xref: /openbmc/linux/kernel/sysctl.c (revision 77a87824)
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	= "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	= "kstack_depth_to_print",
987 		.data		= &kstack_depth_to_print,
988 		.maxlen		= sizeof(int),
989 		.mode		= 0644,
990 		.proc_handler	= proc_dointvec,
991 	},
992 	{
993 		.procname	= "io_delay_type",
994 		.data		= &io_delay_type,
995 		.maxlen		= sizeof(int),
996 		.mode		= 0644,
997 		.proc_handler	= proc_dointvec,
998 	},
999 #endif
1000 #if defined(CONFIG_MMU)
1001 	{
1002 		.procname	= "randomize_va_space",
1003 		.data		= &randomize_va_space,
1004 		.maxlen		= sizeof(int),
1005 		.mode		= 0644,
1006 		.proc_handler	= proc_dointvec,
1007 	},
1008 #endif
1009 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1010 	{
1011 		.procname	= "spin_retry",
1012 		.data		= &spin_retry,
1013 		.maxlen		= sizeof (int),
1014 		.mode		= 0644,
1015 		.proc_handler	= proc_dointvec,
1016 	},
1017 #endif
1018 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1019 	{
1020 		.procname	= "acpi_video_flags",
1021 		.data		= &acpi_realmode_flags,
1022 		.maxlen		= sizeof (unsigned long),
1023 		.mode		= 0644,
1024 		.proc_handler	= proc_doulongvec_minmax,
1025 	},
1026 #endif
1027 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1028 	{
1029 		.procname	= "ignore-unaligned-usertrap",
1030 		.data		= &no_unaligned_warning,
1031 		.maxlen		= sizeof (int),
1032 	 	.mode		= 0644,
1033 		.proc_handler	= proc_dointvec,
1034 	},
1035 #endif
1036 #ifdef CONFIG_IA64
1037 	{
1038 		.procname	= "unaligned-dump-stack",
1039 		.data		= &unaligned_dump_stack,
1040 		.maxlen		= sizeof (int),
1041 		.mode		= 0644,
1042 		.proc_handler	= proc_dointvec,
1043 	},
1044 #endif
1045 #ifdef CONFIG_DETECT_HUNG_TASK
1046 	{
1047 		.procname	= "hung_task_panic",
1048 		.data		= &sysctl_hung_task_panic,
1049 		.maxlen		= sizeof(int),
1050 		.mode		= 0644,
1051 		.proc_handler	= proc_dointvec_minmax,
1052 		.extra1		= &zero,
1053 		.extra2		= &one,
1054 	},
1055 	{
1056 		.procname	= "hung_task_check_count",
1057 		.data		= &sysctl_hung_task_check_count,
1058 		.maxlen		= sizeof(int),
1059 		.mode		= 0644,
1060 		.proc_handler	= proc_dointvec_minmax,
1061 		.extra1		= &zero,
1062 	},
1063 	{
1064 		.procname	= "hung_task_timeout_secs",
1065 		.data		= &sysctl_hung_task_timeout_secs,
1066 		.maxlen		= sizeof(unsigned long),
1067 		.mode		= 0644,
1068 		.proc_handler	= proc_dohung_task_timeout_secs,
1069 		.extra2		= &hung_task_timeout_max,
1070 	},
1071 	{
1072 		.procname	= "hung_task_warnings",
1073 		.data		= &sysctl_hung_task_warnings,
1074 		.maxlen		= sizeof(int),
1075 		.mode		= 0644,
1076 		.proc_handler	= proc_dointvec_minmax,
1077 		.extra1		= &neg_one,
1078 	},
1079 #endif
1080 #ifdef CONFIG_COMPAT
1081 	{
1082 		.procname	= "compat-log",
1083 		.data		= &compat_log,
1084 		.maxlen		= sizeof (int),
1085 	 	.mode		= 0644,
1086 		.proc_handler	= proc_dointvec,
1087 	},
1088 #endif
1089 #ifdef CONFIG_RT_MUTEXES
1090 	{
1091 		.procname	= "max_lock_depth",
1092 		.data		= &max_lock_depth,
1093 		.maxlen		= sizeof(int),
1094 		.mode		= 0644,
1095 		.proc_handler	= proc_dointvec,
1096 	},
1097 #endif
1098 	{
1099 		.procname	= "poweroff_cmd",
1100 		.data		= &poweroff_cmd,
1101 		.maxlen		= POWEROFF_CMD_PATH_LEN,
1102 		.mode		= 0644,
1103 		.proc_handler	= proc_dostring,
1104 	},
1105 #ifdef CONFIG_KEYS
1106 	{
1107 		.procname	= "keys",
1108 		.mode		= 0555,
1109 		.child		= key_sysctls,
1110 	},
1111 #endif
1112 #ifdef CONFIG_PERF_EVENTS
1113 	/*
1114 	 * User-space scripts rely on the existence of this file
1115 	 * as a feature check for perf_events being enabled.
