?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge tag 'for-linus-20190706' of git://git.kernel.dk/linux-block
[linux/fpc-iii.git] / kernel / sysctl.c
blob1beca96fb6252ddc4af07b6292f9dd95c4f53afd
1 // SPDX-License-Identifier: GPL-2.0-only
2 /*
3 * sysctl.c: General linux system control interface
4 *
5 * Begun 24 March 1995, Stephen Tweedie
6 * Added /proc support, Dec 1995
7 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
8 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
9 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
10 * Dynamic registration fixes, Stephen Tweedie.
11 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
12 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
13 * Horn.
14 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
15 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
16 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
17 * Wendling.
18 * The list_for_each() macro wasn't appropriate for the sysctl loop.
19 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
20 */
21
22 #include <linux/module.h>
23 #include <linux/aio.h>
24 #include <linux/mm.h>
25 #include <linux/swap.h>
26 #include <linux/slab.h>
27 #include <linux/sysctl.h>
28 #include <linux/bitmap.h>
29 #include <linux/signal.h>
30 #include <linux/printk.h>
31 #include <linux/proc_fs.h>
32 #include <linux/security.h>
33 #include <linux/ctype.h>
34 #include <linux/kmemleak.h>
35 #include <linux/fs.h>
36 #include <linux/init.h>
37 #include <linux/kernel.h>
38 #include <linux/kobject.h>
39 #include <linux/net.h>
40 #include <linux/sysrq.h>
41 #include <linux/highuid.h>
42 #include <linux/writeback.h>
43 #include <linux/ratelimit.h>
44 #include <linux/compaction.h>
45 #include <linux/hugetlb.h>
46 #include <linux/initrd.h>
47 #include <linux/key.h>
48 #include <linux/times.h>
49 #include <linux/limits.h>
50 #include <linux/dcache.h>
51 #include <linux/dnotify.h>
52 #include <linux/syscalls.h>
53 #include <linux/vmstat.h>
54 #include <linux/nfs_fs.h>
55 #include <linux/acpi.h>
56 #include <linux/reboot.h>
57 #include <linux/ftrace.h>
58 #include <linux/perf_event.h>
59 #include <linux/kprobes.h>
60 #include <linux/pipe_fs_i.h>
61 #include <linux/oom.h>
62 #include <linux/kmod.h>
63 #include <linux/capability.h>
64 #include <linux/binfmts.h>
65 #include <linux/sched/sysctl.h>
66 #include <linux/sched/coredump.h>
67 #include <linux/kexec.h>
68 #include <linux/bpf.h>
69 #include <linux/mount.h>
70 #include <linux/userfaultfd_k.h>
71
72 #include "../lib/kstrtox.h"
73
74 #include <linux/uaccess.h>
75 #include <asm/processor.h>
76
77 #ifdef CONFIG_X86
78 #include <asm/nmi.h>
79 #include <asm/stacktrace.h>
80 #include <asm/io.h>
81 #endif
82 #ifdef CONFIG_SPARC
83 #include <asm/setup.h>
84 #endif
85 #ifdef CONFIG_BSD_PROCESS_ACCT
86 #include <linux/acct.h>
87 #endif
88 #ifdef CONFIG_RT_MUTEXES
89 #include <linux/rtmutex.h>
90 #endif
91 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
92 #include <linux/lockdep.h>
93 #endif
94 #ifdef CONFIG_CHR_DEV_SG
95 #include <scsi/sg.h>
96 #endif
97 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
98 #include <linux/stackleak.h>
99 #endif
100 #ifdef CONFIG_LOCKUP_DETECTOR
101 #include <linux/nmi.h>
102 #endif
103
104 #if defined(CONFIG_SYSCTL)
105
106 /* External variables not in a header file. */
107 extern int suid_dumpable;
108 #ifdef CONFIG_COREDUMP
109 extern int core_uses_pid;
110 extern char core_pattern[];
111 extern unsigned int core_pipe_limit;
112 #endif
113 extern int pid_max;
114 extern int pid_max_min, pid_max_max;
115 extern int percpu_pagelist_fraction;
116 extern int latencytop_enabled;
117 extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
118 #ifndef CONFIG_MMU
119 extern int sysctl_nr_trim_pages;
120 #endif
121
122 /* Constants used for minimum and maximum */
123 #ifdef CONFIG_LOCKUP_DETECTOR
124 static int sixty = 60;
125 #endif
126
127 static int __maybe_unused neg_one = -1;
128
129 static int zero;
130 static int __maybe_unused one = 1;
131 static int __maybe_unused two = 2;
132 static int __maybe_unused four = 4;
133 static unsigned long zero_ul;
134 static unsigned long one_ul = 1;
135 static unsigned long long_max = LONG_MAX;
136 static int one_hundred = 100;
137 static int one_thousand = 1000;
138 #ifdef CONFIG_PRINTK
139 static int ten_thousand = 10000;
140 #endif
141 #ifdef CONFIG_PERF_EVENTS
142 static int six_hundred_forty_kb = 640 * 1024;
143 #endif
144
145 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
146 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
147
148 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
149 static int maxolduid = 65535;
150 static int minolduid;
151
152 static int ngroups_max = NGROUPS_MAX;
153 static const int cap_last_cap = CAP_LAST_CAP;
154
155 /*
156 * This is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs
157 * and hung_task_check_interval_secs
158 */
159 #ifdef CONFIG_DETECT_HUNG_TASK
160 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
161 #endif
162
163 #ifdef CONFIG_INOTIFY_USER
164 #include <linux/inotify.h>
165 #endif
166 #ifdef CONFIG_SPARC
167 #endif
168
169 #ifdef __hppa__
170 extern int pwrsw_enabled;
171 #endif
172
173 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
174 extern int unaligned_enabled;
175 #endif
176
177 #ifdef CONFIG_IA64
178 extern int unaligned_dump_stack;
179 #endif
180
181 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
182 extern int no_unaligned_warning;
183 #endif
184
185 #ifdef CONFIG_PROC_SYSCTL
186
187 /**
188 * enum sysctl_writes_mode - supported sysctl write modes
189 *
190 * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
191 * to be written, and multiple writes on the same sysctl file descriptor
192 * will rewrite the sysctl value, regardless of file position. No warning
193 * is issued when the initial position is not 0.
194 * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
195 * not 0.
196 * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
197 * file position 0 and the value must be fully contained in the buffer
198 * sent to the write syscall. If dealing with strings respect the file
199 * position, but restrict this to the max length of the buffer, anything
200 * passed the max lenght will be ignored. Multiple writes will append
201 * to the buffer.
202 *
203 * These write modes control how current file position affects the behavior of
204 * updating sysctl values through the proc interface on each write.
205 */
206 enum sysctl_writes_mode {
207 SYSCTL_WRITES_LEGACY = -1,
208 SYSCTL_WRITES_WARN = 0,
209 SYSCTL_WRITES_STRICT = 1,
210 };
211
212 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
213
214 static int proc_do_cad_pid(struct ctl_table *table, int write,
215 void __user *buffer, size_t *lenp, loff_t *ppos);
216 static int proc_taint(struct ctl_table *table, int write,
217 void __user *buffer, size_t *lenp, loff_t *ppos);
218 #endif
219
220 #ifdef CONFIG_PRINTK
221 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
222 void __user *buffer, size_t *lenp, loff_t *ppos);
223 #endif
224
225 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
226 void __user *buffer, size_t *lenp, loff_t *ppos);
227 #ifdef CONFIG_COREDUMP
228 static int proc_dostring_coredump(struct ctl_table *table, int write,
229 void __user *buffer, size_t *lenp, loff_t *ppos);
230 #endif
231 static int proc_dopipe_max_size(struct ctl_table *table, int write,
232 void __user *buffer, size_t *lenp, loff_t *ppos);
233
234 #ifdef CONFIG_MAGIC_SYSRQ
235 /* Note: sysrq code uses its own private copy */
236 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
237
238 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
239 void __user *buffer, size_t *lenp,
240 loff_t *ppos)
241 {
242 int error;
243
244 error = proc_dointvec(table, write, buffer, lenp, ppos);
245 if (error)
246 return error;
247
248 if (write)
249 sysrq_toggle_support(__sysrq_enabled);
250
251 return 0;
252 }
253
254 #endif
255
256 static struct ctl_table kern_table[];
257 static struct ctl_table vm_table[];
258 static struct ctl_table fs_table[];
259 static struct ctl_table debug_table[];
260 static struct ctl_table dev_table[];
261 extern struct ctl_table random_table[];
262 #ifdef CONFIG_EPOLL
263 extern struct ctl_table epoll_table[];
264 #endif
265
266 #ifdef CONFIG_FW_LOADER_USER_HELPER
267 extern struct ctl_table firmware_config_table[];
268 #endif
269
270 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
271 int sysctl_legacy_va_layout;
272 #endif
273
274 /* The default sysctl tables: */
275
276 static struct ctl_table sysctl_base_table[] = {
277 {
278 .procname = "kernel",
279 .mode = 0555,
280 .child = kern_table,
281 },
282 {
283 .procname = "vm",
284 .mode = 0555,
285 .child = vm_table,
286 },
287 {
288 .procname = "fs",
289 .mode = 0555,
290 .child = fs_table,
291 },
292 {
293 .procname = "debug",
294 .mode = 0555,
295 .child = debug_table,
296 },
297 {
298 .procname = "dev",
299 .mode = 0555,
300 .child = dev_table,
301 },
302 { }
303 };
304
305 #ifdef CONFIG_SCHED_DEBUG
306 static int min_sched_granularity_ns = 100000; /* 100 usecs */
307 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
308 static int min_wakeup_granularity_ns; /* 0 usecs */
309 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
310 #ifdef CONFIG_SMP
311 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
312 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
313 #endif /* CONFIG_SMP */
314 #endif /* CONFIG_SCHED_DEBUG */
315
316 #ifdef CONFIG_COMPACTION
317 static int min_extfrag_threshold;
318 static int max_extfrag_threshold = 1000;
319 #endif
320
321 static struct ctl_table kern_table[] = {
322 {
323 .procname = "sched_child_runs_first",
324 .data = &sysctl_sched_child_runs_first,
325 .maxlen = sizeof(unsigned int),
326 .mode = 0644,
327 .proc_handler = proc_dointvec,
328 },
329 #ifdef CONFIG_SCHED_DEBUG
330 {
331 .procname = "sched_min_granularity_ns",
332 .data = &sysctl_sched_min_granularity,
333 .maxlen = sizeof(unsigned int),
334 .mode = 0644,
335 .proc_handler = sched_proc_update_handler,
336 .extra1 = &min_sched_granularity_ns,
337 .extra2 = &max_sched_granularity_ns,
