra1 = &dirty_bytes_min,
1376 },
1377 {
1378 .procname = "dirty_writeback_centisecs",
1379 .data = &dirty_writeback_interval,
1380 .maxlen = sizeof(dirty_writeback_interval),
1381 .mode = 0644,
1382 .proc_handler = dirty_writeback_centisecs_handler,
1383 },
1384 {
1385 .procname = "dirty_expire_centisecs",
1386 .data = &dirty_expire_interval,
1387 .maxlen = sizeof(dirty_expire_interval),
1388 .mode = 0644,
1389 .proc_handler = proc_dointvec_minmax,
1390 .extra1 = &zero,
1391 },
1392 {
1393 .procname = "dirtytime_expire_seconds",
1394 .data = &dirtytime_expire_interval,
1395 .maxlen = sizeof(dirtytime_expire_interval),
1396 .mode = 0644,
1397 .proc_handler = dirtytime_interval_handler,
1398 .extra1 = &zero,
1399 },
1400 {
1401 .procname = "swappiness",
1402 .data = &vm_swappiness,
1403 .maxlen = sizeof(vm_swappiness),
1404 .mode = 0644,
1405 .proc_handler = proc_dointvec_minmax,
1406 .extra1 = &zero,
1407 .extra2 = &one_hundred,
1408 },
1409 #ifdef CONFIG_HUGETLB_PAGE
1410 {
1411 .procname = "nr_hugepages",
1412 .data = NULL,
1413 .maxlen = sizeof(unsigned long),
1414 .mode = 0644,
1415 .proc_handler = hugetlb_sysctl_handler,
1416 },
1417 #ifdef CONFIG_NUMA
1418 {
1419 .procname = "nr_hugepages_mempolicy",
1420 .data = NULL,
1421 .maxlen = sizeof(unsigned long),
1422 .mode = 0644,
1423 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1424 },
1425 {
1426 .procname = "numa_stat",
1427 .data = &sysctl_vm_numa_stat,
1428 .maxlen = sizeof(int),
1429 .mode = 0644,
1430 .proc_handler = sysctl_vm_numa_stat_handler,
1431 .extra1 = &zero,
1432 .extra2 = &one,
1433 },
1434 #endif
1435 {
1436 .procname = "hugetlb_shm_group",
1437 .data = &sysctl_hugetlb_shm_group,
1438 .maxlen = sizeof(gid_t),
1439 .mode = 0644,
1440 .proc_handler = proc_dointvec,
1441 },
1442 {
1443 .procname = "nr_overcommit_hugepages",
1444 .data = NULL,
1445 .maxlen = sizeof(unsigned long),
1446 .mode = 0644,
1447 .proc_handler = hugetlb_overcommit_handler,
1448 },
1449 #endif
1450 {
1451 .procname = "lowmem_reserve_ratio",
1452 .data = &sysctl_lowmem_reserve_ratio,
1453 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1454 .mode = 0644,
1455 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1456 },
1457 {
1458 .procname = "drop_caches",
1459 .data = &sysctl_drop_caches,
1460 .maxlen = sizeof(int),
1461 .mode = 0644,
1462 .proc_handler = drop_caches_sysctl_handler,
1463 .extra1 = &one,
1464 .extra2 = &four,
1465 },
1466 #ifdef CONFIG_COMPACTION
1467 {
1468 .procname = "compact_memory",
1469 .data = &sysctl_compact_memory,
1470 .maxlen = sizeof(int),
1471 .mode = 0200,
1472 .proc_handler = sysctl_compaction_handler,
1473 },
1474 {
1475 .procname = "extfrag_threshold",
1476 .data = &sysctl_extfrag_threshold,
1477 .maxlen = sizeof(int),
1478 .mode = 0644,
1479 .proc_handler = proc_dointvec_minmax,
1480 .extra1 = &min_extfrag_threshold,
1481 .extra2 = &max_extfrag_threshold,
1482 },
1483 {
1484 .procname = "compact_unevictable_allowed",
1485 .data = &sysctl_compact_unevictable_allowed,
1486 .maxlen = sizeof(int),
1487 .mode = 0644,
1488 .proc_handler = proc_dointvec,
1489 .extra1 = &zero,
1490 .extra2 = &one,
1491 },
1492
1493 #endif /* CONFIG_COMPACTION */
1494 {
1495 .procname = "min_free_kbytes",
1496 .data = &min_free_kbytes,
1497 .maxlen = sizeof(min_free_kbytes),
1498 .mode = 0644,
1499 .proc_handler = min_free_kbytes_sysctl_handler,
1500 .extra1 = &zero,
1501 },
1502 {
1503 .procname = "watermark_boost_factor",
1504 .data = &watermark_boost_factor,
1505 .maxlen = sizeof(watermark_boost_factor),
1506 .mode = 0644,
1507 .proc_handler = watermark_boost_factor_sysctl_handler,
1508 .extra1 = &zero,
1509 },
1510 {
1511 .procname = "watermark_scale_factor",
1512 .data = &watermark_scale_factor,
1513 .maxlen = sizeof(watermark_scale_factor),
1514 .mode = 0644,
1515 .proc_handler = watermark_scale_factor_sysctl_handler,
1516 .extra1 = &one,
1517 .extra2 = &one_thousand,
1518 },
1519 {
1520 .procname = "percpu_pagelist_fraction",
1521 .data = &percpu_pagelist_fraction,
1522 .maxlen = sizeof(percpu_pagelist_fraction),
1523 .mode = 0644,
1524 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1525 .extra1 = &zero,
1526 },
1527 #ifdef CONFIG_MMU
1528 {
1529 .procname = "max_map_count",
1530 .data = &sysctl_max_map_count,
1531 .maxlen = sizeof(sysctl_max_map_count),
1532 .mode = 0644,
1533 .proc_handler = proc_dointvec_minmax,
1534 .extra1 = &zero,
1535 },
1536 #else
1537 {
1538 .procname = "nr_trim_pages",
1539 .data = &sysctl_nr_trim_pages,
1540 .maxlen = sizeof(sysctl_nr_trim_pages),
1541 .mode = 0644,
1542 .proc_handler = proc_dointvec_minmax,
1543 .extra1 = &zero,
1544 },
1545 #endif
1546 {
1547 .procname = "laptop_mode",
1548 .data = &laptop_mode,
1549 .maxlen = sizeof(laptop_mode),
1550 .mode = 0644,
1551 .proc_handler = proc_dointvec_jiffies,
1552 },
1553 {
1554 .procname = "block_dump",
1555 .data = &block_dump,
1556 .maxlen = sizeof(block_dump),
1557 .mode = 0644,
1558 .proc_handler = proc_dointvec,
1559 .extra1 = &zero,
1560 },
1561 {
1562 .procname = "vfs_cache_pressure",
1563 .data = &sysctl_vfs_cache_pressure,
1564 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1565 .mode = 0644,
1566 .proc_handler = proc_dointvec,
1567 .extra1 = &zero,
1568 },
1569 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1570 {
1571 .procname = "legacy_va_layout",
1572 .data = &sysctl_legacy_va_layout,
1573 .maxlen = sizeof(sysctl_legacy_va_layout),
1574 .mode = 0644,
1575 .proc_handler = proc_dointvec,
1576 .extra1 = &zero,
1577 },
1578 #endif
1579 #ifdef CONFIG_NUMA
1580 {
1581 .procname = "zone_reclaim_mode",
1582 .data = &node_reclaim_mode,
1583 .maxlen = sizeof(node_reclaim_mode),
1584 .mode = 0644,
1585 .proc_handler = proc_dointvec,
1586 .extra1 = &zero,
1587 },
1588 {
1589 .procname = "min_unmapped_ratio",
1590 .data = &sysctl_min_unmapped_ratio,
1591 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1592 .mode = 0644,
1593 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1594 .extra1 = &zero,
