xref: /openbmc/linux/kernel/sysctl.c (revision bc5aa3a0)
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/kmemcheck.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/kexec.h>
67 #include <linux/bpf.h>
68 
69 #include <asm/uaccess.h>
70 #include <asm/processor.h>
71 
72 #ifdef CONFIG_X86
73 #include <asm/nmi.h>
74 #include <asm/stacktrace.h>
75 #include <asm/io.h>
76 #endif
77 #ifdef CONFIG_SPARC
78 #include <asm/setup.h>
79 #endif
80 #ifdef CONFIG_BSD_PROCESS_ACCT
81 #include <linux/acct.h>
82 #endif
83 #ifdef CONFIG_RT_MUTEXES
84 #include <linux/rtmutex.h>
85 #endif
86 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
87 #include <linux/lockdep.h>
88 #endif
89 #ifdef CONFIG_CHR_DEV_SG
90 #include <scsi/sg.h>
91 #endif
92 
93 #ifdef CONFIG_LOCKUP_DETECTOR
94 #include <linux/nmi.h>
95 #endif
96 
97 #if defined(CONFIG_SYSCTL)
98 
99 /* External variables not in a header file. */
100 extern int suid_dumpable;
101 #ifdef CONFIG_COREDUMP
102 extern int core_uses_pid;
103 extern char core_pattern[];
104 extern unsigned int core_pipe_limit;
105 #endif
106 extern int pid_max;
107 extern int pid_max_min, pid_max_max;
108 extern int percpu_pagelist_fraction;
109 extern int compat_log;
110 extern int latencytop_enabled;
111 extern int sysctl_nr_open_min, sysctl_nr_open_max;
112 #ifndef CONFIG_MMU
113 extern int sysctl_nr_trim_pages;
114 #endif
115 
116 /* Constants used for minimum and  maximum */
117 #ifdef CONFIG_LOCKUP_DETECTOR
118 static int sixty = 60;
119 #endif
120 
121 static int __maybe_unused neg_one = -1;
122 
123 static int zero;
124 static int __maybe_unused one = 1;
125 static int __maybe_unused two = 2;
126 static int __maybe_unused four = 4;
127 static unsigned long one_ul = 1;
128 static int one_hundred = 100;
129 static int one_thousand = 1000;
130 #ifdef CONFIG_PRINTK
131 static int ten_thousand = 10000;
132 #endif
133 #ifdef CONFIG_PERF_EVENTS
134 static int six_hundred_forty_kb = 640 * 1024;
135 #endif
136 
137 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
138 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
139 
140 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
141 static int maxolduid = 65535;
142 static int minolduid;
143 
144 static int ngroups_max = NGROUPS_MAX;
145 static const int cap_last_cap = CAP_LAST_CAP;
146 
147 /*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */
148 #ifdef CONFIG_DETECT_HUNG_TASK
149 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
150 #endif
151 
152 #ifdef CONFIG_INOTIFY_USER
153 #include <linux/inotify.h>
154 #endif
155 #ifdef CONFIG_SPARC
156 #endif
157 
158 #ifdef __hppa__
159 extern int pwrsw_enabled;
160 #endif
161 
162 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
163 extern int unaligned_enabled;
164 #endif
165 
166 #ifdef CONFIG_IA64
167 extern int unaligned_dump_stack;
168 #endif
169 
170 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
171 extern int no_unaligned_warning;
172 #endif
173 
174 #ifdef CONFIG_PROC_SYSCTL
175 
176 #define SYSCTL_WRITES_LEGACY	-1
177 #define SYSCTL_WRITES_WARN	 0
178 #define SYSCTL_WRITES_STRICT	 1
179 
180 static int sysctl_writes_strict = SYSCTL_WRITES_STRICT;
181 
182 static int proc_do_cad_pid(struct ctl_table *table, int write,
183 		  void __user *buffer, size_t *lenp, loff_t *ppos);
184 static int proc_taint(struct ctl_table *table, int write,
185 			       void __user *buffer, size_t *lenp, loff_t *ppos);
186 #endif
187 
188 #ifdef CONFIG_PRINTK
189 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
190 				void __user *buffer, size_t *lenp, loff_t *ppos);
191 #endif
192 
193 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
194 		void __user *buffer, size_t *lenp, loff_t *ppos);
195 #ifdef CONFIG_COREDUMP
196 static int proc_dostring_coredump(struct ctl_table *table, int write,
197 		void __user *buffer, size_t *lenp, loff_t *ppos);
198 #endif
199 
200 #ifdef CONFIG_MAGIC_SYSRQ
201 /* Note: sysrq code uses it's own private copy */
202 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
203 
204 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
205 				void __user *buffer, size_t *lenp,
206 				loff_t *ppos)
207 {
208 	int error;
209 
210 	error = proc_dointvec(table, write, buffer, lenp, ppos);
211 	if (error)
212 		return error;
213 
214 	if (write)
215 		sysrq_toggle_support(__sysrq_enabled);
216 
217 	return 0;
218 }
219 
220 #endif
221 
222 static struct ctl_table kern_table[];
223 static struct ctl_table vm_table[];
224 static struct ctl_table fs_table[];
225 static struct ctl_table debug_table[];
226 static struct ctl_table dev_table[];
227 extern struct ctl_table random_table[];
228 #ifdef CONFIG_EPOLL
229 extern struct ctl_table epoll_table[];
230 #endif
231 
232 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
233 int sysctl_legacy_va_layout;
234 #endif
235 
236 /* The default sysctl tables: */
237 
238 static struct ctl_table sysctl_base_table[] = {
239 	{
240 		.procname	= "kernel",
241 		.mode		= 0555,
242 		.child		= kern_table,
243 	},
244 	{
245 		.procname	= "vm",
246 		.mode		= 0555,
247 		.child		= vm_table,
248 	},
249 	{
250 		.procname	= "fs",
251 		.mode		= 0555,
252 		.child		= fs_table,
253 	},
254 	{
255 		.procname	= "debug",
256 		.mode		= 0555,
257 		.child		= debug_table,
258 	},
259 	{
260 		.procname	= "dev",
261 		.mode		= 0555,
262 		.child		= dev_table,
263 	},
264 	{ }
265 };
266 
267 #ifdef CONFIG_SCHED_DEBUG
268 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
269 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
270 static int min_wakeup_granularity_ns;			/* 0 usecs */
271 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
272 #ifdef CONFIG_SMP
273 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
274 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
275 #endif /* CONFIG_SMP */
276 #endif /* CONFIG_SCHED_DEBUG */
277 
278 #ifdef CONFIG_COMPACTION
279 static int min_extfrag_threshold;
280 static int max_extfrag_threshold = 1000;
281 #endif
282 
283 static struct ctl_table kern_table[] = {
284 	{
285 		.procname	= "sched_child_runs_first",
286 		.data		= &sysctl_sched_child_runs_first,
287 		.maxlen		= sizeof(unsigned int),
288 		.mode		= 0644,
289 		.proc_handler	= proc_dointvec,
290 	},
291 #ifdef CONFIG_SCHED_DEBUG
292 	{
293 		.procname	= "sched_min_granularity_ns",
294 		.data		= &sysctl_sched_min_granularity,
295 		.maxlen		= sizeof(unsigned int),
296 		.mode		= 0644,
297 		.proc_handler	= sched_proc_update_handler,
298 		.extra1		= &min_sched_granularity_ns,
299 		.extra2		= &max_sched_granularity_ns,
300 	},
301 	{
302 		.procname	= "sched_latency_ns",
303 		.data		= &sysctl_sched_latency,
304 		.maxlen		= sizeof(unsigned int),
305 		.mode		= 0644,
306 		.proc_handler	= sched_proc_update_handler,
307 		.extra1		= &min_sched_granularity_ns,
308 		.extra2		= &max_sched_granularity_ns,
309 	},
310 	{
311 		.procname	= "sched_wakeup_granularity_ns",
312 		.data		= &sysctl_sched_wakeup_granularity,
313 		.maxlen		= sizeof(unsigned int),
314 		.mode		= 0644,
315 		.proc_handler	= sched_proc_update_handler,
316 		.extra1		= &min_wakeup_granularity_ns,
317 		.extra2		= &max_wakeup_granularity_ns,
318 	},
319 #ifdef CONFIG_SMP
320 	{
321 		.procname	= "sched_tunable_scaling",
322 		.data		= &sysctl_sched_tunable_scaling,
323 		.maxlen		= sizeof(enum sched_tunable_scaling),
324 		.mode		= 0644,
325 		.proc_handler	= sched_proc_update_handler,
326 		.extra1		= &min_sched_tunable_scaling,
327 		.extra2		= &max_sched_tunable_scaling,
328 	},
329 	{
330 		.procname	= "sched_migration_cost_ns",
331 		.data		= &sysctl_sched_migration_cost,
332 		.maxlen		= sizeof(unsigned int),
333 		.mode		= 0644,
334 		.proc_handler	= proc_dointvec,
335 	},
336 	{
337 		.procname	= "sched_nr_migrate",
338 		.data		= &sysctl_sched_nr_migrate,
339 		.maxlen		= sizeof(unsigned int),
340 		.mode		= 0644,
341 		.proc_handler	= proc_dointvec,
342 	},
343 	{
344 		.procname	= "sched_time_avg_ms",
345 		.data		= &sysctl_sched_time_avg,
346 		.maxlen		= sizeof(unsigned int),
347 		.mode		= 0644,
348 		.proc_handler	= proc_dointvec,
349 	},
350 	{
351 		.procname	= "sched_shares_window_ns",
352 		.data		= &sysctl_sched_shares_window,
353 		.maxlen		= sizeof(unsigned int),
354 		.mode		= 0644,
355 		.proc_handler	= proc_dointvec,
356 	},
357 #ifdef CONFIG_SCHEDSTATS
358 	{
359 		.procname	= "sched_schedstats",
360 		.data		= NULL,
361 		.maxlen		= sizeof(unsigned int),
362 		.mode		= 0644,
363 		.proc_handler	= sysctl_schedstats,
364 		.extra1		= &zero,
365 		.extra2		= &one,
366 	},
367 #endif /* CONFIG_SCHEDSTATS */
368 #endif /* CONFIG_SMP */
369 #ifdef CONFIG_NUMA_BALANCING
370 	{
371 		.procname	= "numa_balancing_scan_delay_ms",
372 		.data		= &sysctl_numa_balancing_scan_delay,
373 		.maxlen		= sizeof(unsigned int),
374 		.mode		= 0644,
375 		.proc_handler	= proc_dointvec,
376 	},
377 	{
378 		.procname	= "numa_balancing_scan_period_min_ms",
379 		.data		= &sysctl_numa_balancing_scan_period_min,
380 		.maxlen		= sizeof(unsigned int),
381 		.mode		= 0644,
382 		.proc_handler	= proc_dointvec,
383 	},
384 	{
385 		.procname	= "numa_balancing_scan_period_max_ms",
386 		.data		= &sysctl_numa_balancing_scan_period_max,
387 		.maxlen		= sizeof(unsigned int),
388 		.mode		= 0644,
389 		.proc_handler	= proc_dointvec,
390 	},
391 	{
392 		.procname	= "numa_balancing_scan_size_mb",
393 		.data		= &sysctl_numa_balancing_scan_size,