1116 	 *
1117 	 * So it's an ABI, do not remove!
1118 	 */
1119 	{
1120 		.procname	= "perf_event_paranoid",
1121 		.data		= &sysctl_perf_event_paranoid,
1122 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
1123 		.mode		= 0644,
1124 		.proc_handler	= proc_dointvec,
1125 	},
1126 	{
1127 		.procname	= "perf_event_mlock_kb",
1128 		.data		= &sysctl_perf_event_mlock,
1129 		.maxlen		= sizeof(sysctl_perf_event_mlock),
1130 		.mode		= 0644,
1131 		.proc_handler	= proc_dointvec,
1132 	},
1133 	{
1134 		.procname	= "perf_event_max_sample_rate",
1135 		.data		= &sysctl_perf_event_sample_rate,
1136 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
1137 		.mode		= 0644,
1138 		.proc_handler	= perf_proc_update_handler,
1139 		.extra1		= &one,
1140 	},
1141 	{
1142 		.procname	= "perf_cpu_time_max_percent",
1143 		.data		= &sysctl_perf_cpu_time_max_percent,
1144 		.maxlen		= sizeof(sysctl_perf_cpu_time_max_percent),
1145 		.mode		= 0644,
1146 		.proc_handler	= perf_cpu_time_max_percent_handler,
1147 		.extra1		= &zero,
1148 		.extra2		= &one_hundred,
1149 	},
1150 	{
1151 		.procname	= "perf_event_max_stack",
1152 		.data		= &sysctl_perf_event_max_stack,
1153 		.maxlen		= sizeof(sysctl_perf_event_max_stack),
1154 		.mode		= 0644,
1155 		.proc_handler	= perf_event_max_stack_handler,
1156 		.extra1		= &zero,
1157 		.extra2		= &six_hundred_forty_kb,
1158 	},
1159 	{
1160 		.procname	= "perf_event_max_contexts_per_stack",
1161 		.data		= &sysctl_perf_event_max_contexts_per_stack,
1162 		.maxlen		= sizeof(sysctl_perf_event_max_contexts_per_stack),
1163 		.mode		= 0644,
1164 		.proc_handler	= perf_event_max_stack_handler,
1165 		.extra1		= &zero,
1166 		.extra2		= &one_thousand,
1167 	},
1168 #endif
1169 #ifdef CONFIG_KMEMCHECK
1170 	{
1171 		.procname	= "kmemcheck",
1172 		.data		= &kmemcheck_enabled,
1173 		.maxlen		= sizeof(int),
1174 		.mode		= 0644,
1175 		.proc_handler	= proc_dointvec,
1176 	},
1177 #endif
1178 	{
1179 		.procname	= "panic_on_warn",
1180 		.data		= &panic_on_warn,
1181 		.maxlen		= sizeof(int),
1182 		.mode		= 0644,
1183 		.proc_handler	= proc_dointvec_minmax,
1184 		.extra1		= &zero,
1185 		.extra2		= &one,
1186 	},
1187 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1188 	{
1189 		.procname	= "timer_migration",
1190 		.data		= &sysctl_timer_migration,
1191 		.maxlen		= sizeof(unsigned int),
1192 		.mode		= 0644,
1193 		.proc_handler	= timer_migration_handler,
1194 	},
1195 #endif
1196 #ifdef CONFIG_BPF_SYSCALL
1197 	{
1198 		.procname	= "unprivileged_bpf_disabled",
1199 		.data		= &sysctl_unprivileged_bpf_disabled,
1200 		.maxlen		= sizeof(sysctl_unprivileged_bpf_disabled),
1201 		.mode		= 0644,
1202 		/* only handle a transition from default "0" to "1" */
1203 		.proc_handler	= proc_dointvec_minmax,
1204 		.extra1		= &one,
1205 		.extra2		= &one,
1206 	},
1207 #endif
1208 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1209 	{
1210 		.procname	= "panic_on_rcu_stall",
1211 		.data		= &sysctl_panic_on_rcu_stall,
1212 		.maxlen		= sizeof(sysctl_panic_on_rcu_stall),
1213 		.mode		= 0644,