1595 .extra2 = &one_hundred,
1596 },
1597 {
1598 .procname = "min_slab_ratio",
1599 .data = &sysctl_min_slab_ratio,
1600 .maxlen = sizeof(sysctl_min_slab_ratio),
1601 .mode = 0644,
1602 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1603 .extra1 = &zero,
1604 .extra2 = &one_hundred,
1605 },
1606 #endif
1607 #ifdef CONFIG_SMP
1608 {
1609 .procname = "stat_interval",
1610 .data = &sysctl_stat_interval,
1611 .maxlen = sizeof(sysctl_stat_interval),
1612 .mode = 0644,
1613 .proc_handler = proc_dointvec_jiffies,
1614 },
1615 {
1616 .procname = "stat_refresh",
1617 .data = NULL,
1618 .maxlen = 0,
1619 .mode = 0600,
1620 .proc_handler = vmstat_refresh,
1621 },
1622 #endif
1623 #ifdef CONFIG_MMU
1624 {
1625 .procname = "mmap_min_addr",
1626 .data = &dac_mmap_min_addr,
1627 .maxlen = sizeof(unsigned long),
1628 .mode = 0644,
1629 .proc_handler = mmap_min_addr_handler,
1630 },
1631 #endif
1632 #ifdef CONFIG_NUMA
1633 {
1634 .procname = "numa_zonelist_order",
1635 .data = &numa_zonelist_order,
1636 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1637 .mode = 0644,
1638 .proc_handler = numa_zonelist_order_handler,
1639 },
1640 #endif
1641 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1642 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1643 {
1644 .procname = "vdso_enabled",
1645 #ifdef CONFIG_X86_32
1646 .data = &vdso32_enabled,
1647 .maxlen = sizeof(vdso32_enabled),
1648 #else
1649 .data = &vdso_enabled,
1650 .maxlen = sizeof(vdso_enabled),
1651 #endif
1652 .mode = 0644,
1653 .proc_handler = proc_dointvec,
1654 .extra1 = &zero,
1655 },
1656 #endif
1657 #ifdef CONFIG_HIGHMEM
1658 {
1659 .procname = "highmem_is_dirtyable",
1660 .data = &vm_highmem_is_dirtyable,
1661 .maxlen = sizeof(vm_highmem_is_dirtyable),
1662 .mode = 0644,
1663 .proc_handler = proc_dointvec_minmax,
1664 .extra1 = &zero,
1665 .extra2 = &one,
1666 },
1667 #endif
1668 #ifdef CONFIG_MEMORY_FAILURE
1669 {
1670 .procname = "memory_failure_early_kill",
1671 .data = &sysctl_memory_failure_early_kill,
1672 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1673 .mode = 0644,
1674 .proc_handler = proc_dointvec_minmax,
1675 .extra1 = &zero,
1676 .extra2 = &one,
1677 },
1678 {
1679 .procname = "memory_failure_recovery",
1680 .data = &sysctl_memory_failure_recovery,
1681 .maxlen = sizeof(sysctl_memory_failure_recovery),
1682 .mode = 0644,
1683 .proc_handler = proc_dointvec_minmax,
1684 .extra1 = &zero,
1685 .extra2 = &one,
1686 },
1687 #endif
1688 {
1689 .procname = "user_reserve_kbytes",
1690 .data = &sysctl_user_reserve_kbytes,
1691 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1692 .mode = 0644,
1693 .proc_handler = proc_doulongvec_minmax,
1694 },
1695 {
1696 .procname = "admin_reserve_kbytes",
1697 .data = &sysctl_admin_reserve_kbytes,
1698 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1699 .mode = 0644,
1700 .proc_handler = proc_doulongvec_minmax,
1701 },
1702 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1703 {
1704 .procname = "mmap_rnd_bits",
1705 .data = &mmap_rnd_bits,
1706 .maxlen = sizeof(mmap_rnd_bits),
1707 .mode = 0600,
1708 .proc_handler = proc_dointvec_minmax,
1709 .extra1 = (void *)&mmap_rnd_bits_min,
1710 .extra2 = (void *)&mmap_rnd_bits_max,
1711 },
1712 #endif
1713 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1714 {
1715 .procname = "mmap_rnd_compat_bits",
1716 .data = &mmap_rnd_compat_bits,
1717 .maxlen = sizeof(mmap_rnd_compat_bits),
1718 .mode = 0600,
1719 .proc_handler = proc_dointvec_minmax,
1720 .extra1 = (void *)&mmap_rnd_compat_bits_min,
1721 .extra2 = (void *)&mmap_rnd_compat_bits_max,
1722 },
1723 #endif
1724 #ifdef CONFIG_USERFAULTFD
1725 {
1726 .procname = "unprivileged_userfaultfd",
1727 .data = &sysctl_unprivileged_userfaultfd,
1728 .maxlen = sizeof(sysctl_unprivileged_userfaultfd),
1729 .mode = 0644,
1730 .proc_handler = proc_dointvec_minmax,
1731 .extra1 = &zero,
1732 .extra2 = &one,
1733 },
1734 #endif
1735 { }
1736 };
1737
1738 static struct ctl_table fs_table[] = {
1739 {
1740 .procname = "inode-nr",
1741 .data = &inodes_stat,
1742 .maxlen = 2*sizeof(long),
1743 .mode = 0444,
1744 .proc_handler = proc_nr_inodes,
1745 },
1746 {
1747 .procname = "inode-state",
1748 .data = &inodes_stat,
1749 .maxlen = 7*sizeof(long),
1750 .mode = 0444,
1751 .proc_handler = proc_nr_inodes,
1752 },
1753 {
1754 .procname = "file-nr",
1755 .data = &files_stat,
1756 .maxlen = sizeof(files_stat),
1757 .mode = 0444,
1758 .proc_handler = proc_nr_files,
1759 },
1760 {
1761 .procname = "file-max",
1762 .data = &files_stat.max_files,
1763 .maxlen = sizeof(files_stat.max_files),
1764 .mode = 0644,
1765 .proc_handler = proc_doulongvec_minmax,
1766 .extra1 = &zero_ul,
1767 .extra2 = &long_max,
1768 },
1769 {
1770 .procname = "nr_open",
1771 .data = &sysctl_nr_open,
1772 .maxlen = sizeof(unsigned int),
1773 .mode = 0644,
1774 .proc_handler = proc_dointvec_minmax,
1775 .extra1 = &sysctl_nr_open_min,
1776 .extra2 = &sysctl_nr_open_max,
1777 },
1778 {
1779 .procname = "dentry-state",
1780 .data = &dentry_stat,
1781 .maxlen = 6*sizeof(long),
1782 .mode = 0444,
1783 .proc_handler = proc_nr_dentry,
1784 },
1785 {
1786 .procname = "overflowuid",
1787 .data = &fs_overflowuid,
1788 .maxlen = sizeof(int),
1789 .mode = 0644,
1790 .proc_handler = proc_dointvec_minmax,
1791 .extra1 = &minolduid,
1792 .extra2 = &maxolduid,
1793 },
1794 {
1795 .procname = "overflowgid",
1796 .data = &fs_overflowgid,
1797 .maxlen = sizeof(int),
1798 .mode = 0644,
1799 .proc_handler = proc_dointvec_minmax,
1800 .extra1 = &minolduid,
1801 .extra2 = &maxolduid,
1802 },
1803 #ifdef CONFIG_FILE_LOCKING
1804 {
1805 .procname = "leases-enable",
1806 .data = &leases_enable,
1807 .maxlen = sizeof(int),
1808 .mode = 0644,
1809 .proc_handler = proc_dointvec,
1810 },
1811 #endif
1812 #ifdef CONFIG_DNOTIFY
1813 {
1814 .procname = "dir-notify-enable",
1815 .data = &dir_notify_enable,
1816 .maxlen = sizeof(int),
1817 .mode = 0644,
1818 .proc_handler = proc_dointvec,
1819 },
1820 #endif