394 		.maxlen		= sizeof(unsigned int),
395 		.mode		= 0644,
396 		.proc_handler	= proc_dointvec_minmax,
397 		.extra1		= &one,
398 	},
399 	{
400 		.procname	= "numa_balancing",
401 		.data		= NULL, /* filled in by handler */
402 		.maxlen		= sizeof(unsigned int),
403 		.mode		= 0644,
404 		.proc_handler	= sysctl_numa_balancing,
405 		.extra1		= &zero,
406 		.extra2		= &one,
407 	},
408 #endif /* CONFIG_NUMA_BALANCING */
409 #endif /* CONFIG_SCHED_DEBUG */
410 	{
411 		.procname	= "sched_rt_period_us",
412 		.data		= &sysctl_sched_rt_period,
413 		.maxlen		= sizeof(unsigned int),
414 		.mode		= 0644,
415 		.proc_handler	= sched_rt_handler,
416 	},
417 	{
418 		.procname	= "sched_rt_runtime_us",
419 		.data		= &sysctl_sched_rt_runtime,
420 		.maxlen		= sizeof(int),
421 		.mode		= 0644,
422 		.proc_handler	= sched_rt_handler,
423 	},
424 	{
425 		.procname	= "sched_rr_timeslice_ms",
426 		.data		= &sched_rr_timeslice,
427 		.maxlen		= sizeof(int),
428 		.mode		= 0644,
429 		.proc_handler	= sched_rr_handler,
430 	},
431 #ifdef CONFIG_SCHED_AUTOGROUP
432 	{
433 		.procname	= "sched_autogroup_enabled",
434 		.data		= &sysctl_sched_autogroup_enabled,
435 		.maxlen		= sizeof(unsigned int),
436 		.mode		= 0644,
437 		.proc_handler	= proc_dointvec_minmax,
438 		.extra1		= &zero,
439 		.extra2		= &one,
440 	},
441 #endif
442 #ifdef CONFIG_CFS_BANDWIDTH
443 	{
444 		.procname	= "sched_cfs_bandwidth_slice_us",
445 		.data		= &sysctl_sched_cfs_bandwidth_slice,
446 		.maxlen		= sizeof(unsigned int),
447 		.mode		= 0644,
448 		.proc_handler	= proc_dointvec_minmax,
449 		.extra1		= &one,
450 	},
451 #endif
452 #ifdef CONFIG_PROVE_LOCKING
453 	{
454 		.procname	= "prove_locking",
455 		.data		= &prove_locking,
456 		.maxlen		= sizeof(int),
457 		.mode		= 0644,
458 		.proc_handler	= proc_dointvec,
459 	},
460 #endif
461 #ifdef CONFIG_LOCK_STAT
462 	{
463 		.procname	= "lock_stat",
464 		.data		= &lock_stat,
465 		.maxlen		= sizeof(int),
466 		.mode		= 0644,
467 		.proc_handler	= proc_dointvec,
468 	},
469 #endif
470 	{
471 		.procname	= "panic",
472 		.data		= &panic_timeout,
473 		.maxlen		= sizeof(int),
474 		.mode		= 0644,
475 		.proc_handler	= proc_dointvec,
476 	},
477 #ifdef CONFIG_COREDUMP
478 	{
479 		.procname	= "core_uses_pid",
480 		.data		= &core_uses_pid,
481 		.maxlen		= sizeof(int),
482 		.mode		= 0644,
483 		.proc_handler	= proc_dointvec,
484 	},
485 	{
486 		.procname	= "core_pattern",
487 		.data		= core_pattern,
488 		.maxlen		= CORENAME_MAX_SIZE,
489 		.mode		= 0644,
490 		.proc_handler	= proc_dostring_coredump,
491 	},
492 	{
493 		.procname	= "core_pipe_limit",
494 		.data		= &core_pipe_limit,
495 		.maxlen		= sizeof(unsigned int),
496 		.mode		= 0644,
497 		.proc_handler	= proc_dointvec,
498 	},
499 #endif
500 #ifdef CONFIG_PROC_SYSCTL
501 	{
502 		.procname	= "tainted",
503 		.maxlen 	= sizeof(long),
504 		.mode		= 0644,
505 		.proc_handler	= proc_taint,
506 	},
507 	{
508 		.procname	= "sysctl_writes_strict",
509 		.data		= &sysctl_writes_strict,
510 		.maxlen		= sizeof(int),
511 		.mode		= 0644,
512 		.proc_handler	= proc_dointvec_minmax,
513 		.extra1		= &neg_one,
514 		.extra2		= &one,
515 	},
516 #endif
517 #ifdef CONFIG_LATENCYTOP
518 	{
519 		.procname	= "latencytop",
520 		.data		= &latencytop_enabled,
521 		.maxlen		= sizeof(int),
522 		.mode		= 0644,
523 		.proc_handler	= sysctl_latencytop,
524 	},
525 #endif
526 #ifdef CONFIG_BLK_DEV_INITRD
527 	{
528 		.procname	= "real-root-dev",
529 		.data		= &real_root_dev,
530 		.maxlen		= sizeof(int),
531 		.mode		= 0644,
532 		.proc_handler	= proc_dointvec,
533 	},
534 #endif
535 	{
536 		.procname	= "print-fatal-signals",
537 		.data		= &print_fatal_signals,
538 		.maxlen		= sizeof(int),
539 		.mode		= 0644,
540 		.proc_handler	= proc_dointvec,
541 	},
542 #ifdef CONFIG_SPARC
543 	{
544 		.procname	= "reboot-cmd",
545 		.data		= reboot_command,
546 		.maxlen		= 256,
547 		.mode		= 0644,
548 		.proc_handler	= proc_dostring,
549 	},
550 	{
551 		.procname	= "stop-a",
552 		.data		= &stop_a_enabled,
553 		.maxlen		= sizeof (int),
554 		.mode		= 0644,
555 		.proc_handler	= proc_dointvec,
556 	},
557 	{
558 		.procname	= "scons-poweroff",
559 		.data		= &scons_pwroff,
560 		.maxlen		= sizeof (int),
561 		.mode		= 0644,
562 		.proc_handler	= proc_dointvec,
563 	},
564 #endif
565 #ifdef CONFIG_SPARC64
566 	{
567 		.procname	= "tsb-ratio",
568 		.data		= &sysctl_tsb_ratio,
569 		.maxlen		= sizeof (int),
570 		.mode		= 0644,
571 		.proc_handler	= proc_dointvec,
572 	},
573 #endif
574 #ifdef __hppa__
575 	{
576 		.procname	= "soft-power",
577 		.data		= &pwrsw_enabled,
578 		.maxlen		= sizeof (int),
579 	 	.mode		= 0644,
580 		.proc_handler	= proc_dointvec,
581 	},
582 #endif
583 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
584 	{
585 		.procname	= "unaligned-trap",
586 		.data		= &unaligned_enabled,
587 		.maxlen		= sizeof (int),
588 		.mode		= 0644,
589 		.proc_handler	= proc_dointvec,
590 	},
591 #endif
592 	{
593 		.procname	= "ctrl-alt-del",
594 		.data		= &C_A_D,
595 		.maxlen		= sizeof(int),
596 		.mode		= 0644,
597 		.proc_handler	= proc_dointvec,
598 	},
599 #ifdef CONFIG_FUNCTION_TRACER
600 	{
601 		.procname	= "ftrace_enabled",
602 		.data		= &ftrace_enabled,
603 		.maxlen		= sizeof(int),
604 		.mode		= 0644,
605 		.proc_handler	= ftrace_enable_sysctl,
606 	},
607 #endif
608 #ifdef CONFIG_STACK_TRACER
609 	{
610 		.procname	= "stack_tracer_enabled",
611 		.data		= &stack_tracer_enabled,
612 		.maxlen		= sizeof(int),
613 		.mode		= 0644,
614 		.proc_handler	= stack_trace_sysctl,
615 	},
616 #endif
617 #ifdef CONFIG_TRACING
618 	{
619 		.procname	= "ftrace_dump_on_oops",
620 		.data		= &ftrace_dump_on_oops,
621 		.maxlen		= sizeof(int),
622 		.mode		= 0644,
623 		.proc_handler	= proc_dointvec,
624 	},
625 	{
626 		.procname	= "traceoff_on_warning",
627 		.data		= &__disable_trace_on_warning,
628 		.maxlen		= sizeof(__disable_trace_on_warning),
629 		.mode		= 0644,
630 		.proc_handler	= proc_dointvec,
631 	},
632 	{
633 		.procname	= "tracepoint_printk",
634 		.data		= &tracepoint_printk,
635 		.maxlen		= sizeof(tracepoint_printk),
636 		.mode		= 0644,
637 		.proc_handler	= proc_dointvec,
638 	},
639 #endif
640 #ifdef CONFIG_KEXEC_CORE
641 	{
642 		.procname	= "kexec_load_disabled",
643 		.data		= &kexec_load_disabled,
644 		.maxlen		= sizeof(int),
645 		.mode		= 0644,
646 		/* only handle a transition from default "0" to "1" */
647 		.proc_handler	= proc_dointvec_minmax,
648 		.extra1		= &one,
649 		.extra2		= &one,
650 	},
651 #endif
652 #ifdef CONFIG_MODULES
653 	{
654 		.procname	= "modprobe",
655 		.data		= &modprobe_path,
656 		.maxlen		= KMOD_PATH_LEN,
657 		.mode		= 0644,
658 		.proc_handler	= proc_dostring,
659 	},
660 	{
661 		.procname	= "modules_disabled",
662 		.data		= &modules_disabled,
663 		.maxlen		= sizeof(int),
664 		.mode		= 0644,
665 		/* only handle a transition from default "0" to "1" */
666 		.proc_handler	= proc_dointvec_minmax,
667 		.extra1		= &one,
668 		.extra2		= &one,
669 	},
670 #endif
671 #ifdef CONFIG_UEVENT_HELPER
672 	{
673 		.procname	= "hotplug",
674 		.data		= &uevent_helper,
675 		.maxlen		= UEVENT_HELPER_PATH_LEN,
676 		.mode		= 0644,
677 		.proc_handler	= proc_dostring,
678 	},
679 #endif
680 #ifdef CONFIG_CHR_DEV_SG
681 	{
682 		.procname	= "sg-big-buff",
683 		.data		= &sg_big_buff,
684 		.maxlen		= sizeof (int),
685 		.mode		= 0444,
686 		.proc_handler	= proc_dointvec,
687 	},
688 #endif
689 #ifdef CONFIG_BSD_PROCESS_ACCT
690 	{
691 		.procname	= "acct",
692 		.data		= &acct_parm,
693 		.maxlen		= 3*sizeof(int),
694 		.mode		= 0644,
695 		.proc_handler	= proc_dointvec,
696 	},
697 #endif
698 #ifdef CONFIG_MAGIC_SYSRQ
699 	{
700 		.procname	= "sysrq",
701 		.data		= &__sysrq_enabled,
702 		.maxlen		= sizeof (int),
703 		.mode		= 0644,
704 		.proc_handler	= sysrq_sysctl_handler,
705 	},
706 #endif
707 #ifdef CONFIG_PROC_SYSCTL
708 	{
709 		.procname	= "cad_pid",
710 		.data		= NULL,
711 		.maxlen		= sizeof (int),
712 		.mode		= 0600,
713 		.proc_handler	= proc_do_cad_pid,
714 	},
715 #endif
716 	{
717 		.procname	= "threads-max",
718 		.data		= NULL,
719 		.maxlen		= sizeof(int),
720 		.mode		= 0644,
721 		.proc_handler	= sysctl_max_threads,
722 	},
723 	{
724 		.procname	= "random",
725 		.mode		= 0555,
726 		.child		= random_table,
727 	},
728 	{
729 		.procname	= "usermodehelper",
730 		.mode		= 0555,
731 		.child		= usermodehelper_table,
732 	},
733 	{
734 		.procname	= "overflowuid",
735 		.data		= &overflowuid,
736 		.maxlen		= sizeof(int),
737 		.mode		= 0644,
738 		.proc_handler	= proc_dointvec_minmax,
739 		.extra1		= &minolduid,
740 		.extra2		= &maxolduid,
741 	},
742 	{
743 		.procname	= "overflowgid",
744 		.data		= &overflowgid,
745 		.maxlen		= sizeof(int),
746 		.mode		= 0644,
747 		.proc_handler	= proc_dointvec_minmax,
748 		.extra1		= &minolduid,
749 		.extra2		= &maxolduid,
750 	},
751 #ifdef CONFIG_S390
752 #ifdef CONFIG_MATHEMU
753 	{
754 		.procname	= "ieee_emulation_warnings",
755 		.data		= &sysctl_ieee_emulation_warnings,
756 		.maxlen		= sizeof(int),
757 		.mode		= 0644,
758 		.proc_handler	= proc_dointvec,
759 	},
760 #endif
761 	{
762 		.procname	= "userprocess_debug",
763 		.data		= &show_unhandled_signals,
764 		.maxlen		= sizeof(int),
765 		.mode		= 0644,
766 		.proc_handler	= proc_dointvec,
767 	},
768 #endif
769 	{
770 		.procname	= "pid_max",
771 		.data		= &pid_max,
