xref: /openbmc/linux/kernel/sysctl.c (revision 4beec1d7)
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 #include <linux/mount.h>
69 
70 #include <linux/uaccess.h>
71 #include <asm/processor.h>
72 
73 #ifdef CONFIG_X86
74 #include <asm/nmi.h>
75 #include <asm/stacktrace.h>
76 #include <asm/io.h>
77 #endif
78 #ifdef CONFIG_SPARC
79 #include <asm/setup.h>
80 #endif
81 #ifdef CONFIG_BSD_PROCESS_ACCT
82 #include <linux/acct.h>
83 #endif
84 #ifdef CONFIG_RT_MUTEXES
85 #include <linux/rtmutex.h>
86 #endif
87 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
88 #include <linux/lockdep.h>
89 #endif
90 #ifdef CONFIG_CHR_DEV_SG
91 #include <scsi/sg.h>
92 #endif
93 
94 #ifdef CONFIG_LOCKUP_DETECTOR
95 #include <linux/nmi.h>
96 #endif
97 
98 #if defined(CONFIG_SYSCTL)
99 
100 /* External variables not in a header file. */
101 extern int suid_dumpable;
102 #ifdef CONFIG_COREDUMP
103 extern int core_uses_pid;
104 extern char core_pattern[];
105 extern unsigned int core_pipe_limit;
106 #endif
107 extern int pid_max;
108 extern int pid_max_min, pid_max_max;
109 extern int percpu_pagelist_fraction;
110 extern int latencytop_enabled;
111 extern unsigned 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 #ifdef CONFIG_SCHEDSTATS
351 	{
352 		.procname	= "sched_schedstats",
353 		.data		= NULL,
354 		.maxlen		= sizeof(unsigned int),
355 		.mode		= 0644,
356 		.proc_handler	= sysctl_schedstats,
357 		.extra1		= &zero,
358 		.extra2		= &one,
359 	},
360 #endif /* CONFIG_SCHEDSTATS */
361 #endif /* CONFIG_SMP */
362 #ifdef CONFIG_NUMA_BALANCING
363 	{
364 		.procname	= "numa_balancing_scan_delay_ms",
365 		.data		= &sysctl_numa_balancing_scan_delay,
366 		.maxlen		= sizeof(unsigned int),
367 		.mode		= 0644,
368 		.proc_handler	= proc_dointvec,
369 	},
370 	{
371 		.procname	= "numa_balancing_scan_period_min_ms",
372 		.data		= &sysctl_numa_balancing_scan_period_min,
373 		.maxlen		= sizeof(unsigned int),
374 		.mode		= 0644,
375 		.proc_handler	= proc_dointvec,
376 	},
377 	{
378 		.procname	= "numa_balancing_scan_period_max_ms",
379 		.data		= &sysctl_numa_balancing_scan_period_max,
380 		.maxlen		= sizeof(unsigned int),
381 		.mode		= 0644,
382 		.proc_handler	= proc_dointvec,
383 	},
384 	{
385 		.procname	= "numa_balancing_scan_size_mb",
386 		.data		= &sysctl_numa_balancing_scan_size,
387 		.maxlen		= sizeof(unsigned int),
388 		.mode		= 0644,
389 		.proc_handler	= proc_dointvec_minmax,
390 		.extra1		= &one,
391 	},
392 	{
393 		.procname	= "numa_balancing",
394 		.data		= NULL, /* filled in by handler */
395 		.maxlen		= sizeof(unsigned int),
396 		.mode		= 0644,
397 		.proc_handler	= sysctl_numa_balancing,
398 		.extra1		= &zero,
399 		.extra2		= &one,
400 	},
401 #endif /* CONFIG_NUMA_BALANCING */
402 #endif /* CONFIG_SCHED_DEBUG */
403 	{
404 		.procname	= "sched_rt_period_us",
405 		.data		= &sysctl_sched_rt_period,
406 		.maxlen		= sizeof(unsigned int),
407 		.mode		= 0644,
408 		.proc_handler	= sched_rt_handler,
409 	},
410 	{
411 		.procname	= "sched_rt_runtime_us",
412 		.data		= &sysctl_sched_rt_runtime,
413 		.maxlen		= sizeof(int),
414 		.mode		= 0644,
415 		.proc_handler	= sched_rt_handler,
416 	},
417 	{
418 		.procname	= "sched_rr_timeslice_ms",
419 		.data		= &sched_rr_timeslice,
420 		.maxlen		= sizeof(int),
421 		.mode		= 0644,
422 		.proc_handler	= sched_rr_handler,
423 	},
424 #ifdef CONFIG_SCHED_AUTOGROUP
425 	{
426 		.procname	= "sched_autogroup_enabled",
427 		.data		= &sysctl_sched_autogroup_enabled,
428 		.maxlen		= sizeof(unsigned int),
429 		.mode		= 0644,
430 		.proc_handler	= proc_dointvec_minmax,
431 		.extra1		= &zero,
432 		.extra2		= &one,
433 	},
434 #endif
435 #ifdef CONFIG_CFS_BANDWIDTH
436 	{
437 		.procname	= "sched_cfs_bandwidth_slice_us",
438 		.data		= &sysctl_sched_cfs_bandwidth_slice,
439 		.maxlen		= sizeof(unsigned int),
440 		.mode		= 0644,
441 		.proc_handler	= proc_dointvec_minmax,
442 		.extra1		= &one,
443 	},
444 #endif
445 #ifdef CONFIG_PROVE_LOCKING
446 	{
447 		.procname	= "prove_locking",
448 		.data		= &prove_locking,
449 		.maxlen		= sizeof(int),
450 		.mode		= 0644,
451 		.proc_handler	= proc_dointvec,
452 	},
453 #endif
454 #ifdef CONFIG_LOCK_STAT
455 	{
456 		.procname	= "lock_stat",
457 		.data		= &lock_stat,
458 		.maxlen		= sizeof(int),
459 		.mode		= 0644,
460 		.proc_handler	= proc_dointvec,
461 	},
462 #endif
463 	{
464 		.procname	= "panic",
465 		.data		= &panic_timeout,
466 		.maxlen		= sizeof(int),
467 		.mode		= 0644,
468 		.proc_handler	= proc_dointvec,
469 	},
470 #ifdef CONFIG_COREDUMP
471 	{
472 		.procname	= "core_uses_pid",
473 		.data		= &core_uses_pid,
474 		.maxlen		= sizeof(int),
475 		.mode		= 0644,
476 		.proc_handler	= proc_dointvec,
477 	},
478 	{
479 		.procname	= "core_pattern",
480 		.data		= core_pattern,
481 		.maxlen		= CORENAME_MAX_SIZE,
482 		.mode		= 0644,
483 		.proc_handler	= proc_dostring_coredump,
484 	},
485 	{
486 		.procname	= "core_pipe_limit",
487 		.data		= &core_pipe_limit,
488 		.maxlen		= sizeof(unsigned int),
489 		.mode		= 0644,
490 		.proc_handler	= proc_dointvec,
491 	},
492 #endif
493 #ifdef CONFIG_PROC_SYSCTL
494 	{
495 		.procname	= "tainted",
496 		.maxlen 	= sizeof(long),
497 		.mode		= 0644,
498 		.proc_handler	= proc_taint,
499 	},
500 	{
501 		.procname	= "sysctl_writes_strict",
502 		.data		= &sysctl_writes_strict,
503 		.maxlen		= sizeof(int),
504 		.mode		= 0644,
505 		.proc_handler	= proc_dointvec_minmax,
506 		.extra1		= &neg_one,
507 		.extra2		= &one,
508 	},
509 #endif
510 #ifdef CONFIG_LATENCYTOP
511 	{
512 		.procname	= "latencytop",
513 		.data		= &latencytop_enabled,
514 		.maxlen		= sizeof(int),
515 		.mode		= 0644,
516 		.proc_handler	= sysctl_latencytop,
517 	},
518 #endif
519 #ifdef CONFIG_BLK_DEV_INITRD
520 	{
521 		.procname	= "real-root-dev",
522 		.data		= &real_root_dev,
523 		.maxlen		= sizeof(int),
524 		.mode		= 0644,
525 		.proc_handler	= proc_dointvec,
526 	},
527 #endif
528 	{
529 		.procname	= "print-fatal-signals",
530 		.data		= &print_fatal_signals,
531 		.maxlen		= sizeof(int),
532 		.mode		= 0644,
533 		.proc_handler	= proc_dointvec,
534 	},
535 #ifdef CONFIG_SPARC
536 	{
537 		.procname	= "reboot-cmd",
538 		.data		= reboot_command,
539 		.maxlen		= 256,
540 		.mode		= 0644,
541 		.proc_handler	= proc_dostring,
542 	},
543 	{
544 		.procname	= "stop-a",
545 		.data		= &stop_a_enabled,
546 		.maxlen		= sizeof (int),
547 		.mode		= 0644,
548 		.proc_handler	= proc_dointvec,
549 	},
550 	{
551 		.procname	= "scons-poweroff",
552 		.data		= &scons_pwroff,
553 		.maxlen		= sizeof (int),
554 		.mode		= 0644,
555 		.proc_handler	= proc_dointvec,
556 	},
557 #endif
558 #ifdef CONFIG_SPARC64
559 	{
560 		.procname	= "tsb-ratio",
561 		.data		= &sysctl_tsb_ratio,
562 		.maxlen		= sizeof (int),
563 		.mode		= 0644,
564 		.proc_handler	= proc_dointvec,
565 	},
566 #endif
567 #ifdef __hppa__
568 	{
569 		.procname	= "soft-power",
570 		.data		= &pwrsw_enabled,
571 		.maxlen		= sizeof (int),
572 	 	.mode		= 0644,
573 		.proc_handler	= proc_dointvec,
574 	},
575 #endif
576 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
577 	{
578 		.procname	= "unaligned-trap",
579 		.data		= &unaligned_enabled,
580 		.maxlen		= sizeof (int),
581 		.mode		= 0644,
582 		.proc_handler	= proc_dointvec,
583 	},
584 #endif
585 	{
586 		.procname	= "ctrl-alt-del",
587 		.data		= &C_A_D,
588 		.maxlen		= sizeof(int),
589 		.mode		= 0644,
590 		.proc_handler	= proc_dointvec,
591 	},
592 #ifdef CONFIG_FUNCTION_TRACER
593 	{
594 		.procname	= "ftrace_enabled",
595 		.data		= &ftrace_enabled,
596 		.maxlen		= sizeof(int),
597 		.mode		= 0644,
598 		.proc_handler	= ftrace_enable_sysctl,
599 	},
600 #endif
601 #ifdef CONFIG_STACK_TRACER
602 	{
603 		.procname	= "stack_tracer_enabled",
604 		.data		= &stack_tracer_enabled,
605 		.maxlen		= sizeof(int),
606 		.mode		= 0644,
607 		.proc_handler	= stack_trace_sysctl,
608 	},
609 #endif
610 #ifdef CONFIG_TRACING
611 	{
612 		.procname	= "ftrace_dump_on_oops",
613 		.data		= &ftrace_dump_on_oops,
614 		.maxlen		= sizeof(int),