338 },
339 {
340 .procname = "sched_latency_ns",
341 .data = &sysctl_sched_latency,
342 .maxlen = sizeof(unsigned int),
343 .mode = 0644,
344 .proc_handler = sched_proc_update_handler,
345 .extra1 = &min_sched_granularity_ns,
346 .extra2 = &max_sched_granularity_ns,
347 },
348 {
349 .procname = "sched_wakeup_granularity_ns",
350 .data = &sysctl_sched_wakeup_granularity,
351 .maxlen = sizeof(unsigned int),
352 .mode = 0644,
353 .proc_handler = sched_proc_update_handler,
354 .extra1 = &min_wakeup_granularity_ns,
355 .extra2 = &max_wakeup_granularity_ns,
356 },
357 #ifdef CONFIG_SMP
358 {
359 .procname = "sched_tunable_scaling",
360 .data = &sysctl_sched_tunable_scaling,
361 .maxlen = sizeof(enum sched_tunable_scaling),
362 .mode = 0644,
363 .proc_handler = sched_proc_update_handler,
364 .extra1 = &min_sched_tunable_scaling,
365 .extra2 = &max_sched_tunable_scaling,
366 },
367 {
368 .procname = "sched_migration_cost_ns",
369 .data = &sysctl_sched_migration_cost,
370 .maxlen = sizeof(unsigned int),
371 .mode = 0644,
372 .proc_handler = proc_dointvec,
373 },
374 {
375 .procname = "sched_nr_migrate",
376 .data = &sysctl_sched_nr_migrate,
377 .maxlen = sizeof(unsigned int),
378 .mode = 0644,
379 .proc_handler = proc_dointvec,
380 },
381 #ifdef CONFIG_SCHEDSTATS
382 {
383 .procname = "sched_schedstats",
384 .data = NULL,
385 .maxlen = sizeof(unsigned int),
386 .mode = 0644,
387 .proc_handler = sysctl_schedstats,
388 .extra1 = &zero,
389 .extra2 = &one,
390 },
391 #endif /* CONFIG_SCHEDSTATS */
392 #endif /* CONFIG_SMP */
393 #ifdef CONFIG_NUMA_BALANCING
394 {
395 .procname = "numa_balancing_scan_delay_ms",
396 .data = &sysctl_numa_balancing_scan_delay,
397 .maxlen = sizeof(unsigned int),
398 .mode = 0644,
399 .proc_handler = proc_dointvec,
400 },
401 {
402 .procname = "numa_balancing_scan_period_min_ms",
403 .data = &sysctl_numa_balancing_scan_period_min,
404 .maxlen = sizeof(unsigned int),
405 .mode = 0644,
406 .proc_handler = proc_dointvec,
407 },
408 {
409 .procname = "numa_balancing_scan_period_max_ms",
410 .data = &sysctl_numa_balancing_scan_period_max,
411 .maxlen = sizeof(unsigned int),
412 .mode = 0644,
413 .proc_handler = proc_dointvec,
414 },
415 {
416 .procname = "numa_balancing_scan_size_mb",
417 .data = &sysctl_numa_balancing_scan_size,
418 .maxlen = sizeof(unsigned int),
419 .mode = 0644,
420 .proc_handler = proc_dointvec_minmax,
421 .extra1 = &one,
422 },
423 {
424 .procname = "numa_balancing",
425 .data = NULL, /* filled in by handler */
426 .maxlen = sizeof(unsigned int),
427 .mode = 0644,
428 .proc_handler = sysctl_numa_balancing,
429 .extra1 = &zero,
430 .extra2 = &one,
431 },
432 #endif /* CONFIG_NUMA_BALANCING */
433 #endif /* CONFIG_SCHED_DEBUG */
434 {
435 .procname = "sched_rt_period_us",
436 .data = &sysctl_sched_rt_period,
437 .maxlen = sizeof(unsigned int),
438 .mode = 0644,
439 .proc_handler = sched_rt_handler,
440 },
441 {
442 .procname = "sched_rt_runtime_us",
443 .data = &sysctl_sched_rt_runtime,
444 .maxlen = sizeof(int),
445 .mode = 0644,
446 .proc_handler = sched_rt_handler,
447 },
448 {
449 .procname = "sched_rr_timeslice_ms",
450 .data = &sysctl_sched_rr_timeslice,
451 .maxlen = sizeof(int),
452 .mode = 0644,
453 .proc_handler = sched_rr_handler,
454 },
455 #ifdef CONFIG_SCHED_AUTOGROUP
456 {
457 .procname = "sched_autogroup_enabled",
458 .data = &sysctl_sched_autogroup_enabled,
459 .maxlen = sizeof(unsigned int),
460 .mode = 0644,
461 .proc_handler = proc_dointvec_minmax,
462 .extra1 = &zero,
463 .extra2 = &one,
464 },
465 #endif
466 #ifdef CONFIG_CFS_BANDWIDTH
467 {
468 .procname = "sched_cfs_bandwidth_slice_us",
469 .data = &sysctl_sched_cfs_bandwidth_slice,
470 .maxlen = sizeof(unsigned int),
471 .mode = 0644,
472 .proc_handler = proc_dointvec_minmax,
473 .extra1 = &one,
474 },
475 #endif
476 #if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_CPU_FREQ_GOV_SCHEDUTIL)
477 {
478 .procname = "sched_energy_aware",
479 .data = &sysctl_sched_energy_aware,
480 .maxlen = sizeof(unsigned int),
481 .mode = 0644,
482 .proc_handler = sched_energy_aware_handler,
483 .extra1 = &zero,
484 .extra2 = &one,
485 },
486 #endif
487 #ifdef CONFIG_PROVE_LOCKING
488 {
489 .procname = "prove_locking",
490 .data = &prove_locking,
491 .maxlen = sizeof(int),
492 .mode = 0644,
493 .proc_handler = proc_dointvec,
494 },
495 #endif
496 #ifdef CONFIG_LOCK_STAT
497 {
498 .procname = "lock_stat",
499 .data = &lock_stat,
500 .maxlen = sizeof(int),
501 .mode = 0644,
502 .proc_handler = proc_dointvec,
503 },
504 #endif
505 {
506 .procname = "panic",
507 .data = &panic_timeout,
508 .maxlen = sizeof(int),
509 .mode = 0644,
510 .proc_handler = proc_dointvec,
511 },
512 #ifdef CONFIG_COREDUMP
513 {
514 .procname = "core_uses_pid",
515 .data = &core_uses_pid,
516 .maxlen = sizeof(int),
517 .mode = 0644,
518 .proc_handler = proc_dointvec,
519 },
520 {
521 .procname = "core_pattern",
522 .data = core_pattern,
523 .maxlen = CORENAME_MAX_SIZE,
524 .mode = 0644,
525 .proc_handler = proc_dostring_coredump,
526 },
527 {
528 .procname = "core_pipe_limit",
529 .data = &core_pipe_limit,
530 .maxlen = sizeof(unsigned int),
531 .mode = 0644,
532 .proc_handler = proc_dointvec,
533 },
534 #endif
535 #ifdef CONFIG_PROC_SYSCTL
536 {
537 .procname = "tainted",
538 .maxlen = sizeof(long),
539 .mode = 0644,
540 .proc_handler = proc_taint,
541 },
542 {
543 .procname = "sysctl_writes_strict",
544 .data = &sysctl_writes_strict,
545 .maxlen = sizeof(int),
546 .mode = 0644,
547 .proc_handler = proc_dointvec_minmax,
548 .extra1 = &neg_one,
549 .extra2 = &one,
550 },
551 #endif
552 #ifdef CONFIG_LATENCYTOP
553 {
554 .procname = "latencytop",
555 .data = &latencytop_enabled,
556 .maxlen = sizeof(int),
557 .mode = 0644,
558 .proc_handler = sysctl_latencytop,
559 },
560 #endif
561 #ifdef CONFIG_BLK_DEV_INITRD
562 {
563 .procname = "real-root-dev",
564 .data = &real_root_dev,
565 .maxlen = sizeof(int),
566 .mode = 0644,
567 .proc_handler = proc_dointvec,
568 },
569 #endif
570 {
571 .procname = "print-fatal-signals",
572 .data = &print_fatal_signals,
573 .maxlen = sizeof(int),
574 .mode = 0644,
575 .proc_handler = proc_dointvec,
576 },
577 #ifdef CONFIG_SPARC
578 {
579 .procname = "reboot-cmd",
580 .data = reboot_command,
581 .maxlen = 256,
582 .mode = 0644,
583 .proc_handler = proc_dostring,
584 },
585 {
586 .procname = "stop-a",
587 .data = &stop_a_enabled,
588 .maxlen = sizeof (int),
589 .mode = 0644,
590 .proc_handler = proc_dointvec,
591 },
592 {
593 .procname = "scons-poweroff",
594 .data = &scons_pwroff,
595 .maxlen = sizeof (int),
596 .mode = 0644,
597 .proc_handler = proc_dointvec,
598 },
599 #endif
600 #ifdef CONFIG_SPARC64
601 {
602 .procname = "tsb-ratio",
603 .data = &sysctl_tsb_ratio,
604 .maxlen = sizeof (int),
605 .mode = 0644,
606 .proc_handler = proc_dointvec,
607 },
608 #endif
609 #ifdef __hppa__
610 {
611 .procname = "soft-power",
612 .data = &pwrsw_enabled,
613 .maxlen = sizeof (int),
614 .mode = 0644,
615 .proc_handler = proc_dointvec,
616 },
617 #endif
618 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
619 {
620 .procname = "unaligned-trap",
621 .data = &unaligned_enabled,
622 .maxlen = sizeof (int),
623 .mode = 0644,
624 .proc_handler = proc_dointvec,
625 },
626 #endif
627 {
628 .procname = "ctrl-alt-del",
629 .data = &C_A_D,
630 .maxlen = sizeof(int),
631 .mode = 0644,
632 .proc_handler = proc_dointvec,
633 },
634 #ifdef CONFIG_FUNCTION_TRACER
635 {
636 .procname = "ftrace_enabled",
637 .data = &ftrace_enabled,
638 .maxlen = sizeof(int),
639 .mode = 0644,
640 .proc_handler = ftrace_enable_sysctl,
641 },
642 #endif
643 #ifdef CONFIG_STACK_TRACER
644 {
645 .procname = "stack_tracer_enabled",
646 .data = &stack_tracer_enabled,
647 .maxlen = sizeof(int),
648 .mode = 0644,
649 .proc_handler = stack_trace_sysctl,
650 },
651 #endif
652 #ifdef CONFIG_TRACING
653 {
654 .procname = "ftrace_dump_on_oops",
655 .data = &ftrace_dump_on_oops,
656 .maxlen = sizeof(int),
657 .mode = 0644,
658 .proc_handler = proc_dointvec,
659 },
660 {
661 .procname = "traceoff_on_warning",
662 .data = &__disable_trace_on_warning,
663 .maxlen = sizeof(__disable_trace_on_warning),
664 .mode = 0644,
665 .proc_handler = proc_dointvec,
666 },
667 {
668 .procname = "tracepoint_printk",
669 .data = &tracepoint_printk,
670 .maxlen = sizeof(tracepoint_printk),
671 .mode = 0644,
672 .proc_handler = tracepoint_printk_sysctl,
673 },
674 #endif
675 #ifdef CONFIG_KEXEC_CORE
676 {
677 .procname = "kexec_load_disabled",
678 .data = &kexec_load_disabled,
679 .maxlen = sizeof(int),
680 .mode = 0644,
681 /* only handle a transition from default "0" to "1" */
682 .proc_handler = proc_dointvec_minmax,
683 .extra1 = &one,
684 .extra2 = &one,
685 },
686 #endif
687 #ifdef CONFIG_MODULES
688 {
689 .procname = "modprobe",
690 .data = &modprobe_path,
691 .maxlen = KMOD_PATH_LEN,
692 .mode = 0644,
693 .proc_handler = proc_dostring,
694 },
695 {
696 .procname = "modules_disabled",
697 .data = &modules_disabled,
698 .maxlen = sizeof(int),
699 .mode = 0644,
700 /* only handle a transition from default "0" to "1" */
701 .proc_handler = proc_dointvec_minmax,
702 .extra1 = &one,
703 .extra2 = &one,
704 },
705 #endif
706 #ifdef CONFIG_UEVENT_HELPER
707 {
708 .procname = "hotplug",
709 .data = &uevent_helper,
710 .maxlen = UEVENT_HELPER_PATH_LEN,
711 .mode = 0644,
712 .proc_handler = proc_dostring,
713 },
714 #endif
715 #ifdef CONFIG_CHR_DEV_SG
716 {
717 .procname = "sg-big-buff",
718 .data = &sg_big_buff,
719 .maxlen = sizeof (int),
720 .mode = 0444,
721 .proc_handler = proc_dointvec,
722 },
723 #endif
724 #ifdef CONFIG_BSD_PROCESS_ACCT
725 {
726 .procname = "acct",
727 .data = &acct_parm,
728 .maxlen = 3*sizeof(int),
729 .mode = 0644,
730 .proc_handler = proc_dointvec,
731 },
732 #endif
733 #ifdef CONFIG_MAGIC_SYSRQ
734 {
735 .procname = "sysrq",
736 .data = &__sysrq_enabled,
737 .maxlen = sizeof (int),
738 .mode = 0644,
739 .proc_handler = sysrq_sysctl_handler,
740 },
741 #endif
742 #ifdef CONFIG_PROC_SYSCTL
743 {
744 .procname = "cad_pid",
745 .data = NULL,
746 .maxlen = sizeof (int),
747 .mode = 0600,
748 .proc_handler = proc_do_cad_pid,
749 },
750 #endif
751 {