772 		.maxlen		= sizeof (int),
773 		.mode		= 0644,
774 		.proc_handler	= proc_dointvec_minmax,
775 		.extra1		= &pid_max_min,
776 		.extra2		= &pid_max_max,
777 	},
778 	{
779 		.procname	= "panic_on_oops",
780 		.data		= &panic_on_oops,
781 		.maxlen		= sizeof(int),
782 		.mode		= 0644,
783 		.proc_handler	= proc_dointvec,
784 	},
785 #if defined CONFIG_PRINTK
786 	{
787 		.procname	= "printk",
788 		.data		= &console_loglevel,
789 		.maxlen		= 4*sizeof(int),
790 		.mode		= 0644,
791 		.proc_handler	= proc_dointvec,
792 	},
793 	{
794 		.procname	= "printk_ratelimit",
795 		.data		= &printk_ratelimit_state.interval,
796 		.maxlen		= sizeof(int),
797 		.mode		= 0644,
798 		.proc_handler	= proc_dointvec_jiffies,
799 	},
800 	{
801 		.procname	= "printk_ratelimit_burst",
802 		.data		= &printk_ratelimit_state.burst,
803 		.maxlen		= sizeof(int),
804 		.mode		= 0644,
805 		.proc_handler	= proc_dointvec,
806 	},
807 	{
808 		.procname	= "printk_delay",
809 		.data		= &printk_delay_msec,
810 		.maxlen		= sizeof(int),
811 		.mode		= 0644,
812 		.proc_handler	= proc_dointvec_minmax,
813 		.extra1		= &zero,
814 		.extra2		= &ten_thousand,
815 	},
816 	{
817 		.procname	= "printk_devkmsg",
818 		.data		= devkmsg_log_str,
819 		.maxlen		= DEVKMSG_STR_MAX_SIZE,
820 		.mode		= 0644,
821 		.proc_handler	= devkmsg_sysctl_set_loglvl,
822 	},
823 	{
824 		.procname	= "dmesg_restrict",
825 		.data		= &dmesg_restrict,
826 		.maxlen		= sizeof(int),
827 		.mode		= 0644,
828 		.proc_handler	= proc_dointvec_minmax_sysadmin,
829 		.extra1		= &zero,
830 		.extra2		= &one,
831 	},
832 	{
833 		.procname	= "kptr_restrict",
834 		.data		= &kptr_restrict,
835 		.maxlen		= sizeof(int),
836 		.mode		= 0644,
837 		.proc_handler	= proc_dointvec_minmax_sysadmin,
838 		.extra1		= &zero,
839 		.extra2		= &two,
840 	},
841 #endif
842 	{
843 		.procname	= "ngroups_max",
844 		.data		= &ngroups_max,
845 		.maxlen		= sizeof (int),
846 		.mode		= 0444,
847 		.proc_handler	= proc_dointvec,
848 	},
849 	{
850 		.procname	= "cap_last_cap",
851 		.data		= (void *)&cap_last_cap,
852 		.maxlen		= sizeof(int),
853 		.mode		= 0444,
854 		.proc_handler	= proc_dointvec,
855 	},
856 #if defined(CONFIG_LOCKUP_DETECTOR)
857 	{
858 		.procname       = "watchdog",
859 		.data           = &watchdog_user_enabled,
860 		.maxlen         = sizeof (int),
861 		.mode           = 0644,
862 		.proc_handler   = proc_watchdog,
863 		.extra1		= &zero,
864 		.extra2		= &one,
865 	},
866 	{
867 		.procname	= "watchdog_thresh",
868 		.data		= &watchdog_thresh,
869 		.maxlen		= sizeof(int),
870 		.mode		= 0644,
871 		.proc_handler	= proc_watchdog_thresh,
872 		.extra1		= &zero,
873 		.extra2		= &sixty,
874 	},
875 	{
876 		.procname       = "nmi_watchdog",
877 		.data           = &nmi_watchdog_enabled,
878 		.maxlen         = sizeof (int),
879 		.mode           = 0644,
880 		.proc_handler   = proc_nmi_watchdog,
881 		.extra1		= &zero,
882 #if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR)
883 		.extra2		= &one,
884 #else
885 		.extra2		= &zero,
886 #endif
887 	},
888 	{
889 		.procname       = "soft_watchdog",
890 		.data           = &soft_watchdog_enabled,
891 		.maxlen         = sizeof (int),
892 		.mode           = 0644,
893 		.proc_handler   = proc_soft_watchdog,
894 		.extra1		= &zero,
895 		.extra2		= &one,
896 	},
897 	{
898 		.procname	= "watchdog_cpumask",
899 		.data		= &watchdog_cpumask_bits,
900 		.maxlen		= NR_CPUS,
901 		.mode		= 0644,
902 		.proc_handler	= proc_watchdog_cpumask,
903 	},
904 	{
905 		.procname	= "softlockup_panic",
906 		.data		= &softlockup_panic,
907 		.maxlen		= sizeof(int),
908 		.mode		= 0644,
909 		.proc_handler	= proc_dointvec_minmax,
910 		.extra1		= &zero,
911 		.extra2		= &one,
912 	},
913 #ifdef CONFIG_HARDLOCKUP_DETECTOR
914 	{
915 		.procname	= "hardlockup_panic",
916 		.data		= &hardlockup_panic,
917 		.maxlen		= sizeof(int),
918 		.mode		= 0644,
919 		.proc_handler	= proc_dointvec_minmax,
920 		.extra1		= &zero,
921 		.extra2		= &one,
922 	},
923 #endif
924 #ifdef CONFIG_SMP
925 	{
926 		.procname	= "softlockup_all_cpu_backtrace",
927 		.data		= &sysctl_softlockup_all_cpu_backtrace,
928 		.maxlen		= sizeof(int),
929 		.mode		= 0644,
930 		.proc_handler	= proc_dointvec_minmax,
931 		.extra1		= &zero,
932 		.extra2		= &one,
933 	},
934 	{
935 		.procname	= "hardlockup_all_cpu_backtrace",
936 		.data		= &sysctl_hardlockup_all_cpu_backtrace,
937 		.maxlen		= sizeof(int),
938 		.mode		= 0644,
939 		.proc_handler	= proc_dointvec_minmax,
940 		.extra1		= &zero,
941 		.extra2		= &one,
942 	},
943 #endif /* CONFIG_SMP */
944 #endif
945 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
946 	{
947 		.procname       = "unknown_nmi_panic",
948 		.data           = &unknown_nmi_panic,
949 		.maxlen         = sizeof (int),
950 		.mode           = 0644,
951 		.proc_handler   = proc_dointvec,
952 	},
953 #endif
954 #if defined(CONFIG_X86)
955 	{
956 		.procname	= "panic_on_unrecovered_nmi",
957 		.data		= &panic_on_unrecovered_nmi,
958 		.maxlen		= sizeof(int),
959 		.mode		= 0644,
960 		.proc_handler	= proc_dointvec,
961 	},
962 	{
963 		.procname	= "panic_on_io_nmi",
964 		.data		= &panic_on_io_nmi,
965 		.maxlen		= sizeof(int),
966 		.mode		= 0644,
967 		.proc_handler	= proc_dointvec,
968 	},
969 #ifdef CONFIG_DEBUG_STACKOVERFLOW
970 	{
971 		.procname	= "panic_on_stackoverflow",
972 		.data		= &sysctl_panic_on_stackoverflow,
973 		.maxlen		= sizeof(int),
974 		.mode		= 0644,
975 		.proc_handler	= proc_dointvec,
976 	},
977 #endif
978 	{
979 		.procname	= "bootloader_type",
980 		.data		= &bootloader_type,
981 		.maxlen		= sizeof (int),
982 		.mode		= 0444,
983 		.proc_handler	= proc_dointvec,
984 	},
985 	{
986 		.procname	= "bootloader_version",
987 		.data		= &bootloader_version,
988 		.maxlen		= sizeof (int),
989 		.mode		= 0444,
990 		.proc_handler	= proc_dointvec,
991 	},
992 	{
993 		.procname	= "kstack_depth_to_print",
994 		.data		= &kstack_depth_to_print,
995 		.maxlen		= sizeof(int),
996 		.mode		= 0644,
997 		.proc_handler	= proc_dointvec,
998 	},
999 	{
1000 		.procname	= "io_delay_type",
1001 		.data		= &io_delay_type,
1002 		.maxlen		= sizeof(int),
1003 		.mode		= 0644,
1004 		.proc_handler	= proc_dointvec,
1005 	},
1006 #endif
1007 #if defined(CONFIG_MMU)
1008 	{
1009 		.procname	= "randomize_va_space",
1010 		.data		= &randomize_va_space,
1011 		.maxlen		= sizeof(int),
1012 		.mode		= 0644,
1013 		.proc_handler	= proc_dointvec,
1014 	},
1015 #endif
1016 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1017 	{
1018 		.procname	= "spin_retry",
1019 		.data		= &spin_retry,
1020 		.maxlen		= sizeof (int),
1021 		.mode		= 0644,
1022 		.proc_handler	= proc_dointvec,
1023 	},
1024 #endif
1025 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1026 	{
1027 		.procname	= "acpi_video_flags",
1028 		.data		= &acpi_realmode_flags,
1029 		.maxlen		= sizeof (unsigned long),
1030 		.mode		= 0644,
1031 		.proc_handler	= proc_doulongvec_minmax,
1032 	},
1033 #endif
1034 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1035 	{
1036 		.procname	= "ignore-unaligned-usertrap",
1037 		.data		= &no_unaligned_warning,
1038 		.maxlen		= sizeof (int),
1039 	 	.mode		= 0644,
1040 		.proc_handler	= proc_dointvec,
1041 	},
1042 #endif
1043 #ifdef CONFIG_IA64
1044 	{
1045 		.procname	= "unaligned-dump-stack",
1046 		.data		= &unaligned_dump_stack,
1047 		.maxlen		= sizeof (int),
1048 		.mode		= 0644,
1049 		.proc_handler	= proc_dointvec,
1050 	},
1051 #endif
1052 #ifdef CONFIG_DETECT_HUNG_TASK
1053 	{
1054 		.procname	= "hung_task_panic",
1055 		.data		= &sysctl_hung_task_panic,
1056 		.maxlen		= sizeof(int),
1057 		.mode		= 0644,
1058 		.proc_handler	= proc_dointvec_minmax,
1059 		.extra1		= &zero,
1060 		.extra2		= &one,
1061 	},
1062 	{
1063 		.procname	= "hung_task_check_count",
1064 		.data		= &sysctl_hung_task_check_count,
1065 		.maxlen		= sizeof(int),
1066 		.mode		= 0644,
1067 		.proc_handler	= proc_dointvec_minmax,
1068 		.extra1		= &zero,
1069 	},
1070 	{
1071 		.procname	= "hung_task_timeout_secs",
1072 		.data		= &sysctl_hung_task_timeout_secs,
1073 		.maxlen		= sizeof(unsigned long),
1074 		.mode		= 0644,
1075 		.proc_handler	= proc_dohung_task_timeout_secs,
1076 		.extra2		= &hung_task_timeout_max,
1077 	},
1078 	{
1079 		.procname	= "hung_task_warnings",
1080 		.data		= &sysctl_hung_task_warnings,
1081 		.maxlen		= sizeof(int),
1082 		.mode		= 0644,
1083 		.proc_handler	= proc_dointvec_minmax,
1084 		.extra1		= &neg_one,
1085 	},
1086 #endif
1087 #ifdef CONFIG_COMPAT
1088 	{
1089 		.procname	= "compat-log",
1090 		.data		= &compat_log,
1091 		.maxlen		= sizeof (int),
1092 	 	.mode		= 0644,
1093 		.proc_handler	= proc_dointvec,
1094 	},
1095 #endif
1096 #ifdef CONFIG_RT_MUTEXES
1097 	{
1098 		.procname	= "max_lock_depth",
1099 		.data		= &max_lock_depth,
1100 		.maxlen		= sizeof(int),
1101 		.mode		= 0644,
1102 		.proc_handler	= proc_dointvec,
1103 	},
1104 #endif
1105 	{
1106 		.procname	= "poweroff_cmd",
1107 		.data		= &poweroff_cmd,
1108 		.maxlen		= POWEROFF_CMD_PATH_LEN,
1109 		.mode		= 0644,
1110 		.proc_handler	= proc_dostring,
1111 	},
1112 #ifdef CONFIG_KEYS
1113 	{
1114 		.procname	= "keys",
1115 		.mode		= 0555,
1116 		.child		= key_sysctls,
1117 	},
1118 #endif
1119 #ifdef CONFIG_PERF_EVENTS
1120 	/*
1121 	 * User-space scripts rely on the existence of this file
1122 	 * as a feature check for perf_events being enabled.