615 		.mode		= 0644,
616 		.proc_handler	= proc_dointvec,
617 	},
618 	{
619 		.procname	= "traceoff_on_warning",
620 		.data		= &__disable_trace_on_warning,
621 		.maxlen		= sizeof(__disable_trace_on_warning),
622 		.mode		= 0644,
623 		.proc_handler	= proc_dointvec,
624 	},
625 	{
626 		.procname	= "tracepoint_printk",
627 		.data		= &tracepoint_printk,
628 		.maxlen		= sizeof(tracepoint_printk),
629 		.mode		= 0644,
630 		.proc_handler	= tracepoint_printk_sysctl,
631 	},
632 #endif
633 #ifdef CONFIG_KEXEC_CORE
634 	{
635 		.procname	= "kexec_load_disabled",
636 		.data		= &kexec_load_disabled,
637 		.maxlen		= sizeof(int),
638 		.mode		= 0644,
639 		/* only handle a transition from default "0" to "1" */
640 		.proc_handler	= proc_dointvec_minmax,
641 		.extra1		= &one,
642 		.extra2		= &one,
643 	},
644 #endif
645 #ifdef CONFIG_MODULES
646 	{
647 		.procname	= "modprobe",
648 		.data		= &modprobe_path,
649 		.maxlen		= KMOD_PATH_LEN,
650 		.mode		= 0644,
651 		.proc_handler	= proc_dostring,
652 	},
653 	{
654 		.procname	= "modules_disabled",
655 		.data		= &modules_disabled,
656 		.maxlen		= sizeof(int),
657 		.mode		= 0644,
658 		/* only handle a transition from default "0" to "1" */
659 		.proc_handler	= proc_dointvec_minmax,
660 		.extra1		= &one,
661 		.extra2		= &one,
662 	},
663 #endif
664 #ifdef CONFIG_UEVENT_HELPER
665 	{
666 		.procname	= "hotplug",
667 		.data		= &uevent_helper,
668 		.maxlen		= UEVENT_HELPER_PATH_LEN,
669 		.mode		= 0644,
670 		.proc_handler	= proc_dostring,
671 	},
672 #endif
673 #ifdef CONFIG_CHR_DEV_SG
674 	{
675 		.procname	= "sg-big-buff",
676 		.data		= &sg_big_buff,
677 		.maxlen		= sizeof (int),
678 		.mode		= 0444,
679 		.proc_handler	= proc_dointvec,
680 	},
681 #endif
682 #ifdef CONFIG_BSD_PROCESS_ACCT
683 	{
684 		.procname	= "acct",
685 		.data		= &acct_parm,
686 		.maxlen		= 3*sizeof(int),
687 		.mode		= 0644,
688 		.proc_handler	= proc_dointvec,
689 	},
690 #endif
691 #ifdef CONFIG_MAGIC_SYSRQ
692 	{
693 		.procname	= "sysrq",
694 		.data		= &__sysrq_enabled,
695 		.maxlen		= sizeof (int),
696 		.mode		= 0644,
697 		.proc_handler	= sysrq_sysctl_handler,
698 	},
699 #endif
700 #ifdef CONFIG_PROC_SYSCTL
701 	{
702 		.procname	= "cad_pid",
703 		.data		= NULL,
704 		.maxlen		= sizeof (int),
705 		.mode		= 0600,
706 		.proc_handler	= proc_do_cad_pid,
707 	},
708 #endif
709 	{
710 		.procname	= "threads-max",
711 		.data		= NULL,
712 		.maxlen		= sizeof(int),
713 		.mode		= 0644,
714 		.proc_handler	= sysctl_max_threads,
715 	},
716 	{
717 		.procname	= "random",
718 		.mode		= 0555,
719 		.child		= random_table,
720 	},
721 	{
722 		.procname	= "usermodehelper",
723 		.mode		= 0555,
724 		.child		= usermodehelper_table,
725 	},
726 	{
727 		.procname	= "overflowuid",
728 		.data		= &overflowuid,
729 		.maxlen		= sizeof(int),
730 		.mode		= 0644,
731 		.proc_handler	= proc_dointvec_minmax,
732 		.extra1		= &minolduid,
733 		.extra2		= &maxolduid,
734 	},
735 	{
736 		.procname	= "overflowgid",
737 		.data		= &overflowgid,
738 		.maxlen		= sizeof(int),
739 		.mode		= 0644,
740 		.proc_handler	= proc_dointvec_minmax,
741 		.extra1		= &minolduid,
742 		.extra2		= &maxolduid,
743 	},
744 #ifdef CONFIG_S390
745 #ifdef CONFIG_MATHEMU
746 	{
747 		.procname	= "ieee_emulation_warnings",
748 		.data		= &sysctl_ieee_emulation_warnings,
749 		.maxlen		= sizeof(int),
750 		.mode		= 0644,
751 		.proc_handler	= proc_dointvec,
752 	},
753 #endif
754 	{
755 		.procname	= "userprocess_debug",
756 		.data		= &show_unhandled_signals,
757 		.maxlen		= sizeof(int),
758 		.mode		= 0644,
759 		.proc_handler	= proc_dointvec,
760 	},
761 #endif
762 	{
763 		.procname	= "pid_max",
764 		.data		= &pid_max,
765 		.maxlen		= sizeof (int),
766 		.mode		= 0644,
767 		.proc_handler	= proc_dointvec_minmax,
768 		.extra1		= &pid_max_min,
769 		.extra2		= &pid_max_max,
770 	},
771 	{
772 		.procname	= "panic_on_oops",
773 		.data		= &panic_on_oops,
774 		.maxlen		= sizeof(int),
775 		.mode		= 0644,
776 		.proc_handler	= proc_dointvec,
777 	},
778 #if defined CONFIG_PRINTK
779 	{
780 		.procname	= "printk",
781 		.data		= &console_loglevel,
782 		.maxlen		= 4*sizeof(int),
783 		.mode		= 0644,
784 		.proc_handler	= proc_dointvec,
785 	},
786 	{
787 		.procname	= "printk_ratelimit",
788 		.data		= &printk_ratelimit_state.interval,
789 		.maxlen		= sizeof(int),
790 		.mode		= 0644,
791 		.proc_handler	= proc_dointvec_jiffies,
792 	},
793 	{
794 		.procname	= "printk_ratelimit_burst",
795 		.data		= &printk_ratelimit_state.burst,
796 		.maxlen		= sizeof(int),
797 		.mode		= 0644,
798 		.proc_handler	= proc_dointvec,
799 	},
800 	{
801 		.procname	= "printk_delay",
802 		.data		= &printk_delay_msec,
803 		.maxlen		= sizeof(int),
804 		.mode		= 0644,
805 		.proc_handler	= proc_dointvec_minmax,
806 		.extra1		= &zero,
807 		.extra2		= &ten_thousand,
808 	},
809 	{
810 		.procname	= "printk_devkmsg",
811 		.data		= devkmsg_log_str,
812 		.maxlen		= DEVKMSG_STR_MAX_SIZE,
813 		.mode		= 0644,
814 		.proc_handler	= devkmsg_sysctl_set_loglvl,
815 	},
816 	{
817 		.procname	= "dmesg_restrict",
818 		.data		= &dmesg_restrict,
819 		.maxlen		= sizeof(int),
820 		.mode		= 0644,
821 		.proc_handler	= proc_dointvec_minmax_sysadmin,
822 		.extra1		= &zero,
823 		.extra2		= &one,
824 	},
825 	{
826 		.procname	= "kptr_restrict",
827 		.data		= &kptr_restrict,
828 		.maxlen		= sizeof(int),
829 		.mode		= 0644,
830 		.proc_handler	= proc_dointvec_minmax_sysadmin,
831 		.extra1		= &zero,
832 		.extra2		= &two,
833 	},
834 #endif
835 	{
836 		.procname	= "ngroups_max",
837 		.data		= &ngroups_max,
838 		.maxlen		= sizeof (int),
839 		.mode		= 0444,
840 		.proc_handler	= proc_dointvec,
841 	},
842 	{
843 		.procname	= "cap_last_cap",
844 		.data		= (void *)&cap_last_cap,
845 		.maxlen		= sizeof(int),
846 		.mode		= 0444,
847 		.proc_handler	= proc_dointvec,
848 	},
849 #if defined(CONFIG_LOCKUP_DETECTOR)
850 	{
851 		.procname       = "watchdog",
852 		.data           = &watchdog_user_enabled,
853 		.maxlen         = sizeof (int),
854 		.mode           = 0644,
855 		.proc_handler   = proc_watchdog,
856 		.extra1		= &zero,
857 		.extra2		= &one,
858 	},
859 	{
860 		.procname	= "watchdog_thresh",
861 		.data		= &watchdog_thresh,
862 		.maxlen		= sizeof(int),
863 		.mode		= 0644,
864 		.proc_handler	= proc_watchdog_thresh,
865 		.extra1		= &zero,
866 		.extra2		= &sixty,
867 	},
868 	{
869 		.procname       = "nmi_watchdog",
870 		.data           = &nmi_watchdog_enabled,
871 		.maxlen         = sizeof (int),
872 		.mode           = 0644,
873 		.proc_handler   = proc_nmi_watchdog,
874 		.extra1		= &zero,
875 #if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR)
876 		.extra2		= &one,
877 #else
878 		.extra2		= &zero,
879 #endif
880 	},
881 	{
882 		.procname       = "soft_watchdog",
883 		.data           = &soft_watchdog_enabled,
884 		.maxlen         = sizeof (int),
885 		.mode           = 0644,
886 		.proc_handler   = proc_soft_watchdog,
887 		.extra1		= &zero,
888 		.extra2		= &one,
889 	},
890 	{
891 		.procname	= "watchdog_cpumask",
892 		.data		= &watchdog_cpumask_bits,
893 		.maxlen		= NR_CPUS,
894 		.mode		= 0644,
895 		.proc_handler	= proc_watchdog_cpumask,
896 	},
897 	{
898 		.procname	= "softlockup_panic",
899 		.data		= &softlockup_panic,
900 		.maxlen		= sizeof(int),
901 		.mode		= 0644,
902 		.proc_handler	= proc_dointvec_minmax,
903 		.extra1		= &zero,
904 		.extra2		= &one,
905 	},
906 #ifdef CONFIG_HARDLOCKUP_DETECTOR
907 	{
908 		.procname	= "hardlockup_panic",
909 		.data		= &hardlockup_panic,
910 		.maxlen		= sizeof(int),
911 		.mode		= 0644,
912 		.proc_handler	= proc_dointvec_minmax,
913 		.extra1		= &zero,
914 		.extra2		= &one,
915 	},
916 #endif
917 #ifdef CONFIG_SMP
918 	{
919 		.procname	= "softlockup_all_cpu_backtrace",
920 		.data		= &sysctl_softlockup_all_cpu_backtrace,
921 		.maxlen		= sizeof(int),
922 		.mode		= 0644,
923 		.proc_handler	= proc_dointvec_minmax,
924 		.extra1		= &zero,
925 		.extra2		= &one,
926 	},
927 	{
928 		.procname	= "hardlockup_all_cpu_backtrace",