752 .procname = "threads-max",
753 .data = NULL,
754 .maxlen = sizeof(int),
755 .mode = 0644,
756 .proc_handler = sysctl_max_threads,
757 },
758 {
759 .procname = "random",
760 .mode = 0555,
761 .child = random_table,
762 },
763 {
764 .procname = "usermodehelper",
765 .mode = 0555,
766 .child = usermodehelper_table,
767 },
768 #ifdef CONFIG_FW_LOADER_USER_HELPER
769 {
770 .procname = "firmware_config",
771 .mode = 0555,
772 .child = firmware_config_table,
773 },
774 #endif
775 {
776 .procname = "overflowuid",
777 .data = &overflowuid,
778 .maxlen = sizeof(int),
779 .mode = 0644,
780 .proc_handler = proc_dointvec_minmax,
781 .extra1 = &minolduid,
782 .extra2 = &maxolduid,
783 },
784 {
785 .procname = "overflowgid",
786 .data = &overflowgid,
787 .maxlen = sizeof(int),
788 .mode = 0644,
789 .proc_handler = proc_dointvec_minmax,
790 .extra1 = &minolduid,
791 .extra2 = &maxolduid,
792 },
793 #ifdef CONFIG_S390
794 #ifdef CONFIG_MATHEMU
795 {
796 .procname = "ieee_emulation_warnings",
797 .data = &sysctl_ieee_emulation_warnings,
798 .maxlen = sizeof(int),
799 .mode = 0644,
800 .proc_handler = proc_dointvec,
801 },
802 #endif
803 {
804 .procname = "userprocess_debug",
805 .data = &show_unhandled_signals,
806 .maxlen = sizeof(int),
807 .mode = 0644,
808 .proc_handler = proc_dointvec,
809 },
810 #endif
811 {
812 .procname = "pid_max",
813 .data = &pid_max,
814 .maxlen = sizeof (int),
815 .mode = 0644,
816 .proc_handler = proc_dointvec_minmax,
817 .extra1 = &pid_max_min,
818 .extra2 = &pid_max_max,
819 },
820 {
821 .procname = "panic_on_oops",
822 .data = &panic_on_oops,
823 .maxlen = sizeof(int),
824 .mode = 0644,
825 .proc_handler = proc_dointvec,
826 },
827 {
828 .procname = "panic_print",
829 .data = &panic_print,
830 .maxlen = sizeof(unsigned long),
831 .mode = 0644,
832 .proc_handler = proc_doulongvec_minmax,
833 },
834 #if defined CONFIG_PRINTK
835 {
836 .procname = "printk",
837 .data = &console_loglevel,
838 .maxlen = 4*sizeof(int),
839 .mode = 0644,
840 .proc_handler = proc_dointvec,
841 },
842 {
843 .procname = "printk_ratelimit",
844 .data = &printk_ratelimit_state.interval,
845 .maxlen = sizeof(int),
846 .mode = 0644,
847 .proc_handler = proc_dointvec_jiffies,
848 },
849 {
850 .procname = "printk_ratelimit_burst",
851 .data = &printk_ratelimit_state.burst,
852 .maxlen = sizeof(int),
853 .mode = 0644,
854 .proc_handler = proc_dointvec,
855 },
856 {
857 .procname = "printk_delay",
858 .data = &printk_delay_msec,
859 .maxlen = sizeof(int),
860 .mode = 0644,
861 .proc_handler = proc_dointvec_minmax,
862 .extra1 = &zero,
863 .extra2 = &ten_thousand,
864 },
865 {
866 .procname = "printk_devkmsg",
867 .data = devkmsg_log_str,
868 .maxlen = DEVKMSG_STR_MAX_SIZE,
869 .mode = 0644,
870 .proc_handler = devkmsg_sysctl_set_loglvl,
871 },
872 {
873 .procname = "dmesg_restrict",
874 .data = &dmesg_restrict,
875 .maxlen = sizeof(int),
876 .mode = 0644,
877 .proc_handler = proc_dointvec_minmax_sysadmin,
878 .extra1 = &zero,
879 .extra2 = &one,
880 },
881 {
882 .procname = "kptr_restrict",
883 .data = &kptr_restrict,
884 .maxlen = sizeof(int),
885 .mode = 0644,
886 .proc_handler = proc_dointvec_minmax_sysadmin,
887 .extra1 = &zero,
888 .extra2 = &two,
889 },
890 #endif
891 {
892 .procname = "ngroups_max",
893 .data = &ngroups_max,
894 .maxlen = sizeof (int),
895 .mode = 0444,
896 .proc_handler = proc_dointvec,
897 },
898 {
899 .procname = "cap_last_cap",
900 .data = (void *)&cap_last_cap,
901 .maxlen = sizeof(int),
902 .mode = 0444,
903 .proc_handler = proc_dointvec,
904 },
905 #if defined(CONFIG_LOCKUP_DETECTOR)
906 {
907 .procname = "watchdog",
908 .data = &watchdog_user_enabled,
909 .maxlen = sizeof(int),
910 .mode = 0644,
911 .proc_handler = proc_watchdog,
912 .extra1 = &zero,
913 .extra2 = &one,
914 },
915 {
916 .procname = "watchdog_thresh",
917 .data = &watchdog_thresh,
918 .maxlen = sizeof(int),
919 .mode = 0644,
920 .proc_handler = proc_watchdog_thresh,
921 .extra1 = &zero,
922 .extra2 = &sixty,
923 },
924 {
925 .procname = "nmi_watchdog",
926 .data = &nmi_watchdog_user_enabled,
927 .maxlen = sizeof(int),
928 .mode = NMI_WATCHDOG_SYSCTL_PERM,
929 .proc_handler = proc_nmi_watchdog,
930 .extra1 = &zero,
931 .extra2 = &one,
932 },
933 {
934 .procname = "watchdog_cpumask",
935 .data = &watchdog_cpumask_bits,
936 .maxlen = NR_CPUS,
937 .mode = 0644,
938 .proc_handler = proc_watchdog_cpumask,
939 },
940 #ifdef CONFIG_SOFTLOCKUP_DETECTOR
941 {
942 .procname = "soft_watchdog",
943 .data = &soft_watchdog_user_enabled,
944 .maxlen = sizeof(int),
945 .mode = 0644,
946 .proc_handler = proc_soft_watchdog,
947 .extra1 = &zero,
948 .extra2 = &one,
949 },
950 {
951 .procname = "softlockup_panic",
952 .data = &softlockup_panic,
953 .maxlen = sizeof(int),
954 .mode = 0644,
955 .proc_handler = proc_dointvec_minmax,
956 .extra1 = &zero,
957 .extra2 = &one,
958 },
959 #ifdef CONFIG_SMP
960 {
961 .procname = "softlockup_all_cpu_backtrace",
962 .data = &sysctl_softlockup_all_cpu_backtrace,
963 .maxlen = sizeof(int),
964 .mode = 0644,
965 .proc_handler = proc_dointvec_minmax,
966 .extra1 = &zero,
967 .extra2 = &one,
968 },
969 #endif /* CONFIG_SMP */
970 #endif
971 #ifdef CONFIG_HARDLOCKUP_DETECTOR
972 {
973 .procname = "hardlockup_panic",
974 .data = &hardlockup_panic,
975 .maxlen = sizeof(int),
976 .mode = 0644,
977 .proc_handler = proc_dointvec_minmax,
978 .extra1 = &zero,
979 .extra2 = &one,
980 },
981 #ifdef CONFIG_SMP
982 {
983 .procname = "hardlockup_all_cpu_backtrace",
984 .data = &sysctl_hardlockup_all_cpu_backtrace,
985 .maxlen = sizeof(int),
986 .mode = 0644,
987 .proc_handler = proc_dointvec_minmax,
988 .extra1 = &zero,
989 .extra2 = &one,
990 },
991 #endif /* CONFIG_SMP */
992 #endif
993 #endif
994
995 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
996 {
997 .procname = "unknown_nmi_panic",
998 .data = &unknown_nmi_panic,
999 .maxlen = sizeof (int),
1000 .mode = 0644,
1001 .proc_handler = proc_dointvec,
1002 },
1003 #endif
1004 #if defined(CONFIG_X86)
1005 {
1006 .procname = "panic_on_unrecovered_nmi",
1007 .data = &panic_on_unrecovered_nmi,
1008 .maxlen = sizeof(int),
1009 .mode = 0644,
1010 .proc_handler = proc_dointvec,
1011 },
1012 {
1013 .procname = "panic_on_io_nmi",
1014 .data = &panic_on_io_nmi,
1015 .maxlen = sizeof(int),
1016 .mode = 0644,
1017 .proc_handler = proc_dointvec,
1018 },
1019 #ifdef CONFIG_DEBUG_STACKOVERFLOW
1020 {
1021 .procname = "panic_on_stackoverflow",
1022 .data = &sysctl_panic_on_stackoverflow,
1023 .maxlen = sizeof(int),
1024 .mode = 0644,
1025 .proc_handler = proc_dointvec,
1026 },
1027 #endif
1028 {
1029 .procname = "bootloader_type",
1030 .data = &bootloader_type,
1031 .maxlen = sizeof (int),
1032 .mode = 0444,
1033 .proc_handler = proc_dointvec,
1034 },
1035 {
1036 .procname = "bootloader_version",
1037 .data = &bootloader_version,
1038 .maxlen = sizeof (int),
1039 .mode = 0444,
1040 .proc_handler = proc_dointvec,
1041 },
1042 {
1043 .procname = "io_delay_type",
1044 .data = &io_delay_type,
1045 .maxlen = sizeof(int),
1046 .mode = 0644,
1047 .proc_handler = proc_dointvec,
1048 },
1049 #endif
1050 #if defined(CONFIG_MMU)
1051 {
1052 .procname = "randomize_va_space",
1053 .data = &randomize_va_space,
1054 .maxlen = sizeof(int),
1055 .mode = 0644,
1056 .proc_handler = proc_dointvec,
1057 },
1058 #endif
1059 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1060 {
1061 .procname = "spin_retry",
1062 .data = &spin_retry,
1063 .maxlen = sizeof (int),
1064 .mode = 0644,
1065 .proc_handler = proc_dointvec,
1066 },
1067 #endif
1068 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1069 {
1070 .procname = "acpi_video_flags",
1071 .data = &acpi_realmode_flags,
1072 .maxlen = sizeof (unsigned long),
1073 .mode = 0644,
1074 .proc_handler = proc_doulongvec_minmax,
1075 },
1076 #endif
1077 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1078 {
1079 .procname = "ignore-unaligned-usertrap",
1080 .data = &no_unaligned_warning,
1081 .maxlen = sizeof (int),
1082 .mode = 0644,
1083 .proc_handler = proc_dointvec,
1084 },
1085 #endif
1086 #ifdef CONFIG_IA64
1087 {
1088 .procname = "unaligned-dump-stack",
1089 .data = &unaligned_dump_stack,
1090 .maxlen = sizeof (int),
1091 .mode = 0644,
1092 .proc_handler = proc_dointvec,
1093 },
1094 #endif
1095 #ifdef CONFIG_DETECT_HUNG_TASK
1096 {
1097 .procname = "hung_task_panic",
1098 .data = &sysctl_hung_task_panic,
1099 .maxlen = sizeof(int),
1100 .mode = 0644,
1101 .proc_handler = proc_dointvec_minmax,
1102 .extra1 = &zero,
1103 .extra2 = &one,
1104 },
1105 {
1106 .procname = "hung_task_check_count",
1107 .data = &sysctl_hung_task_check_count,
1108 .maxlen = sizeof(int),
1109 .mode = 0644,
1110 .proc_handler = proc_dointvec_minmax,
1111 .extra1 = &zero,
1112 },
1113 {
1114 .procname = "hung_task_timeout_secs",
1115 .data = &sysctl_hung_task_timeout_secs,
1116 .maxlen = sizeof(unsigned long),
1117 .mode = 0644,
1118 .proc_handler = proc_dohung_task_timeout_secs,
1119 .extra2 = &hung_task_timeout_max,
1120 },
1121 {
1122 .procname = "hung_task_check_interval_secs",
1123 .data = &sysctl_hung_task_check_interval_secs,
1124 .maxlen = sizeof(unsigned long),
1125 .mode = 0644,
1126 .proc_handler = proc_dohung_task_timeout_secs,
1127 .extra2 = &hung_task_timeout_max,
1128 },
1129 {
1130 .procname = "hung_task_warnings",
1131 .data = &sysctl_hung_task_warnings,
1132 .maxlen = sizeof(int),
1133 .mode = 0644,
1134 .proc_handler = proc_dointvec_minmax,
1135 .extra1 = &neg_one,
1136 },
1137 #endif
1138 #ifdef CONFIG_RT_MUTEXES
1139 {
1140 .procname = "max_lock_depth",
1141 .data = &max_lock_depth,
1142 .maxlen = sizeof(int),
1143 .mode = 0644,
1144 .proc_handler = proc_dointvec,
1145 },
1146 #endif
1147 {
1148 .procname = "poweroff_cmd",
1149 .data = &poweroff_cmd,
1150 .maxlen = POWEROFF_CMD_PATH_LEN,
1151 .mode = 0644,
1152 .proc_handler = proc_dostring,
1153 },
1154 #ifdef CONFIG_KEYS
1155 {
1156 .procname = "keys",
1157 .mode = 0555,
1158 .child = key_sysctls,
1159 },
1160 #endif
1161 #ifdef CONFIG_PERF_EVENTS
1162 /*
1163 * User-space scripts rely on the existence of this file
1164 * as a feature check for perf_events being enabled.
1165 *
1166 * So it's an ABI, do not remove!