1214 		.proc_handler	= proc_dointvec_minmax,
1215 		.extra1		= &zero,
1216 		.extra2		= &one,
1217 	},
1218 #endif
1219 	{ }
1220 };
1221 
1222 static struct ctl_table vm_table[] = {
1223 	{
1224 		.procname	= "overcommit_memory",
1225 		.data		= &sysctl_overcommit_memory,
1226 		.maxlen		= sizeof(sysctl_overcommit_memory),
1227 		.mode		= 0644,
1228 		.proc_handler	= proc_dointvec_minmax,
1229 		.extra1		= &zero,
1230 		.extra2		= &two,
1231 	},
1232 	{
1233 		.procname	= "panic_on_oom",
1234 		.data		= &sysctl_panic_on_oom,
1235 		.maxlen		= sizeof(sysctl_panic_on_oom),
1236 		.mode		= 0644,
1237 		.proc_handler	= proc_dointvec_minmax,
1238 		.extra1		= &zero,
1239 		.extra2		= &two,
1240 	},
1241 	{
1242 		.procname	= "oom_kill_allocating_task",
1243 		.data		= &sysctl_oom_kill_allocating_task,
1244 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1245 		.mode		= 0644,
1246 		.proc_handler	= proc_dointvec,
1247 	},
1248 	{
1249 		.procname	= "oom_dump_tasks",
1250 		.data		= &sysctl_oom_dump_tasks,
1251 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1252 		.mode		= 0644,
1253 		.proc_handler	= proc_dointvec,
1254 	},
1255 	{
1256 		.procname	= "overcommit_ratio",
1257 		.data		= &sysctl_overcommit_ratio,
1258 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1259 		.mode		= 0644,
1260 		.proc_handler	= overcommit_ratio_handler,
1261 	},
1262 	{
1263 		.procname	= "overcommit_kbytes",
1264 		.data		= &sysctl_overcommit_kbytes,
1265 		.maxlen		= sizeof(sysctl_overcommit_kbytes),
1266 		.mode		= 0644,
1267 		.proc_handler	= overcommit_kbytes_handler,
1268 	},
1269 	{
1270 		.procname	= "page-cluster",
1271 		.data		= &page_cluster,
1272 		.maxlen		= sizeof(int),
1273 		.mode		= 0644,
1274 		.proc_handler	= proc_dointvec_minmax,
1275 		.extra1		= &zero,
1276 	},
1277 	{
1278 		.procname	= "dirty_background_ratio",
1279 		.data		= &dirty_background_ratio,
1280 		.maxlen		= sizeof(dirty_background_ratio),
1281 		.mode		= 0644,
1282 		.proc_handler	= dirty_background_ratio_handler,
1283 		.extra1		= &zero,
1284 		.extra2		= &one_hundred,
1285 	},
1286 	{
1287 		.procname	= "dirty_background_bytes",
1288 		.data		= &dirty_background_bytes,
1289 		.maxlen		= sizeof(dirty_background_bytes),
1290 		.mode		= 0644,
1291 		.proc_handler	= dirty_background_bytes_handler,
1292 		.extra1		= &one_ul,
1293 	},
1294 	{
1295 		.procname	= "dirty_ratio",
1296 		.data		= &vm_dirty_ratio,
1297 		.maxlen		= sizeof(vm_dirty_ratio),
1298 		.mode		= 0644,
1299 		.proc_handler	= dirty_ratio_handler,
1300 		.extra1		= &zero,
1301 		.extra2		= &one_hundred,
1302 	},
1303 	{
1304 		.procname	= "dirty_bytes",
1305 		.data		= &vm_dirty_bytes,
1306 		.maxlen		= sizeof(vm_dirty_bytes),
1307 		.mode		= 0644,
1308 		.proc_handler	= dirty_bytes_handler,
1309 		.extra1		= &dirty_bytes_min,
1310 	},
1311 	{
1312 		.procname	= "dirty_writeback_centisecs",
1313 		.data		= &dirty_writeback_interval,
1314 		.maxlen		= sizeof(dirty_writeback_interval),
1315 		.mode		= 0644,
1316 		.proc_handler	= dirty_writeback_centisecs_handler,