1821 #ifdef CONFIG_MMU
1822 #ifdef CONFIG_FILE_LOCKING
1823 {
1824 .procname = "lease-break-time",
1825 .data = &lease_break_time,
1826 .maxlen = sizeof(int),
1827 .mode = 0644,
1828 .proc_handler = proc_dointvec,
1829 },
1830 #endif
1831 #ifdef CONFIG_AIO
1832 {
1833 .procname = "aio-nr",
1834 .data = &aio_nr,
1835 .maxlen = sizeof(aio_nr),
1836 .mode = 0444,
1837 .proc_handler = proc_doulongvec_minmax,
1838 },
1839 {
1840 .procname = "aio-max-nr",
1841 .data = &aio_max_nr,
1842 .maxlen = sizeof(aio_max_nr),
1843 .mode = 0644,
1844 .proc_handler = proc_doulongvec_minmax,
1845 },
1846 #endif /* CONFIG_AIO */
1847 #ifdef CONFIG_INOTIFY_USER
1848 {
1849 .procname = "inotify",
1850 .mode = 0555,
1851 .child = inotify_table,
1852 },
1853 #endif
1854 #ifdef CONFIG_EPOLL
1855 {
1856 .procname = "epoll",
1857 .mode = 0555,
1858 .child = epoll_table,
1859 },
1860 #endif
1861 #endif
1862 {
1863 .procname = "protected_symlinks",
1864 .data = &sysctl_protected_symlinks,
1865 .maxlen = sizeof(int),
1866 .mode = 0600,
1867 .proc_handler = proc_dointvec_minmax,
1868 .extra1 = &zero,
1869 .extra2 = &one,
1870 },
1871 {
1872 .procname = "protected_hardlinks",
1873 .data = &sysctl_protected_hardlinks,
1874 .maxlen = sizeof(int),
1875 .mode = 0600,
1876 .proc_handler = proc_dointvec_minmax,
1877 .extra1 = &zero,
1878 .extra2 = &one,
1879 },
1880 {
1881 .procname = "protected_fifos",
1882 .data = &sysctl_protected_fifos,
1883 .maxlen = sizeof(int),
1884 .mode = 0600,
1885 .proc_handler = proc_dointvec_minmax,
1886 .extra1 = &zero,
1887 .extra2 = &two,
1888 },
1889 {
1890 .procname = "protected_regular",
1891 .data = &sysctl_protected_regular,
1892 .maxlen = sizeof(int),
1893 .mode = 0600,
1894 .proc_handler = proc_dointvec_minmax,
1895 .extra1 = &zero,
1896 .extra2 = &two,
1897 },
1898 {
1899 .procname = "suid_dumpable",
1900 .data = &suid_dumpable,
1901 .maxlen = sizeof(int),
1902 .mode = 0644,
1903 .proc_handler = proc_dointvec_minmax_coredump,
1904 .extra1 = &zero,
1905 .extra2 = &two,
1906 },
1907 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1908 {
1909 .procname = "binfmt_misc",
1910 .mode = 0555,
1911 .child = sysctl_mount_point,
1912 },
1913 #endif
1914 {
1915 .procname = "pipe-max-size",
1916 .data = &pipe_max_size,
1917 .maxlen = sizeof(pipe_max_size),
1918 .mode = 0644,
1919 .proc_handler = proc_dopipe_max_size,
1920 },
1921 {
1922 .procname = "pipe-user-pages-hard",
1923 .data = &pipe_user_pages_hard,
1924 .maxlen = sizeof(pipe_user_pages_hard),
1925 .mode = 0644,
1926 .proc_handler = proc_doulongvec_minmax,
1927 },
1928 {
1929 .procname = "pipe-user-pages-soft",
1930 .data = &pipe_user_pages_soft,
1931 .maxlen = sizeof(pipe_user_pages_soft),
1932 .mode = 0644,
1933 .proc_handler = proc_doulongvec_minmax,
1934 },
1935 {
1936 .procname = "mount-max",
1937 .data = &sysctl_mount_max,
1938 .maxlen = sizeof(unsigned int),
1939 .mode = 0644,
1940 .proc_handler = proc_dointvec_minmax,
1941 .extra1 = &one,
1942 },
1943 { }
1944 };
1945
1946 static struct ctl_table debug_table[] = {
1947 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1948 {
1949 .procname = "exception-trace",
1950 .data = &show_unhandled_signals,
1951 .maxlen = sizeof(int),
1952 .mode = 0644,
1953 .proc_handler = proc_dointvec
1954 },
1955 #endif
1956 #if defined(CONFIG_OPTPROBES)
1957 {
1958 .procname = "kprobes-optimization",
1959 .data = &sysctl_kprobes_optimization,
1960 .maxlen = sizeof(int),
1961 .mode = 0644,
1962 .proc_handler = proc_kprobes_optimization_handler,
1963 .extra1 = &zero,
1964 .extra2 = &one,
1965 },
1966 #endif
1967 { }
1968 };
1969
1970 static struct ctl_table dev_table[] = {
1971 { }
1972 };
1973
1974 int __init sysctl_init(void)
1975 {
1976 struct ctl_table_header *hdr;
1977
1978 hdr = register_sysctl_table(sysctl_base_table);
1979 kmemleak_not_leak(hdr);
1980 return 0;
1981 }
1982
1983 #endif /* CONFIG_SYSCTL */
1984
1985 /*
1986 * /proc/sys support
1987 */
1988
1989 #ifdef CONFIG_PROC_SYSCTL
1990
1991 static int _proc_do_string(char *data, int maxlen, int write,
1992 char __user *buffer,
1993 size_t *lenp, loff_t *ppos)
1994 {
1995 size_t len;
1996 char __user *p;
1997 char c;
1998
1999 if (!data || !maxlen || !*lenp) {
2000 *lenp = 0;
2001 return 0;
2002 }
2003
2004 if (write) {
2005 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
2006 /* Only continue writes not past the end of buffer. */
2007 len = strlen(data);
2008 if (len > maxlen - 1)
2009 len = maxlen - 1;
2010
2011 if (*ppos > len)
2012 return 0;
2013 len = *ppos;
2014 } else {
2015 /* Start writing from beginning of buffer. */
2016 len = 0;
2017 }
2018
2019 *ppos += *lenp;
2020 p = buffer;
2021 while ((p - buffer) < *lenp && len < maxlen - 1) {
2022 if (get_user(c, p++))
2023 return -EFAULT;
2024 if (c == 0 || c == '\n')
2025 break;
2026 data[len++] = c;
2027 }
2028 data[len] = 0;
2029 } else {
2030 len = strlen(data);
2031 if (len > maxlen)
2032 len = maxlen;
2033
2034 if (*ppos > len) {
2035 *lenp = 0;
2036 return 0;
2037 }
2038
2039 data += *ppos;
2040 len -= *ppos;
2041
2042 if (len > *lenp)
2043 len = *lenp;
2044 if (len)
2045 if (copy_to_user(buffer, data, len))
2046 return -EFAULT;
2047 if (len < *lenp) {
2048 if (put_user('\n', buffer + len))
2049 return -EFAULT;
2050 len++;
2051 }
2052 *lenp = len;
2053 *ppos += len;
2054 }
2055 return 0;
2056 }
2057
2058 static void warn_sysctl_write(struct ctl_table *table)
2059 {
2060 pr_warn_once("%s wrote to %s when file position was not 0!\n"
2061 "This will not be supported in the future. To silence this\n"
2062 "warning, set kernel.sysctl_writes_strict = -1\n",
2063 current->comm, table->procname);
2064 }
2065
2066 /**
2067 * proc_first_pos_non_zero_ignore - check if first position is allowed
2068 * @ppos: file position
2069 * @table: the sysctl table
2070 *
2071 * Returns true if the first position is non-zero and the sysctl_writes_strict
2072 * mode indicates this is not allowed for numeric input types. String proc
2073 * handlers can ignore the return value.