1167 */
1168 {
1169 .procname = "perf_event_paranoid",
1170 .data = &sysctl_perf_event_paranoid,
1171 .maxlen = sizeof(sysctl_perf_event_paranoid),
1172 .mode = 0644,
1173 .proc_handler = proc_dointvec,
1174 },
1175 {
1176 .procname = "perf_event_mlock_kb",
1177 .data = &sysctl_perf_event_mlock,
1178 .maxlen = sizeof(sysctl_perf_event_mlock),
1179 .mode = 0644,
1180 .proc_handler = proc_dointvec,
1181 },
1182 {
1183 .procname = "perf_event_max_sample_rate",
1184 .data = &sysctl_perf_event_sample_rate,
1185 .maxlen = sizeof(sysctl_perf_event_sample_rate),
1186 .mode = 0644,
1187 .proc_handler = perf_proc_update_handler,
1188 .extra1 = &one,
1189 },
1190 {
1191 .procname = "perf_cpu_time_max_percent",
1192 .data = &sysctl_perf_cpu_time_max_percent,
1193 .maxlen = sizeof(sysctl_perf_cpu_time_max_percent),
1194 .mode = 0644,
1195 .proc_handler = perf_cpu_time_max_percent_handler,
1196 .extra1 = &zero,
1197 .extra2 = &one_hundred,
1198 },
1199 {
1200 .procname = "perf_event_max_stack",
1201 .data = &sysctl_perf_event_max_stack,
1202 .maxlen = sizeof(sysctl_perf_event_max_stack),
1203 .mode = 0644,
1204 .proc_handler = perf_event_max_stack_handler,
1205 .extra1 = &zero,
1206 .extra2 = &six_hundred_forty_kb,
1207 },
1208 {
1209 .procname = "perf_event_max_contexts_per_stack",
1210 .data = &sysctl_perf_event_max_contexts_per_stack,
1211 .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack),
1212 .mode = 0644,
1213 .proc_handler = perf_event_max_stack_handler,
1214 .extra1 = &zero,
1215 .extra2 = &one_thousand,
1216 },
1217 #endif
1218 {
1219 .procname = "panic_on_warn",
1220 .data = &panic_on_warn,
1221 .maxlen = sizeof(int),
1222 .mode = 0644,
1223 .proc_handler = proc_dointvec_minmax,
1224 .extra1 = &zero,
1225 .extra2 = &one,
1226 },
1227 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1228 {
1229 .procname = "timer_migration",
1230 .data = &sysctl_timer_migration,
1231 .maxlen = sizeof(unsigned int),
1232 .mode = 0644,
1233 .proc_handler = timer_migration_handler,
1234 .extra1 = &zero,
1235 .extra2 = &one,
1236 },
1237 #endif
1238 #ifdef CONFIG_BPF_SYSCALL
1239 {
1240 .procname = "unprivileged_bpf_disabled",
1241 .data = &sysctl_unprivileged_bpf_disabled,
1242 .maxlen = sizeof(sysctl_unprivileged_bpf_disabled),
1243 .mode = 0644,
1244 /* only handle a transition from default "0" to "1" */
1245 .proc_handler = proc_dointvec_minmax,
1246 .extra1 = &one,
1247 .extra2 = &one,
1248 },
1249 {
1250 .procname = "bpf_stats_enabled",
1251 .data = &bpf_stats_enabled_key.key,
1252 .maxlen = sizeof(bpf_stats_enabled_key),
1253 .mode = 0644,
1254 .proc_handler = proc_do_static_key,
1255 },
1256 #endif
1257 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1258 {
1259 .procname = "panic_on_rcu_stall",
1260 .data = &sysctl_panic_on_rcu_stall,
1261 .maxlen = sizeof(sysctl_panic_on_rcu_stall),
1262 .mode = 0644,
1263 .proc_handler = proc_dointvec_minmax,
1264 .extra1 = &zero,
1265 .extra2 = &one,
1266 },
1267 #endif
1268 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
1269 {
1270 .procname = "stack_erasing",
1271 .data = NULL,
1272 .maxlen = sizeof(int),
1273 .mode = 0600,
1274 .proc_handler = stack_erasing_sysctl,
1275 .extra1 = &zero,
1276 .extra2 = &one,
1277 },
1278 #endif
1279 { }
1280 };
1281
1282 static struct ctl_table vm_table[] = {
1283 {
1284 .procname = "overcommit_memory",
1285 .data = &sysctl_overcommit_memory,
1286 .maxlen = sizeof(sysctl_overcommit_memory),
1287 .mode = 0644,
1288 .proc_handler = proc_dointvec_minmax,
1289 .extra1 = &zero,
1290 .extra2 = &two,
1291 },
1292 {
1293 .procname = "panic_on_oom",
1294 .data = &sysctl_panic_on_oom,
1295 .maxlen = sizeof(sysctl_panic_on_oom),
1296 .mode = 0644,
1297 .proc_handler = proc_dointvec_minmax,
1298 .extra1 = &zero,
1299 .extra2 = &two,
1300 },
1301 {
1302 .procname = "oom_kill_allocating_task",
1303 .data = &sysctl_oom_kill_allocating_task,
1304 .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1305 .mode = 0644,
1306 .proc_handler = proc_dointvec,
1307 },
1308 {
1309 .procname = "oom_dump_tasks",
1310 .data = &sysctl_oom_dump_tasks,
1311 .maxlen = sizeof(sysctl_oom_dump_tasks),
1312 .mode = 0644,
1313 .proc_handler = proc_dointvec,
1314 },
1315 {
1316 .procname = "overcommit_ratio",
1317 .data = &sysctl_overcommit_ratio,
1318 .maxlen = sizeof(sysctl_overcommit_ratio),
1319 .mode = 0644,
1320 .proc_handler = overcommit_ratio_handler,
1321 },
1322 {
1323 .procname = "overcommit_kbytes",
1324 .data = &sysctl_overcommit_kbytes,
1325 .maxlen = sizeof(sysctl_overcommit_kbytes),
1326 .mode = 0644,
1327 .proc_handler = overcommit_kbytes_handler,
1328 },
1329 {
1330 .procname = "page-cluster",
1331 .data = &page_cluster,
1332 .maxlen = sizeof(int),
1333 .mode = 0644,
1334 .proc_handler = proc_dointvec_minmax,
1335 .extra1 = &zero,
1336 },
1337 {
1338 .procname = "dirty_background_ratio",
1339 .data = &dirty_background_ratio,
1340 .maxlen = sizeof(dirty_background_ratio),
1341 .mode = 0644,
1342 .proc_handler = dirty_background_ratio_handler,
1343 .extra1 = &zero,
1344 .extra2 = &one_hundred,
1345 },
1346 {
1347 .procname = "dirty_background_bytes",
1348 .data = &dirty_background_bytes,
1349 .maxlen = sizeof(dirty_background_bytes),
1350 .mode = 0644,
1351 .proc_handler = dirty_background_bytes_handler,
1352 .extra1 = &one_ul,
1353 },
1354 {
1355 .procname = "dirty_ratio",
1356 .data = &vm_dirty_ratio,
1357 .maxlen = sizeof(vm_dirty_ratio),
1358 .mode = 0644,
1359 .proc_handler = dirty_ratio_handler,
1360 .extra1 = &zero,
1361 .extra2 = &one_hundred,
1362 },
1363 {
1364 .procname = "dirty_bytes",
1365 .data = &vm_dirty_bytes,
1366 .maxlen = sizeof(vm_dirty_bytes),
1367 .mode = 0644,
1368 .proc_handler = dirty_bytes_handler,
1369 .extra1 = &dirty_bytes_min,
1370 },
1371 {
1372 .procname = "dirty_writeback_centisecs",
1373 .data = &dirty_writeback_interval,
1374 .maxlen = sizeof(dirty_writeback_interval),
1375 .mode = 0644,
1376 .proc_handler = dirty_writeback_centisecs_handler,
1377 },
1378 {
1379 .procname = "dirty_expire_centisecs",
1380 .data = &dirty_expire_interval,
1381 .maxlen = sizeof(dirty_expire_interval),
1382 .mode = 0644,
1383 .proc_handler = proc_dointvec_minmax,
1384 .extra1 = &zero,
1385 },
1386 {
1387 .procname = "dirtytime_expire_seconds",
1388 .data = &dirtytime_expire_interval,
1389 .maxlen = sizeof(dirtytime_expire_interval),
1390 .mode = 0644,
1391 .proc_handler = dirtytime_interval_handler,
1392 .extra1 = &zero,
1393 },
1394 {
1395 .procname = "swappiness",
1396 .data = &vm_swappiness,
1397 .maxlen = sizeof(vm_swappiness),
1398 .mode = 0644,
1399 .proc_handler = proc_dointvec_minmax,
1400 .extra1 = &zero,
1401 .extra2 = &one_hundred,
1402 },
1403 #ifdef CONFIG_HUGETLB_PAGE
1404 {
1405 .procname = "nr_hugepages",
1406 .data = NULL,
1407 .maxlen = sizeof(unsigned long),
1408 .mode = 0644,
1409 .proc_handler = hugetlb_sysctl_handler,
1410 },
1411 #ifdef CONFIG_NUMA
1412 {
1413 .procname = "nr_hugepages_mempolicy",
1414 .data = NULL,
1415 .maxlen = sizeof(unsigned long),
1416 .mode = 0644,
1417 .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1418 },
1419 {
1420 .procname = "numa_stat",
1421 .data = &sysctl_vm_numa_stat,
1422 .maxlen = sizeof(int),
1423 .mode = 0644,
1424 .proc_handler = sysctl_vm_numa_stat_handler,
1425 .extra1 = &zero,
1426 .extra2 = &one,
1427 },
1428 #endif
1429 {
1430 .procname = "hugetlb_shm_group",
1431 .data = &sysctl_hugetlb_shm_group,
1432 .maxlen = sizeof(gid_t),
1433 .mode = 0644,
1434 .proc_handler = proc_dointvec,
1435 },
1436 {
1437 .procname = "nr_overcommit_hugepages",
1438 .data = NULL,
1439 .maxlen = sizeof(unsigned long),
1440 .mode = 0644,
1441 .proc_handler = hugetlb_overcommit_handler,
1442 },
1443 #endif
1444 {
1445 .procname = "lowmem_reserve_ratio",
1446 .data = &sysctl_lowmem_reserve_ratio,
1447 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1448 .mode = 0644,
1449 .proc_handler = lowmem_reserve_ratio_sysctl_handler,
1450 },
1451 {
1452 .procname = "drop_caches",
1453 .data = &sysctl_drop_caches,
1454 .maxlen = sizeof(int),
1455 .mode = 0644,
1456 .proc_handler = drop_caches_sysctl_handler,
1457 .extra1 = &one,
1458 .extra2 = &four,
1459 },
1460 #ifdef CONFIG_COMPACTION
1461 {
1462 .procname = "compact_memory",
1463 .data = &sysctl_compact_memory,
1464 .maxlen = sizeof(int),
1465 .mode = 0200,
1466 .proc_handler = sysctl_compaction_handler,
1467 },
1468 {
1469 .procname = "extfrag_threshold",
1470 .data = &sysctl_extfrag_threshold,
1471 .maxlen = sizeof(int),
1472 .mode = 0644,
1473 .proc_handler = proc_dointvec_minmax,
1474 .extra1 = &min_extfrag_threshold,
1475 .extra2 = &max_extfrag_threshold,
1476 },
1477 {
1478 .procname = "compact_unevictable_allowed",
1479 .data = &sysctl_compact_unevictable_allowed,
1480 .maxlen = sizeof(int),
1481 .mode = 0644,
1482 .proc_handler = proc_dointvec,
1483 .extra1 = &zero,
1484 .extra2 = &one,
1485 },
1486
1487 #endif /* CONFIG_COMPACTION */
1488 {
1489 .procname = "min_free_kbytes",
1490 .data = &min_free_kbytes,
1491 .maxlen = sizeof(min_free_kbytes),
1492 .mode = 0644,
1493 .proc_handler = min_free_kbytes_sysctl_handler,
1494 .extra1 = &zero,
1495 },
1496 {
1497 .procname = "watermark_boost_factor",
1498 .data = &watermark_boost_factor,
1499 .maxlen = sizeof(watermark_boost_factor),
1500 .mode = 0644,
1501 .proc_handler = watermark_boost_factor_sysctl_handler,
1502 .extra1 = &zero,
1503 },
1504 {
1505 .procname = "watermark_scale_factor",
1506 .data = &watermark_scale_factor,
1507 .maxlen = sizeof(watermark_scale_factor),
1508 .mode = 0644,
1509 .proc_handler = watermark_scale_factor_sysctl_handler,
1510 .extra1 = &one,
1511 .extra2 = &one_thousand,
1512 },
1513 {
1514 .procname = "percpu_pagelist_fraction",
1515 .data = &percpu_pagelist_fraction,
1516 .maxlen = sizeof(percpu_pagelist_fraction),
1517 .mode = 0644,
1518 .proc_handler = percpu_pagelist_fraction_sysctl_handler,
1519 .extra1 = &zero,
1520 },
1521 #ifdef CONFIG_MMU
1522 {
1523 .procname = "max_map_count",
1524 .data = &sysctl_max_map_count,
1525 .maxlen = sizeof(sysctl_max_map_count),
1526 .mode = 0644,
1527 .proc_handler = proc_dointvec_minmax,
1528 .extra1 = &zero,
1529 },
1530 #else
1531 {
1532 .procname = "nr_trim_pages",
1533 .data = &sysctl_nr_trim_pages,
1534 .maxlen = sizeof(sysctl_nr_trim_pages),
1535 .mode = 0644,
1536 .proc_handler = proc_dointvec_minmax,
1537 .extra1 = &zero,
1538 },
1539 #endif
1540 {
1541 .procname = "laptop_mode",
1542 .data = &laptop_mode,
1543 .maxlen = sizeof(laptop_mode),
1544 .mode = 0644,
1545 .proc_handler = proc_dointvec_jiffies,
1546 },
1547 {
1548 .procname = "block_dump",
1549 .data = &block_dump,
1550 .maxlen = sizeof(block_dump),
1551 .mode = 0644,
1552 .proc_handler = proc_dointvec,
1553 .extra1 = &zero,
1554 },
1555 {
1556 .procname = "vfs_cache_pressure",
1557 .data = &sysctl_vfs_cache_pressure,
1558 .maxlen = sizeof(sysctl_vfs_cache_pressure),
1559 .mode = 0644,
1560 .proc_handler = proc_dointvec,
1561 .extra1 = &zero,
1562 },
1563 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1564 {
1565 .procname = "legacy_va_layout",
1566 .data = &sysctl_legacy_va_layout,
1567 .maxlen = sizeof(sysctl_legacy_va_layout),
1568 .mode = 0644,
1569 .proc_handler = proc_dointvec,
1570 .extra1 = &zero,
1571 },
1572 #endif
1573 #ifdef CONFIG_NUMA
1574 {
1575 .procname = "zone_reclaim_mode",
1576 .data = &node_reclaim_mode,
1577 .maxlen = sizeof(node_reclaim_mode),
1578 .mode = 0644,
1579 .proc_handler = proc_dointvec,
1580 .extra1 = &zero,
1581 },
1582 {
1583 .procname = "min_unmapped_ratio",
1584 .data = &sysctl_min_unmapped_ratio,
1585 .maxlen = sizeof(sysctl_min_unmapped_ratio),
1586 .mode = 0644,
1587 .proc_handler = sysctl_min_unmapped_ratio_sysctl_handler,
1588 .extra1 = &zero,
1589 .extra2 = &one_hundred,
1590 },
1591 {
1592 .procname = "min_slab_ratio",
1593 .data = &sysctl_min_slab_ratio,