929 		.data		= &sysctl_hardlockup_all_cpu_backtrace,
930 		.maxlen		= sizeof(int),
931 		.mode		= 0644,
932 		.proc_handler	= proc_dointvec_minmax,
933 		.extra1		= &zero,
934 		.extra2		= &one,
935 	},
936 #endif /* CONFIG_SMP */
937 #endif
938 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
939 	{
940 		.procname       = "unknown_nmi_panic",
941 		.data           = &unknown_nmi_panic,
942 		.maxlen         = sizeof (int),
943 		.mode           = 0644,
944 		.proc_handler   = proc_dointvec,
945 	},
946 #endif
947 #if defined(CONFIG_X86)
948 	{
949 		.procname	= "panic_on_unrecovered_nmi",
950 		.data		= &panic_on_unrecovered_nmi,
951 		.maxlen		= sizeof(int),
952 		.mode		= 0644,
953 		.proc_handler	= proc_dointvec,
954 	},
955 	{
956 		.procname	= "panic_on_io_nmi",
957 		.data		= &panic_on_io_nmi,
958 		.maxlen		= sizeof(int),
959 		.mode		= 0644,
960 		.proc_handler	= proc_dointvec,
961 	},
962 #ifdef CONFIG_DEBUG_STACKOVERFLOW
963 	{
964 		.procname	= "panic_on_stackoverflow",
965 		.data		= &sysctl_panic_on_stackoverflow,
966 		.maxlen		= sizeof(int),
967 		.mode		= 0644,
968 		.proc_handler	= proc_dointvec,
969 	},
970 #endif
971 	{
972 		.procname	= "bootloader_type",
973 		.data		= &bootloader_type,
974 		.maxlen		= sizeof (int),
975 		.mode		= 0444,
976 		.proc_handler	= proc_dointvec,
977 	},
978 	{
979 		.procname	= "bootloader_version",
980 		.data		= &bootloader_version,
981 		.maxlen		= sizeof (int),
982 		.mode		= 0444,
983 		.proc_handler	= proc_dointvec,
984 	},
985 	{
986 		.procname	= "io_delay_type",
987 		.data		= &io_delay_type,
988 		.maxlen		= sizeof(int),
989 		.mode		= 0644,
990 		.proc_handler	= proc_dointvec,
991 	},
992 #endif
993 #if defined(CONFIG_MMU)
994 	{
995 		.procname	= "randomize_va_space",
996 		.data		= &randomize_va_space,
997 		.maxlen		= sizeof(int),
998 		.mode		= 0644,
999 		.proc_handler	= proc_dointvec,
1000 	},
1001 #endif
1002 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1003 	{
1004 		.procname	= "spin_retry",
1005 		.data		= &spin_retry,
1006 		.maxlen		= sizeof (int),
1007 		.mode		= 0644,
1008 		.proc_handler	= proc_dointvec,
1009 	},
1010 #endif
1011 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1012 	{
1013 		.procname	= "acpi_video_flags",
1014 		.data		= &acpi_realmode_flags,
1015 		.maxlen		= sizeof (unsigned long),
1016 		.mode		= 0644,
1017 		.proc_handler	= proc_doulongvec_minmax,
1018 	},
1019 #endif
1020 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1021 	{
1022 		.procname	= "ignore-unaligned-usertrap",
1023 		.data		= &no_unaligned_warning,
1024 		.maxlen		= sizeof (int),
1025 	 	.mode		= 0644,
1026 		.proc_handler	= proc_dointvec,
1027 	},
1028 #endif
1029 #ifdef CONFIG_IA64
1030 	{
1031 		.procname	= "unaligned-dump-stack",
1032 		.data		= &unaligned_dump_stack,
1033 		.maxlen		= sizeof (int),
1034 		.mode		= 0644,
1035 		.proc_handler	= proc_dointvec,
1036 	},
1037 #endif
1038 #ifdef CONFIG_DETECT_HUNG_TASK
1039 	{
1040 		.procname	= "hung_task_panic",
1041 		.data		= &sysctl_hung_task_panic,
1042 		.maxlen		= sizeof(int),
1043 		.mode		= 0644,
1044 		.proc_handler	= proc_dointvec_minmax,
1045 		.extra1		= &zero,
1046 		.extra2		= &one,
1047 	},
1048 	{
1049 		.procname	= "hung_task_check_count",
1050 		.data		= &sysctl_hung_task_check_count,
1051 		.maxlen		= sizeof(int),
1052 		.mode		= 0644,
1053 		.proc_handler	= proc_dointvec_minmax,
1054 		.extra1		= &zero,
1055 	},
1056 	{
1057 		.procname	= "hung_task_timeout_secs",
1058 		.data		= &sysctl_hung_task_timeout_secs,
1059 		.maxlen		= sizeof(unsigned long),
1060 		.mode		= 0644,
1061 		.proc_handler	= proc_dohung_task_timeout_secs,
1062 		.extra2		= &hung_task_timeout_max,
1063 	},
1064 	{
1065 		.procname	= "hung_task_warnings",
1066 		.data		= &sysctl_hung_task_warnings,
1067 		.maxlen		= sizeof(int),
1068 		.mode		= 0644,
1069 		.proc_handler	= proc_dointvec_minmax,
1070 		.extra1		= &neg_one,
1071 	},
1072 #endif
1073 #ifdef CONFIG_RT_MUTEXES
1074 	{
1075 		.procname	= "max_lock_depth",
1076 		.data		= &max_lock_depth,
1077 		.maxlen		= sizeof(int),
1078 		.mode		= 0644,
1079 		.proc_handler	= proc_dointvec,
1080 	},
1081 #endif
1082 	{
1083 		.procname	= "poweroff_cmd",
1084 		.data		= &poweroff_cmd,
1085 		.maxlen		= POWEROFF_CMD_PATH_LEN,
1086 		.mode		= 0644,
1087 		.proc_handler	= proc_dostring,
1088 	},
1089 #ifdef CONFIG_KEYS
1090 	{
1091 		.procname	= "keys",
1092 		.mode		= 0555,
1093 		.child		= key_sysctls,
1094 	},
1095 #endif
1096 #ifdef CONFIG_PERF_EVENTS
1097 	/*
1098 	 * User-space scripts rely on the existence of this file
1099 	 * as a feature check for perf_events being enabled.
1100 	 *
1101 	 * So it's an ABI, do not remove!
1102 	 */
1103 	{
1104 		.procname	= "perf_event_paranoid",
1105 		.data		= &sysctl_perf_event_paranoid,
1106 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
1107 		.mode		= 0644,
1108 		.proc_handler	= proc_dointvec,
1109 	},
1110 	{
1111 		.procname	= "perf_event_mlock_kb",
1112 		.data		= &sysctl_perf_event_mlock,
1113 		.maxlen		= sizeof(sysctl_perf_event_mlock),
1114 		.mode		= 0644,
1115 		.proc_handler	= proc_dointvec,
1116 	},
1117 	{
1118 		.procname	= "perf_event_max_sample_rate",
1119 		.data		= &sysctl_perf_event_sample_rate,
1120 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
1121 		.mode		= 0644,
1122 		.proc_handler	= perf_proc_update_handler,
1123 		.extra1		= &one,
1124 	},
1125 	{
1126 		.procname	= "perf_cpu_time_max_percent",
1127 		.data		= &sysctl_perf_cpu_time_max_percent,
1128 		.maxlen		= sizeof(sysctl_perf_cpu_time_max_percent),
1129 		.mode		= 0644,
1130 		.proc_handler	= perf_cpu_time_max_percent_handler,
1131 		.extra1		= &zero,
1132 		.extra2		= &one_hundred,
1133 	},
1134 	{
1135 		.procname	= "perf_event_max_stack",
1136 		.data		= &sysctl_perf_event_max_stack,
1137 		.maxlen		= sizeof(sysctl_perf_event_max_stack),
1138 		.mode		= 0644,
1139 		.proc_handler	= perf_event_max_stack_handler,
1140 		.extra1		= &zero,
1141 		.extra2		= &six_hundred_forty_kb,
1142 	},
1143 	{
1144 		.procname	= "perf_event_max_contexts_per_stack",
1145 		.data		= &sysctl_perf_event_max_contexts_per_stack,
1146 		.maxlen		= sizeof(sysctl_perf_event_max_contexts_per_stack),
1147 		.mode		= 0644,
1148 		.proc_handler	= perf_event_max_stack_handler,
1149 		.extra1		= &zero,
1150 		.extra2		= &one_thousand,
1151 	},
1152 #endif
1153 #ifdef CONFIG_KMEMCHECK
1154 	{
1155 		.procname	= "kmemcheck",
1156 		.data		= &kmemcheck_enabled,
1157 		.maxlen		= sizeof(int),
1158 		.mode		= 0644,
1159 		.proc_handler	= proc_dointvec,
1160 	},
1161 #endif
1162 	{
1163 		.procname	= "panic_on_warn",
1164 		.data		= &panic_on_warn,
1165 		.maxlen		= sizeof(int),
1166 		.mode		= 0644,
1167 		.proc_handler	= proc_dointvec_minmax,
1168 		.extra1		= &zero,
1169 		.extra2		= &one,
1170 	},
1171 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1172 	{
1173 		.procname	= "timer_migration",
1174 		.data		= &sysctl_timer_migration,
1175 		.maxlen		= sizeof(unsigned int),
1176 		.mode		= 0644,
1177 		.proc_handler	= timer_migration_handler,
1178 	},
1179 #endif
1180 #ifdef CONFIG_BPF_SYSCALL
1181 	{
1182 		.procname	= "unprivileged_bpf_disabled",
1183 		.data		= &sysctl_unprivileged_bpf_disabled,
1184 		.maxlen		= sizeof(sysctl_unprivileged_bpf_disabled),
1185 		.mode		= 0644,
1186 		/* only handle a transition from default "0" to "1" */
1187 		.proc_handler	= proc_dointvec_minmax,
1188 		.extra1		= &one,
1189 		.extra2		= &one,
1190 	},
1191 #endif
1192 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1193 	{
1194 		.procname	= "panic_on_rcu_stall",
1195 		.data		= &sysctl_panic_on_rcu_stall,
1196 		.maxlen		= sizeof(sysctl_panic_on_rcu_stall),
1197 		.mode		= 0644,
1198 		.proc_handler	= proc_dointvec_minmax,
1199 		.extra1		= &zero,
1200 		.extra2		= &one,
1201 	},
1202 #endif
1203 	{ }
1204 };
1205 
1206 static struct ctl_table vm_table[] = {
1207 	{
1208 		.procname	= "overcommit_memory",
1209 		.data		= &sysctl_overcommit_memory,
1210 		.maxlen		= sizeof(sysctl_overcommit_memory),