1123 	 *
1124 	 * So it's an ABI, do not remove!
1125 	 */
1126 	{
1127 		.procname	= "perf_event_paranoid",
1128 		.data		= &sysctl_perf_event_paranoid,
1129 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
1130 		.mode		= 0644,
1131 		.proc_handler	= proc_dointvec,
1132 	},
1133 	{
1134 		.procname	= "perf_event_mlock_kb",
1135 		.data		= &sysctl_perf_event_mlock,
1136 		.maxlen		= sizeof(sysctl_perf_event_mlock),
1137 		.mode		= 0644,
1138 		.proc_handler	= proc_dointvec,
1139 	},
1140 	{
1141 		.procname	= "perf_event_max_sample_rate",
1142 		.data		= &sysctl_perf_event_sample_rate,
1143 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
1144 		.mode		= 0644,
1145 		.proc_handler	= perf_proc_update_handler,
1146 		.extra1		= &one,
1147 	},
1148 	{
1149 		.procname	= "perf_cpu_time_max_percent",
1150 		.data		= &sysctl_perf_cpu_time_max_percent,
1151 		.maxlen		= sizeof(sysctl_perf_cpu_time_max_percent),
1152 		.mode		= 0644,
1153 		.proc_handler	= perf_cpu_time_max_percent_handler,
1154 		.extra1		= &zero,
1155 		.extra2		= &one_hundred,
1156 	},
1157 	{
1158 		.procname	= "perf_event_max_stack",
1159 		.data		= &sysctl_perf_event_max_stack,
1160 		.maxlen		= sizeof(sysctl_perf_event_max_stack),
1161 		.mode		= 0644,
1162 		.proc_handler	= perf_event_max_stack_handler,
1163 		.extra1		= &zero,
1164 		.extra2		= &six_hundred_forty_kb,
1165 	},
1166 	{
1167 		.procname	= "perf_event_max_contexts_per_stack",
1168 		.data		= &sysctl_perf_event_max_contexts_per_stack,
1169 		.maxlen		= sizeof(sysctl_perf_event_max_contexts_per_stack),
1170 		.mode		= 0644,
1171 		.proc_handler	= perf_event_max_stack_handler,
1172 		.extra1		= &zero,
1173 		.extra2		= &one_thousand,
1174 	},
1175 #endif
1176 #ifdef CONFIG_KMEMCHECK
1177 	{
1178 		.procname	= "kmemcheck",
1179 		.data		= &kmemcheck_enabled,
1180 		.maxlen		= sizeof(int),
1181 		.mode		= 0644,
1182 		.proc_handler	= proc_dointvec,
1183 	},
1184 #endif
1185 	{
1186 		.procname	= "panic_on_warn",
1187 		.data		= &panic_on_warn,
1188 		.maxlen		= sizeof(int),
1189 		.mode		= 0644,
1190 		.proc_handler	= proc_dointvec_minmax,
1191 		.extra1		= &zero,
1192 		.extra2		= &one,
1193 	},
1194 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1195 	{
1196 		.procname	= "timer_migration",
1197 		.data		= &sysctl_timer_migration,
1198 		.maxlen		= sizeof(unsigned int),
1199 		.mode		= 0644,
1200 		.proc_handler	= timer_migration_handler,
1201 	},
1202 #endif
1203 #ifdef CONFIG_BPF_SYSCALL
1204 	{
1205 		.procname	= "unprivileged_bpf_disabled",
1206 		.data		= &sysctl_unprivileged_bpf_disabled,
1207 		.maxlen		= sizeof(sysctl_unprivileged_bpf_disabled),
1208 		.mode		= 0644,
1209 		/* only handle a transition from default "0" to "1" */
1210 		.proc_handler	= proc_dointvec_minmax,
1211 		.extra1		= &one,
1212 		.extra2		= &one,
1213 	},
1214 #endif
1215 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1216 	{
1217 		.procname	= "panic_on_rcu_stall",
1218 		.data		= &sysctl_panic_on_rcu_stall,
1219 		.maxlen		= sizeof(sysctl_panic_on_rcu_stall),
1220 		.mode		= 0644,
1221 		.proc_handler	= proc_dointvec_minmax,
1222 		.extra1		= &zero,
1223 		.extra2		= &one,
1224 	},
1225 #endif
1226 	{ }
1227 };
1228 
1229 static struct ctl_table vm_table[] = {
1230 	{
1231 		.procname	= "overcommit_memory",
1232 		.data		= &sysctl_overcommit_memory,
1233 		.maxlen		= sizeof(sysctl_overcommit_memory),
1234 		.mode		= 0644,
1235 		.proc_handler	= proc_dointvec_minmax,
1236 		.extra1		= &zero,
1237 		.extra2		= &two,
1238 	},
1239 	{
1240 		.procname	= "panic_on_oom",
1241 		.data		= &sysctl_panic_on_oom,
1242 		.maxlen		= sizeof(sysctl_panic_on_oom),
1243 		.mode		= 0644,
1244 		.proc_handler	= proc_dointvec_minmax,
1245 		.extra1		= &zero,
1246 		.extra2		= &two,
1247 	},
1248 	{
1249 		.procname	= "oom_kill_allocating_task",
1250 		.data		= &sysctl_oom_kill_allocating_task,
1251 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1252 		.mode		= 0644,
1253 		.proc_handler	= proc_dointvec,
1254 	},
1255 	{
1256 		.procname	= "oom_dump_tasks",
1257 		.data		= &sysctl_oom_dump_tasks,
1258 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1259 		.mode		= 0644,
1260 		.proc_handler	= proc_dointvec,
1261 	},
1262 	{
1263 		.procname	= "overcommit_ratio",
1264 		.data		= &sysctl_overcommit_ratio,
1265 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1266 		.mode		= 0644,
1267 		.proc_handler	= overcommit_ratio_handler,
1268 	},
1269 	{
1270 		.procname	= "overcommit_kbytes",
1271 		.data		= &sysctl_overcommit_kbytes,
1272 		.maxlen		= sizeof(sysctl_overcommit_kbytes),
1273 		.mode		= 0644,
1274 		.proc_handler	= overcommit_kbytes_handler,
1275 	},
1276 	{
1277 		.procname	= "page-cluster",
1278 		.data		= &page_cluster,
1279 		.maxlen		= sizeof(int),
1280 		.mode		= 0644,
1281 		.proc_handler	= proc_dointvec_minmax,
1282 		.extra1		= &zero,
1283 	},
1284 	{
1285 		.procname	= "dirty_background_ratio",
1286 		.data		= &dirty_background_ratio,
1287 		.maxlen		= sizeof(dirty_background_ratio),
1288 		.mode		= 0644,
1289 		.proc_handler	= dirty_background_ratio_handler,
1290 		.extra1		= &zero,
1291 		.extra2		= &one_hundred,
1292 	},
1293 	{
1294 		.procname	= "dirty_background_bytes",
1295 		.data		= &dirty_background_bytes,
1296 		.maxlen		= sizeof(dirty_background_bytes),
1297 		.mode		= 0644,
1298 		.proc_handler	= dirty_background_bytes_handler,
1299 		.extra1		= &one_ul,
1300 	},
1301 	{
1302 		.procname	= "dirty_ratio",
1303 		.data		= &vm_dirty_ratio,
1304 		.maxlen		= sizeof(vm_dirty_ratio),
1305 		.mode		= 0644,
1306 		.proc_handler	= dirty_ratio_handler,
1307 		.extra1		= &zero,
1308 		.extra2		= &one_hundred,
1309 	},
1310 	{
1311 		.procname	= "dirty_bytes",
1312 		.data		= &vm_dirty_bytes,
1313 		.maxlen		= sizeof(vm_dirty_bytes),
1314 		.mode		= 0644,
1315 		.proc_handler	= dirty_bytes_handler,
1316 		.extra1		= &dirty_bytes_min,
1317 	},
1318 	{
1319 		.procname	= "dirty_writeback_centisecs",
1320 		.data		= &dirty_writeback_interval,
1321 		.maxlen		= sizeof(dirty_writeback_interval),
1322 		.mode		= 0644,
1323 		.proc_handler	= dirty_writeback_centisecs_handler,
1324 	},
1325 	{
1326 		.procname	= "dirty_expire_centisecs",
1327 		.data		= &dirty_expire_interval,
1328 		.maxlen		= sizeof(dirty_expire_interval),
1329 		.mode		= 0644,
1330 		.proc_handler	= proc_dointvec_minmax,
1331 		.extra1		= &zero,
1332 	},
1333 	{
1334 		.procname	= "dirtytime_expire_seconds",
1335 		.data		= &dirtytime_expire_interval,
1336 		.maxlen		= sizeof(dirty_expire_interval),
1337 		.mode		= 0644,
1338 		.proc_handler	= dirtytime_interval_handler,
1339 		.extra1		= &zero,
1340 	},
1341 	{
1342 		.procname       = "nr_pdflush_threads",
1343 		.mode           = 0444 /* read-only */,
1344 		.proc_handler   = pdflush_proc_obsolete,
1345 	},
1346 	{
1347 		.procname	= "swappiness",
1348 		.data		= &vm_swappiness,
1349 		.maxlen		= sizeof(vm_swappiness),
1350 		.mode		= 0644,
1351 		.proc_handler	= proc_dointvec_minmax,
1352 		.extra1		= &zero,
1353 		.extra2		= &one_hundred,
1354 	},
1355 #ifdef CONFIG_HUGETLB_PAGE
1356 	{
1357 		.procname	= "nr_hugepages",
1358 		.data		= NULL,
1359 		.maxlen		= sizeof(unsigned long),
1360 		.mode		= 0644,
1361 		.proc_handler	= hugetlb_sysctl_handler,
1362 	},
1363 #ifdef CONFIG_NUMA
1364 	{
1365 		.procname       = "nr_hugepages_mempolicy",
1366 		.data           = NULL,
1367 		.maxlen         = sizeof(unsigned long),
1368 		.mode           = 0644,
1369 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1370 	},
1371 #endif
1372 	 {
1373 		.procname	= "hugetlb_shm_group",
1374 		.data		= &sysctl_hugetlb_shm_group,
1375 		.maxlen		= sizeof(gid_t),
1376 		.mode		= 0644,
1377 		.proc_handler	= proc_dointvec,
1378 	 },
1379 	 {
1380 		.procname	= "hugepages_treat_as_movable",
1381 		.data		= &hugepages_treat_as_movable,
1382 		.maxlen		= sizeof(int),
1383 		.mode		= 0644,
1384 		.proc_handler	= proc_dointvec,
1385 	},
1386 	{
1387 		.procname	= "nr_overcommit_hugepages",
1388 		.data		= NULL,
1389 		.maxlen		= sizeof(unsigned long),
1390 		.mode		= 0644,
1391 		.proc_handler	= hugetlb_overcommit_handler,
1392 	},
1393 #endif
1394 	{
1395 		.procname	= "lowmem_reserve_ratio",
1396 		.data		= &sysctl_lowmem_reserve_ratio,
1397 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1398 		.mode		= 0644,
1399 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1400 	},
1401 	{
1402 		.procname	= "drop_caches",
1403 		.data		= &sysctl_drop_caches,
1404 		.maxlen		= sizeof(int),
1405 		.mode		= 0644,
1406 		.proc_handler	= drop_caches_sysctl_handler,
1407 		.extra1		= &one,
1408 		.extra2		= &four,
1409 	},
1410 #ifdef CONFIG_COMPACTION
1411 	{
1412 		.procname	= "compact_memory",
1413 		.data		= &sysctl_compact_memory,
1414 		.maxlen		= sizeof(int),
1415 		.mode		= 0200,
1416 		.proc_handler	= sysctl_compaction_handler,
1417 	},
1418 	{
1419 		.procname	= "extfrag_threshold",
1420 		.data		= &sysctl_extfrag_threshold,
1421 		.maxlen		= sizeof(int),
1422 		.mode		= 0644,
1423 		.proc_handler	= sysctl_extfrag_handler,
1424 		.extra1		= &min_extfrag_threshold,