1594 .maxlen = sizeof(sysctl_min_slab_ratio),
1595 .mode = 0644,
1596 .proc_handler = sysctl_min_slab_ratio_sysctl_handler,
1597 .extra1 = &zero,
1598 .extra2 = &one_hundred,
1599 },
1600 #endif
1601 #ifdef CONFIG_SMP
1602 {
1603 .procname = "stat_interval",
1604 .data = &sysctl_stat_interval,
1605 .maxlen = sizeof(sysctl_stat_interval),
1606 .mode = 0644,
1607 .proc_handler = proc_dointvec_jiffies,
1608 },
1609 {
1610 .procname = "stat_refresh",
1611 .data = NULL,
1612 .maxlen = 0,
1613 .mode = 0600,
1614 .proc_handler = vmstat_refresh,
1615 },
1616 #endif
1617 #ifdef CONFIG_MMU
1618 {
1619 .procname = "mmap_min_addr",
1620 .data = &dac_mmap_min_addr,
1621 .maxlen = sizeof(unsigned long),
1622 .mode = 0644,
1623 .proc_handler = mmap_min_addr_handler,
1624 },
1625 #endif
1626 #ifdef CONFIG_NUMA
1627 {
1628 .procname = "numa_zonelist_order",
1629 .data = &numa_zonelist_order,
1630 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1631 .mode = 0644,
1632 .proc_handler = numa_zonelist_order_handler,
1633 },
1634 #endif
1635 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1636 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1637 {
1638 .procname = "vdso_enabled",
1639 #ifdef CONFIG_X86_32
1640 .data = &vdso32_enabled,
1641 .maxlen = sizeof(vdso32_enabled),
1642 #else
1643 .data = &vdso_enabled,
1644 .maxlen = sizeof(vdso_enabled),
1645 #endif
1646 .mode = 0644,
1647 .proc_handler = proc_dointvec,
1648 .extra1 = &zero,
1649 },
1650 #endif
1651 #ifdef CONFIG_HIGHMEM
1652 {
1653 .procname = "highmem_is_dirtyable",
1654 .data = &vm_highmem_is_dirtyable,
1655 .maxlen = sizeof(vm_highmem_is_dirtyable),
1656 .mode = 0644,
1657 .proc_handler = proc_dointvec_minmax,
1658 .extra1 = &zero,
1659 .extra2 = &one,
1660 },
1661 #endif
1662 #ifdef CONFIG_MEMORY_FAILURE
1663 {
1664 .procname = "memory_failure_early_kill",
1665 .data = &sysctl_memory_failure_early_kill,
1666 .maxlen = sizeof(sysctl_memory_failure_early_kill),
1667 .mode = 0644,
1668 .proc_handler = proc_dointvec_minmax,
1669 .extra1 = &zero,
1670 .extra2 = &one,
1671 },
1672 {
1673 .procname = "memory_failure_recovery",
1674 .data = &sysctl_memory_failure_recovery,
1675 .maxlen = sizeof(sysctl_memory_failure_recovery),
1676 .mode = 0644,
1677 .proc_handler = proc_dointvec_minmax,
1678 .extra1 = &zero,
1679 .extra2 = &one,
1680 },
1681 #endif
1682 {
1683 .procname = "user_reserve_kbytes",
1684 .data = &sysctl_user_reserve_kbytes,
1685 .maxlen = sizeof(sysctl_user_reserve_kbytes),
1686 .mode = 0644,
1687 .proc_handler = proc_doulongvec_minmax,
1688 },
1689 {
1690 .procname = "admin_reserve_kbytes",
1691 .data = &sysctl_admin_reserve_kbytes,
1692 .maxlen = sizeof(sysctl_admin_reserve_kbytes),
1693 .mode = 0644,
1694 .proc_handler = proc_doulongvec_minmax,
1695 },
1696 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1697 {
1698 .procname = "mmap_rnd_bits",
1699 .data = &mmap_rnd_bits,
1700 .maxlen = sizeof(mmap_rnd_bits),
1701 .mode = 0600,
1702 .proc_handler = proc_dointvec_minmax,
1703 .extra1 = (void *)&mmap_rnd_bits_min,
1704 .extra2 = (void *)&mmap_rnd_bits_max,
1705 },
1706 #endif
1707 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1708 {
1709 .procname = "mmap_rnd_compat_bits",
1710 .data = &mmap_rnd_compat_bits,
1711 .maxlen = sizeof(mmap_rnd_compat_bits),
1712 .mode = 0600,
1713 .proc_handler = proc_dointvec_minmax,
1714 .extra1 = (void *)&mmap_rnd_compat_bits_min,
1715 .extra2 = (void *)&mmap_rnd_compat_bits_max,
1716 },
1717 #endif
1718 #ifdef CONFIG_USERFAULTFD
1719 {
1720 .procname = "unprivileged_userfaultfd",
1721 .data = &sysctl_unprivileged_userfaultfd,
1722 .maxlen = sizeof(sysctl_unprivileged_userfaultfd),
1723 .mode = 0644,
1724 .proc_handler = proc_dointvec_minmax,
1725 .extra1 = &zero,
1726 .extra2 = &one,
1727 },
1728 #endif
1729 { }
1730 };
1731
1732 static struct ctl_table fs_table[] = {
1733 {
1734 .procname = "inode-nr",
1735 .data = &inodes_stat,
1736 .maxlen = 2*sizeof(long),
1737 .mode = 0444,
1738 .proc_handler = proc_nr_inodes,
1739 },
1740 {
1741 .procname = "inode-state",
1742 .data = &inodes_stat,
1743 .maxlen = 7*sizeof(long),
1744 .mode = 0444,
1745 .proc_handler = proc_nr_inodes,
1746 },
1747 {
1748 .procname = "file-nr",
1749 .data = &files_stat,
1750 .maxlen = sizeof(files_stat),
1751 .mode = 0444,
1752 .proc_handler = proc_nr_files,
1753 },
1754 {
1755 .procname = "file-max",
1756 .data = &files_stat.max_files,
1757 .maxlen = sizeof(files_stat.max_files),
1758 .mode = 0644,
1759 .proc_handler = proc_doulongvec_minmax,
1760 .extra1 = &zero_ul,
1761 .extra2 = &long_max,
1762 },
1763 {
1764 .procname = "nr_open",
1765 .data = &sysctl_nr_open,
1766 .maxlen = sizeof(unsigned int),
1767 .mode = 0644,
1768 .proc_handler = proc_dointvec_minmax,
1769 .extra1 = &sysctl_nr_open_min,
1770 .extra2 = &sysctl_nr_open_max,
1771 },
1772 {
1773 .procname = "dentry-state",
1774 .data = &dentry_stat,
1775 .maxlen = 6*sizeof(long),
1776 .mode = 0444,
1777 .proc_handler = proc_nr_dentry,
1778 },
1779 {
1780 .procname = "overflowuid",
1781 .data = &fs_overflowuid,
1782 .maxlen = sizeof(int),
1783 .mode = 0644,
1784 .proc_handler = proc_dointvec_minmax,
1785 .extra1 = &minolduid,
1786 .extra2 = &maxolduid,
1787 },
1788 {
1789 .procname = "overflowgid",
1790 .data = &fs_overflowgid,
1791 .maxlen = sizeof(int),
1792 .mode = 0644,
1793 .proc_handler = proc_dointvec_minmax,
1794 .extra1 = &minolduid,
1795 .extra2 = &maxolduid,
1796 },
1797 #ifdef CONFIG_FILE_LOCKING
1798 {
1799 .procname = "leases-enable",
1800 .data = &leases_enable,
1801 .maxlen = sizeof(int),
1802 .mode = 0644,
1803 .proc_handler = proc_dointvec,
1804 },
1805 #endif
1806 #ifdef CONFIG_DNOTIFY
1807 {
1808 .procname = "dir-notify-enable",
1809 .data = &dir_notify_enable,
1810 .maxlen = sizeof(int),
1811 .mode = 0644,
1812 .proc_handler = proc_dointvec,
1813 },
1814 #endif
1815 #ifdef CONFIG_MMU
1816 #ifdef CONFIG_FILE_LOCKING
1817 {
1818 .procname = "lease-break-time",
1819 .data = &lease_break_time,
1820 .maxlen = sizeof(int),
1821 .mode = 0644,
1822 .proc_handler = proc_dointvec,
1823 },
1824 #endif
1825 #ifdef CONFIG_AIO
1826 {
1827 .procname = "aio-nr",
1828 .data = &aio_nr,
1829 .maxlen = sizeof(aio_nr),
1830 .mode = 0444,
1831 .proc_handler = proc_doulongvec_minmax,
1832 },
1833 {
1834 .procname = "aio-max-nr",
1835 .data = &aio_max_nr,
1836 .maxlen = sizeof(aio_max_nr),
1837 .mode = 0644,
1838 .proc_handler = proc_doulongvec_minmax,
1839 },
1840 #endif /* CONFIG_AIO */
1841 #ifdef CONFIG_INOTIFY_USER
1842 {
1843 .procname = "inotify",
1844 .mode = 0555,
1845 .child = inotify_table,
1846 },
1847 #endif
1848 #ifdef CONFIG_EPOLL
1849 {
1850 .procname = "epoll",
1851 .mode = 0555,
1852 .child = epoll_table,
1853 },
1854 #endif
1855 #endif
1856 {
1857 .procname = "protected_symlinks",
1858 .data = &sysctl_protected_symlinks,
1859 .maxlen = sizeof(int),
1860 .mode = 0600,
1861 .proc_handler = proc_dointvec_minmax,
1862 .extra1 = &zero,
1863 .extra2 = &one,
1864 },
1865 {
1866 .procname = "protected_hardlinks",
1867 .data = &sysctl_protected_hardlinks,
1868 .maxlen = sizeof(int),
1869 .mode = 0600,
1870 .proc_handler = proc_dointvec_minmax,
1871 .extra1 = &zero,
1872 .extra2 = &one,
1873 },
1874 {
1875 .procname = "protected_fifos",
1876 .data = &sysctl_protected_fifos,
1877 .maxlen = sizeof(int),
1878 .mode = 0600,
1879 .proc_handler = proc_dointvec_minmax,
1880 .extra1 = &zero,
1881 .extra2 = &two,
1882 },
1883 {
1884 .procname = "protected_regular",
1885 .data = &sysctl_protected_regular,
1886 .maxlen = sizeof(int),
1887 .mode = 0600,
1888 .proc_handler = proc_dointvec_minmax,
1889 .extra1 = &zero,
1890 .extra2 = &two,
1891 },
1892 {
1893 .procname = "suid_dumpable",
1894 .data = &suid_dumpable,
1895 .maxlen = sizeof(int),
1896 .mode = 0644,
1897 .proc_handler = proc_dointvec_minmax_coredump,
1898 .extra1 = &zero,
1899 .extra2 = &two,
1900 },
1901 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1902 {
1903 .procname = "binfmt_misc",
1904 .mode = 0555,
1905 .child = sysctl_mount_point,
1906 },
1907 #endif
1908 {
1909 .procname = "pipe-max-size",
1910 .data = &pipe_max_size,
1911 .maxlen = sizeof(pipe_max_size),
1912 .mode = 0644,
1913 .proc_handler = proc_dopipe_max_size,
1914 },
1915 {
1916 .procname = "pipe-user-pages-hard",
1917 .data = &pipe_user_pages_hard,
1918 .maxlen = sizeof(pipe_user_pages_hard),
1919 .mode = 0644,
1920 .proc_handler = proc_doulongvec_minmax,
1921 },
1922 {
1923 .procname = "pipe-user-pages-soft",
1924 .data = &pipe_user_pages_soft,
1925 .maxlen = sizeof(pipe_user_pages_soft),
1926 .mode = 0644,
1927 .proc_handler = proc_doulongvec_minmax,
1928 },
1929 {
1930 .procname = "mount-max",
1931 .data = &sysctl_mount_max,
1932 .maxlen = sizeof(unsigned int),
1933 .mode = 0644,
1934 .proc_handler = proc_dointvec_minmax,
1935 .extra1 = &one,
1936 },
1937 { }
1938 };
1939
1940 static struct ctl_table debug_table[] = {
1941 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1942 {
1943 .procname = "exception-trace",
1944 .data = &show_unhandled_signals,
1945 .maxlen = sizeof(int),
1946 .mode = 0644,
1947 .proc_handler = proc_dointvec
1948 },
1949 #endif
1950 #if defined(CONFIG_OPTPROBES)
1951 {
1952 .procname = "kprobes-optimization",
1953 .data = &sysctl_kprobes_optimization,
1954 .maxlen = sizeof(int),
1955 .mode = 0644,
1956 .proc_handler = proc_kprobes_optimization_handler,
1957 .extra1 = &zero,
1958 .extra2 = &one,
1959 },
1960 #endif
1961 { }
1962 };
1963
1964 static struct ctl_table dev_table[] = {
1965 { }
1966 };
1967
1968 int __init sysctl_init(void)
1969 {
1970 struct ctl_table_header *hdr;
1971
1972 hdr = register_sysctl_table(sysctl_base_table);
1973 kmemleak_not_leak(hdr);
1974 return 0;
1975 }
1976
1977 #endif /* CONFIG_SYSCTL */
1978
1979 /*
1980 * /proc/sys support
1981 */
1982
1983 #ifdef CONFIG_PROC_SYSCTL
1984
1985 static int _proc_do_string(char *data, int maxlen, int write,
1986 char __user *buffer,
1987 size_t *lenp, loff_t *ppos)
1988 {
1989 size_t len;
1990 char __user *p;
1991 char c;
1992
1993 if (!data || !maxlen || !*lenp) {
1994 *lenp = 0;
1995 return 0;
1996 }
1997
1998 if (write) {
1999 if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
2000 /* Only continue writes not past the end of buffer. */
2001 len = strlen(data);
2002 if (len > maxlen - 1)
2003 len = maxlen - 1;
2004
2005 if (*ppos > len)
2006 return 0;
2007 len = *ppos;
2008 } else {
2009 /* Start writing from beginning of buffer. */
2010 len = 0;
2011 }
2012
2013 *ppos += *lenp;
2014 p = buffer;
2015 while ((p - buffer) < *lenp && len < maxlen - 1) {
2016 if (get_user(c, p++))
2017 return -EFAULT;
2018 if (c == 0 || c == '\n')
2019 break;
2020 data[len++] = c;
2021 }
2022 data[len] = 0;
2023 } else {
2024 len = strlen(data);
2025 if (len > maxlen)
2026 len = maxlen;
2027
2028 if (*ppos > len) {
2029 *lenp = 0;
2030 return 0;
2031 }
2032
2033 data += *ppos;
2034 len -= *ppos;
2035
2036 if (len > *lenp)
2037 len = *lenp;
2038 if (len)
2039 if (copy_to_user(buffer, data, len))
2040 return -EFAULT;
2041 if (len < *lenp) {
2042 if (put_user('\n', buffer + len))
2043 return -EFAULT;
2044 len++;
2045 }
2046 *lenp = len;
2047 *ppos += len;
2048 }
2049 return 0;
2050 }
2051
2052 static void warn_sysctl_write(struct ctl_table *table)
2053 {
2054 pr_warn_once("%s wrote to %s when file position was not 0!\n"
2055 "This will not be supported in the future. To silence this\n"
2056 "warning, set kernel.sysctl_writes_strict = -1\n",
2057 current->comm, table->procname);
2058 }
2059
2060 /**
2061 * proc_first_pos_non_zero_ignore - check if first position is allowed
2062 * @ppos: file position
2063 * @table: the sysctl table
2064 *
2065 * Returns true if the first position is non-zero and the sysctl_writes_strict
2066 * mode indicates this is not allowed for numeric input types. String proc
2067 * handlers can ignore the return value.