1211 		.mode		= 0644,
1212 		.proc_handler	= proc_dointvec_minmax,
1213 		.extra1		= &zero,
1214 		.extra2		= &two,
1215 	},
1216 	{
1217 		.procname	= "panic_on_oom",
1218 		.data		= &sysctl_panic_on_oom,
1219 		.maxlen		= sizeof(sysctl_panic_on_oom),
1220 		.mode		= 0644,
1221 		.proc_handler	= proc_dointvec_minmax,
1222 		.extra1		= &zero,
1223 		.extra2		= &two,
1224 	},
1225 	{
1226 		.procname	= "oom_kill_allocating_task",
1227 		.data		= &sysctl_oom_kill_allocating_task,
1228 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1229 		.mode		= 0644,
1230 		.proc_handler	= proc_dointvec,
1231 	},
1232 	{
1233 		.procname	= "oom_dump_tasks",
1234 		.data		= &sysctl_oom_dump_tasks,
1235 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1236 		.mode		= 0644,
1237 		.proc_handler	= proc_dointvec,
1238 	},
1239 	{
1240 		.procname	= "overcommit_ratio",
1241 		.data		= &sysctl_overcommit_ratio,
1242 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1243 		.mode		= 0644,
1244 		.proc_handler	= overcommit_ratio_handler,
1245 	},
1246 	{
1247 		.procname	= "overcommit_kbytes",
1248 		.data		= &sysctl_overcommit_kbytes,
1249 		.maxlen		= sizeof(sysctl_overcommit_kbytes),
1250 		.mode		= 0644,
1251 		.proc_handler	= overcommit_kbytes_handler,
1252 	},
1253 	{
1254 		.procname	= "page-cluster",
1255 		.data		= &page_cluster,
1256 		.maxlen		= sizeof(int),
1257 		.mode		= 0644,
1258 		.proc_handler	= proc_dointvec_minmax,
1259 		.extra1		= &zero,
1260 	},
1261 	{
1262 		.procname	= "dirty_background_ratio",
1263 		.data		= &dirty_background_ratio,
1264 		.maxlen		= sizeof(dirty_background_ratio),
1265 		.mode		= 0644,
1266 		.proc_handler	= dirty_background_ratio_handler,
1267 		.extra1		= &zero,
1268 		.extra2		= &one_hundred,
1269 	},
1270 	{
1271 		.procname	= "dirty_background_bytes",
1272 		.data		= &dirty_background_bytes,
1273 		.maxlen		= sizeof(dirty_background_bytes),
1274 		.mode		= 0644,
1275 		.proc_handler	= dirty_background_bytes_handler,
1276 		.extra1		= &one_ul,
1277 	},
1278 	{
1279 		.procname	= "dirty_ratio",
1280 		.data		= &vm_dirty_ratio,
1281 		.maxlen		= sizeof(vm_dirty_ratio),
1282 		.mode		= 0644,
1283 		.proc_handler	= dirty_ratio_handler,
1284 		.extra1		= &zero,
1285 		.extra2		= &one_hundred,
1286 	},
1287 	{
1288 		.procname	= "dirty_bytes",
1289 		.data		= &vm_dirty_bytes,
1290 		.maxlen		= sizeof(vm_dirty_bytes),
1291 		.mode		= 0644,
1292 		.proc_handler	= dirty_bytes_handler,
1293 		.extra1		= &dirty_bytes_min,
1294 	},
1295 	{
1296 		.procname	= "dirty_writeback_centisecs",
1297 		.data		= &dirty_writeback_interval,
1298 		.maxlen		= sizeof(dirty_writeback_interval),
1299 		.mode		= 0644,
1300 		.proc_handler	= dirty_writeback_centisecs_handler,
1301 	},
1302 	{
1303 		.procname	= "dirty_expire_centisecs",
1304 		.data		= &dirty_expire_interval,
1305 		.maxlen		= sizeof(dirty_expire_interval),
1306 		.mode		= 0644,
1307 		.proc_handler	= proc_dointvec_minmax,
1308 		.extra1		= &zero,
1309 	},
1310 	{
1311 		.procname	= "dirtytime_expire_seconds",
1312 		.data		= &dirtytime_expire_interval,
1313 		.maxlen		= sizeof(dirty_expire_interval),
1314 		.mode		= 0644,
1315 		.proc_handler	= dirtytime_interval_handler,
1316 		.extra1		= &zero,
1317 	},
1318 	{
1319 		.procname       = "nr_pdflush_threads",
1320 		.mode           = 0444 /* read-only */,
1321 		.proc_handler   = pdflush_proc_obsolete,
1322 	},
1323 	{
1324 		.procname	= "swappiness",
1325 		.data		= &vm_swappiness,
1326 		.maxlen		= sizeof(vm_swappiness),
1327 		.mode		= 0644,
1328 		.proc_handler	= proc_dointvec_minmax,
1329 		.extra1		= &zero,
1330 		.extra2		= &one_hundred,
1331 	},
1332 #ifdef CONFIG_HUGETLB_PAGE
1333 	{
1334 		.procname	= "nr_hugepages",
1335 		.data		= NULL,
1336 		.maxlen		= sizeof(unsigned long),
1337 		.mode		= 0644,
1338 		.proc_handler	= hugetlb_sysctl_handler,
1339 	},
1340 #ifdef CONFIG_NUMA
1341 	{
1342 		.procname       = "nr_hugepages_mempolicy",
1343 		.data           = NULL,
1344 		.maxlen         = sizeof(unsigned long),
1345 		.mode           = 0644,
1346 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1347 	},
1348 #endif
1349 	 {
1350 		.procname	= "hugetlb_shm_group",
1351 		.data		= &sysctl_hugetlb_shm_group,
1352 		.maxlen		= sizeof(gid_t),
1353 		.mode		= 0644,
1354 		.proc_handler	= proc_dointvec,
1355 	 },
1356 	 {
1357 		.procname	= "hugepages_treat_as_movable",
1358 		.data		= &hugepages_treat_as_movable,
1359 		.maxlen		= sizeof(int),
1360 		.mode		= 0644,
1361 		.proc_handler	= proc_dointvec,
1362 	},
1363 	{
1364 		.procname	= "nr_overcommit_hugepages",
1365 		.data		= NULL,
1366 		.maxlen		= sizeof(unsigned long),
1367 		.mode		= 0644,
1368 		.proc_handler	= hugetlb_overcommit_handler,
1369 	},
1370 #endif
1371 	{
1372 		.procname	= "lowmem_reserve_ratio",
1373 		.data		= &sysctl_lowmem_reserve_ratio,
1374 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1375 		.mode		= 0644,
1376 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1377 	},
1378 	{
1379 		.procname	= "drop_caches",
1380 		.data		= &sysctl_drop_caches,
1381 		.maxlen		= sizeof(int),
1382 		.mode		= 0644,
1383 		.proc_handler	= drop_caches_sysctl_handler,
1384 		.extra1		= &one,
1385 		.extra2		= &four,
1386 	},
1387 #ifdef CONFIG_COMPACTION
1388 	{
1389 		.procname	= "compact_memory",
1390 		.data		= &sysctl_compact_memory,
1391 		.maxlen		= sizeof(int),
1392 		.mode		= 0200,
1393 		.proc_handler	= sysctl_compaction_handler,
1394 	},
1395 	{
1396 		.procname	= "extfrag_threshold",
1397 		.data		= &sysctl_extfrag_threshold,
1398 		.maxlen		= sizeof(int),
1399 		.mode		= 0644,
1400 		.proc_handler	= sysctl_extfrag_handler,
1401 		.extra1		= &min_extfrag_threshold,
1402 		.extra2		= &max_extfrag_threshold,
1403 	},
1404 	{
1405 		.procname	= "compact_unevictable_allowed",
1406 		.data		= &sysctl_compact_unevictable_allowed,
1407 		.maxlen		= sizeof(int),
1408 		.mode		= 0644,
1409 		.proc_handler	= proc_dointvec,
1410 		.extra1		= &zero,
1411 		.extra2		= &one,
1412 	},
1413 
1414 #endif /* CONFIG_COMPACTION */
1415 	{
1416 		.procname	= "min_free_kbytes",
1417 		.data		= &min_free_kbytes,
1418 		.maxlen		= sizeof(min_free_kbytes),
1419 		.mode		= 0644,
1420 		.proc_handler	= min_free_kbytes_sysctl_handler,
1421 		.extra1		= &zero,
1422 	},
1423 	{
1424 		.procname	= "watermark_scale_factor",
1425 		.data		= &watermark_scale_factor,
1426 		.maxlen		= sizeof(watermark_scale_factor),
1427 		.mode		= 0644,
1428 		.proc_handler	= watermark_scale_factor_sysctl_handler,
1429 		.extra1		= &one,
1430 		.extra2		= &one_thousand,
1431 	},
1432 	{
1433 		.procname	= "percpu_pagelist_fraction",
1434 		.data		= &percpu_pagelist_fraction,
1435 		.maxlen		= sizeof(percpu_pagelist_fraction),
1436 		.mode		= 0644,
1437 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1438 		.extra1		= &zero,
1439 	},
1440 #ifdef CONFIG_MMU
1441 	{
1442 		.procname	= "max_map_count",
1443 		.data		= &sysctl_max_map_count,
1444 		.maxlen		= sizeof(sysctl_max_map_count),
1445 		.mode		= 0644,
1446 		.proc_handler	= proc_dointvec_minmax,
1447 		.extra1		= &zero,
1448 	},
1449 #else
1450 	{
1451 		.procname	= "nr_trim_pages",
1452 		.data		= &sysctl_nr_trim_pages,
1453 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1454 		.mode		= 0644,
1455 		.proc_handler	= proc_dointvec_minmax,
1456 		.extra1		= &zero,
1457 	},
1458 #endif
1459 	{
1460 		.procname	= "laptop_mode",
1461 		.data		= &laptop_mode,
1462 		.maxlen		= sizeof(laptop_mode),
1463 		.mode		= 0644,
1464 		.proc_handler	= proc_dointvec_jiffies,
1465 	},
1466 	{
1467 		.procname	= "block_dump",
1468 		.data		= &block_dump,
1469 		.maxlen		= sizeof(block_dump),
1470 		.mode		= 0644,
1471 		.proc_handler	= proc_dointvec,
1472 		.extra1		= &zero,
1473 	},
1474 	{
1475 		.procname	= "vfs_cache_pressure",
1476 		.data		= &sysctl_vfs_cache_pressure,
1477 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1478 		.mode		= 0644,
1479 		.proc_handler	= proc_dointvec,
1480 		.extra1		= &zero,
1481 	},
1482 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1483 	{