2074 */
2075 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
2076 struct ctl_table *table)
2077 {
2078 if (!*ppos)
2079 return false;
2080
2081 switch (sysctl_writes_strict) {
2082 case SYSCTL_WRITES_STRICT:
2083 return true;
2084 case SYSCTL_WRITES_WARN:
2085 warn_sysctl_write(table);
2086 return false;
2087 default:
2088 return false;
2089 }
2090 }
2091
2092 /**
2093 * proc_dostring - read a string sysctl
2094 * @table: the sysctl table
2095 * @write: %TRUE if this is a write to the sysctl file
2096 * @buffer: the user buffer
2097 * @lenp: the size of the user buffer
2098 * @ppos: file position
2099 *
2100 * Reads/writes a string from/to the user buffer. If the kernel
2101 * buffer provided is not large enough to hold the string, the
2102 * string is truncated. The copied string is %NULL-terminated.
2103 * If the string is being read by the user process, it is copied
2104 * and a newline '\n' is added. It is truncated if the buffer is
2105 * not large enough.
2106 *
2107 * Returns 0 on success.
2108 */
2109 int proc_dostring(struct ctl_table *table, int write,
2110 void __user *buffer, size_t *lenp, loff_t *ppos)
2111 {
2112 if (write)
2113 proc_first_pos_non_zero_ignore(ppos, table);
2114
2115 return _proc_do_string((char *)(table->data), table->maxlen, write,
2116 (char __user *)buffer, lenp, ppos);
2117 }
2118
2119 static size_t proc_skip_spaces(char **buf)
2120 {
2121 size_t ret;
2122 char *tmp = skip_spaces(*buf);
2123 ret = tmp - *buf;
2124 *buf = tmp;
2125 return ret;
2126 }
2127
2128 static void proc_skip_char(char **buf, size_t *size, const char v)
2129 {
2130 while (*size) {
2131 if (**buf != v)
2132 break;
2133 (*size)--;
2134 (*buf)++;
2135 }
2136 }
2137
2138 /**
2139 * strtoul_lenient - parse an ASCII formatted integer from a buffer and only
2140 * fail on overflow
2141 *
2142 * @cp: kernel buffer containing the string to parse
2143 * @endp: pointer to store the trailing characters
2144 * @base: the base to use
2145 * @res: where the parsed integer will be stored
2146 *
2147 * In case of success 0 is returned and @res will contain the parsed integer,
2148 * @endp will hold any trailing characters.
2149 * This function will fail the parse on overflow. If there wasn't an overflow
2150 * the function will defer the decision what characters count as invalid to the
2151 * caller.
2152 */
2153 static int strtoul_lenient(const char *cp, char **endp, unsigned int base,
2154 unsigned long *res)
2155 {
2156 unsigned long long result;
2157 unsigned int rv;
2158
2159 cp = _parse_integer_fixup_radix(cp, &base);
2160 rv = _parse_integer(cp, base, &result);
2161 if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result))
2162 return -ERANGE;
2163
2164 cp += rv;
2165
2166 if (endp)
2167 *endp = (char *)cp;
2168
2169 *res = (unsigned long)result;
2170 return 0;
2171 }
2172
2173 #define TMPBUFLEN 22
2174 /**
2175 * proc_get_long - reads an ASCII formatted integer from a user buffer
2176 *
2177 * @buf: a kernel buffer
2178 * @size: size of the kernel buffer
2179 * @val: this is where the number will be stored
2180 * @neg: set to %TRUE if number is negative
2181 * @perm_tr: a vector which contains the allowed trailers
2182 * @perm_tr_len: size of the perm_tr vector
2183 * @tr: pointer to store the trailer character
2184 *
2185 * In case of success %0 is returned and @buf and @size are updated with
2186 * the amount of bytes read. If @tr is non-NULL and a trailing
2187 * character exists (size is non-zero after returning from this
2188 * function), @tr is updated with the trailing character.
2189 */
2190 static int proc_get_long(char **buf, size_t *size,
2191 unsigned long *val, bool *neg,
2192 const char *perm_tr, unsigned perm_tr_len, char *tr)
2193 {
2194 int len;
2195 char *p, tmp[TMPBUFLEN];
2196
2197 if (!*size)
2198 return -EINVAL;
2199
2200 len = *size;
2201 if (len > TMPBUFLEN - 1)
2202 len = TMPBUFLEN - 1;
2203
2204 memcpy(tmp, *buf, len);
2205
2206 tmp[len] = 0;
2207 p = tmp;
2208 if (*p == '-' && *size > 1) {
2209 *neg = true;
2210 p++;
2211 } else
2212 *neg = false;
2213 if (!isdigit(*p))
2214 return -EINVAL;
2215
2216 if (strtoul_lenient(p, &p, 0, val))
2217 return -EINVAL;
2218
2219 len = p - tmp;
2220
2221 /* We don't know if the next char is whitespace thus we may accept
2222 * invalid integers (e.g. 1234...a) or two integers instead of one
2223 * (e.g. 123...1). So lets not allow such large numbers. */
2224 if (len == TMPBUFLEN - 1)
2225 return -EINVAL;
2226
2227 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2228 return -EINVAL;
2229
2230 if (tr && (len < *size))
2231 *tr = *p;
2232
2233 *buf += len;
2234 *size -= len;
2235
2236 return 0;
2237 }
2238
2239 /**
2240 * proc_put_long - converts an integer to a decimal ASCII formatted string
2241 *
2242 * @buf: the user buffer
2243 * @size: the size of the user buffer
2244 * @val: the integer to be converted
2245 * @neg: sign of the number, %TRUE for negative
2246 *
2247 * In case of success %0 is returned and @buf and @size are updated with
2248 * the amount of bytes written.
2249 */
2250 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2251 bool neg)
2252 {
2253 int len;
2254 char tmp[TMPBUFLEN], *p = tmp;
2255
2256 sprintf(p, "%s%lu", neg ? "-" : "", val);
2257 len = strlen(tmp);
2258 if (len > *size)
2259 len = *size;
2260 if (copy_to_user(*buf, tmp, len))
2261 return -EFAULT;
2262 *size -= len;
2263 *buf += len;
2264 return 0;
2265 }
2266 #undef TMPBUFLEN
2267
2268 static int proc_put_char(void __user **buf, size_t *size, char c)
2269 {
2270 if (*size) {
2271 char __user **buffer = (char __user **)buf;
2272 if (put_user(c, *buffer))
2273 return -EFAULT;
2274 (*size)--, (*buffer)++;
2275 *buf = *buffer;
2276 }
2277 return 0;
2278 }
2279
2280 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2281 int *valp,
2282 int write, void *data)
2283 {
2284 if (write) {
2285 if (*negp) {
2286 if (*lvalp > (unsigned long) INT_MAX + 1)
2287 return -EINVAL;
2288 *valp = -*lvalp;
2289 } else {
2290 if (*lvalp > (unsigned long) INT_MAX)
2291 return -EINVAL;
2292 *valp = *lvalp;
2293 }
2294 } else {
2295 int val = *valp;
2296 if (val < 0) {
2297 *negp = true;
2298 *lvalp = -(unsigned long)val;
2299 } else {
2300 *negp = false;
2301 *lvalp = (unsigned long)val;
2302 }
2303 }
2304 return 0;
2305 }
2306
2307 static int do_proc_douintvec_conv(unsigned long *lvalp,
2308 unsigned int *valp,
2309 int write, void *data)
2310 {
2311 if (write) {
2312 if (*lvalp > UINT_MAX)
2313 return -EINVAL;
2314 *valp = *lvalp;
2315 } else {
2316 unsigned int val = *valp;
2317 *lvalp = (unsigned long)val;
2318 }
2319 return 0;
2320 }
2321
2322 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2323
2324 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2325 int write, void __user *buffer,
2326 size_t *lenp, loff_t *ppos,
2327 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2328 int write, void *data),
2329 void *data)
2330 {
2331 int *i, vleft, first = 1, err = 0;
2332 size_t left;
2333 char *kbuf = NULL, *p;
2334
2335 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2336 *lenp = 0;
2337 return 0;
2338 }
2339
2340 i = (int *) tbl_data;
2341 vleft = table->maxlen / sizeof(*i);
2342 left = *lenp;
2343
2344 if (!conv)
2345 conv = do_proc_dointvec_conv;
2346
2347 if (write) {
2348 if (proc_first_pos_non_zero_ignore(ppos, table))
2349 goto out;
2350