2068 */
2069 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
2070 struct ctl_table *table)
2071 {
2072 if (!*ppos)
2073 return false;
2074
2075 switch (sysctl_writes_strict) {
2076 case SYSCTL_WRITES_STRICT:
2077 return true;
2078 case SYSCTL_WRITES_WARN:
2079 warn_sysctl_write(table);
2080 return false;
2081 default:
2082 return false;
2083 }
2084 }
2085
2086 /**
2087 * proc_dostring - read a string sysctl
2088 * @table: the sysctl table
2089 * @write: %TRUE if this is a write to the sysctl file
2090 * @buffer: the user buffer
2091 * @lenp: the size of the user buffer
2092 * @ppos: file position
2093 *
2094 * Reads/writes a string from/to the user buffer. If the kernel
2095 * buffer provided is not large enough to hold the string, the
2096 * string is truncated. The copied string is %NULL-terminated.
2097 * If the string is being read by the user process, it is copied
2098 * and a newline '\n' is added. It is truncated if the buffer is
2099 * not large enough.
2100 *
2101 * Returns 0 on success.
2102 */
2103 int proc_dostring(struct ctl_table *table, int write,
2104 void __user *buffer, size_t *lenp, loff_t *ppos)
2105 {
2106 if (write)
2107 proc_first_pos_non_zero_ignore(ppos, table);
2108
2109 return _proc_do_string((char *)(table->data), table->maxlen, write,
2110 (char __user *)buffer, lenp, ppos);
2111 }
2112
2113 static size_t proc_skip_spaces(char **buf)
2114 {
2115 size_t ret;
2116 char *tmp = skip_spaces(*buf);
2117 ret = tmp - *buf;
2118 *buf = tmp;
2119 return ret;
2120 }
2121
2122 static void proc_skip_char(char **buf, size_t *size, const char v)
2123 {
2124 while (*size) {
2125 if (**buf != v)
2126 break;
2127 (*size)--;
2128 (*buf)++;
2129 }
2130 }
2131
2132 /**
2133 * strtoul_lenient - parse an ASCII formatted integer from a buffer and only
2134 * fail on overflow
2135 *
2136 * @cp: kernel buffer containing the string to parse
2137 * @endp: pointer to store the trailing characters
2138 * @base: the base to use
2139 * @res: where the parsed integer will be stored
2140 *
2141 * In case of success 0 is returned and @res will contain the parsed integer,
2142 * @endp will hold any trailing characters.
2143 * This function will fail the parse on overflow. If there wasn't an overflow
2144 * the function will defer the decision what characters count as invalid to the
2145 * caller.
2146 */
2147 static int strtoul_lenient(const char *cp, char **endp, unsigned int base,
2148 unsigned long *res)
2149 {
2150 unsigned long long result;
2151 unsigned int rv;
2152
2153 cp = _parse_integer_fixup_radix(cp, &base);
2154 rv = _parse_integer(cp, base, &result);
2155 if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result))
2156 return -ERANGE;
2157
2158 cp += rv;
2159
2160 if (endp)
2161 *endp = (char *)cp;
2162
2163 *res = (unsigned long)result;
2164 return 0;
2165 }
2166
2167 #define TMPBUFLEN 22
2168 /**
2169 * proc_get_long - reads an ASCII formatted integer from a user buffer
2170 *
2171 * @buf: a kernel buffer
2172 * @size: size of the kernel buffer
2173 * @val: this is where the number will be stored
2174 * @neg: set to %TRUE if number is negative
2175 * @perm_tr: a vector which contains the allowed trailers
2176 * @perm_tr_len: size of the perm_tr vector
2177 * @tr: pointer to store the trailer character
2178 *
2179 * In case of success %0 is returned and @buf and @size are updated with
2180 * the amount of bytes read. If @tr is non-NULL and a trailing
2181 * character exists (size is non-zero after returning from this
2182 * function), @tr is updated with the trailing character.
2183 */
2184 static int proc_get_long(char **buf, size_t *size,
2185 unsigned long *val, bool *neg,
2186 const char *perm_tr, unsigned perm_tr_len, char *tr)
2187 {
2188 int len;
2189 char *p, tmp[TMPBUFLEN];
2190
2191 if (!*size)
2192 return -EINVAL;
2193
2194 len = *size;
2195 if (len > TMPBUFLEN - 1)
2196 len = TMPBUFLEN - 1;
2197
2198 memcpy(tmp, *buf, len);
2199
2200 tmp[len] = 0;
2201 p = tmp;
2202 if (*p == '-' && *size > 1) {
2203 *neg = true;
2204 p++;
2205 } else
2206 *neg = false;
2207 if (!isdigit(*p))
2208 return -EINVAL;
2209
2210 if (strtoul_lenient(p, &p, 0, val))
2211 return -EINVAL;
2212
2213 len = p - tmp;
2214
2215 /* We don't know if the next char is whitespace thus we may accept
2216 * invalid integers (e.g. 1234...a) or two integers instead of one
2217 * (e.g. 123...1). So lets not allow such large numbers. */
2218 if (len == TMPBUFLEN - 1)
2219 return -EINVAL;
2220
2221 if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2222 return -EINVAL;
2223
2224 if (tr && (len < *size))
2225 *tr = *p;
2226
2227 *buf += len;
2228 *size -= len;
2229
2230 return 0;
2231 }
2232
2233 /**
2234 * proc_put_long - converts an integer to a decimal ASCII formatted string
2235 *
2236 * @buf: the user buffer
2237 * @size: the size of the user buffer
2238 * @val: the integer to be converted
2239 * @neg: sign of the number, %TRUE for negative
2240 *
2241 * In case of success %0 is returned and @buf and @size are updated with
2242 * the amount of bytes written.
2243 */
2244 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2245 bool neg)
2246 {
2247 int len;
2248 char tmp[TMPBUFLEN], *p = tmp;
2249
2250 sprintf(p, "%s%lu", neg ? "-" : "", val);
2251 len = strlen(tmp);
2252 if (len > *size)
2253 len = *size;
2254 if (copy_to_user(*buf, tmp, len))
2255 return -EFAULT;
2256 *size -= len;
2257 *buf += len;
2258 return 0;
2259 }
2260 #undef TMPBUFLEN
2261
2262 static int proc_put_char(void __user **buf, size_t *size, char c)
2263 {
2264 if (*size) {
2265 char __user **buffer = (char __user **)buf;
2266 if (put_user(c, *buffer))
2267 return -EFAULT;
2268 (*size)--, (*buffer)++;
2269 *buf = *buffer;
2270 }
2271 return 0;
2272 }
2273
2274 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2275 int *valp,
2276 int write, void *data)
2277 {
2278 if (write) {
2279 if (*negp) {
2280 if (*lvalp > (unsigned long) INT_MAX + 1)
2281 return -EINVAL;
2282 *valp = -*lvalp;
2283 } else {
2284 if (*lvalp > (unsigned long) INT_MAX)
2285 return -EINVAL;
2286 *valp = *lvalp;
2287 }
2288 } else {
2289 int val = *valp;
2290 if (val < 0) {
2291 *negp = true;
2292 *lvalp = -(unsigned long)val;
2293 } else {
2294 *negp = false;
2295 *lvalp = (unsigned long)val;
2296 }
2297 }
2298 return 0;
2299 }
2300
2301 static int do_proc_douintvec_conv(unsigned long *lvalp,
2302 unsigned int *valp,
2303 int write, void *data)
2304 {
2305 if (write) {
2306 if (*lvalp > UINT_MAX)
2307 return -EINVAL;
2308 *valp = *lvalp;
2309 } else {
2310 unsigned int val = *valp;
2311 *lvalp = (unsigned long)val;
2312 }
2313 return 0;
2314 }
2315
2316 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2317
2318 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2319 int write, void __user *buffer,
2320 size_t *lenp, loff_t *ppos,
2321 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2322 int write, void *data),
2323 void *data)
2324 {
2325 int *i, vleft, first = 1, err = 0;
2326 size_t left;
2327 char *kbuf = NULL, *p;
2328
2329 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2330 *lenp = 0;
2331 return 0;
2332 }
2333
2334 i = (int *) tbl_data;
2335 vleft = table->maxlen / sizeof(*i);
2336 left = *lenp;
2337
2338 if (!conv)
2339 conv = do_proc_dointvec_conv;
2340
2341 if (write) {
2342 if (proc_first_pos_non_zero_ignore(ppos, table))
2343 goto out;
2344
2345 if (left > PAGE_SIZE - 1)
2346 left = PAGE_SIZE - 1;
2347 p = kbuf = memdup_user_nul(buffer, left);
2348 if (IS_ERR(kbuf))
2349 return PTR_ERR(kbuf);
2350 }
2351
2352 for (; left && vleft--; i++, first=0) {
2353 unsigned long lval;
2354 bool neg;
2355
2356 if (write) {
2357 left -= proc_skip_spaces(&p);
2358
2359 if (!left)
2360 break;
2361 err = proc_get_long(&p, &left, &lval, &neg,
2362 proc_wspace_sep,
2363 sizeof(proc_wspace_sep), NULL);
2364 if (err)
2365 break;
2366 if (conv(&neg, &lval, i, 1, data)) {
2367 err = -EINVAL;
2368 break;
2369 }
2370 } else {
2371 if (conv(&neg, &lval, i, 0, data)) {
2372 err = -EINVAL;
2373 break;
2374 }
2375 if (!first)
2376 err = proc_put_char(&buffer, &left, '\t');
2377 if (err)
2378 break;
2379 err = proc_put_long(&buffer, &left, lval, neg);
2380 if (err)
2381 break;
2382 }
2383 }
2384
2385 if (!write && !first && left && !err)
2386 err = proc_put_char(&buffer, &left, '\n');
2387 if (write && !err && left)
2388 left -= proc_skip_spaces(&p);
2389 if (write) {
2390 kfree(kbuf);
2391 if (first)
2392 return err ? : -EINVAL;
2393 }
2394 *lenp -= left;
2395 out:
2396 *ppos += *lenp;
2397 return err;
2398 }
2399
2400 static int do_proc_dointvec(struct ctl_table *table, int write,
2401 void __user *buffer, size_t *lenp, loff_t *ppos,
2402 int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2403 int write, void *data),
2404 void *data)
2405 {
2406 return __do_proc_dointvec(table->data, table, write,
2407 buffer, lenp, ppos, conv, data);
2408 }
2409
2410 static int do_proc_douintvec_w(unsigned int *tbl_data,
2411 struct ctl_table *table,
2412 void __user *buffer,
2413 size_t *lenp, loff_t *ppos,
2414 int (*conv)(unsigned long *lvalp,
2415 unsigned int *valp,
2416 int write, void *data),
2417 void *data)
2418 {
2419 unsigned long lval;
2420 int err = 0;
2421 size_t left;
2422 bool neg;
2423 char *kbuf = NULL, *p;
2424
2425 left = *lenp;
2426
2427 if (proc_first_pos_non_zero_ignore(ppos, table))
2428 goto bail_early;
2429
2430 if (left > PAGE_SIZE - 1)
2431 left = PAGE_SIZE - 1;
2432
2433 p = kbuf = memdup_user_nul(buffer, left);
2434 if (IS_ERR(kbuf))
2435 return -EINVAL;
2436
2437 left -= proc_skip_spaces(&p);
2438 if (!left) {
2439 err = -EINVAL;
2440 goto out_free;
2441 }
2442
2443 err = proc_get_long(&p, &left, &lval, &neg,
2444 proc_wspace_sep,
2445 sizeof(proc_wspace_sep), NULL);
2446 if (err || neg) {
2447 err = -EINVAL;
2448 goto out_free;
2449 }
2450
2451 if (conv(&lval, tbl_data, 1, data)) {
2452 err = -EINVAL;
2453 goto out_free;
2454 }
2455
2456 if (!err && left)
2457 left -= proc_skip_spaces(&p);
2458
2459 out_free:
2460 kfree(kbuf);
2461 if (err)
2462 return -EINVAL;
2463
2464 return 0;
2465
2466 /* This is in keeping with old __do_proc_dointvec() */
2467 bail_early:
2468 *ppos += *lenp;
2469 return err;
2470 }
2471
2472 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2473 size_t *lenp, loff_t *ppos,
2474 int (*conv)(unsigned long *lvalp,
2475 unsigned int *valp,
2476 int write, void *data),
2477 void *data)
2478 {
2479 unsigned long lval;
2480 int err = 0;
2481 size_t left;
2482
2483 left = *lenp;
2484
2485 if (conv(&lval, tbl_data, 0, data)) {
2486 err = -EINVAL;
2487 goto out;
2488 }
2489
2490 err = proc_put_long(&buffer, &left, lval, false);
2491 if (err || !left)
2492 goto out;
2493
2494 err = proc_put_char(&buffer, &left, '\n');
2495
2496 out:
2497 *lenp -= left;
2498 *ppos += *lenp;
2499
2500 return err;
2501 }
2502
2503 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2504 int write, void __user *buffer,
2505 size_t *lenp, loff_t *ppos,
2506 int (*conv)(unsigned long *lvalp,
2507 unsigned int *valp,
2508 int write, void *data),
2509 void *data)
2510 {
2511 unsigned int *i, vleft;
2512
2513 if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2514 *lenp = 0;
2515 return 0;
2516 }
2517
2518 i = (unsigned int *) tbl_data;
2519 vleft = table->maxlen / sizeof(*i);
2520
2521 /*
2522 * Arrays are not supported, keep this simple. *Do not* add
2523 * support for them.