1425 		.extra2		= &max_extfrag_threshold,
1426 	},
1427 	{
1428 		.procname	= "compact_unevictable_allowed",
1429 		.data		= &sysctl_compact_unevictable_allowed,
1430 		.maxlen		= sizeof(int),
1431 		.mode		= 0644,
1432 		.proc_handler	= proc_dointvec,
1433 		.extra1		= &zero,
1434 		.extra2		= &one,
1435 	},
1436 
1437 #endif /* CONFIG_COMPACTION */
1438 	{
1439 		.procname	= "min_free_kbytes",
1440 		.data		= &min_free_kbytes,
1441 		.maxlen		= sizeof(min_free_kbytes),
1442 		.mode		= 0644,
1443 		.proc_handler	= min_free_kbytes_sysctl_handler,
1444 		.extra1		= &zero,
1445 	},
1446 	{
1447 		.procname	= "watermark_scale_factor",
1448 		.data		= &watermark_scale_factor,
1449 		.maxlen		= sizeof(watermark_scale_factor),
1450 		.mode		= 0644,
1451 		.proc_handler	= watermark_scale_factor_sysctl_handler,
1452 		.extra1		= &one,
1453 		.extra2		= &one_thousand,
1454 	},
1455 	{
1456 		.procname	= "percpu_pagelist_fraction",
1457 		.data		= &percpu_pagelist_fraction,
1458 		.maxlen		= sizeof(percpu_pagelist_fraction),
1459 		.mode		= 0644,
1460 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1461 		.extra1		= &zero,
1462 	},
1463 #ifdef CONFIG_MMU
1464 	{
1465 		.procname	= "max_map_count",
1466 		.data		= &sysctl_max_map_count,
1467 		.maxlen		= sizeof(sysctl_max_map_count),
1468 		.mode		= 0644,
1469 		.proc_handler	= proc_dointvec_minmax,
1470 		.extra1		= &zero,
1471 	},
1472 #else
1473 	{
1474 		.procname	= "nr_trim_pages",
1475 		.data		= &sysctl_nr_trim_pages,
1476 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1477 		.mode		= 0644,
1478 		.proc_handler	= proc_dointvec_minmax,
1479 		.extra1		= &zero,
1480 	},
1481 #endif
1482 	{
1483 		.procname	= "laptop_mode",
1484 		.data		= &laptop_mode,
1485 		.maxlen		= sizeof(laptop_mode),
1486 		.mode		= 0644,
1487 		.proc_handler	= proc_dointvec_jiffies,
1488 	},
1489 	{
1490 		.procname	= "block_dump",
1491 		.data		= &block_dump,
1492 		.maxlen		= sizeof(block_dump),
1493 		.mode		= 0644,
1494 		.proc_handler	= proc_dointvec,
1495 		.extra1		= &zero,
1496 	},
1497 	{
1498 		.procname	= "vfs_cache_pressure",
1499 		.data		= &sysctl_vfs_cache_pressure,
1500 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1501 		.mode		= 0644,
1502 		.proc_handler	= proc_dointvec,
1503 		.extra1		= &zero,
1504 	},
1505 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1506 	{
1507 		.procname	= "legacy_va_layout",
1508 		.data		= &sysctl_legacy_va_layout,
1509 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1510 		.mode		= 0644,
1511 		.proc_handler	= proc_dointvec,
1512 		.extra1		= &zero,
1513 	},
1514 #endif
1515 #ifdef CONFIG_NUMA
1516 	{
1517 		.procname	= "zone_reclaim_mode",
1518 		.data		= &node_reclaim_mode,
1519 		.maxlen		= sizeof(node_reclaim_mode),
1520 		.mode		= 0644,
1521 		.proc_handler	= proc_dointvec,
1522 		.extra1		= &zero,
1523 	},
1524 	{
1525 		.procname	= "min_unmapped_ratio",
1526 		.data		= &sysctl_min_unmapped_ratio,
1527 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1528 		.mode		= 0644,
1529 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1530 		.extra1		= &zero,
1531 		.extra2		= &one_hundred,
1532 	},
1533 	{
1534 		.procname	= "min_slab_ratio",
1535 		.data		= &sysctl_min_slab_ratio,
1536 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1537 		.mode		= 0644,
1538 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1539 		.extra1		= &zero,
1540 		.extra2		= &one_hundred,
1541 	},
1542 #endif
1543 #ifdef CONFIG_SMP
1544 	{
1545 		.procname	= "stat_interval",
1546 		.data		= &sysctl_stat_interval,
1547 		.maxlen		= sizeof(sysctl_stat_interval),
1548 		.mode		= 0644,
1549 		.proc_handler	= proc_dointvec_jiffies,
1550 	},
1551 	{
1552 		.procname	= "stat_refresh",
1553 		.data		= NULL,
1554 		.maxlen		= 0,
1555 		.mode		= 0600,
1556 		.proc_handler	= vmstat_refresh,
1557 	},
1558 #endif
1559 #ifdef CONFIG_MMU
1560 	{
1561 		.procname	= "mmap_min_addr",
1562 		.data		= &dac_mmap_min_addr,
1563 		.maxlen		= sizeof(unsigned long),
1564 		.mode		= 0644,
1565 		.proc_handler	= mmap_min_addr_handler,
1566 	},
1567 #endif
1568 #ifdef CONFIG_NUMA
1569 	{
1570 		.procname	= "numa_zonelist_order",
1571 		.data		= &numa_zonelist_order,
1572 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1573 		.mode		= 0644,
1574 		.proc_handler	= numa_zonelist_order_handler,
1575 	},
1576 #endif
1577 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1578    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1579 	{
1580 		.procname	= "vdso_enabled",
1581 #ifdef CONFIG_X86_32
1582 		.data		= &vdso32_enabled,
1583 		.maxlen		= sizeof(vdso32_enabled),
1584 #else
1585 		.data		= &vdso_enabled,
1586 		.maxlen		= sizeof(vdso_enabled),
1587 #endif
1588 		.mode		= 0644,
1589 		.proc_handler	= proc_dointvec,
1590 		.extra1		= &zero,
1591 	},
1592 #endif
1593 #ifdef CONFIG_HIGHMEM
1594 	{
1595 		.procname	= "highmem_is_dirtyable",
1596 		.data		= &vm_highmem_is_dirtyable,
1597 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1598 		.mode		= 0644,
1599 		.proc_handler	= proc_dointvec_minmax,
1600 		.extra1		= &zero,
1601 		.extra2		= &one,
1602 	},
1603 #endif
1604 #ifdef CONFIG_MEMORY_FAILURE
1605 	{
1606 		.procname	= "memory_failure_early_kill",
1607 		.data		= &sysctl_memory_failure_early_kill,
1608 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1609 		.mode		= 0644,
1610 		.proc_handler	= proc_dointvec_minmax,
1611 		.extra1		= &zero,
1612 		.extra2		= &one,
1613 	},
1614 	{
1615 		.procname	= "memory_failure_recovery",
1616 		.data		= &sysctl_memory_failure_recovery,
1617 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1618 		.mode		= 0644,
1619 		.proc_handler	= proc_dointvec_minmax,
1620 		.extra1		= &zero,
1621 		.extra2		= &one,
1622 	},
1623 #endif
1624 	{
1625 		.procname	= "user_reserve_kbytes",
1626 		.data		= &sysctl_user_reserve_kbytes,
1627 		.maxlen		= sizeof(sysctl_user_reserve_kbytes),
1628 		.mode		= 0644,
1629 		.proc_handler	= proc_doulongvec_minmax,
1630 	},
1631 	{
1632 		.procname	= "admin_reserve_kbytes",
1633 		.data		= &sysctl_admin_reserve_kbytes,
1634 		.maxlen		= sizeof(sysctl_admin_reserve_kbytes),
1635 		.mode		= 0644,
1636 		.proc_handler	= proc_doulongvec_minmax,
1637 	},
1638 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1639 	{
1640 		.procname	= "mmap_rnd_bits",
1641 		.data		= &mmap_rnd_bits,
1642 		.maxlen		= sizeof(mmap_rnd_bits),
1643 		.mode		= 0600,
1644 		.proc_handler	= proc_dointvec_minmax,
1645 		.extra1		= (void *)&mmap_rnd_bits_min,
1646 		.extra2		= (void *)&mmap_rnd_bits_max,
1647 	},
1648 #endif
1649 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1650 	{
1651 		.procname	= "mmap_rnd_compat_bits",
1652 		.data		= &mmap_rnd_compat_bits,
1653 		.maxlen		= sizeof(mmap_rnd_compat_bits),
1654 		.mode		= 0600,
1655 		.proc_handler	= proc_dointvec_minmax,
1656 		.extra1		= (void *)&mmap_rnd_compat_bits_min,
1657 		.extra2		= (void *)&mmap_rnd_compat_bits_max,
1658 	},
1659 #endif
1660 	{ }
1661 };
1662 
1663 static struct ctl_table fs_table[] = {
1664 	{
1665 		.procname	= "inode-nr",
1666 		.data		= &inodes_stat,
1667 		.maxlen		= 2*sizeof(long),
1668 		.mode		= 0444,
1669 		.proc_handler	= proc_nr_inodes,
1670 	},
1671 	{
1672 		.procname	= "inode-state",
1673 		.data		= &inodes_stat,
1674 		.maxlen		= 7*sizeof(long),
1675 		.mode		= 0444,
1676 		.proc_handler	= proc_nr_inodes,
1677 	},
1678 	{
1679 		.procname	= "file-nr",
1680 		.data		= &files_stat,
1681 		.maxlen		= sizeof(files_stat),
1682 		.mode		= 0444,
1683 		.proc_handler	= proc_nr_files,
1684 	},
1685 	{
1686 		.procname	= "file-max",
1687 		.data		= &files_stat.max_files,
1688 		.maxlen		= sizeof(files_stat.max_files),
1689 		.mode		= 0644,
1690 		.proc_handler	= proc_doulongvec_minmax,
1691 	},
1692 	{
1693 		.procname	= "nr_open",
1694 		.data		= &sysctl_nr_open,
1695 		.maxlen		= sizeof(int),
1696 		.mode		= 0644,
1697 		.proc_handler	= proc_dointvec_minmax,
1698 		.extra1		= &sysctl_nr_open_min,
1699 		.extra2		= &sysctl_nr_open_max,
1700 	},
1701 	{
1702 		.procname	= "dentry-state",
1703 		.data		= &dentry_stat,
1704 		.maxlen		= 6*sizeof(long),
1705 		.mode		= 0444,
1706 		.proc_handler	= proc_nr_dentry,
1707 	},
1708 	{
1709 		.procname	= "overflowuid",
1710 		.data		= &fs_overflowuid,
1711 		.maxlen		= sizeof(int),
1712 		.mode		= 0644,
1713 		.proc_handler	= proc_dointvec_minmax,
1714 		.extra1		= &minolduid,
1715 		.extra2		= &maxolduid,
1716 	},
1717 	{
1718 		.procname	= "overflowgid",
1719 		.data		= &fs_overflowgid,
1720 		.maxlen		= sizeof(int),
1721 		.mode		= 0644,
1722 		.proc_handler	= proc_dointvec_minmax,
1723 		.extra1		= &minolduid,
1724 		.extra2		= &maxolduid,
1725 	},
1726 #ifdef CONFIG_FILE_LOCKING
1727 	{
1728 		.procname	= "leases-enable",
1729 		.data		= &leases_enable,
1730 		.maxlen		= sizeof(int),
1731 		.mode		= 0644,
1732 		.proc_handler	= proc_dointvec,
1733 	},
1734 #endif
1735 #ifdef CONFIG_DNOTIFY
1736 	{
1737 		.procname	= "dir-notify-enable",
1738 		.data		= &dir_notify_enable,
1739 		.maxlen		= sizeof(int),
1740 		.mode		= 0644,
1741 		.proc_handler	= proc_dointvec,
1742 	},