1317 	},
1318 	{
1319 		.procname	= "dirty_expire_centisecs",
1320 		.data		= &dirty_expire_interval,
1321 		.maxlen		= sizeof(dirty_expire_interval),
1322 		.mode		= 0644,
1323 		.proc_handler	= proc_dointvec_minmax,
1324 		.extra1		= &zero,
1325 	},
1326 	{
1327 		.procname	= "dirtytime_expire_seconds",
1328 		.data		= &dirtytime_expire_interval,
1329 		.maxlen		= sizeof(dirty_expire_interval),
1330 		.mode		= 0644,
1331 		.proc_handler	= dirtytime_interval_handler,
1332 		.extra1		= &zero,
1333 	},
1334 	{
1335 		.procname       = "nr_pdflush_threads",
1336 		.mode           = 0444 /* read-only */,
1337 		.proc_handler   = pdflush_proc_obsolete,
1338 	},
1339 	{
1340 		.procname	= "swappiness",
1341 		.data		= &vm_swappiness,
1342 		.maxlen		= sizeof(vm_swappiness),
1343 		.mode		= 0644,
1344 		.proc_handler	= proc_dointvec_minmax,
1345 		.extra1		= &zero,
1346 		.extra2		= &one_hundred,
1347 	},
1348 #ifdef CONFIG_HUGETLB_PAGE
1349 	{
1350 		.procname	= "nr_hugepages",
1351 		.data		= NULL,
1352 		.maxlen		= sizeof(unsigned long),
1353 		.mode		= 0644,
1354 		.proc_handler	= hugetlb_sysctl_handler,
1355 	},
1356 #ifdef CONFIG_NUMA
1357 	{
1358 		.procname       = "nr_hugepages_mempolicy",
1359 		.data           = NULL,
1360 		.maxlen         = sizeof(unsigned long),
1361 		.mode           = 0644,
1362 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1363 	},
1364 #endif
1365 	 {
1366 		.procname	= "hugetlb_shm_group",
1367 		.data		= &sysctl_hugetlb_shm_group,
1368 		.maxlen		= sizeof(gid_t),
1369 		.mode		= 0644,
1370 		.proc_handler	= proc_dointvec,
1371 	 },
1372 	 {
1373 		.procname	= "hugepages_treat_as_movable",
1374 		.data		= &hugepages_treat_as_movable,
1375 		.maxlen		= sizeof(int),
1376 		.mode		= 0644,
1377 		.proc_handler	= proc_dointvec,
1378 	},
1379 	{
1380 		.procname	= "nr_overcommit_hugepages",
1381 		.data		= NULL,
1382 		.maxlen		= sizeof(unsigned long),
1383 		.mode		= 0644,
1384 		.proc_handler	= hugetlb_overcommit_handler,
1385 	},
1386 #endif
1387 	{
1388 		.procname	= "lowmem_reserve_ratio",
1389 		.data		= &sysctl_lowmem_reserve_ratio,
1390 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1391 		.mode		= 0644,
1392 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1393 	},
1394 	{
1395 		.procname	= "drop_caches",
1396 		.data		= &sysctl_drop_caches,
1397 		.maxlen		= sizeof(int),
1398 		.mode		= 0644,
1399 		.proc_handler	= drop_caches_sysctl_handler,
1400 		.extra1		= &one,
1401 		.extra2		= &four,
1402 	},
1403 #ifdef CONFIG_COMPACTION
1404 	{
1405 		.procname	= "compact_memory",
1406 		.data		= &sysctl_compact_memory,
1407 		.maxlen		= sizeof(int),
1408 		.mode		= 0200,
1409 		.proc_handler	= sysctl_compaction_handler,
1410 	},
1411 	{
1412 		.procname	= "extfrag_threshold",
1413 		.data		= &sysctl_extfrag_threshold,
1414 		.maxlen		= sizeof(int),
1415 		.mode		= 0644,
1416 		.proc_handler	= sysctl_extfrag_handler,
1417 		.extra1		= &min_extfrag_threshold,
1418 		.extra2		= &max_extfrag_threshold,
1419 	},
1420 	{
1421 		.procname	= "compact_unevictable_allowed",