2524 */
2525 if (vleft != 1) {
2526 *lenp = 0;
2527 return -EINVAL;
2528 }
2529
2530 if (!conv)
2531 conv = do_proc_douintvec_conv;
2532
2533 if (write)
2534 return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2535 conv, data);
2536 return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2537 }
2538
2539 static int do_proc_douintvec(struct ctl_table *table, int write,
2540 void __user *buffer, size_t *lenp, loff_t *ppos,
2541 int (*conv)(unsigned long *lvalp,
2542 unsigned int *valp,
2543 int write, void *data),
2544 void *data)
2545 {
2546 return __do_proc_douintvec(table->data, table, write,
2547 buffer, lenp, ppos, conv, data);
2548 }
2549
2550 /**
2551 * proc_dointvec - read a vector of integers
2552 * @table: the sysctl table
2553 * @write: %TRUE if this is a write to the sysctl file
2554 * @buffer: the user buffer
2555 * @lenp: the size of the user buffer
2556 * @ppos: file position
2557 *
2558 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2559 * values from/to the user buffer, treated as an ASCII string.
2560 *
2561 * Returns 0 on success.
2562 */
2563 int proc_dointvec(struct ctl_table *table, int write,
2564 void __user *buffer, size_t *lenp, loff_t *ppos)
2565 {
2566 return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2567 }
2568
2569 /**
2570 * proc_douintvec - read a vector of unsigned integers
2571 * @table: the sysctl table
2572 * @write: %TRUE if this is a write to the sysctl file
2573 * @buffer: the user buffer
2574 * @lenp: the size of the user buffer
2575 * @ppos: file position
2576 *
2577 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2578 * values from/to the user buffer, treated as an ASCII string.
2579 *
2580 * Returns 0 on success.
2581 */
2582 int proc_douintvec(struct ctl_table *table, int write,
2583 void __user *buffer, size_t *lenp, loff_t *ppos)
2584 {
2585 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2586 do_proc_douintvec_conv, NULL);
2587 }
2588
2589 /*
2590 * Taint values can only be increased
2591 * This means we can safely use a temporary.
2592 */
2593 static int proc_taint(struct ctl_table *table, int write,
2594 void __user *buffer, size_t *lenp, loff_t *ppos)
2595 {
2596 struct ctl_table t;
2597 unsigned long tmptaint = get_taint();
2598 int err;
2599
2600 if (write && !capable(CAP_SYS_ADMIN))
2601 return -EPERM;
2602
2603 t = *table;
2604 t.data = &tmptaint;
2605 err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2606 if (err < 0)
2607 return err;
2608
2609 if (write) {
2610 /*
2611 * Poor man's atomic or. Not worth adding a primitive
2612 * to everyone's atomic.h for this
2613 */
2614 int i;
2615 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2616 if ((tmptaint >> i) & 1)
2617 add_taint(i, LOCKDEP_STILL_OK);
2618 }
2619 }
2620
2621 return err;
2622 }
2623
2624 #ifdef CONFIG_PRINTK
2625 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2626 void __user *buffer, size_t *lenp, loff_t *ppos)
2627 {
2628 if (write && !capable(CAP_SYS_ADMIN))
2629 return -EPERM;
2630
2631 return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2632 }
2633 #endif
2634
2635 /**
2636 * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
2637 * @min: pointer to minimum allowable value
2638 * @max: pointer to maximum allowable value
2639 *
2640 * The do_proc_dointvec_minmax_conv_param structure provides the
2641 * minimum and maximum values for doing range checking for those sysctl
2642 * parameters that use the proc_dointvec_minmax() handler.
2643 */
2644 struct do_proc_dointvec_minmax_conv_param {
2645 int *min;
2646 int *max;
2647 };
2648
2649 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2650 int *valp,
2651 int write, void *data)
2652 {
2653 int tmp, ret;
2654 struct do_proc_dointvec_minmax_conv_param *param = data;
2655 /*
2656 * If writing, first do so via a temporary local int so we can
2657 * bounds-check it before touching *valp.
2658 */
2659 int *ip = write ? &tmp : valp;
2660
2661 ret = do_proc_dointvec_conv(negp, lvalp, ip, write, data);
2662 if (ret)
2663 return ret;
2664
2665 if (write) {
2666 if ((param->min && *param->min > tmp) ||
2667 (param->max && *param->max < tmp))
2668 return -EINVAL;
2669 *valp = tmp;
2670 }
2671
2672 return 0;
2673 }
2674
2675 /**
2676 * proc_dointvec_minmax - read a vector of integers with min/max values
2677 * @table: the sysctl table
2678 * @write: %TRUE if this is a write to the sysctl file
2679 * @buffer: the user buffer
2680 * @lenp: the size of the user buffer
2681 * @ppos: file position
2682 *
2683 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2684 * values from/to the user buffer, treated as an ASCII string.
2685 *
2686 * This routine will ensure the values are within the range specified by
2687 * table->extra1 (min) and table->extra2 (max).
2688 *
2689 * Returns 0 on success or -EINVAL on write when the range check fails.
2690 */
2691 int proc_dointvec_minmax(struct ctl_table *table, int write,
2692 void __user *buffer, size_t *lenp, loff_t *ppos)
2693 {
2694 struct do_proc_dointvec_minmax_conv_param param = {
2695 .min = (int *) table->extra1,
2696 .max = (int *) table->extra2,
2697 };
2698 return do_proc_dointvec(table, write, buffer, lenp, ppos,
2699 do_proc_dointvec_minmax_conv, &param);
2700 }
2701
2702 /**
2703 * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
2704 * @min: pointer to minimum allowable value
2705 * @max: pointer to maximum allowable value
2706 *
2707 * The do_proc_douintvec_minmax_conv_param structure provides the
2708 * minimum and maximum values for doing range checking for those sysctl
2709 * parameters that use the proc_douintvec_minmax() handler.
2710 */
2711 struct do_proc_douintvec_minmax_conv_param {
2712 unsigned int *min;
2713 unsigned int *max;
2714 };
2715
2716 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2717 unsigned int *valp,
2718 int write, void *data)
2719 {
2720 int ret;
2721 unsigned int tmp;
2722 struct do_proc_douintvec_minmax_conv_param *param = data;
2723 /* write via temporary local uint for bounds-checking */
2724 unsigned int *up = write ? &tmp : valp;
2725
2726 ret = do_proc_douintvec_conv(lvalp, up, write, data);
2727 if (ret)
2728 return ret;
2729
2730 if (write) {
2731 if ((param->min && *param->min > tmp) ||
2732 (param->max && *param->max < tmp))
2733 return -ERANGE;
2734
2735 *valp = tmp;
2736 }
2737
2738 return 0;
2739 }
2740
2741 /**
2742 * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2743 * @table: the sysctl table
2744 * @write: %TRUE if this is a write to the sysctl file
2745 * @buffer: the user buffer
2746 * @lenp: the size of the user buffer
2747 * @ppos: file position
2748 *
2749 * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2750 * values from/to the user buffer, treated as an ASCII string. Negative
2751 * strings are not allowed.
2752 *
2753 * This routine will ensure the values are within the range specified by
2754 * table->extra1 (min) and table->extra2 (max). There is a final sanity
2755 * check for UINT_MAX to avoid having to support wrap around uses from
2756 * userspace.
2757 *
2758 * Returns 0 on success or -ERANGE on write when the range check fails.
2759 */
2760 int proc_douintvec_minmax(struct ctl_table *table, int write,
2761 void __user *buffer, size_t *lenp, loff_t *ppos)
2762 {
2763 struct do_proc_douintvec_minmax_conv_param param = {
2764 .min = (unsigned int *) table->extra1,
2765 .max = (unsigned int *) table->extra2,
2766 };
2767 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2768 do_proc_douintvec_minmax_conv, &param);
2769 }
2770
2771 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
2772 unsigned int *valp,
2773 int write, void *data)
2774 {
2775 if (write) {
2776 unsigned int val;
2777
2778 val = round_pipe_size(*lvalp);
2779 if (val == 0)
2780 return -EINVAL;
2781
2782 *valp = val;
2783 } else {
2784 unsigned int val = *valp;
2785 *lvalp = (unsigned long) val;
2786 }
2787
2788 return 0;
2789 }
2790
2791 static int proc_dopipe_max_size(struct ctl_table *table, int write,
2792 void __user *buffer, size_t *lenp, loff_t *ppos)
2793 {
2794 return do_proc_douintvec(table, write, buffer, lenp, ppos,
2795 do_proc_dopipe_max_size_conv, NULL);
2796 }
2797
2798 static void validate_coredump_safety(void)
2799 {
2800 #ifdef CONFIG_COREDUMP
2801 if (suid_dumpable == SUID_DUMP_ROOT &&
2802 core_pattern[0] != '/' && core_pattern[0] != '|') {
2803 printk(KERN_WARNING
2804 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2805 "Pipe handler or fully qualified core dump path required.\n"
2806 "Set kernel.core_pattern before fs.suid_dumpable.\n"
2807 );
2808 }
2809 #endif
2810 }
2811
2812 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2813 void __user *buffer, size_t *lenp, loff_t *ppos)
2814 {
2815 int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2816 if (!error)
2817 validate_coredump_safety();
2818 return error;
2819 }
2820
2821 #ifdef CONFIG_COREDUMP
2822 static int proc_dostring_coredump(struct ctl_table *table, int write,
2823 void __user *buffer, size_t *lenp, loff_t *ppos)
2824 {
2825 int error = proc_dostring(table, write, buffer, lenp, ppos);
2826 if (!error)
2827 validate_coredump_safety();
2828 return error;
2829 }
2830 #endif
2831
2832 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2833 void __user *buffer,
2834 size_t *lenp, loff_t *ppos,
2835 unsigned long convmul,
2836 unsigned long convdiv)
2837 {
2838 unsigned long *i, *min, *max;
2839 int vleft, first = 1, err = 0;
2840 size_t left;
2841 char *kbuf = NULL, *p;
2842
2843 if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2844 *lenp = 0;
2845 return 0;
2846 }
2847
2848 i = (unsigned long *) data;
2849 min = (unsigned long *) table->extra1;
2850 max = (unsigned long *) table->extra2;
2851 vleft = table->maxlen / sizeof(unsigned long);
2852 left = *lenp;
2853
2854 if (write) {
2855 if (proc_first_pos_non_zero_ignore(ppos, table))
2856 goto out;
2857
2858 if (left > PAGE_SIZE - 1)
2859 left = PAGE_SIZE - 1;
2860 p = kbuf = memdup_user_nul(buffer, left);
2861 if (IS_ERR(kbuf))
2862 return PTR_ERR(kbuf);
2863 }
2864
2865 for (; left && vleft--; i++, first = 0) {
2866 unsigned long val;
2867
2868 if (write) {
2869 bool neg;
2870
2871 left -= proc_skip_spaces(&p);
2872 if (!left)
2873 break;
2874
2875 err = proc_get_long(&p, &left, &val, &neg,
2876 proc_wspace_sep,
2877 sizeof(proc_wspace_sep), NULL);
2878 if (err)
2879 break;
2880 if (neg)
2881 continue;
2882 val = convmul * val / convdiv;
2883 if ((min && val < *min) || (max && val > *max)) {
2884 err = -EINVAL;
2885 break;
2886 }
2887 *i = val;
2888 } else {
2889 val = convdiv * (*i) / convmul;
2890 if (!first) {
2891 err = proc_put_char(&buffer, &left, '\t');
2892 if (err)
2893 break;
2894 }
2895 err = proc_put_long(&buffer, &left, val, false);
2896 if (err)
2897 break;
2898 }
2899 }
2900
2901 if (!write && !first && left && !err)
2902 err = proc_put_char(&buffer, &left, '\n');
2903 if (write && !err)
2904 left -= proc_skip_spaces(&p);
2905 if (write) {
2906 kfree(kbuf);
2907 if (first)
2908 return err ? : -EINVAL;
2909 }
2910 *lenp -= left;
2911 out:
2912 *ppos += *lenp;
2913 return err;
2914 }
2915
2916 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2917 void __user *buffer,
2918 size_t *lenp, loff_t *ppos,
2919 unsigned long convmul,
2920 unsigned long convdiv)
2921 {
2922 return __do_proc_doulongvec_minmax(table->data, table, write,
2923 buffer, lenp, ppos, convmul, convdiv);
2924 }
2925
2926 /**
2927 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2928 * @table: the sysctl table
2929 * @write: %TRUE if this is a write to the sysctl file
2930 * @buffer: the user buffer
2931 * @lenp: the size of the user buffer
2932 * @ppos: file position
2933 *
2934 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2935 * values from/to the user buffer, treated as an ASCII string.