1484 		.procname	= "legacy_va_layout",
1485 		.data		= &sysctl_legacy_va_layout,
1486 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1487 		.mode		= 0644,
1488 		.proc_handler	= proc_dointvec,
1489 		.extra1		= &zero,
1490 	},
1491 #endif
1492 #ifdef CONFIG_NUMA
1493 	{
1494 		.procname	= "zone_reclaim_mode",
1495 		.data		= &node_reclaim_mode,
1496 		.maxlen		= sizeof(node_reclaim_mode),
1497 		.mode		= 0644,
1498 		.proc_handler	= proc_dointvec,
1499 		.extra1		= &zero,
1500 	},
1501 	{
1502 		.procname	= "min_unmapped_ratio",
1503 		.data		= &sysctl_min_unmapped_ratio,
1504 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1505 		.mode		= 0644,
1506 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1507 		.extra1		= &zero,
1508 		.extra2		= &one_hundred,
1509 	},
1510 	{
1511 		.procname	= "min_slab_ratio",
1512 		.data		= &sysctl_min_slab_ratio,
1513 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1514 		.mode		= 0644,
1515 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1516 		.extra1		= &zero,
1517 		.extra2		= &one_hundred,
1518 	},
1519 #endif
1520 #ifdef CONFIG_SMP
1521 	{
1522 		.procname	= "stat_interval",
1523 		.data		= &sysctl_stat_interval,
1524 		.maxlen		= sizeof(sysctl_stat_interval),
1525 		.mode		= 0644,
1526 		.proc_handler	= proc_dointvec_jiffies,
1527 	},
1528 	{
1529 		.procname	= "stat_refresh",
1530 		.data		= NULL,
1531 		.maxlen		= 0,
1532 		.mode		= 0600,
1533 		.proc_handler	= vmstat_refresh,
1534 	},
1535 #endif
1536 #ifdef CONFIG_MMU
1537 	{
1538 		.procname	= "mmap_min_addr",
1539 		.data		= &dac_mmap_min_addr,
1540 		.maxlen		= sizeof(unsigned long),
1541 		.mode		= 0644,
1542 		.proc_handler	= mmap_min_addr_handler,
1543 	},
1544 #endif
1545 #ifdef CONFIG_NUMA
1546 	{
1547 		.procname	= "numa_zonelist_order",
1548 		.data		= &numa_zonelist_order,
1549 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1550 		.mode		= 0644,
1551 		.proc_handler	= numa_zonelist_order_handler,
1552 	},
1553 #endif
1554 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1555    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1556 	{
1557 		.procname	= "vdso_enabled",
1558 #ifdef CONFIG_X86_32
1559 		.data		= &vdso32_enabled,
1560 		.maxlen		= sizeof(vdso32_enabled),
1561 #else
1562 		.data		= &vdso_enabled,
1563 		.maxlen		= sizeof(vdso_enabled),
1564 #endif
1565 		.mode		= 0644,
1566 		.proc_handler	= proc_dointvec,
1567 		.extra1		= &zero,
1568 	},
1569 #endif
1570 #ifdef CONFIG_HIGHMEM
1571 	{
1572 		.procname	= "highmem_is_dirtyable",
1573 		.data		= &vm_highmem_is_dirtyable,
1574 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1575 		.mode		= 0644,
1576 		.proc_handler	= proc_dointvec_minmax,
1577 		.extra1		= &zero,
1578 		.extra2		= &one,
1579 	},
1580 #endif
1581 #ifdef CONFIG_MEMORY_FAILURE
1582 	{
1583 		.procname	= "memory_failure_early_kill",
1584 		.data		= &sysctl_memory_failure_early_kill,
1585 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1586 		.mode		= 0644,
1587 		.proc_handler	= proc_dointvec_minmax,
1588 		.extra1		= &zero,
1589 		.extra2		= &one,
1590 	},
1591 	{
1592 		.procname	= "memory_failure_recovery",
1593 		.data		= &sysctl_memory_failure_recovery,
1594 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1595 		.mode		= 0644,
1596 		.proc_handler	= proc_dointvec_minmax,
1597 		.extra1		= &zero,
1598 		.extra2		= &one,
1599 	},
1600 #endif
1601 	{
1602 		.procname	= "user_reserve_kbytes",
1603 		.data		= &sysctl_user_reserve_kbytes,
1604 		.maxlen		= sizeof(sysctl_user_reserve_kbytes),
1605 		.mode		= 0644,
1606 		.proc_handler	= proc_doulongvec_minmax,
1607 	},
1608 	{
1609 		.procname	= "admin_reserve_kbytes",
1610 		.data		= &sysctl_admin_reserve_kbytes,
1611 		.maxlen		= sizeof(sysctl_admin_reserve_kbytes),
1612 		.mode		= 0644,
1613 		.proc_handler	= proc_doulongvec_minmax,
1614 	},
1615 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1616 	{
1617 		.procname	= "mmap_rnd_bits",
1618 		.data		= &mmap_rnd_bits,
1619 		.maxlen		= sizeof(mmap_rnd_bits),
1620 		.mode		= 0600,
1621 		.proc_handler	= proc_dointvec_minmax,
1622 		.extra1		= (void *)&mmap_rnd_bits_min,
1623 		.extra2		= (void *)&mmap_rnd_bits_max,
1624 	},
1625 #endif
1626 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1627 	{
1628 		.procname	= "mmap_rnd_compat_bits",
1629 		.data		= &mmap_rnd_compat_bits,
1630 		.maxlen		= sizeof(mmap_rnd_compat_bits),
1631 		.mode		= 0600,
1632 		.proc_handler	= proc_dointvec_minmax,
1633 		.extra1		= (void *)&mmap_rnd_compat_bits_min,
1634 		.extra2		= (void *)&mmap_rnd_compat_bits_max,
1635 	},
1636 #endif
1637 	{ }
1638 };
1639 
1640 static struct ctl_table fs_table[] = {
1641 	{
1642 		.procname	= "inode-nr",
1643 		.data		= &inodes_stat,
1644 		.maxlen		= 2*sizeof(long),
1645 		.mode		= 0444,
1646 		.proc_handler	= proc_nr_inodes,
1647 	},
1648 	{
1649 		.procname	= "inode-state",
1650 		.data		= &inodes_stat,
1651 		.maxlen		= 7*sizeof(long),
1652 		.mode		= 0444,
1653 		.proc_handler	= proc_nr_inodes,
1654 	},
1655 	{
1656 		.procname	= "file-nr",
1657 		.data		= &files_stat,
1658 		.maxlen		= sizeof(files_stat),
1659 		.mode		= 0444,
1660 		.proc_handler	= proc_nr_files,
1661 	},
1662 	{
1663 		.procname	= "file-max",
1664 		.data		= &files_stat.max_files,
1665 		.maxlen		= sizeof(files_stat.max_files),
1666 		.mode		= 0644,
1667 		.proc_handler	= proc_doulongvec_minmax,
1668 	},
1669 	{
1670 		.procname	= "nr_open",
1671 		.data		= &sysctl_nr_open,
1672 		.maxlen		= sizeof(unsigned int),
1673 		.mode		= 0644,
1674 		.proc_handler	= proc_dointvec_minmax,
1675 		.extra1		= &sysctl_nr_open_min,
1676 		.extra2		= &sysctl_nr_open_max,
1677 	},
1678 	{
1679 		.procname	= "dentry-state",
1680 		.data		= &dentry_stat,
1681 		.maxlen		= 6*sizeof(long),
1682 		.mode		= 0444,
1683 		.proc_handler	= proc_nr_dentry,
1684 	},
1685 	{
1686 		.procname	= "overflowuid",
1687 		.data		= &fs_overflowuid,
1688 		.maxlen		= sizeof(int),
1689 		.mode		= 0644,
1690 		.proc_handler	= proc_dointvec_minmax,
1691 		.extra1		= &minolduid,
1692 		.extra2		= &maxolduid,
1693 	},
1694 	{
1695 		.procname	= "overflowgid",
1696 		.data		= &fs_overflowgid,
1697 		.maxlen		= sizeof(int),
1698 		.mode		= 0644,
1699 		.proc_handler	= proc_dointvec_minmax,
1700 		.extra1		= &minolduid,
1701 		.extra2		= &maxolduid,
1702 	},
1703 #ifdef CONFIG_FILE_LOCKING
1704 	{
1705 		.procname	= "leases-enable",
1706 		.data		= &leases_enable,
1707 		.maxlen		= sizeof(int),
1708 		.mode		= 0644,
1709 		.proc_handler	= proc_dointvec,
1710 	},
1711 #endif
1712 #ifdef CONFIG_DNOTIFY
1713 	{
1714 		.procname	= "dir-notify-enable",
1715 		.data		= &dir_notify_enable,
1716 		.maxlen		= sizeof(int),
1717 		.mode		= 0644,
1718 		.proc_handler	= proc_dointvec,
1719 	},
1720 #endif
1721 #ifdef CONFIG_MMU
1722 #ifdef CONFIG_FILE_LOCKING
1723 	{
1724 		.procname	= "lease-break-time",
1725 		.data		= &lease_break_time,
1726 		.maxlen		= sizeof(int),
1727 		.mode		= 0644,
1728 		.proc_handler	= proc_dointvec,
1729 	},
1730 #endif
1731 #ifdef CONFIG_AIO
1732 	{
1733 		.procname	= "aio-nr",
1734 		.data		= &aio_nr,
1735 		.maxlen		= sizeof(aio_nr),
1736 		.mode		= 0444,
1737 		.proc_handler	= proc_doulongvec_minmax,
1738 	},
1739 	{
1740 		.procname	= "aio-max-nr",
1741 		.data		= &aio_max_nr,
1742 		.maxlen		= sizeof(aio_max_nr),
1743 		.mode		= 0644,
1744 		.proc_handler	= proc_doulongvec_minmax,
1745 	},
1746 #endif /* CONFIG_AIO */
1747 #ifdef CONFIG_INOTIFY_USER
1748 	{
1749 		.procname	= "inotify",
1750 		.mode		= 0555,
1751 		.child		= inotify_table,
1752 	},
1753 #endif
1754 #ifdef CONFIG_EPOLL
1755 	{
1756 		.procname	= "epoll",
1757 		.mode		= 0555,
1758 		.child		= epoll_table,
1759 	},
1760 #endif
1761 #endif
1762 	{
1763 		.procname	= "protected_symlinks",
1764 		.data		= &sysctl_protected_symlinks,
1765 		.maxlen		= sizeof(int),
1766 		.mode		= 0600,
1767 		.proc_handler	= proc_dointvec_minmax,
1768 		.extra1		= &zero,
1769 		.extra2		= &one,
1770 	},
1771 	{
1772 		.procname	= "protected_hardlinks",