1743 #endif
1744 #ifdef CONFIG_MMU
1745 #ifdef CONFIG_FILE_LOCKING
1746 	{
1747 		.procname	= "lease-break-time",
1748 		.data		= &lease_break_time,
1749 		.maxlen		= sizeof(int),
1750 		.mode		= 0644,
1751 		.proc_handler	= proc_dointvec,
1752 	},
1753 #endif
1754 #ifdef CONFIG_AIO
1755 	{
1756 		.procname	= "aio-nr",
1757 		.data		= &aio_nr,
1758 		.maxlen		= sizeof(aio_nr),
1759 		.mode		= 0444,
1760 		.proc_handler	= proc_doulongvec_minmax,
1761 	},
1762 	{
1763 		.procname	= "aio-max-nr",
1764 		.data		= &aio_max_nr,
1765 		.maxlen		= sizeof(aio_max_nr),
1766 		.mode		= 0644,
1767 		.proc_handler	= proc_doulongvec_minmax,
1768 	},
1769 #endif /* CONFIG_AIO */
1770 #ifdef CONFIG_INOTIFY_USER
1771 	{
1772 		.procname	= "inotify",
1773 		.mode		= 0555,
1774 		.child		= inotify_table,
1775 	},
1776 #endif
1777 #ifdef CONFIG_EPOLL
1778 	{
1779 		.procname	= "epoll",
1780 		.mode		= 0555,
1781 		.child		= epoll_table,
1782 	},
1783 #endif
1784 #endif
1785 	{
1786 		.procname	= "protected_symlinks",
1787 		.data		= &sysctl_protected_symlinks,
1788 		.maxlen		= sizeof(int),
1789 		.mode		= 0600,
1790 		.proc_handler	= proc_dointvec_minmax,
1791 		.extra1		= &zero,
1792 		.extra2		= &one,
1793 	},
1794 	{
1795 		.procname	= "protected_hardlinks",
1796 		.data		= &sysctl_protected_hardlinks,
1797 		.maxlen		= sizeof(int),
1798 		.mode		= 0600,
1799 		.proc_handler	= proc_dointvec_minmax,
1800 		.extra1		= &zero,
1801 		.extra2		= &one,
1802 	},
1803 	{
1804 		.procname	= "suid_dumpable",
1805 		.data		= &suid_dumpable,
1806 		.maxlen		= sizeof(int),
1807 		.mode		= 0644,
1808 		.proc_handler	= proc_dointvec_minmax_coredump,
1809 		.extra1		= &zero,
1810 		.extra2		= &two,
1811 	},
1812 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1813 	{
1814 		.procname	= "binfmt_misc",
1815 		.mode		= 0555,
1816 		.child		= sysctl_mount_point,
1817 	},
1818 #endif
1819 	{
1820 		.procname	= "pipe-max-size",
1821 		.data		= &pipe_max_size,
1822 		.maxlen		= sizeof(int),
1823 		.mode		= 0644,
1824 		.proc_handler	= &pipe_proc_fn,
1825 		.extra1		= &pipe_min_size,
1826 	},
1827 	{
1828 		.procname	= "pipe-user-pages-hard",
1829 		.data		= &pipe_user_pages_hard,
1830 		.maxlen		= sizeof(pipe_user_pages_hard),
1831 		.mode		= 0644,
1832 		.proc_handler	= proc_doulongvec_minmax,
1833 	},
1834 	{
1835 		.procname	= "pipe-user-pages-soft",
1836 		.data		= &pipe_user_pages_soft,
1837 		.maxlen		= sizeof(pipe_user_pages_soft),
1838 		.mode		= 0644,
1839 		.proc_handler	= proc_doulongvec_minmax,
1840 	},
1841 	{ }
1842 };
1843 
1844 static struct ctl_table debug_table[] = {
1845 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1846 	{
1847 		.procname	= "exception-trace",
1848 		.data		= &show_unhandled_signals,
1849 		.maxlen		= sizeof(int),
1850 		.mode		= 0644,
1851 		.proc_handler	= proc_dointvec
1852 	},
1853 #endif
1854 #if defined(CONFIG_OPTPROBES)
1855 	{
1856 		.procname	= "kprobes-optimization",
1857 		.data		= &sysctl_kprobes_optimization,
1858 		.maxlen		= sizeof(int),
1859 		.mode		= 0644,
1860 		.proc_handler	= proc_kprobes_optimization_handler,
1861 		.extra1		= &zero,
1862 		.extra2		= &one,
1863 	},
1864 #endif
1865 	{ }
1866 };
1867 
1868 static struct ctl_table dev_table[] = {
1869 	{ }
1870 };
1871 
1872 int __init sysctl_init(void)
1873 {
1874 	struct ctl_table_header *hdr;
1875 
1876 	hdr = register_sysctl_table(sysctl_base_table);
1877 	kmemleak_not_leak(hdr);
1878 	return 0;
1879 }
1880 
1881 #endif /* CONFIG_SYSCTL */
1882 
1883 /*
1884  * /proc/sys support
1885  */
1886 
1887 #ifdef CONFIG_PROC_SYSCTL
1888 
1889 static int _proc_do_string(char *data, int maxlen, int write,
1890 			   char __user *buffer,
1891 			   size_t *lenp, loff_t *ppos)
1892 {
1893 	size_t len;
1894 	char __user *p;
1895 	char c;
1896 
1897 	if (!data || !maxlen || !*lenp) {
1898 		*lenp = 0;
1899 		return 0;
1900 	}
1901 
1902 	if (write) {
1903 		if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1904 			/* Only continue writes not past the end of buffer. */
1905 			len = strlen(data);
1906 			if (len > maxlen - 1)
1907 				len = maxlen - 1;
1908 
1909 			if (*ppos > len)
1910 				return 0;
1911 			len = *ppos;
1912 		} else {
1913 			/* Start writing from beginning of buffer. */
1914 			len = 0;
1915 		}
1916 
1917 		*ppos += *lenp;
1918 		p = buffer;
1919 		while ((p - buffer) < *lenp && len < maxlen - 1) {
1920 			if (get_user(c, p++))
1921 				return -EFAULT;
1922 			if (c == 0 || c == '\n')
1923 				break;
1924 			data[len++] = c;
1925 		}
1926 		data[len] = 0;
1927 	} else {
1928 		len = strlen(data);
1929 		if (len > maxlen)
1930 			len = maxlen;
1931 
1932 		if (*ppos > len) {
1933 			*lenp = 0;
1934 			return 0;
1935 		}
1936 
1937 		data += *ppos;
1938 		len  -= *ppos;
1939 
1940 		if (len > *lenp)
1941 			len = *lenp;
1942 		if (len)
1943 			if (copy_to_user(buffer, data, len))
1944 				return -EFAULT;
1945 		if (len < *lenp) {
1946 			if (put_user('\n', buffer + len))
1947 				return -EFAULT;
1948 			len++;
1949 		}
1950 		*lenp = len;
1951 		*ppos += len;
1952 	}
1953 	return 0;
1954 }
1955 
1956 static void warn_sysctl_write(struct ctl_table *table)
1957 {
1958 	pr_warn_once("%s wrote to %s when file position was not 0!\n"
1959 		"This will not be supported in the future. To silence this\n"
1960 		"warning, set kernel.sysctl_writes_strict = -1\n",
1961 		current->comm, table->procname);
1962 }
1963 
1964 /**
1965  * proc_dostring - read a string sysctl
1966  * @table: the sysctl table
1967  * @write: %TRUE if this is a write to the sysctl file
1968  * @buffer: the user buffer
1969  * @lenp: the size of the user buffer
1970  * @ppos: file position
1971  *
1972  * Reads/writes a string from/to the user buffer. If the kernel
1973  * buffer provided is not large enough to hold the string, the
1974  * string is truncated. The copied string is %NULL-terminated.
1975  * If the string is being read by the user process, it is copied
1976  * and a newline '\n' is added. It is truncated if the buffer is
1977  * not large enough.
1978  *
1979  * Returns 0 on success.
1980  */
1981 int proc_dostring(struct ctl_table *table, int write,
1982 		  void __user *buffer, size_t *lenp, loff_t *ppos)
1983 {
1984 	if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN)
1985 		warn_sysctl_write(table);
1986 
1987 	return _proc_do_string((char *)(table->data), table->maxlen, write,
1988 			       (char __user *)buffer, lenp, ppos);
1989 }
1990 
1991 static size_t proc_skip_spaces(char **buf)
1992 {
1993 	size_t ret;
1994 	char *tmp = skip_spaces(*buf);
1995 	ret = tmp - *buf;
1996 	*buf = tmp;
1997 	return ret;
1998 }
1999 
2000 static void proc_skip_char(char **buf, size_t *size, const char v)
2001 {
2002 	while (*size) {
2003 		if (**buf != v)
2004 			break;
2005 		(*size)--;
2006 		(*buf)++;
2007 	}
2008 }
2009 
2010 #define TMPBUFLEN 22
2011 /**
2012  * proc_get_long - reads an ASCII formatted integer from a user buffer
2013  *
2014  * @buf: a kernel buffer
2015  * @size: size of the kernel buffer
2016  * @val: this is where the number will be stored
2017  * @neg: set to %TRUE if number is negative
2018  * @perm_tr: a vector which contains the allowed trailers
2019  * @perm_tr_len: size of the perm_tr vector
2020  * @tr: pointer to store the trailer character
2021  *
2022  * In case of success %0 is returned and @buf and @size are updated with
2023  * the amount of bytes read. If @tr is non-NULL and a trailing
2024  * character exists (size is non-zero after returning from this
2025  * function), @tr is updated with the trailing character.
2026  */
2027 static int proc_get_long(char **buf, size_t *size,
2028 			  unsigned long *val, bool *neg,
2029 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
2030 {
2031 	int len;
2032 	char *p, tmp[TMPBUFLEN];
2033 
2034 	if (!*size)
2035 		return -EINVAL;
2036 
2037 	len = *size;
2038 	if (len > TMPBUFLEN - 1)
2039 		len = TMPBUFLEN - 1;
2040 
2041 	memcpy(tmp, *buf, len);
2042 
2043 	tmp[len] = 0;
2044 	p = tmp;
2045 	if (*p == '-' && *size > 1) {
2046 		*neg = true;
2047 		p++;
2048 	} else
2049 		*neg = false;
2050 	if (!isdigit(*p))
2051 		return -EINVAL;
2052 
2053 	*val = simple_strtoul(p, &p, 0);
2054 
2055 	len = p - tmp;
2056 
2057 	/* We don't know if the next char is whitespace thus we may accept
2058 	 * invalid integers (e.g. 1234...a) or two integers instead of one
2059 	 * (e.g. 123...1). So lets not allow such large numbers. */
2060 	if (len == TMPBUFLEN - 1)
2061 		return -EINVAL;
2062 
2063 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2064 		return -EINVAL;
2065 
2066 	if (tr && (len < *size))
2067 		*tr = *p;
2068 
2069 	*buf += len;
2070 	*size -= len;
2071 
2072 	return 0;
2073 }
2074 
2075 /**
2076  * proc_put_long - converts an integer to a decimal ASCII formatted string
2077  *
2078  * @buf: the user buffer
2079  * @size: the size of the user buffer
2080  * @val: the integer to be converted
2081  * @neg: sign of the number, %TRUE for negative
2082  *
2083  * In case of success %0 is returned and @buf and @size are updated with
2084  * the amount of bytes written.