2936 *
2937 * This routine will ensure the values are within the range specified by
2938 * table->extra1 (min) and table->extra2 (max).
2939 *
2940 * Returns 0 on success.
2941 */
2942 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2943 void __user *buffer, size_t *lenp, loff_t *ppos)
2944 {
2945 return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2946 }
2947
2948 /**
2949 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2950 * @table: the sysctl table
2951 * @write: %TRUE if this is a write to the sysctl file
2952 * @buffer: the user buffer
2953 * @lenp: the size of the user buffer
2954 * @ppos: file position
2955 *
2956 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2957 * values from/to the user buffer, treated as an ASCII string. The values
2958 * are treated as milliseconds, and converted to jiffies when they are stored.
2959 *
2960 * This routine will ensure the values are within the range specified by
2961 * table->extra1 (min) and table->extra2 (max).
2962 *
2963 * Returns 0 on success.
2964 */
2965 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2966 void __user *buffer,
2967 size_t *lenp, loff_t *ppos)
2968 {
2969 return do_proc_doulongvec_minmax(table, write, buffer,
2970 lenp, ppos, HZ, 1000l);
2971 }
2972
2973
2974 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2975 int *valp,
2976 int write, void *data)
2977 {
2978 if (write) {
2979 if (*lvalp > INT_MAX / HZ)
2980 return 1;
2981 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2982 } else {
2983 int val = *valp;
2984 unsigned long lval;
2985 if (val < 0) {
2986 *negp = true;
2987 lval = -(unsigned long)val;
2988 } else {
2989 *negp = false;
2990 lval = (unsigned long)val;
2991 }
2992 *lvalp = lval / HZ;
2993 }
2994 return 0;
2995 }
2996
2997 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2998 int *valp,
2999 int write, void *data)
3000 {
3001 if (write) {
3002 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
3003 return 1;
3004 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
3005 } else {
3006 int val = *valp;
3007 unsigned long lval;
3008 if (val < 0) {
3009 *negp = true;
3010 lval = -(unsigned long)val;
3011 } else {
3012 *negp = false;
3013 lval = (unsigned long)val;
3014 }
3015 *lvalp = jiffies_to_clock_t(lval);
3016 }
3017 return 0;
3018 }
3019
3020 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
3021 int *valp,
3022 int write, void *data)
3023 {
3024 if (write) {
3025 unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
3026
3027 if (jif > INT_MAX)
3028 return 1;
3029 *valp = (int)jif;
3030 } else {
3031 int val = *valp;
3032 unsigned long lval;
3033 if (val < 0) {
3034 *negp = true;
3035 lval = -(unsigned long)val;
3036 } else {
3037 *negp = false;
3038 lval = (unsigned long)val;
3039 }
3040 *lvalp = jiffies_to_msecs(lval);
3041 }
3042 return 0;
3043 }
3044
3045 /**
3046 * proc_dointvec_jiffies - read a vector of integers as seconds
3047 * @table: the sysctl table
3048 * @write: %TRUE if this is a write to the sysctl file
3049 * @buffer: the user buffer
3050 * @lenp: the size of the user buffer
3051 * @ppos: file position
3052 *
3053 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3054 * values from/to the user buffer, treated as an ASCII string.
3055 * The values read are assumed to be in seconds, and are converted into
3056 * jiffies.
3057 *
3058 * Returns 0 on success.
3059 */
3060 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3061 void __user *buffer, size_t *lenp, loff_t *ppos)
3062 {
3063 return do_proc_dointvec(table,write,buffer,lenp,ppos,
3064 do_proc_dointvec_jiffies_conv,NULL);
3065 }
3066
3067 /**
3068 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
3069 * @table: the sysctl table
3070 * @write: %TRUE if this is a write to the sysctl file
3071 * @buffer: the user buffer
3072 * @lenp: the size of the user buffer
3073 * @ppos: pointer to the file position
3074 *
3075 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3076 * values from/to the user buffer, treated as an ASCII string.
3077 * The values read are assumed to be in 1/USER_HZ seconds, and
3078 * are converted into jiffies.
3079 *
3080 * Returns 0 on success.
3081 */
3082 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3083 void __user *buffer, size_t *lenp, loff_t *ppos)
3084 {
3085 return do_proc_dointvec(table,write,buffer,lenp,ppos,
3086 do_proc_dointvec_userhz_jiffies_conv,NULL);
3087 }
3088
3089 /**
3090 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
3091 * @table: the sysctl table
3092 * @write: %TRUE if this is a write to the sysctl file
3093 * @buffer: the user buffer
3094 * @lenp: the size of the user buffer
3095 * @ppos: file position
3096 * @ppos: the current position in the file
3097 *
3098 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3099 * values from/to the user buffer, treated as an ASCII string.
3100 * The values read are assumed to be in 1/1000 seconds, and
3101 * are converted into jiffies.
3102 *
3103 * Returns 0 on success.
3104 */
3105 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3106 void __user *buffer, size_t *lenp, loff_t *ppos)
3107 {
3108 return do_proc_dointvec(table, write, buffer, lenp, ppos,
3109 do_proc_dointvec_ms_jiffies_conv, NULL);
3110 }
3111
3112 static int proc_do_cad_pid(struct ctl_table *table, int write,
3113 void __user *buffer, size_t *lenp, loff_t *ppos)
3114 {
3115 struct pid *new_pid;
3116 pid_t tmp;
3117 int r;
3118
3119 tmp = pid_vnr(cad_pid);
3120
3121 r = __do_proc_dointvec(&tmp, table, write, buffer,
3122 lenp, ppos, NULL, NULL);
3123 if (r || !write)
3124 return r;
3125
3126 new_pid = find_get_pid(tmp);
3127 if (!new_pid)
3128 return -ESRCH;
3129
3130 put_pid(xchg(&cad_pid, new_pid));
3131 return 0;
3132 }
3133
3134 /**
3135 * proc_do_large_bitmap - read/write from/to a large bitmap
3136 * @table: the sysctl table
3137 * @write: %TRUE if this is a write to the sysctl file
3138 * @buffer: the user buffer
3139 * @lenp: the size of the user buffer
3140 * @ppos: file position
3141 *
3142 * The bitmap is stored at table->data and the bitmap length (in bits)
3143 * in table->maxlen.
3144 *
3145 * We use a range comma separated format (e.g. 1,3-4,10-10) so that
3146 * large bitmaps may be represented in a compact manner. Writing into
3147 * the file will clear the bitmap then update it with the given input.
3148 *
3149 * Returns 0 on success.
3150 */
3151 int proc_do_large_bitmap(struct ctl_table *table, int write,
3152 void __user *buffer, size_t *lenp, loff_t *ppos)
3153 {
3154 int err = 0;
3155 bool first = 1;
3156 size_t left = *lenp;
3157 unsigned long bitmap_len = table->maxlen;
3158 unsigned long *bitmap = *(unsigned long **) table->data;
3159 unsigned long *tmp_bitmap = NULL;
3160 char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
3161
3162 if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
3163 *lenp = 0;
3164 return 0;
3165 }
3166
3167 if (write) {
3168 char *kbuf, *p;
3169 size_t skipped = 0;
3170
3171 if (left > PAGE_SIZE - 1) {
3172 left = PAGE_SIZE - 1;
3173 /* How much of the buffer we'll skip this pass */
3174 skipped = *lenp - left;
3175 }
3176
3177 p = kbuf = memdup_user_nul(buffer, left);
3178 if (IS_ERR(kbuf))
3179 return PTR_ERR(kbuf);
3180
3181 tmp_bitmap = bitmap_zalloc(bitmap_len, GFP_KERNEL);
3182 if (!tmp_bitmap) {
3183 kfree(kbuf);
3184 return -ENOMEM;
3185 }
3186 proc_skip_char(&p, &left, '\n');
3187 while (!err && left) {
3188 unsigned long val_a, val_b;
3189 bool neg;
3190 size_t saved_left;
3191
3192 /* In case we stop parsing mid-number, we can reset */
3193 saved_left = left;
3194 err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3195 sizeof(tr_a), &c);
3196 /*
3197 * If we consumed the entirety of a truncated buffer or
3198 * only one char is left (may be a "-"), then stop here,
3199 * reset, & come back for more.
3200 */
3201 if ((left <= 1) && skipped) {
3202 left = saved_left;
3203 break;
3204 }
3205
3206 if (err)
3207 break;
3208 if (val_a >= bitmap_len || neg) {
3209 err = -EINVAL;
3210 break;
3211 }
3212
3213 val_b = val_a;
3214 if (left) {
3215 p++;
3216 left--;
3217 }
3218
3219 if (c == '-') {
3220 err = proc_get_long(&p, &left, &val_b,
3221 &neg, tr_b, sizeof(tr_b),
3222 &c);
3223 /*
3224 * If we consumed all of a truncated buffer or
3225 * then stop here, reset, & come back for more.
3226 */
3227 if (!left && skipped) {
3228 left = saved_left;
3229 break;
3230 }
3231
3232 if (err)
3233 break;
3234 if (val_b >= bitmap_len || neg ||
3235 val_a > val_b) {
3236 err = -EINVAL;
3237 break;
3238 }
3239 if (left) {
3240 p++;
3241 left--;
3242 }
3243 }
3244
3245 bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3246 first = 0;
3247 proc_skip_char(&p, &left, '\n');
3248 }
3249 kfree(kbuf);
3250 left += skipped;
3251 } else {
3252 unsigned long bit_a, bit_b = 0;
3253
3254 while (left) {
3255 bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3256 if (bit_a >= bitmap_len)
3257 break;
3258 bit_b = find_next_zero_bit(bitmap, bitmap_len,
3259 bit_a + 1) - 1;
3260
3261 if (!first) {
3262 err = proc_put_char(&buffer, &left, ',');
3263 if (err)
3264 break;
3265 }
3266 err = proc_put_long(&buffer, &left, bit_a, false);
3267 if (err)
3268 break;
3269 if (bit_a != bit_b) {
3270 err = proc_put_char(&buffer, &left, '-');
3271 if (err)
3272 break;
3273 err = proc_put_long(&buffer, &left, bit_b, false);
3274 if (err)
3275 break;
3276 }
3277
3278 first = 0; bit_b++;
3279 }
3280 if (!err)
3281 err = proc_put_char(&buffer, &left, '\n');
3282 }
3283
3284 if (!err) {
3285 if (write) {
3286 if (*ppos)
3287 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3288 else
3289 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3290 }
3291 *lenp -= left;
3292 *ppos += *lenp;
3293 }
3294
3295 bitmap_free(tmp_bitmap);
3296 return err;
3297 }
3298
3299 #else /* CONFIG_PROC_SYSCTL */
3300
3301 int proc_dostring(struct ctl_table *table, int write,
3302 void __user *buffer, size_t *lenp, loff_t *ppos)
3303 {
3304 return -ENOSYS;
3305 }
3306
3307 int proc_dointvec(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_douintvec(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_dointvec_minmax(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_douintvec_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_dointvec_jiffies(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_userhz_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_ms_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_doulongvec_minmax(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_ms_jiffies_minmax(struct ctl_table *table, int write,
3356 void __user *buffer,
3357 size_t *lenp, loff_t *ppos)
3358 {
3359 return -ENOSYS;
3360 }
3361
3362 int proc_do_large_bitmap(struct ctl_table *table, int write,
3363 void __user *buffer, size_t *lenp, loff_t *ppos)
3364 {
3365 return -ENOSYS;
3366 }
3367
3368 #endif /* CONFIG_PROC_SYSCTL */
3369
3370 #if defined(CONFIG_SYSCTL)
3371 int proc_do_static_key(struct ctl_table *table, int write,
3372 void __user *buffer, size_t *lenp,
3373 loff_t *ppos)
3374 {
3375 struct static_key *key = (struct static_key *)table->data;
3376 static DEFINE_MUTEX(static_key_mutex);
3377 int val, ret;
3378 struct ctl_table tmp = {
3379 .data = &val,
3380 .maxlen = sizeof(val),
3381 .mode = table->mode,
3382 .extra1 = &zero,
3383 .extra2 = &one,
3384 };
3385
3386 if (write && !capable(CAP_SYS_ADMIN))
3387 return -EPERM;
3388
3389 mutex_lock(&static_key_mutex);
3390 val = static_key_enabled(key);
3391 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
3392 if (write && !ret) {
3393 if (val)
3394 static_key_enable(key);
3395 else
3396 static_key_disable(key);
3397 }
3398 mutex_unlock(&static_key_mutex);
3399 return ret;
3400 }
3401 #endif
3402 /*
3403 * No sense putting this after each symbol definition, twice,
3404 * exception granted :-)
3405 */
3406 EXPORT_SYMBOL(proc_dointvec);
3407 EXPORT_SYMBOL(proc_douintvec);
3408 EXPORT_SYMBOL(proc_dointvec_jiffies);
3409 EXPORT_SYMBOL(proc_dointvec_minmax);
3410 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3411 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3412 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3413 EXPORT_SYMBOL(proc_dostring);
3414 EXPORT_SYMBOL(proc_doulongvec_minmax);
3415 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3416 EXPORT_SYMBOL(proc_do_large_bitmap);
Linux with added FPC-III support