2351 if (left > PAGE_SIZE - 1)
2352 left = PAGE_SIZE - 1;
2353 p = kbuf = memdup_user_nul(buffer, left);
2354 if (IS_ERR(kbuf))
2355 return PTR_ERR(kbuf);
2356 }
2357
2358 for (; left && vleft--; i++, first=0) {
2359 unsigned long lval;
2360 bool neg;
2361
2362 if (write) {
2363 left -= proc_skip_spaces(&p);
2364
2365 if (!left)
2366 break;
2367 err = proc_get_long(&p, &left, &lval, &neg,
2368 proc_wspace_sep,
2369 sizeof(proc_wspace_sep), NULL);
2370 if (err)
2371 break;
2372 if (conv(&neg, &lval, i, 1, data)) {
2373 err = -EINVAL;
2374 break;
2375 }
2376 } else {
2377 if (conv(&neg, &lval, i, 0, data)) {
2378 err = -EINVAL;
2379 break;
2380 }
2381 if (!first)
2382 err = proc_put_char(&buffer, &left, '\t');
2383 if (err)
2384 break;
2385 err = proc_put_long(&buffer, &left, lval, neg);
2386 if (err)
2387 break;
2388 }
2389 }
2390
2391 if (!write && !first && left && !err)
2392 err = proc_put_char(&buffer, &left, '\n');
2393 if (write && !err && left)
2394 left -= proc_skip_spaces(&p);
2395 if (write) {
2396 kfree(kbuf);
2397 if (first)
2398 return err ? : -EINVAL;
2399 }
2400 *lenp -= left;
2401 out:
2402 *ppos += *lenp;
2403 return err;
2404 }
2405
2406 static int do_proc_dointvec(struct ctl_table *table, int write,
2407 void __user *buffer, size_t *lenp, loff_t *ppos,
2408 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2409 int write, void *data),
2410 void *data)
2411 {
2412 return __do_proc_dointvec(table->data, table, write,
2413 buffer, lenp, ppos, conv, data);
2414 }
2415
2416 static int do_proc_douintvec_w(unsigned int *tbl_data,
2417 struct ctl_table *table,
2418 void __user *buffer,
2419 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 unsigned long lval;
2426 int err = 0;
2427 size_t left;
2428 bool neg;
2429 char *kbuf = NULL, *p;
2430
2431 left = *lenp;
2432
2433 if (proc_first_pos_non_zero_ignore(ppos, table))
2434 goto bail_early;
2435
2436 if (left > PAGE_SIZE - 1)
2437 left = PAGE_SIZE - 1;
2438
2439 p = kbuf = memdup_user_nul(buffer, left);
2440 if (IS_ERR(kbuf))
2441 return -EINVAL;
2442
2443 left -= proc_skip_spaces(&p);
2444 if (!left) {
2445 err = -EINVAL;
2446 goto out_free;
2447 }
2448
2449 err = proc_get_long(&p, &left, &lval, &neg,
2450 proc_wspace_sep,
2451 sizeof(proc_wspace_sep), NULL);
2452 if (err || neg) {
2453 err = -EINVAL;
2454 goto out_free;
2455 }
2456
2457 if (conv(&lval, tbl_data, 1, data)) {
2458 err = -EINVAL;
2459 goto out_free;
2460 }
2461
2462 if (!err && left)
2463 left -= proc_skip_spaces(&p);
2464
2465 out_free:
2466 kfree(kbuf);
2467 if (err)
2468 return -EINVAL;
2469
2470 return 0;
2471
2472 /* This is in keeping with old __do_proc_dointvec() */
2473 bail_early:
2474 *ppos += *lenp;
2475 return err;
2476 }
2477
2478 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2479 size_t *lenp, loff_t *ppos,
2480 int (*conv)(unsigned long *lvalp,
2481 unsigned int *valp,
2482 int write, void *data),
2483 void *data)
2484 {
2485 unsigned long lval;
2486 int err = 0;
2487 size_t left;
2488
2489 left = *lenp;
2490
2491 if (conv(&lval, tbl_data, 0, data)) {
2492 err = -EINVAL;
2493 goto out;
2494 }
2495
2496 err = proc_put_long(&buffer, &left, lval, false);
2497 if (err || !left)
2498 goto out;
2499
2500 err = proc_put_char(&buffer, &left, '\n');
2501
2502 out:
2503 *lenp -= left;
2504 *ppos += *lenp;
2505
2506 return err;
2507 }
2508
2509 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2510 int write, void __user *buffer,
2511 size_t *lenp, loff_t *ppos,
2512 int (*conv)(unsigned long *lvalp,
2513 unsigned int *valp,
2514 int write, void *data),
2515 void *data)
2516 {
2517 unsigned int *i, vleft;
2518
2519 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2520 *lenp = 0;
2521 return 0;
2522 }
2523
2524 i = (unsigned int *) tbl_data;
2525 vleft = table->maxlen / sizeof(*i);
2526
2527 /*
2528 * Arrays are not supported, keep this simple. *Do not* add
2529 * support for them.
2530 */
2531 if (vleft != 1) {
2532 *lenp = 0;
2533 return -EINVAL;
2534 }
2535
2536 if (!conv)
2537 conv = do_proc_douintvec_conv;
2538
2539 if (write)
2540 return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2541 conv, data);
2542 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2543 }
2544
2545 static int do_proc_douintvec(struct ctl_table *table, int write,
2546 void __user *buffer, size_t *lenp, loff_t *ppos,
2547 int (*conv)(unsigned long *lvalp,
2548 unsigned int *valp,
2549 int write, void *data),
2550 void *data)
2551 {
2552 return __do_proc_douintvec(table->data, table, write,
2553 buffer, lenp, ppos, conv, data);
2554 }
2555
2556 /**
2557 * proc_dointvec - read a vector of integers
2558 * @table: the sysctl table
2559 * @write: %TRUE if this is a write to the sysctl file
2560 * @buffer: the user buffer
2561 * @lenp: the size of the user buffer
2562 * @ppos: file position
2563 *
2564 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2565 * values from/to the user buffer, treated as an ASCII string.
2566 *
2567 * Returns 0 on success.
2568 */
2569 int proc_dointvec(struct ctl_table *table, int write,
2570 void __user *buffer, size_t *lenp, loff_t *ppos)
2571 {
2572 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2573 }
2574
2575 /**
2576 * proc_douintvec - read a vector of unsigned integers
2577 * @table: the sysctl table
2578 * @write: %TRUE if this is a write to the sysctl file
2579 * @buffer: the user buffer
2580 * @lenp: the size of the user buffer
2581 * @ppos: file position
2582 *
2583 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2584 * values from/to the user buffer, treated as an ASCII string.
2585 *
2586 * Returns 0 on success.
2587 */
2588 int proc_douintvec(struct ctl_table *table, int write,
2589 void __user *buffer, size_t *lenp, loff_t *ppos)
2590 {
2591 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2592 do_proc_douintvec_conv, NULL);
2593 }
2594
2595 /*
2596 * Taint values can only be increased
2597 * This means we can safely use a temporary.
2598 */
2599 static int proc_taint(struct ctl_table *table, int write,
2600 void __user *buffer, size_t *lenp, loff_t *ppos)
2601 {
2602 struct ctl_table t;
2603 unsigned long tmptaint = get_taint();
2604 int err;
2605
2606 if (write && !capable(CAP_SYS_ADMIN))
2607 return -EPERM;
2608
2609 t = *table;
2610 t.data = &tmptaint;
2611 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2612 if (err < 0)
2613 return err;
2614
2615 if (write) {
2616 /*
2617 * Poor man's atomic or. Not worth adding a primitive
2618 * to everyone's atomic.h for this
2619 */
2620 int i;
2621 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2622 if ((tmptaint >> i) & 1)
2623 add_taint(i, LOCKDEP_STILL_OK);
2624 }
2625 }
2626
2627 return err;
2628 }
2629
2630 #ifdef CONFIG_PRINTK
2631 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2632 void __user *buffer, size_t *lenp, loff_t *ppos)
2633 {
2634 if (write && !capable(CAP_SYS_ADMIN))
2635 return -EPERM;
2636
2637 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2638 }
2639 #endif
2640
2641 /**
2642 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
2643 * @min: pointer to minimum allowable value
2644 * @max: pointer to maximum allowable value
2645 *
2646 * The do_proc_dointvec_minmax_conv_param structure provides the
2647 * minimum and maximum values for doing range checking for those sysctl
2648 * parameters that use the proc_dointvec_minmax() handler.