2085  */
2086 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2087 			  bool neg)
2088 {
2089 	int len;
2090 	char tmp[TMPBUFLEN], *p = tmp;
2091 
2092 	sprintf(p, "%s%lu", neg ? "-" : "", val);
2093 	len = strlen(tmp);
2094 	if (len > *size)
2095 		len = *size;
2096 	if (copy_to_user(*buf, tmp, len))
2097 		return -EFAULT;
2098 	*size -= len;
2099 	*buf += len;
2100 	return 0;
2101 }
2102 #undef TMPBUFLEN
2103 
2104 static int proc_put_char(void __user **buf, size_t *size, char c)
2105 {
2106 	if (*size) {
2107 		char __user **buffer = (char __user **)buf;
2108 		if (put_user(c, *buffer))
2109 			return -EFAULT;
2110 		(*size)--, (*buffer)++;
2111 		*buf = *buffer;
2112 	}
2113 	return 0;
2114 }
2115 
2116 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2117 				 int *valp,
2118 				 int write, void *data)
2119 {
2120 	if (write) {
2121 		if (*negp) {
2122 			if (*lvalp > (unsigned long) INT_MAX + 1)
2123 				return -EINVAL;
2124 			*valp = -*lvalp;
2125 		} else {
2126 			if (*lvalp > (unsigned long) INT_MAX)
2127 				return -EINVAL;
2128 			*valp = *lvalp;
2129 		}
2130 	} else {
2131 		int val = *valp;
2132 		if (val < 0) {
2133 			*negp = true;
2134 			*lvalp = -(unsigned long)val;
2135 		} else {
2136 			*negp = false;
2137 			*lvalp = (unsigned long)val;
2138 		}
2139 	}
2140 	return 0;
2141 }
2142 
2143 static int do_proc_douintvec_conv(bool *negp, unsigned long *lvalp,
2144 				 int *valp,
2145 				 int write, void *data)
2146 {
2147 	if (write) {
2148 		if (*negp)
2149 			return -EINVAL;
2150 		*valp = *lvalp;
2151 	} else {
2152 		unsigned int val = *valp;
2153 		*lvalp = (unsigned long)val;
2154 	}
2155 	return 0;
2156 }
2157 
2158 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2159 
2160 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2161 		  int write, void __user *buffer,
2162 		  size_t *lenp, loff_t *ppos,
2163 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2164 			      int write, void *data),
2165 		  void *data)
2166 {
2167 	int *i, vleft, first = 1, err = 0;
2168 	size_t left;
2169 	char *kbuf = NULL, *p;
2170 
2171 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2172 		*lenp = 0;
2173 		return 0;
2174 	}
2175 
2176 	i = (int *) tbl_data;
2177 	vleft = table->maxlen / sizeof(*i);
2178 	left = *lenp;
2179 
2180 	if (!conv)
2181 		conv = do_proc_dointvec_conv;
2182 
2183 	if (write) {
2184 		if (*ppos) {
2185 			switch (sysctl_writes_strict) {
2186 			case SYSCTL_WRITES_STRICT:
2187 				goto out;
2188 			case SYSCTL_WRITES_WARN:
2189 				warn_sysctl_write(table);
2190 				break;
2191 			default:
2192 				break;
2193 			}
2194 		}
2195 
2196 		if (left > PAGE_SIZE - 1)
2197 			left = PAGE_SIZE - 1;
2198 		p = kbuf = memdup_user_nul(buffer, left);
2199 		if (IS_ERR(kbuf))
2200 			return PTR_ERR(kbuf);
2201 	}
2202 
2203 	for (; left && vleft--; i++, first=0) {
2204 		unsigned long lval;
2205 		bool neg;
2206 
2207 		if (write) {
2208 			left -= proc_skip_spaces(&p);
2209 
2210 			if (!left)
2211 				break;
2212 			err = proc_get_long(&p, &left, &lval, &neg,
2213 					     proc_wspace_sep,
2214 					     sizeof(proc_wspace_sep), NULL);
2215 			if (err)
2216 				break;
2217 			if (conv(&neg, &lval, i, 1, data)) {
2218 				err = -EINVAL;
2219 				break;
2220 			}
2221 		} else {
2222 			if (conv(&neg, &lval, i, 0, data)) {
2223 				err = -EINVAL;
2224 				break;
2225 			}
2226 			if (!first)
2227 				err = proc_put_char(&buffer, &left, '\t');
2228 			if (err)
2229 				break;
2230 			err = proc_put_long(&buffer, &left, lval, neg);
2231 			if (err)
2232 				break;
2233 		}
2234 	}
2235 
2236 	if (!write && !first && left && !err)
2237 		err = proc_put_char(&buffer, &left, '\n');
2238 	if (write && !err && left)
2239 		left -= proc_skip_spaces(&p);
2240 	if (write) {
2241 		kfree(kbuf);
2242 		if (first)
2243 			return err ? : -EINVAL;
2244 	}
2245 	*lenp -= left;
2246 out:
2247 	*ppos += *lenp;
2248 	return err;
2249 }
2250 
2251 static int do_proc_dointvec(struct ctl_table *table, int write,
2252 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2253 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2254 			      int write, void *data),
2255 		  void *data)
2256 {
2257 	return __do_proc_dointvec(table->data, table, write,
2258 			buffer, lenp, ppos, conv, data);
2259 }
2260 
2261 /**
2262  * proc_dointvec - read a vector of integers
2263  * @table: the sysctl table
2264  * @write: %TRUE if this is a write to the sysctl file
2265  * @buffer: the user buffer
2266  * @lenp: the size of the user buffer
2267  * @ppos: file position
2268  *
2269  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2270  * values from/to the user buffer, treated as an ASCII string.
2271  *
2272  * Returns 0 on success.
2273  */
2274 int proc_dointvec(struct ctl_table *table, int write,
2275 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2276 {
2277 	return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2278 }
2279 
2280 /**
2281  * proc_douintvec - read a vector of unsigned integers
2282  * @table: the sysctl table
2283  * @write: %TRUE if this is a write to the sysctl file
2284  * @buffer: the user buffer
2285  * @lenp: the size of the user buffer
2286  * @ppos: file position
2287  *
2288  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2289  * values from/to the user buffer, treated as an ASCII string.
2290  *
2291  * Returns 0 on success.
2292  */
2293 int proc_douintvec(struct ctl_table *table, int write,
2294 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2295 {
2296 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2297 				do_proc_douintvec_conv, NULL);
2298 }
2299 
2300 /*
2301  * Taint values can only be increased
2302  * This means we can safely use a temporary.
2303  */
2304 static int proc_taint(struct ctl_table *table, int write,
2305 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2306 {
2307 	struct ctl_table t;
2308 	unsigned long tmptaint = get_taint();
2309 	int err;
2310 
2311 	if (write && !capable(CAP_SYS_ADMIN))
2312 		return -EPERM;
2313 
2314 	t = *table;
2315 	t.data = &tmptaint;
2316 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2317 	if (err < 0)
2318 		return err;
2319 
2320 	if (write) {
2321 		/*
2322 		 * Poor man's atomic or. Not worth adding a primitive
2323 		 * to everyone's atomic.h for this
2324 		 */
2325 		int i;
2326 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2327 			if ((tmptaint >> i) & 1)
2328 				add_taint(i, LOCKDEP_STILL_OK);
2329 		}
2330 	}
2331 
2332 	return err;
2333 }
2334 
2335 #ifdef CONFIG_PRINTK
2336 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2337 				void __user *buffer, size_t *lenp, loff_t *ppos)
2338 {
2339 	if (write && !capable(CAP_SYS_ADMIN))
2340 		return -EPERM;
2341 
2342 	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2343 }
2344 #endif
2345 
2346 struct do_proc_dointvec_minmax_conv_param {
2347 	int *min;
2348 	int *max;
2349 };
2350 
2351 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2352 					int *valp,
2353 					int write, void *data)
2354 {
2355 	struct do_proc_dointvec_minmax_conv_param *param = data;
2356 	if (write) {
2357 		int val = *negp ? -*lvalp : *lvalp;
2358 		if ((param->min && *param->min > val) ||
2359 		    (param->max && *param->max < val))
2360 			return -EINVAL;
2361 		*valp = val;
2362 	} else {
2363 		int val = *valp;
2364 		if (val < 0) {
2365 			*negp = true;
2366 			*lvalp = -(unsigned long)val;
2367 		} else {
2368 			*negp = false;
2369 			*lvalp = (unsigned long)val;
2370 		}
2371 	}
2372 	return 0;
2373 }
2374 
2375 /**
2376  * proc_dointvec_minmax - read a vector of integers with min/max values
2377  * @table: the sysctl table
2378  * @write: %TRUE if this is a write to the sysctl file
2379  * @buffer: the user buffer
2380  * @lenp: the size of the user buffer
2381  * @ppos: file position
2382  *
2383  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2384  * values from/to the user buffer, treated as an ASCII string.
2385  *
2386  * This routine will ensure the values are within the range specified by
2387  * table->extra1 (min) and table->extra2 (max).
2388  *
2389  * Returns 0 on success.
2390  */
2391 int proc_dointvec_minmax(struct ctl_table *table, int write,
2392 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2393 {
2394 	struct do_proc_dointvec_minmax_conv_param param = {
2395 		.min = (int *) table->extra1,
2396 		.max = (int *) table->extra2,
2397 	};
2398 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2399 				do_proc_dointvec_minmax_conv, &param);
2400 }
2401 
2402 static void validate_coredump_safety(void)
2403 {
2404 #ifdef CONFIG_COREDUMP
2405 	if (suid_dumpable == SUID_DUMP_ROOT &&
2406 	    core_pattern[0] != '/' && core_pattern[0] != '|') {
2407 		printk(KERN_WARNING "Unsafe core_pattern used with "\
2408 			"suid_dumpable=2. Pipe handler or fully qualified "\
2409 			"core dump path required.\n");
2410 	}
2411 #endif
2412 }
2413 
2414 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2415 		void __user *buffer, size_t *lenp, loff_t *ppos)
2416 {
2417 	int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2418 	if (!error)
2419 		validate_coredump_safety();
2420 	return error;
2421 }
2422 
2423 #ifdef CONFIG_COREDUMP
2424 static int proc_dostring_coredump(struct ctl_table *table, int write,
2425 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2426 {
2427 	int error = proc_dostring(table, write, buffer, lenp, ppos);
2428 	if (!error)
2429 		validate_coredump_safety();
2430 	return error;
2431 }
2432 #endif
2433 
2434 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2435 				     void __user *buffer,
2436 				     size_t *lenp, loff_t *ppos,
2437 				     unsigned long convmul,
2438 				     unsigned long convdiv)
2439 {
2440 	unsigned long *i, *min, *max;
2441 	int vleft, first = 1, err = 0;
2442 	size_t left;
2443 	char *kbuf = NULL, *p;
2444 
2445 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2446 		*lenp = 0;
2447 		return 0;
2448 	}
2449 
2450 	i = (unsigned long *) data;
2451 	min = (unsigned long *) table->extra1;
2452 	max = (unsigned long *) table->extra2;
2453 	vleft = table->maxlen / sizeof(unsigned long);
2454 	left = *lenp;
2455 
2456 	if (write) {
2457 		if (*ppos) {
2458 			switch (sysctl_writes_strict) {
2459 			case SYSCTL_WRITES_STRICT:
2460 				goto out;
2461 			case SYSCTL_WRITES_WARN:
2462 				warn_sysctl_write(table);
2463 				break;
2464 			default:
2465 				break;
2466 			}
2467 		}
2468 
2469 		if (left > PAGE_SIZE - 1)
2470 			left = PAGE_SIZE - 1;
2471 		p = kbuf = memdup_user_nul(buffer, left);
2472 		if (IS_ERR(kbuf))
2473 			return PTR_ERR(kbuf);
2474 	}
2475 
2476 	for (; left && vleft--; i++, first = 0) {
2477 		unsigned long val;
2478 
2479 		if (write) {
2480 			bool neg;
2481 
2482 			left -= proc_skip_spaces(&p);
2483 
2484 			err = proc_get_long(&p, &left, &val, &neg,
2485 					     proc_wspace_sep,
2486 					     sizeof(proc_wspace_sep), NULL);
2487 			if (err)
2488 				break;
2489 			if (neg)
2490 				continue;
2491 			if ((min && val < *min) || (max && val > *max))
2492 				continue;
2493 			*i = val;
2494 		} else {
2495 			val = convdiv * (*i) / convmul;
2496 			if (!first) {
2497 				err = proc_put_char(&buffer, &left, '\t');
2498 				if (err)
2499 					break;
2500 			}
2501 			err = proc_put_long(&buffer, &left, val, false);
2502 			if (err)
2503 				break;
2504 		}
2505 	}
2506 
2507 	if (!write && !first && left && !err)
2508 		err = proc_put_char(&buffer, &left, '\n');
2509 	if (write && !err)
2510 		left -= proc_skip_spaces(&p);
2511 	if (write) {
2512 		kfree(kbuf);
2513 		if (first)
2514 			return err ? : -EINVAL;
2515 	}
2516 	*lenp -= left;
2517 out:
2518 	*ppos += *lenp;
2519 	return err;
2520 }
2521 
2522 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2523 				     void __user *buffer,
2524 				     size_t *lenp, loff_t *ppos,
2525 				     unsigned long convmul,
2526 				     unsigned long convdiv)
2527 {
2528 	return __do_proc_doulongvec_minmax(table->data, table, write,
2529 			buffer, lenp, ppos, convmul, convdiv);
2530 }
2531 
2532 /**
2533  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2534  * @table: the sysctl table
2535  * @write: %TRUE if this is a write to the sysctl file
2536  * @buffer: the user buffer
2537  * @lenp: the size of the user buffer
2538  * @ppos: file position
2539  *
2540  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2541  * values from/to the user buffer, treated as an ASCII string.