1773 		.data		= &sysctl_protected_hardlinks,
1774 		.maxlen		= sizeof(int),
1775 		.mode		= 0600,
1776 		.proc_handler	= proc_dointvec_minmax,
1777 		.extra1		= &zero,
1778 		.extra2		= &one,
1779 	},
1780 	{
1781 		.procname	= "suid_dumpable",
1782 		.data		= &suid_dumpable,
1783 		.maxlen		= sizeof(int),
1784 		.mode		= 0644,
1785 		.proc_handler	= proc_dointvec_minmax_coredump,
1786 		.extra1		= &zero,
1787 		.extra2		= &two,
1788 	},
1789 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1790 	{
1791 		.procname	= "binfmt_misc",
1792 		.mode		= 0555,
1793 		.child		= sysctl_mount_point,
1794 	},
1795 #endif
1796 	{
1797 		.procname	= "pipe-max-size",
1798 		.data		= &pipe_max_size,
1799 		.maxlen		= sizeof(int),
1800 		.mode		= 0644,
1801 		.proc_handler	= &pipe_proc_fn,
1802 		.extra1		= &pipe_min_size,
1803 	},
1804 	{
1805 		.procname	= "pipe-user-pages-hard",
1806 		.data		= &pipe_user_pages_hard,
1807 		.maxlen		= sizeof(pipe_user_pages_hard),
1808 		.mode		= 0644,
1809 		.proc_handler	= proc_doulongvec_minmax,
1810 	},
1811 	{
1812 		.procname	= "pipe-user-pages-soft",
1813 		.data		= &pipe_user_pages_soft,
1814 		.maxlen		= sizeof(pipe_user_pages_soft),
1815 		.mode		= 0644,
1816 		.proc_handler	= proc_doulongvec_minmax,
1817 	},
1818 	{
1819 		.procname	= "mount-max",
1820 		.data		= &sysctl_mount_max,
1821 		.maxlen		= sizeof(unsigned int),
1822 		.mode		= 0644,
1823 		.proc_handler	= proc_dointvec_minmax,
1824 		.extra1		= &one,
1825 	},
1826 	{ }
1827 };
1828 
1829 static struct ctl_table debug_table[] = {
1830 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1831 	{
1832 		.procname	= "exception-trace",
1833 		.data		= &show_unhandled_signals,
1834 		.maxlen		= sizeof(int),
1835 		.mode		= 0644,
1836 		.proc_handler	= proc_dointvec
1837 	},
1838 #endif
1839 #if defined(CONFIG_OPTPROBES)
1840 	{
1841 		.procname	= "kprobes-optimization",
1842 		.data		= &sysctl_kprobes_optimization,
1843 		.maxlen		= sizeof(int),
1844 		.mode		= 0644,
1845 		.proc_handler	= proc_kprobes_optimization_handler,
1846 		.extra1		= &zero,
1847 		.extra2		= &one,
1848 	},
1849 #endif
1850 	{ }
1851 };
1852 
1853 static struct ctl_table dev_table[] = {
1854 	{ }
1855 };
1856 
1857 int __init sysctl_init(void)
1858 {
1859 	struct ctl_table_header *hdr;
1860 
1861 	hdr = register_sysctl_table(sysctl_base_table);
1862 	kmemleak_not_leak(hdr);
1863 	return 0;
1864 }
1865 
1866 #endif /* CONFIG_SYSCTL */
1867 
1868 /*
1869  * /proc/sys support
1870  */
1871 
1872 #ifdef CONFIG_PROC_SYSCTL
1873 
1874 static int _proc_do_string(char *data, int maxlen, int write,
1875 			   char __user *buffer,
1876 			   size_t *lenp, loff_t *ppos)
1877 {
1878 	size_t len;
1879 	char __user *p;
1880 	char c;
1881 
1882 	if (!data || !maxlen || !*lenp) {
1883 		*lenp = 0;
1884 		return 0;
1885 	}
1886 
1887 	if (write) {
1888 		if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1889 			/* Only continue writes not past the end of buffer. */
1890 			len = strlen(data);
1891 			if (len > maxlen - 1)
1892 				len = maxlen - 1;
1893 
1894 			if (*ppos > len)
1895 				return 0;
1896 			len = *ppos;
1897 		} else {
1898 			/* Start writing from beginning of buffer. */
1899 			len = 0;
1900 		}
1901 
1902 		*ppos += *lenp;
1903 		p = buffer;
1904 		while ((p - buffer) < *lenp && len < maxlen - 1) {
1905 			if (get_user(c, p++))
1906 				return -EFAULT;
1907 			if (c == 0 || c == '\n')
1908 				break;
1909 			data[len++] = c;
1910 		}
1911 		data[len] = 0;
1912 	} else {
1913 		len = strlen(data);
1914 		if (len > maxlen)
1915 			len = maxlen;
1916 
1917 		if (*ppos > len) {
1918 			*lenp = 0;
1919 			return 0;
1920 		}
1921 
1922 		data += *ppos;
1923 		len  -= *ppos;
1924 
1925 		if (len > *lenp)
1926 			len = *lenp;
1927 		if (len)
1928 			if (copy_to_user(buffer, data, len))
1929 				return -EFAULT;
1930 		if (len < *lenp) {
1931 			if (put_user('\n', buffer + len))
1932 				return -EFAULT;
1933 			len++;
1934 		}
1935 		*lenp = len;
1936 		*ppos += len;
1937 	}
1938 	return 0;
1939 }
1940 
1941 static void warn_sysctl_write(struct ctl_table *table)
1942 {
1943 	pr_warn_once("%s wrote to %s when file position was not 0!\n"
1944 		"This will not be supported in the future. To silence this\n"
1945 		"warning, set kernel.sysctl_writes_strict = -1\n",
1946 		current->comm, table->procname);
1947 }
1948 
1949 /**
1950  * proc_dostring - read a string sysctl
1951  * @table: the sysctl table
1952  * @write: %TRUE if this is a write to the sysctl file
1953  * @buffer: the user buffer
1954  * @lenp: the size of the user buffer
1955  * @ppos: file position
1956  *
1957  * Reads/writes a string from/to the user buffer. If the kernel
1958  * buffer provided is not large enough to hold the string, the
1959  * string is truncated. The copied string is %NULL-terminated.
1960  * If the string is being read by the user process, it is copied
1961  * and a newline '\n' is added. It is truncated if the buffer is
1962  * not large enough.
1963  *
1964  * Returns 0 on success.
1965  */
1966 int proc_dostring(struct ctl_table *table, int write,
1967 		  void __user *buffer, size_t *lenp, loff_t *ppos)
1968 {
1969 	if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN)
1970 		warn_sysctl_write(table);
1971 
1972 	return _proc_do_string((char *)(table->data), table->maxlen, write,
1973 			       (char __user *)buffer, lenp, ppos);
1974 }
1975 
1976 static size_t proc_skip_spaces(char **buf)
1977 {
1978 	size_t ret;
1979 	char *tmp = skip_spaces(*buf);
1980 	ret = tmp - *buf;
1981 	*buf = tmp;
1982 	return ret;
1983 }
1984 
1985 static void proc_skip_char(char **buf, size_t *size, const char v)
1986 {
1987 	while (*size) {
1988 		if (**buf != v)
1989 			break;
1990 		(*size)--;
1991 		(*buf)++;
1992 	}
1993 }
1994 
1995 #define TMPBUFLEN 22
1996 /**
1997  * proc_get_long - reads an ASCII formatted integer from a user buffer
1998  *
1999  * @buf: a kernel buffer
2000  * @size: size of the kernel buffer
2001  * @val: this is where the number will be stored
2002  * @neg: set to %TRUE if number is negative
2003  * @perm_tr: a vector which contains the allowed trailers
2004  * @perm_tr_len: size of the perm_tr vector
2005  * @tr: pointer to store the trailer character
2006  *
2007  * In case of success %0 is returned and @buf and @size are updated with
2008  * the amount of bytes read. If @tr is non-NULL and a trailing
2009  * character exists (size is non-zero after returning from this
2010  * function), @tr is updated with the trailing character.
2011  */
2012 static int proc_get_long(char **buf, size_t *size,
2013 			  unsigned long *val, bool *neg,
2014 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
2015 {
2016 	int len;
2017 	char *p, tmp[TMPBUFLEN];
2018 
2019 	if (!*size)
2020 		return -EINVAL;
2021 
2022 	len = *size;
2023 	if (len > TMPBUFLEN - 1)
2024 		len = TMPBUFLEN - 1;
2025 
2026 	memcpy(tmp, *buf, len);
2027 
2028 	tmp[len] = 0;
2029 	p = tmp;
2030 	if (*p == '-' && *size > 1) {
2031 		*neg = true;
2032 		p++;
2033 	} else
2034 		*neg = false;
2035 	if (!isdigit(*p))
2036 		return -EINVAL;
2037 
2038 	*val = simple_strtoul(p, &p, 0);
2039 
2040 	len = p - tmp;
2041 
2042 	/* We don't know if the next char is whitespace thus we may accept
2043 	 * invalid integers (e.g. 1234...a) or two integers instead of one
2044 	 * (e.g. 123...1). So lets not allow such large numbers. */
2045 	if (len == TMPBUFLEN - 1)
2046 		return -EINVAL;
2047 
2048 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2049 		return -EINVAL;
2050 
2051 	if (tr && (len < *size))
2052 		*tr = *p;
2053 
2054 	*buf += len;
2055 	*size -= len;
2056 
2057 	return 0;
2058 }
2059 
2060 /**
2061  * proc_put_long - converts an integer to a decimal ASCII formatted string
2062  *
2063  * @buf: the user buffer
2064  * @size: the size of the user buffer
2065  * @val: the integer to be converted
2066  * @neg: sign of the number, %TRUE for negative
2067  *
2068  * In case of success %0 is returned and @buf and @size are updated with
2069  * the amount of bytes written.