2649 */
2650 struct do_proc_dointvec_minmax_conv_param {
2651 int *min;
2652 int *max;
2653 };
2654
2655 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2656 int *valp,
2657 int write, void *data)
2658 {
2659 int tmp, ret;
2660 struct do_proc_dointvec_minmax_conv_param *param = data;
2661 /*
2662 * If writing, first do so via a temporary local int so we can
2663 * bounds-check it before touching *valp.
2664 */
2665 int *ip = write ? &tmp : valp;
2666
2667 ret = do_proc_dointvec_conv(negp, lvalp, ip, write, data);
2668 if (ret)
2669 return ret;
2670
2671 if (write) {
2672 if ((param->min && *param->min > tmp) ||
2673 (param->max && *param->max < tmp))
2674 return -EINVAL;
2675 *valp = tmp;
2676 }
2677
2678 return 0;
2679 }
2680
2681 /**
2682 * proc_dointvec_minmax - read a vector of integers with min/max values
2683 * @table: the sysctl table
2684 * @write: %TRUE if this is a write to the sysctl file
2685 * @buffer: the user buffer
2686 * @lenp: the size of the user buffer
2687 * @ppos: file position
2688 *
2689 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2690 * values from/to the user buffer, treated as an ASCII string.
2691 *
2692 * This routine will ensure the values are within the range specified by
2693 * table->extra1 (min) and table->extra2 (max).
2694 *
2695 * Returns 0 on success or -EINVAL on write when the range check fails.
2696 */
2697 int proc_dointvec_minmax(struct ctl_table *table, int write,
2698 void __user *buffer, size_t *lenp, loff_t *ppos)
2699 {
2700 struct do_proc_dointvec_minmax_conv_param param = {
2701 .min = (int *) table->extra1,
2702 .max = (int *) table->extra2,
2703 };
2704 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2705 do_proc_dointvec_minmax_conv, &param);
2706 }
2707
2708 /**
2709 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
2710 * @min: pointer to minimum allowable value
2711 * @max: pointer to maximum allowable value
2712 *
2713 * The do_proc_douintvec_minmax_conv_param structure provides the
2714 * minimum and maximum values for doing range checking for those sysctl
2715 * parameters that use the proc_douintvec_minmax() handler.
2716 */
2717 struct do_proc_douintvec_minmax_conv_param {
2718 unsigned int *min;
2719 unsigned int *max;
2720 };
2721
2722 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2723 unsigned int *valp,
2724 int write, void *data)
2725 {
2726 int ret;
2727 unsigned int tmp;
2728 struct do_proc_douintvec_minmax_conv_param *param = data;
2729 /* write via temporary local uint for bounds-checking */
2730 unsigned int *up = write ? &tmp : valp;
2731
2732 ret = do_proc_douintvec_conv(lvalp, up, write, data);
2733 if (ret)
2734 return ret;
2735
2736 if (write) {
2737 if ((param->min && *param->min > tmp) ||
2738 (param->max && *param->max < tmp))
2739 return -ERANGE;
2740
2741 *valp = tmp;
2742 }
2743
2744 return 0;
2745 }
2746
2747 /**
2748 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2749 * @table: the sysctl table
2750 * @write: %TRUE if this is a write to the sysctl file
2751 * @buffer: the user buffer
2752 * @lenp: the size of the user buffer
2753 * @ppos: file position
2754 *
2755 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2756 * values from/to the user buffer, treated as an ASCII string. Negative
2757 * strings are not allowed.
2758 *
2759 * This routine will ensure the values are within the range specified by
2760 * table->extra1 (min) and table->extra2 (max). There is a final sanity
2761 * check for UINT_MAX to avoid having to support wrap around uses from
2762 * userspace.
2763 *
2764 * Returns 0 on success or -ERANGE on write when the range check fails.
2765 */
2766 int proc_douintvec_minmax(struct ctl_table *table, int write,
2767 void __user *buffer, size_t *lenp, loff_t *ppos)
2768 {
2769 struct do_proc_douintvec_minmax_conv_param param = {
2770 .min = (unsigned int *) table->extra1,
2771 .max = (unsigned int *) table->extra2,
2772 };
2773 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2774 do_proc_douintvec_minmax_conv, &param);
2775 }
2776
2777 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
2778 unsigned int *valp,
2779 int write, void *data)
2780 {
2781 if (write) {
2782 unsigned int val;
2783
2784 val = round_pipe_size(*lvalp);
2785 if (val == 0)
2786 return -EINVAL;
2787
2788 *valp = val;
2789 } else {
2790 unsigned int val = *valp;
2791 *lvalp = (unsigned long) val;
2792 }
2793
2794 return 0;
2795 }
2796
2797 static int proc_dopipe_max_size(struct ctl_table *table, int write,
2798 void __user *buffer, size_t *lenp, loff_t *ppos)
2799 {
2800 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2801 do_proc_dopipe_max_size_conv, NULL);
2802 }
2803
2804 static void validate_coredump_safety(void)
2805 {
2806 #ifdef CONFIG_COREDUMP
2807 if (suid_dumpable == SUID_DUMP_ROOT &&
2808 core_pattern[0] != '/' && core_pattern[0] != '|') {
2809 printk(KERN_WARNING
2810 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2811 "Pipe handler or fully qualified core dump path required.\n"
2812 "Set kernel.core_pattern before fs.suid_dumpable.\n"
2813 );
2814 }
2815 #endif
2816 }
2817
2818 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2819 void __user *buffer, size_t *lenp, loff_t *ppos)
2820 {
2821 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2822 if (!error)
2823 validate_coredump_safety();
2824 return error;
2825 }
2826
2827 #ifdef CONFIG_COREDUMP
2828 static int proc_dostring_coredump(struct ctl_table *table, int write,
2829 void __user *buffer, size_t *lenp, loff_t *ppos)
2830 {
2831 int error = proc_dostring(table, write, buffer, lenp, ppos);
2832 if (!error)
2833 validate_coredump_safety();
2834 return error;
2835 }
2836 #endif
2837
2838 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2839 void __user *buffer,
2840 size_t *lenp, loff_t *ppos,
2841 unsigned long convmul,
2842 unsigned long convdiv)
2843 {
2844 unsigned long *i, *min, *max;
2845 int vleft, first = 1, err = 0;
2846 size_t left;
2847 char *kbuf = NULL, *p;
2848
2849 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2850 *lenp = 0;
2851 return 0;
2852 }
2853
2854 i = (unsigned long *) data;
2855 min = (unsigned long *) table->extra1;
2856 max = (unsigned long *) table->extra2;
2857 vleft = table->maxlen / sizeof(unsigned long);
2858 left = *lenp;
2859
2860 if (write) {
2861 if (proc_first_pos_non_zero_ignore(ppos, table))
2862 goto out;
2863
2864 if (left > PAGE_SIZE - 1)
2865 left = PAGE_SIZE - 1;
2866 p = kbuf = memdup_user_nul(buffer, left);
2867 if (IS_ERR(kbuf))
2868 return PTR_ERR(kbuf);
2869 }
2870
2871 for (; left && vleft--; i++, first = 0) {
2872 unsigned long val;
2873
2874 if (write) {
2875 bool neg;
2876
2877 left -= proc_skip_spaces(&p);
2878 if (!left)
2879 break;
2880
2881 err = proc_get_long(&p, &left, &val, &neg,
2882 proc_wspace_sep,
2883 sizeof(proc_wspace_sep), NULL);
2884 if (err)
2885 break;
2886 if (neg)
2887 continue;
2888 val = convmul * val / convdiv;
2889 if ((min && val < *min) || (max && val > *max)) {
2890 err = -EINVAL;
2891 break;
2892 }
2893 *i = val;
2894 } else {
2895 val = convdiv * (*i) / convmul;
2896 if (!first) {
2897 err = proc_put_char(&buffer, &left, '\t');
2898 if (err)
2899 break;
2900 }
2901 err = proc_put_long(&buffer, &left, val, false);
2902 if (err)
2903 break;
2904 }
2905 }
2906
2907 if (!write && !first && left && !err)
2908 err = proc_put_char(&buffer, &left, '\n');
2909 if (write && !err)
2910 left -= proc_skip_spaces(&p);
2911 if (write) {
2912 kfree(kbuf);
2913 if (first)
2914 return err ? : -EINVAL;
2915 }
2916 *lenp -= left;
2917 out:
2918 *ppos += *lenp;
2919 return err;
2920 }
2921
2922 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2923 void __user *buffer,
2924 size_t *lenp, loff_t *ppos,
2925 unsigned long convmul,
2926 unsigned long convdiv)
2927 {
2928 return __do_proc_doulongvec_minmax(table->data, table, write,
2929 buffer, lenp, ppos, convmul, convdiv);
2930 }
2931
2932 /**
2933 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2934 * @table: the sysctl table
2935 * @write: %TRUE if this is a write to the sysctl file
2936 * @buffer: the user buffer
2937 * @lenp: the size of the user buffer
2938 * @ppos: file position
2939 *
2940 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2941 * values from/to the user buffer, treated as an ASCII string.