2542  *
2543  * This routine will ensure the values are within the range specified by
2544  * table->extra1 (min) and table->extra2 (max).
2545  *
2546  * Returns 0 on success.
2547  */
2548 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2549 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2550 {
2551     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2552 }
2553 
2554 /**
2555  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2556  * @table: the sysctl table
2557  * @write: %TRUE if this is a write to the sysctl file
2558  * @buffer: the user buffer
2559  * @lenp: the size of the user buffer
2560  * @ppos: file position
2561  *
2562  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2563  * values from/to the user buffer, treated as an ASCII string. The values
2564  * are treated as milliseconds, and converted to jiffies when they are stored.
2565  *
2566  * This routine will ensure the values are within the range specified by
2567  * table->extra1 (min) and table->extra2 (max).
2568  *
2569  * Returns 0 on success.
2570  */
2571 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2572 				      void __user *buffer,
2573 				      size_t *lenp, loff_t *ppos)
2574 {
2575     return do_proc_doulongvec_minmax(table, write, buffer,
2576 				     lenp, ppos, HZ, 1000l);
2577 }
2578 
2579 
2580 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2581 					 int *valp,
2582 					 int write, void *data)
2583 {
2584 	if (write) {
2585 		if (*lvalp > LONG_MAX / HZ)
2586 			return 1;
2587 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2588 	} else {
2589 		int val = *valp;
2590 		unsigned long lval;
2591 		if (val < 0) {
2592 			*negp = true;
2593 			lval = -(unsigned long)val;
2594 		} else {
2595 			*negp = false;
2596 			lval = (unsigned long)val;
2597 		}
2598 		*lvalp = lval / HZ;
2599 	}
2600 	return 0;
2601 }
2602 
2603 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2604 						int *valp,
2605 						int write, void *data)
2606 {
2607 	if (write) {
2608 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2609 			return 1;
2610 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2611 	} else {
2612 		int val = *valp;
2613 		unsigned long lval;
2614 		if (val < 0) {
2615 			*negp = true;
2616 			lval = -(unsigned long)val;
2617 		} else {
2618 			*negp = false;
2619 			lval = (unsigned long)val;
2620 		}
2621 		*lvalp = jiffies_to_clock_t(lval);
2622 	}
2623 	return 0;
2624 }
2625 
2626 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2627 					    int *valp,
2628 					    int write, void *data)
2629 {
2630 	if (write) {
2631 		unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2632 
2633 		if (jif > INT_MAX)
2634 			return 1;
2635 		*valp = (int)jif;
2636 	} else {
2637 		int val = *valp;
2638 		unsigned long lval;
2639 		if (val < 0) {
2640 			*negp = true;
2641 			lval = -(unsigned long)val;
2642 		} else {
2643 			*negp = false;
2644 			lval = (unsigned long)val;
2645 		}
2646 		*lvalp = jiffies_to_msecs(lval);
2647 	}
2648 	return 0;
2649 }
2650 
2651 /**
2652  * proc_dointvec_jiffies - read a vector of integers as seconds
2653  * @table: the sysctl table
2654  * @write: %TRUE if this is a write to the sysctl file
2655  * @buffer: the user buffer
2656  * @lenp: the size of the user buffer
2657  * @ppos: file position
2658  *
2659  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2660  * values from/to the user buffer, treated as an ASCII string.
2661  * The values read are assumed to be in seconds, and are converted into
2662  * jiffies.
2663  *
2664  * Returns 0 on success.
2665  */
2666 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2667 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2668 {
2669     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2670 		    	    do_proc_dointvec_jiffies_conv,NULL);
2671 }
2672 
2673 /**
2674  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2675  * @table: the sysctl table
2676  * @write: %TRUE if this is a write to the sysctl file
2677  * @buffer: the user buffer
2678  * @lenp: the size of the user buffer
2679  * @ppos: pointer to the file position
2680  *
2681  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2682  * values from/to the user buffer, treated as an ASCII string.
2683  * The values read are assumed to be in 1/USER_HZ seconds, and
2684  * are converted into jiffies.
2685  *
2686  * Returns 0 on success.
2687  */
2688 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2689 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2690 {
2691     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2692 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2693 }
2694 
2695 /**
2696  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2697  * @table: the sysctl table
2698  * @write: %TRUE if this is a write to the sysctl file
2699  * @buffer: the user buffer
2700  * @lenp: the size of the user buffer
2701  * @ppos: file position
2702  * @ppos: the current position in the file
2703  *
2704  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2705  * values from/to the user buffer, treated as an ASCII string.
2706  * The values read are assumed to be in 1/1000 seconds, and
2707  * are converted into jiffies.
2708  *
2709  * Returns 0 on success.
2710  */
2711 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2712 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2713 {
2714 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2715 				do_proc_dointvec_ms_jiffies_conv, NULL);
2716 }
2717 
2718 static int proc_do_cad_pid(struct ctl_table *table, int write,
2719 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2720 {
2721 	struct pid *new_pid;
2722 	pid_t tmp;
2723 	int r;
2724 
2725 	tmp = pid_vnr(cad_pid);
2726 
2727 	r = __do_proc_dointvec(&tmp, table, write, buffer,
2728 			       lenp, ppos, NULL, NULL);
2729 	if (r || !write)
2730 		return r;
2731 
2732 	new_pid = find_get_pid(tmp);
2733 	if (!new_pid)
2734 		return -ESRCH;
2735 
2736 	put_pid(xchg(&cad_pid, new_pid));
2737 	return 0;
2738 }
2739 
2740 /**
2741  * proc_do_large_bitmap - read/write from/to a large bitmap
2742  * @table: the sysctl table
2743  * @write: %TRUE if this is a write to the sysctl file
2744  * @buffer: the user buffer
2745  * @lenp: the size of the user buffer
2746  * @ppos: file position
2747  *
2748  * The bitmap is stored at table->data and the bitmap length (in bits)
2749  * in table->maxlen.
2750  *
2751  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2752  * large bitmaps may be represented in a compact manner. Writing into
2753  * the file will clear the bitmap then update it with the given input.
2754  *
2755  * Returns 0 on success.
2756  */
2757 int proc_do_large_bitmap(struct ctl_table *table, int write,
2758 			 void __user *buffer, size_t *lenp, loff_t *ppos)
2759 {
2760 	int err = 0;
2761 	bool first = 1;
2762 	size_t left = *lenp;
2763 	unsigned long bitmap_len = table->maxlen;
2764 	unsigned long *bitmap = *(unsigned long **) table->data;
2765 	unsigned long *tmp_bitmap = NULL;
2766 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2767 
2768 	if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2769 		*lenp = 0;
2770 		return 0;
2771 	}
2772 
2773 	if (write) {
2774 		char *kbuf, *p;
2775 
2776 		if (left > PAGE_SIZE - 1)
2777 			left = PAGE_SIZE - 1;
2778 
2779 		p = kbuf = memdup_user_nul(buffer, left);
2780 		if (IS_ERR(kbuf))
2781 			return PTR_ERR(kbuf);
2782 
2783 		tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2784 				     GFP_KERNEL);
2785 		if (!tmp_bitmap) {
2786 			kfree(kbuf);
2787 			return -ENOMEM;
2788 		}
2789 		proc_skip_char(&p, &left, '\n');
2790 		while (!err && left) {
2791 			unsigned long val_a, val_b;
2792 			bool neg;
2793 
2794 			err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
2795 					     sizeof(tr_a), &c);
2796 			if (err)
2797 				break;
2798 			if (val_a >= bitmap_len || neg) {
2799 				err = -EINVAL;
2800 				break;
2801 			}
2802 
2803 			val_b = val_a;
2804 			if (left) {
2805 				p++;
2806 				left--;
2807 			}
2808 
2809 			if (c == '-') {
2810 				err = proc_get_long(&p, &left, &val_b,
2811 						     &neg, tr_b, sizeof(tr_b),
2812 						     &c);
2813 				if (err)
2814 					break;
2815 				if (val_b >= bitmap_len || neg ||
2816 				    val_a > val_b) {
2817 					err = -EINVAL;
2818 					break;
2819 				}
2820 				if (left) {
2821 					p++;
2822 					left--;
2823 				}
2824 			}
2825 
2826 			bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2827 			first = 0;
2828 			proc_skip_char(&p, &left, '\n');
2829 		}
2830 		kfree(kbuf);
2831 	} else {
2832 		unsigned long bit_a, bit_b = 0;
2833 
2834 		while (left) {
2835 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2836 			if (bit_a >= bitmap_len)
2837 				break;
2838 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
2839 						   bit_a + 1) - 1;
2840 
2841 			if (!first) {
2842 				err = proc_put_char(&buffer, &left, ',');
2843 				if (err)
2844 					break;
2845 			}
2846 			err = proc_put_long(&buffer, &left, bit_a, false);
2847 			if (err)
2848 				break;
2849 			if (bit_a != bit_b) {
2850 				err = proc_put_char(&buffer, &left, '-');
2851 				if (err)
2852 					break;
2853 				err = proc_put_long(&buffer, &left, bit_b, false);
2854 				if (err)
2855 					break;
2856 			}
2857 
2858 			first = 0; bit_b++;
2859 		}
2860 		if (!err)
2861 			err = proc_put_char(&buffer, &left, '\n');
2862 	}
2863 
2864 	if (!err) {
2865 		if (write) {
2866 			if (*ppos)
2867 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2868 			else
2869 				bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2870 		}
2871 		kfree(tmp_bitmap);
2872 		*lenp -= left;
2873 		*ppos += *lenp;
2874 		return 0;
2875 	} else {
2876 		kfree(tmp_bitmap);
2877 		return err;
2878 	}
2879 }
2880 
2881 #else /* CONFIG_PROC_SYSCTL */
2882 
2883 int proc_dostring(struct ctl_table *table, int write,
2884 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2885 {
2886 	return -ENOSYS;
2887 }
2888 
2889 int proc_dointvec(struct ctl_table *table, int write,
2890 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2891 {
2892 	return -ENOSYS;
2893 }
2894 
2895 int proc_douintvec(struct ctl_table *table, int write,
2896 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2897 {
2898 	return -ENOSYS;
2899 }
2900 
2901 int proc_dointvec_minmax(struct ctl_table *table, int write,
2902 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2903 {
2904 	return -ENOSYS;
2905 }
2906 
2907 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2908 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2909 {
2910 	return -ENOSYS;
2911 }
2912 
2913 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2914 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2915 {
2916 	return -ENOSYS;
2917 }
2918 
2919 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2920 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2921 {
2922 	return -ENOSYS;
2923 }
2924 
2925 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2926 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2927 {
2928 	return -ENOSYS;
2929 }
2930 
2931 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2932 				      void __user *buffer,
2933 				      size_t *lenp, loff_t *ppos)
2934 {
2935     return -ENOSYS;
2936 }
2937 
2938 
2939 #endif /* CONFIG_PROC_SYSCTL */
2940 
2941 /*
2942  * No sense putting this after each symbol definition, twice,
2943  * exception granted :-)
2944  */
2945 EXPORT_SYMBOL(proc_dointvec);
2946 EXPORT_SYMBOL(proc_douintvec);
2947 EXPORT_SYMBOL(proc_dointvec_jiffies);
2948 EXPORT_SYMBOL(proc_dointvec_minmax);
2949 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2950 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2951 EXPORT_SYMBOL(proc_dostring);
2952 EXPORT_SYMBOL(proc_doulongvec_minmax);
2953 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2954