2070  */
2071 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2072 			  bool neg)
2073 {
2074 	int len;
2075 	char tmp[TMPBUFLEN], *p = tmp;
2076 
2077 	sprintf(p, "%s%lu", neg ? "-" : "", val);
2078 	len = strlen(tmp);
2079 	if (len > *size)
2080 		len = *size;
2081 	if (copy_to_user(*buf, tmp, len))
2082 		return -EFAULT;
2083 	*size -= len;
2084 	*buf += len;
2085 	return 0;
2086 }
2087 #undef TMPBUFLEN
2088 
2089 static int proc_put_char(void __user **buf, size_t *size, char c)
2090 {
2091 	if (*size) {
2092 		char __user **buffer = (char __user **)buf;
2093 		if (put_user(c, *buffer))
2094 			return -EFAULT;
2095 		(*size)--, (*buffer)++;
2096 		*buf = *buffer;
2097 	}
2098 	return 0;
2099 }
2100 
2101 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2102 				 int *valp,
2103 				 int write, void *data)
2104 {
2105 	if (write) {
2106 		if (*negp) {
2107 			if (*lvalp > (unsigned long) INT_MAX + 1)
2108 				return -EINVAL;
2109 			*valp = -*lvalp;
2110 		} else {
2111 			if (*lvalp > (unsigned long) INT_MAX)
2112 				return -EINVAL;
2113 			*valp = *lvalp;
2114 		}
2115 	} else {
2116 		int val = *valp;
2117 		if (val < 0) {
2118 			*negp = true;
2119 			*lvalp = -(unsigned long)val;
2120 		} else {
2121 			*negp = false;
2122 			*lvalp = (unsigned long)val;
2123 		}
2124 	}
2125 	return 0;
2126 }
2127 
2128 static int do_proc_douintvec_conv(bool *negp, unsigned long *lvalp,
2129 				 int *valp,
2130 				 int write, void *data)
2131 {
2132 	if (write) {
2133 		if (*negp)
2134 			return -EINVAL;
2135 		*valp = *lvalp;
2136 	} else {
2137 		unsigned int val = *valp;
2138 		*lvalp = (unsigned long)val;
2139 	}
2140 	return 0;
2141 }
2142 
2143 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2144 
2145 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2146 		  int write, void __user *buffer,
2147 		  size_t *lenp, loff_t *ppos,
2148 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2149 			      int write, void *data),
2150 		  void *data)
2151 {
2152 	int *i, vleft, first = 1, err = 0;
2153 	size_t left;
2154 	char *kbuf = NULL, *p;
2155 
2156 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2157 		*lenp = 0;
2158 		return 0;
2159 	}
2160 
2161 	i = (int *) tbl_data;
2162 	vleft = table->maxlen / sizeof(*i);
2163 	left = *lenp;
2164 
2165 	if (!conv)
2166 		conv = do_proc_dointvec_conv;
2167 
2168 	if (write) {
2169 		if (*ppos) {
2170 			switch (sysctl_writes_strict) {
2171 			case SYSCTL_WRITES_STRICT:
2172 				goto out;
2173 			case SYSCTL_WRITES_WARN:
2174 				warn_sysctl_write(table);
2175 				break;
2176 			default:
2177 				break;
2178 			}
2179 		}
2180 
2181 		if (left > PAGE_SIZE - 1)
2182 			left = PAGE_SIZE - 1;
2183 		p = kbuf = memdup_user_nul(buffer, left);
2184 		if (IS_ERR(kbuf))
2185 			return PTR_ERR(kbuf);
2186 	}
2187 
2188 	for (; left && vleft--; i++, first=0) {
2189 		unsigned long lval;
2190 		bool neg;
2191 
2192 		if (write) {
2193 			left -= proc_skip_spaces(&p);
2194 
2195 			if (!left)
2196 				break;
2197 			err = proc_get_long(&p, &left, &lval, &neg,
2198 					     proc_wspace_sep,
2199 					     sizeof(proc_wspace_sep), NULL);
2200 			if (err)
2201 				break;
2202 			if (conv(&neg, &lval, i, 1, data)) {
2203 				err = -EINVAL;
2204 				break;
2205 			}
2206 		} else {
2207 			if (conv(&neg, &lval, i, 0, data)) {
2208 				err = -EINVAL;
2209 				break;
2210 			}
2211 			if (!first)
2212 				err = proc_put_char(&buffer, &left, '\t');
2213 			if (err)
2214 				break;
2215 			err = proc_put_long(&buffer, &left, lval, neg);
2216 			if (err)
2217 				break;
2218 		}
2219 	}
2220 
2221 	if (!write && !first && left && !err)
2222 		err = proc_put_char(&buffer, &left, '\n');
2223 	if (write && !err && left)
2224 		left -= proc_skip_spaces(&p);
2225 	if (write) {
2226 		kfree(kbuf);
2227 		if (first)
2228 			return err ? : -EINVAL;
2229 	}
2230 	*lenp -= left;
2231 out:
2232 	*ppos += *lenp;
2233 	return err;
2234 }
2235 
2236 static int do_proc_dointvec(struct ctl_table *table, int write,
2237 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2238 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2239 			      int write, void *data),
2240 		  void *data)
2241 {
2242 	return __do_proc_dointvec(table->data, table, write,
2243 			buffer, lenp, ppos, conv, data);
2244 }
2245 
2246 /**
2247  * proc_dointvec - read a vector of integers
2248  * @table: the sysctl table
2249  * @write: %TRUE if this is a write to the sysctl file
2250  * @buffer: the user buffer
2251  * @lenp: the size of the user buffer
2252  * @ppos: file position
2253  *
2254  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2255  * values from/to the user buffer, treated as an ASCII string.
2256  *
2257  * Returns 0 on success.
2258  */
2259 int proc_dointvec(struct ctl_table *table, int write,
2260 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2261 {
2262 	return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2263 }
2264 
2265 /**
2266  * proc_douintvec - read a vector of unsigned integers
2267  * @table: the sysctl table
2268  * @write: %TRUE if this is a write to the sysctl file
2269  * @buffer: the user buffer
2270  * @lenp: the size of the user buffer
2271  * @ppos: file position
2272  *
2273  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2274  * values from/to the user buffer, treated as an ASCII string.
2275  *
2276  * Returns 0 on success.
2277  */
2278 int proc_douintvec(struct ctl_table *table, int write,
2279 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2280 {
2281 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2282 				do_proc_douintvec_conv, NULL);
2283 }
2284 
2285 /*
2286  * Taint values can only be increased
2287  * This means we can safely use a temporary.
2288  */
2289 static int proc_taint(struct ctl_table *table, int write,
2290 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2291 {
2292 	struct ctl_table t;
2293 	unsigned long tmptaint = get_taint();
2294 	int err;
2295 
2296 	if (write && !capable(CAP_SYS_ADMIN))
2297 		return -EPERM;
2298 
2299 	t = *table;
2300 	t.data = &tmptaint;
2301 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2302 	if (err < 0)
2303 		return err;
2304 
2305 	if (write) {
2306 		/*
2307 		 * Poor man's atomic or. Not worth adding a primitive
2308 		 * to everyone's atomic.h for this
2309 		 */
2310 		int i;
2311 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2312 			if ((tmptaint >> i) & 1)
2313 				add_taint(i, LOCKDEP_STILL_OK);
2314 		}
2315 	}
2316 
2317 	return err;
2318 }
2319 
2320 #ifdef CONFIG_PRINTK
2321 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2322 				void __user *buffer, size_t *lenp, loff_t *ppos)
2323 {
2324 	if (write && !capable(CAP_SYS_ADMIN))
2325 		return -EPERM;
2326 
2327 	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2328 }
2329 #endif
2330 
2331 struct do_proc_dointvec_minmax_conv_param {
2332 	int *min;
2333 	int *max;
2334 };
2335 
2336 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2337 					int *valp,
2338 					int write, void *data)
2339 {
2340 	struct do_proc_dointvec_minmax_conv_param *param = data;
2341 	if (write) {
2342 		int val = *negp ? -*lvalp : *lvalp;
2343 		if ((param->min && *param->min > val) ||
2344 		    (param->max && *param->max < val))
2345 			return -EINVAL;
2346 		*valp = val;
2347 	} else {
2348 		int val = *valp;
2349 		if (val < 0) {
2350 			*negp = true;
2351 			*lvalp = -(unsigned long)val;
2352 		} else {
2353 			*negp = false;
2354 			*lvalp = (unsigned long)val;
2355 		}
2356 	}
2357 	return 0;
2358 }
2359 
2360 /**
2361  * proc_dointvec_minmax - read a vector of integers with min/max values
2362  * @table: the sysctl table
2363  * @write: %TRUE if this is a write to the sysctl file
2364  * @buffer: the user buffer
2365  * @lenp: the size of the user buffer
2366  * @ppos: file position
2367  *
2368  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2369  * values from/to the user buffer, treated as an ASCII string.
2370  *
2371  * This routine will ensure the values are within the range specified by
2372  * table->extra1 (min) and table->extra2 (max).
2373  *
2374  * Returns 0 on success.
2375  */
2376 int proc_dointvec_minmax(struct ctl_table *table, int write,
2377 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2378 {
2379 	struct do_proc_dointvec_minmax_conv_param param = {
2380 		.min = (int *) table->extra1,
2381 		.max = (int *) table->extra2,
2382 	};
2383 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2384 				do_proc_dointvec_minmax_conv, &param);
2385 }
2386 
2387 static void validate_coredump_safety(void)
2388 {
2389 #ifdef CONFIG_COREDUMP
2390 	if (suid_dumpable == SUID_DUMP_ROOT &&
2391 	    core_pattern[0] != '/' && core_pattern[0] != '|') {
2392 		printk(KERN_WARNING
2393 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2394 "Pipe handler or fully qualified core dump path required.\n"
2395 "Set kernel.core_pattern before fs.suid_dumpable.\n"
2396 		);
2397 	}
2398 #endif
2399 }
2400 
2401 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2402 		void __user *buffer, size_t *lenp, loff_t *ppos)
2403 {
2404 	int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2405 	if (!error)
2406 		validate_coredump_safety();
2407 	return error;
2408 }
2409 
2410 #ifdef CONFIG_COREDUMP
2411 static int proc_dostring_coredump(struct ctl_table *table, int write,
2412 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2413 {
2414 	int error = proc_dostring(table, write, buffer, lenp, ppos);
2415 	if (!error)
2416 		validate_coredump_safety();
2417 	return error;
2418 }
2419 #endif
2420 
2421 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2422 				     void __user *buffer,
2423 				     size_t *lenp, loff_t *ppos,
2424 				     unsigned long convmul,
2425 				     unsigned long convdiv)
2426 {
2427 	unsigned long *i, *min, *max;
2428 	int vleft, first = 1, err = 0;
2429 	size_t left;
2430 	char *kbuf = NULL, *p;
2431 
2432 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2433 		*lenp = 0;
2434 		return 0;
2435 	}
2436 
2437 	i = (unsigned long *) data;
2438 	min = (unsigned long *) table->extra1;
2439 	max = (unsigned long *) table->extra2;
2440 	vleft = table->maxlen / sizeof(unsigned long);
2441 	left = *lenp;
2442 
2443 	if (write) {
2444 		if (*ppos) {
2445 			switch (sysctl_writes_strict) {
2446 			case SYSCTL_WRITES_STRICT:
2447 				goto out;
2448 			case SYSCTL_WRITES_WARN:
2449 				warn_sysctl_write(table);
2450 				break;
2451 			default:
2452 				break;
2453 			}
2454 		}
2455 
2456 		if (left > PAGE_SIZE - 1)
2457 			left = PAGE_SIZE - 1;
2458 		p = kbuf = memdup_user_nul(buffer, left);
2459 		if (IS_ERR(kbuf))
2460 			return PTR_ERR(kbuf);
2461 	}
2462 
2463 	for (; left && vleft--; i++, first = 0) {
2464 		unsigned long val;
2465 
2466 		if (write) {
2467 			bool neg;
2468 
2469 			left -= proc_skip_spaces(&p);
2470 
2471 			err = proc_get_long(&p, &left, &val, &neg,
2472 					     proc_wspace_sep,
2473 					     sizeof(proc_wspace_sep), NULL);
2474 			if (err)
2475 				break;
2476 			if (neg)
2477 				continue;
2478 			val = convmul * val / convdiv;
2479 			if ((min && val < *min) || (max && val > *max))
2480 				continue;
2481 			*i = val;
2482 		} else {
2483 			val = convdiv * (*i) / convmul;
2484 			if (!first) {
2485 				err = proc_put_char(&buffer, &left, '\t');
2486 				if (err)
2487 					break;
2488 			}
2489 			err = proc_put_long(&buffer, &left, val, false);
2490 			if (err)
2491 				break;
2492 		}
2493 	}
2494 
2495 	if (!write && !first && left && !err)
2496 		err = proc_put_char(&buffer, &left, '\n');
2497 	if (write && !err)
2498 		left -= proc_skip_spaces(&p);
2499 	if (write) {
2500 		kfree(kbuf);
2501 		if (first)
2502 			return err ? : -EINVAL;
2503 	}
2504 	*lenp -= left;
2505 out:
2506 	*ppos += *lenp;
2507 	return err;
2508 }
2509 
2510 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2511 				     void __user *buffer,
2512 				     size_t *lenp, loff_t *ppos,
2513 				     unsigned long convmul,
2514 				     unsigned long convdiv)
2515 {
2516 	return __do_proc_doulongvec_minmax(table->data, table, write,
2517 			buffer, lenp, ppos, convmul, convdiv);
2518 }
2519 
2520 /**
2521  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2522  * @table: the sysctl table
2523  * @write: %TRUE if this is a write to the sysctl file
2524  * @buffer: the user buffer
2525  * @lenp: the size of the user buffer
2526  * @ppos: file position
2527  *
2528  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2529  * values from/to the user buffer, treated as an ASCII string.