2942 *
2943 * This routine will ensure the values are within the range specified by
2944 * table->extra1 (min) and table->extra2 (max).
2945 *
2946 * Returns 0 on success.
2947 */
2948 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2949 void __user *buffer, size_t *lenp, loff_t *ppos)
2950 {
2951 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2952 }
2953
2954 /**
2955 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2956 * @table: the sysctl table
2957 * @write: %TRUE if this is a write to the sysctl file
2958 * @buffer: the user buffer
2959 * @lenp: the size of the user buffer
2960 * @ppos: file position
2961 *
2962 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2963 * values from/to the user buffer, treated as an ASCII string. The values
2964 * are treated as milliseconds, and converted to jiffies when they are stored.
2965 *
2966 * This routine will ensure the values are within the range specified by
2967 * table->extra1 (min) and table->extra2 (max).
2968 *
2969 * Returns 0 on success.
2970 */
2971 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2972 void __user *buffer,
2973 size_t *lenp, loff_t *ppos)
2974 {
2975 return do_proc_doulongvec_minmax(table, write, buffer,
2976 lenp, ppos, HZ, 1000l);
2977 }
2978
2979
2980 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2981 int *valp,
2982 int write, void *data)
2983 {
2984 if (write) {
2985 if (*lvalp > INT_MAX / HZ)
2986 return 1;
2987 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2988 } else {
2989 int val = *valp;
2990 unsigned long lval;
2991 if (val < 0) {
2992 *negp = true;
2993 lval = -(unsigned long)val;
2994 } else {
2995 *negp = false;
2996 lval = (unsigned long)val;
2997 }
2998 *lvalp = lval / HZ;
2999 }
3000 return 0;
3001 }
3002
3003 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
3004 int *valp,
3005 int write, void *data)
3006 {
3007 if (write) {
3008 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
3009 return 1;
3010 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
3011 } else {
3012 int val = *valp;
3013 unsigned long lval;
3014 if (val < 0) {
3015 *negp = true;
3016 lval = -(unsigned long)val;
3017 } else {
3018 *negp = false;
3019 lval = (unsigned long)val;
3020 }
3021 *lvalp = jiffies_to_clock_t(lval);
3022 }
3023 return 0;
3024 }
3025
3026 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
3027 int *valp,
3028 int write, void *data)
3029 {
3030 if (write) {
3031 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
3032
3033 if (jif > INT_MAX)
3034 return 1;
3035 *valp = (int)jif;
3036 } else {
3037 int val = *valp;
3038 unsigned long lval;
3039 if (val < 0) {
3040 *negp = true;
3041 lval = -(unsigned long)val;
3042 } else {
3043 *negp = false;
3044 lval = (unsigned long)val;
3045 }
3046 *lvalp = jiffies_to_msecs(lval);
3047 }
3048 return 0;
3049 }
3050
3051 /**
3052 * proc_dointvec_jiffies - read a vector of integers as seconds
3053 * @table: the sysctl table
3054 * @write: %TRUE if this is a write to the sysctl file
3055 * @buffer: the user buffer
3056 * @lenp: the size of the user buffer
3057 * @ppos: file position
3058 *
3059 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3060 * values from/to the user buffer, treated as an ASCII string.
3061 * The values read are assumed to be in seconds, and are converted into
3062 * jiffies.
3063 *
3064 * Returns 0 on success.
3065 */
3066 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3067 void __user *buffer, size_t *lenp, loff_t *ppos)
3068 {
3069 return do_proc_dointvec(table,write,buffer,lenp,ppos,
3070 do_proc_dointvec_jiffies_conv,NULL);
3071 }
3072
3073 /**
3074 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
3075 * @table: the sysctl table
3076 * @write: %TRUE if this is a write to the sysctl file
3077 * @buffer: the user buffer
3078 * @lenp: the size of the user buffer
3079 * @ppos: pointer to the file position
3080 *
3081 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3082 * values from/to the user buffer, treated as an ASCII string.
3083 * The values read are assumed to be in 1/USER_HZ seconds, and
3084 * are converted into jiffies.
3085 *
3086 * Returns 0 on success.
3087 */
3088 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3089 void __user *buffer, size_t *lenp, loff_t *ppos)
3090 {
3091 return do_proc_dointvec(table,write,buffer,lenp,ppos,
3092 do_proc_dointvec_userhz_jiffies_conv,NULL);
3093 }
3094
3095 /**
3096 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
3097 * @table: the sysctl table
3098 * @write: %TRUE if this is a write to the sysctl file
3099 * @buffer: the user buffer
3100 * @lenp: the size of the user buffer
3101 * @ppos: file position
3102 * @ppos: the current position in the file
3103 *
3104 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3105 * values from/to the user buffer, treated as an ASCII string.
3106 * The values read are assumed to be in 1/1000 seconds, and
3107 * are converted into jiffies.
3108 *
3109 * Returns 0 on success.
3110 */
3111 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3112 void __user *buffer, size_t *lenp, loff_t *ppos)
3113 {
3114 return do_proc_dointvec(table, write, buffer, lenp, ppos,
3115 do_proc_dointvec_ms_jiffies_conv, NULL);
3116 }
3117
3118 static int proc_do_cad_pid(struct ctl_table *table, int write,
3119 void __user *buffer, size_t *lenp, loff_t *ppos)
3120 {
3121 struct pid *new_pid;
3122 pid_t tmp;
3123 int r;
3124
3125 tmp = pid_vnr(cad_pid);
3126
3127 r = __do_proc_dointvec(&tmp, table, write, buffer,
3128 lenp, ppos, NULL, NULL);
3129 if (r || !write)
3130 return r;
3131
3132 new_pid = find_get_pid(tmp);
3133 if (!new_pid)
3134 return -ESRCH;
3135
3136 put_pid(xchg(&cad_pid, new_pid));
3137 return 0;
3138 }
3139
3140 /**
3141 * proc_do_large_bitmap - read/write from/to a large bitmap
3142 * @table: the sysctl table
3143 * @write: %TRUE if this is a write to the sysctl file
3144 * @buffer: the user buffer
3145 * @lenp: the size of the user buffer
3146 * @ppos: file position
3147 *
3148 * The bitmap is stored at table->data and the bitmap length (in bits)
3149 * in table->maxlen.
3150 *
3151 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
3152 * large bitmaps may be represented in a compact manner. Writing into
3153 * the file will clear the bitmap then update it with the given input.