2530  *
2531  * This routine will ensure the values are within the range specified by
2532  * table->extra1 (min) and table->extra2 (max).
2533  *
2534  * Returns 0 on success.
2535  */
2536 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2537 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2538 {
2539     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2540 }
2541 
2542 /**
2543  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2544  * @table: the sysctl table
2545  * @write: %TRUE if this is a write to the sysctl file
2546  * @buffer: the user buffer
2547  * @lenp: the size of the user buffer
2548  * @ppos: file position
2549  *
2550  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2551  * values from/to the user buffer, treated as an ASCII string. The values
2552  * are treated as milliseconds, and converted to jiffies when they are stored.
2553  *
2554  * This routine will ensure the values are within the range specified by
2555  * table->extra1 (min) and table->extra2 (max).
2556  *
2557  * Returns 0 on success.
2558  */
2559 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2560 				      void __user *buffer,
2561 				      size_t *lenp, loff_t *ppos)
2562 {
2563     return do_proc_doulongvec_minmax(table, write, buffer,
2564 				     lenp, ppos, HZ, 1000l);
2565 }
2566 
2567 
2568 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2569 					 int *valp,
2570 					 int write, void *data)
2571 {
2572 	if (write) {
2573 		if (*lvalp > LONG_MAX / HZ)
2574 			return 1;
2575 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2576 	} else {
2577 		int val = *valp;
2578 		unsigned long lval;
2579 		if (val < 0) {
2580 			*negp = true;
2581 			lval = -(unsigned long)val;
2582 		} else {
2583 			*negp = false;
2584 			lval = (unsigned long)val;
2585 		}
2586 		*lvalp = lval / HZ;
2587 	}
2588 	return 0;
2589 }
2590 
2591 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2592 						int *valp,
2593 						int write, void *data)
2594 {
2595 	if (write) {
2596 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2597 			return 1;
2598 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2599 	} else {
2600 		int val = *valp;
2601 		unsigned long lval;
2602 		if (val < 0) {
2603 			*negp = true;
2604 			lval = -(unsigned long)val;
2605 		} else {
2606 			*negp = false;
2607 			lval = (unsigned long)val;
2608 		}
2609 		*lvalp = jiffies_to_clock_t(lval);
2610 	}
2611 	return 0;
2612 }
2613 
2614 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2615 					    int *valp,
2616 					    int write, void *data)
2617 {
2618 	if (write) {
2619 		unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2620 
2621 		if (jif > INT_MAX)
2622 			return 1;
2623 		*valp = (int)jif;
2624 	} else {
2625 		int val = *valp;
2626 		unsigned long lval;
2627 		if (val < 0) {
2628 			*negp = true;
2629 			lval = -(unsigned long)val;
2630 		} else {
2631 			*negp = false;
2632 			lval = (unsigned long)val;
2633 		}
2634 		*lvalp = jiffies_to_msecs(lval);
2635 	}
2636 	return 0;
2637 }
2638 
2639 /**
2640  * proc_dointvec_jiffies - read a vector of integers as seconds
2641  * @table: the sysctl table
2642  * @write: %TRUE if this is a write to the sysctl file
2643  * @buffer: the user buffer
2644  * @lenp: the size of the user buffer
2645  * @ppos: file position
2646  *
2647  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2648  * values from/to the user buffer, treated as an ASCII string.
2649  * The values read are assumed to be in seconds, and are converted into
2650  * jiffies.
2651  *
2652  * Returns 0 on success.
2653  */
2654 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2655 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2656 {
2657     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2658 		    	    do_proc_dointvec_jiffies_conv,NULL);
2659 }
2660 
2661 /**
2662  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2663  * @table: the sysctl table
2664  * @write: %TRUE if this is a write to the sysctl file
2665  * @buffer: the user buffer
2666  * @lenp: the size of the user buffer
2667  * @ppos: pointer to the file position
2668  *
2669  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2670  * values from/to the user buffer, treated as an ASCII string.
2671  * The values read are assumed to be in 1/USER_HZ seconds, and
2672  * are converted into jiffies.
2673  *
2674  * Returns 0 on success.
2675  */
2676 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2677 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2678 {
2679     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2680 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2681 }
2682 
2683 /**
2684  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2685  * @table: the sysctl table
2686  * @write: %TRUE if this is a write to the sysctl file
2687  * @buffer: the user buffer
2688  * @lenp: the size of the user buffer
2689  * @ppos: file position
2690  * @ppos: the current position in the file
2691  *
2692  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2693  * values from/to the user buffer, treated as an ASCII string.
2694  * The values read are assumed to be in 1/1000 seconds, and
2695  * are converted into jiffies.
2696  *
2697  * Returns 0 on success.
2698  */
2699 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2700 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2701 {
2702 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2703 				do_proc_dointvec_ms_jiffies_conv, NULL);
2704 }
2705 
2706 static int proc_do_cad_pid(struct ctl_table *table, int write,
2707 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2708 {
2709 	struct pid *new_pid;
2710 	pid_t tmp;
2711 	int r;
2712 
2713 	tmp = pid_vnr(cad_pid);
2714 
2715 	r = __do_proc_dointvec(&tmp, table, write, buffer,
2716 			       lenp, ppos, NULL, NULL);
2717 	if (r || !write)
2718 		return r;
2719 
2720 	new_pid = find_get_pid(tmp);
2721 	if (!new_pid)
2722 		return -ESRCH;
2723 
2724 	put_pid(xchg(&cad_pid, new_pid));
2725 	return 0;
2726 }
2727 
2728 /**
2729  * proc_do_large_bitmap - read/write from/to a large bitmap
2730  * @table: the sysctl table
2731  * @write: %TRUE if this is a write to the sysctl file
2732  * @buffer: the user buffer
2733  * @lenp: the size of the user buffer
2734  * @ppos: file position
2735  *
2736  * The bitmap is stored at table->data and the bitmap length (in bits)
2737  * in table->maxlen.
2738  *
2739  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2740  * large bitmaps may be represented in a compact manner. Writing into
2741  * the file will clear the bitmap then update it with the given input.
2742  *
2743  * Returns 0 on success.
2744  */
2745 int proc_do_large_bitmap(struct ctl_table *table, int write,
2746 			 void __user *buffer, size_t *lenp, loff_t *ppos)
2747 {
2748 	int err = 0;
2749 	bool first = 1;
2750 	size_t left = *lenp;
2751 	unsigned long bitmap_len = table->maxlen;
2752 	unsigned long *bitmap = *(unsigned long **) table->data;
2753 	unsigned long *tmp_bitmap = NULL;
2754 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2755 
2756 	if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2757 		*lenp = 0;
2758 		return 0;
2759 	}
2760 
2761 	if (write) {
2762 		char *kbuf, *p;
2763 
2764 		if (left > PAGE_SIZE - 1)
2765 			left = PAGE_SIZE - 1;
2766 
2767 		p = kbuf = memdup_user_nul(buffer, left);
2768 		if (IS_ERR(kbuf))
2769 			return PTR_ERR(kbuf);
2770 
2771 		tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2772 				     GFP_KERNEL);
2773 		if (!tmp_bitmap) {
2774 			kfree(kbuf);
2775 			return -ENOMEM;
2776 		}
2777 		proc_skip_char(&p, &left, '\n');
2778 		while (!err && left) {
2779 			unsigned long val_a, val_b;
2780 			bool neg;
2781 
2782 			err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
2783 					     sizeof(tr_a), &c);
2784 			if (err)
2785 				break;
2786 			if (val_a >= bitmap_len || neg) {
2787 				err = -EINVAL;
2788 				break;
2789 			}
2790 
2791 			val_b = val_a;
2792 			if (left) {
2793 				p++;
2794 				left--;
2795 			}
2796 
2797 			if (c == '-') {
2798 				err = proc_get_long(&p, &left, &val_b,
2799 						     &neg, tr_b, sizeof(tr_b),
2800 						     &c);
2801 				if (err)
2802 					break;
2803 				if (val_b >= bitmap_len || neg ||
2804 				    val_a > val_b) {
2805 					err = -EINVAL;
2806 					break;
2807 				}
2808 				if (left) {
2809 					p++;
2810 					left--;
2811 				}
2812 			}
2813 
2814 			bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2815 			first = 0;
2816 			proc_skip_char(&p, &left, '\n');
2817 		}
2818 		kfree(kbuf);
2819 	} else {
2820 		unsigned long bit_a, bit_b = 0;
2821 
2822 		while (left) {
2823 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2824 			if (bit_a >= bitmap_len)
2825 				break;
2826 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
2827 						   bit_a + 1) - 1;
2828 
2829 			if (!first) {
2830 				err = proc_put_char(&buffer, &left, ',');
2831 				if (err)
2832 					break;
2833 			}
2834 			err = proc_put_long(&buffer, &left, bit_a, false);
2835 			if (err)
2836 				break;
2837 			if (bit_a != bit_b) {
2838 				err = proc_put_char(&buffer, &left, '-');
2839 				if (err)
2840 					break;
2841 				err = proc_put_long(&buffer, &left, bit_b, false);
2842 				if (err)
2843 					break;
2844 			}
2845 
2846 			first = 0; bit_b++;
2847 		}
2848 		if (!err)
2849 			err = proc_put_char(&buffer, &left, '\n');
2850 	}
2851 
2852 	if (!err) {
2853 		if (write) {
2854 			if (*ppos)
2855 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2856 			else
2857 				bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2858 		}
2859 		kfree(tmp_bitmap);
2860 		*lenp -= left;
2861 		*ppos += *lenp;
2862 		return 0;
2863 	} else {
2864 		kfree(tmp_bitmap);
2865 		return err;
2866 	}
2867 }
2868 
2869 #else /* CONFIG_PROC_SYSCTL */
2870 
2871 int proc_dostring(struct ctl_table *table, int write,
2872 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2873 {
2874 	return -ENOSYS;
2875 }
2876 
2877 int proc_dointvec(struct ctl_table *table, int write,
2878 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2879 {
2880 	return -ENOSYS;
2881 }
2882 
2883 int proc_douintvec(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_minmax(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_dointvec_jiffies(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_userhz_jiffies(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_ms_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_doulongvec_minmax(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_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2920 				      void __user *buffer,
2921 				      size_t *lenp, loff_t *ppos)
2922 {
2923     return -ENOSYS;
2924 }
2925 
2926 
2927 #endif /* CONFIG_PROC_SYSCTL */
2928 
2929 /*
2930  * No sense putting this after each symbol definition, twice,
2931  * exception granted :-)
2932  */
2933 EXPORT_SYMBOL(proc_dointvec);
2934 EXPORT_SYMBOL(proc_douintvec);
2935 EXPORT_SYMBOL(proc_dointvec_jiffies);
2936 EXPORT_SYMBOL(proc_dointvec_minmax);
2937 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2938 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2939 EXPORT_SYMBOL(proc_dostring);
2940 EXPORT_SYMBOL(proc_doulongvec_minmax);
2941 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2942