3154 *
3155 * Returns 0 on success.
3156 */
3157 int proc_do_large_bitmap(struct ctl_table *table, int write,
3158 void __user *buffer, size_t *lenp, loff_t *ppos)
3159 {
3160 int err = 0;
3161 bool first = 1;
3162 size_t left = *lenp;
3163 unsigned long bitmap_len = table->maxlen;
3164 unsigned long *bitmap = *(unsigned long **) table->data;
3165 unsigned long *tmp_bitmap = NULL;
3166 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
3167
3168 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
3169 *lenp = 0;
3170 return 0;
3171 }
3172
3173 if (write) {
3174 char *kbuf, *p;
3175 size_t skipped = 0;
3176
3177 if (left > PAGE_SIZE - 1) {
3178 left = PAGE_SIZE - 1;
3179 /* How much of the buffer we'll skip this pass */
3180 skipped = *lenp - left;
3181 }
3182
3183 p = kbuf = memdup_user_nul(buffer, left);
3184 if (IS_ERR(kbuf))
3185 return PTR_ERR(kbuf);
3186
3187 tmp_bitmap = bitmap_zalloc(bitmap_len, GFP_KERNEL);
3188 if (!tmp_bitmap) {
3189 kfree(kbuf);
3190 return -ENOMEM;
3191 }
3192 proc_skip_char(&p, &left, '\n');
3193 while (!err && left) {
3194 unsigned long val_a, val_b;
3195 bool neg;
3196 size_t saved_left;
3197
3198 /* In case we stop parsing mid-number, we can reset */
3199 saved_left = left;
3200 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3201 sizeof(tr_a), &c);
3202 /*
3203 * If we consumed the entirety of a truncated buffer or
3204 * only one char is left (may be a "-"), then stop here,
3205 * reset, & come back for more.
3206 */
3207 if ((left <= 1) && skipped) {
3208 left = saved_left;
3209 break;
3210 }
3211
3212 if (err)
3213 break;
3214 if (val_a >= bitmap_len || neg) {
3215 err = -EINVAL;
3216 break;
3217 }
3218
3219 val_b = val_a;
3220 if (left) {
3221 p++;
3222 left--;
3223 }
3224
3225 if (c == '-') {
3226 err = proc_get_long(&p, &left, &val_b,
3227 &neg, tr_b, sizeof(tr_b),
3228 &c);
3229 /*
3230 * If we consumed all of a truncated buffer or
3231 * then stop here, reset, & come back for more.
3232 */
3233 if (!left && skipped) {
3234 left = saved_left;
3235 break;
3236 }
3237
3238 if (err)
3239 break;
3240 if (val_b >= bitmap_len || neg ||
3241 val_a > val_b) {
3242 err = -EINVAL;
3243 break;
3244 }
3245 if (left) {
3246 p++;
3247 left--;
3248 }
3249 }
3250
3251 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3252 first = 0;
3253 proc_skip_char(&p, &left, '\n');
3254 }
3255 kfree(kbuf);
3256 left += skipped;
3257 } else {
3258 unsigned long bit_a, bit_b = 0;
3259
3260 while (left) {
3261 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3262 if (bit_a >= bitmap_len)
3263 break;
3264 bit_b = find_next_zero_bit(bitmap, bitmap_len,
3265 bit_a + 1) - 1;
3266
3267 if (!first) {
3268 err = proc_put_char(&buffer, &left, ',');
3269 if (err)
3270 break;
3271 }
3272 err = proc_put_long(&buffer, &left, bit_a, false);
3273 if (err)
3274 break;
3275 if (bit_a != bit_b) {
3276 err = proc_put_char(&buffer, &left, '-');
3277 if (err)
3278 break;
3279 err = proc_put_long(&buffer, &left, bit_b, false);
3280 if (err)
3281 break;
3282 }
3283
3284 first = 0; bit_b++;
3285 }
3286 if (!err)
3287 err = proc_put_char(&buffer, &left, '\n');
3288 }
3289
3290 if (!err) {
3291 if (write) {
3292 if (*ppos)
3293 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3294 else
3295 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3296 }
3297 *lenp -= left;
3298 *ppos += *lenp;
3299 }
3300
3301 bitmap_free(tmp_bitmap);
3302 return err;
3303 }
3304
3305 #else /* CONFIG_PROC_SYSCTL */
3306
3307 int proc_dostring(struct ctl_table *table, int write,
3308 void __user *buffer, size_t *lenp, loff_t *ppos)
3309 {
3310 return -ENOSYS;
3311 }
3312
3313 int proc_dointvec(struct ctl_table *table, int write,
3314 void __user *buffer, size_t *lenp, loff_t *ppos)
3315 {
3316 return -ENOSYS;
3317 }
3318
3319 int proc_douintvec(struct ctl_table *table, int write,
3320 void __user *buffer, size_t *lenp, loff_t *ppos)
3321 {
3322 return -ENOSYS;
3323 }
3324
3325 int proc_dointvec_minmax(struct ctl_table *table, int write,
3326 void __user *buffer, size_t *lenp, loff_t *ppos)
3327 {
3328 return -ENOSYS;
3329 }
3330
3331 int proc_douintvec_minmax(struct ctl_table *table, int write,
3332 void __user *buffer, size_t *lenp, loff_t *ppos)
3333 {
3334 return -ENOSYS;
3335 }
3336
3337 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3338 void __user *buffer, size_t *lenp, loff_t *ppos)
3339 {
3340 return -ENOSYS;
3341 }
3342
3343 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3344 void __user *buffer, size_t *lenp, loff_t *ppos)
3345 {
3346 return -ENOSYS;
3347 }
3348
3349 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3350 void __user *buffer, size_t *lenp, loff_t *ppos)
3351 {
3352 return -ENOSYS;
3353 }
3354
3355 int proc_doulongvec_minmax(struct ctl_table *table, int write,
3356 void __user *buffer, size_t *lenp, loff_t *ppos)
3357 {
3358 return -ENOSYS;
3359 }
3360
3361 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3362 void __user *buffer,
3363 size_t *lenp, loff_t *ppos)
3364 {
3365 return -ENOSYS;
3366 }
3367
3368 int proc_do_large_bitmap(struct ctl_table *table, int write,
3369 void __user *buffer, size_t *lenp, loff_t *ppos)
3370 {
3371 return -ENOSYS;
3372 }
3373
3374 #endif /* CONFIG_PROC_SYSCTL */
3375
3376 #if defined(CONFIG_BPF_SYSCALL) && defined(CONFIG_SYSCTL)
3377 static int proc_dointvec_minmax_bpf_stats(struct ctl_table *table, int write,
3378 void __user *buffer, size_t *lenp,
3379 loff_t *ppos)
3380 {
3381 int ret, bpf_stats = *(int *)table->data;
3382 struct ctl_table tmp = *table;
3383
3384 if (write && !capable(CAP_SYS_ADMIN))
3385 return -EPERM;
3386
3387 tmp.data = &bpf_stats;
3388 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
3389 if (write && !ret) {
3390 *(int *)table->data = bpf_stats;
3391 if (bpf_stats)
3392 static_branch_enable(&bpf_stats_enabled_key);
3393 else
3394 static_branch_disable(&bpf_stats_enabled_key);
3395 }
3396 return ret;
3397 }
3398 #endif
3399 /*
3400 * No sense putting this after each symbol definition, twice,
3401 * exception granted :-)
3402 */
3403 EXPORT_SYMBOL(proc_dointvec);
3404 EXPORT_SYMBOL(proc_douintvec);
3405 EXPORT_SYMBOL(proc_dointvec_jiffies);
3406 EXPORT_SYMBOL(proc_dointvec_minmax);
3407 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3408 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3409 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3410 EXPORT_SYMBOL(proc_dostring);
3411 EXPORT_SYMBOL(proc_doulongvec_minmax);
3412 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3413 EXPORT_SYMBOL(proc_do_large_bitmap);
A mirror of Linus' kernel repository