1422 		.data		= &sysctl_compact_unevictable_allowed,
1423 		.maxlen		= sizeof(int),
1424 		.mode		= 0644,
1425 		.proc_handler	= proc_dointvec,
1426 		.extra1		= &zero,
1427 		.extra2		= &one,
1428 	},
1429 
1430 #endif /* CONFIG_COMPACTION */
1431 	{
1432 		.procname	= "min_free_kbytes",
1433 		.data		= &min_free_kbytes,
1434 		.maxlen		= sizeof(min_free_kbytes),
1435 		.mode		= 0644,
1436 		.proc_handler	= min_free_kbytes_sysctl_handler,
1437 		.extra1		= &zero,
1438 	},
1439 	{
1440 		.procname	= "watermark_scale_factor",
1441 		.data		= &watermark_scale_factor,
1442 		.maxlen		= sizeof(watermark_scale_factor),
1443 		.mode		= 0644,
1444 		.proc_handler	= watermark_scale_factor_sysctl_handler,
1445 		.extra1		= &one,
1446 		.extra2		= &one_thousand,
1447 	},
1448 	{
1449 		.procname	= "percpu_pagelist_fraction",
1450 		.data		= &percpu_pagelist_fraction,
1451 		.maxlen		= sizeof(percpu_pagelist_fraction),
1452 		.mode		= 0644,
1453 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1454 		.extra1		= &zero,
1455 	},
1456 #ifdef CONFIG_MMU
1457 	{
1458 		.procname	= "max_map_count",
1459 		.data		= &sysctl_max_map_count,
1460 		.maxlen		= sizeof(sysctl_max_map_count),
1461 		.mode		= 0644,
1462 		.proc_handler	= proc_dointvec_minmax,
1463 		.extra1		= &zero,
1464 	},
1465 #else
1466 	{
1467 		.procname	= "nr_trim_pages",
1468 		.data		= &sysctl_nr_trim_pages,
1469 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1470 		.mode		= 0644,
1471 		.proc_handler	= proc_dointvec_minmax,
1472 		.extra1		= &zero,
1473 	},
1474 #endif
1475 	{
1476 		.procname	= "laptop_mode",
1477 		.data		= &laptop_mode,
1478 		.maxlen		= sizeof(laptop_mode),
1479 		.mode		= 0644,
1480 		.proc_handler	= proc_dointvec_jiffies,
1481 	},
1482 	{
1483 		.procname	= "block_dump",
1484 		.data		= &block_dump,
1485 		.maxlen		= sizeof(block_dump),
1486 		.mode		= 0644,
1487 		.proc_handler	= proc_dointvec,
1488 		.extra1		= &zero,
1489 	},
1490 	{
1491 		.procname	= "vfs_cache_pressure",
1492 		.data		= &sysctl_vfs_cache_pressure,
1493 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1494 		.mode		= 0644,
1495 		.proc_handler	= proc_dointvec,
1496 		.extra1		= &zero,
1497 	},
1498 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1499 	{
1500 		.procname	= "legacy_va_layout",
1501 		.data		= &sysctl_legacy_va_layout,
1502 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1503 		.mode		= 0644,
1504 		.proc_handler	= proc_dointvec,
1505 		.extra1		= &zero,
1506 	},
1507 #endif
1508 #ifdef CONFIG_NUMA
1509 	{
1510 		.procname	= "zone_reclaim_mode",
1511 		.data		= &node_reclaim_mode,
1512 		.maxlen		= sizeof(node_reclaim_mode),
1513 		.mode		= 0644,
1514 		.proc_handler	= proc_dointvec,
1515 		.extra1		= &zero,
1516 	},
1517 	{
1518 		.procname	= "min_unmapped_ratio",
1519 		.data		= &sysctl_min_unmapped_ratio,
1520 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1521 		.mode		= 0644,
1522 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1523 		.extra1		= &zero,
1524 		.extra2		= &one_hundred,
1525 	},
1526 	{
1527 		.procname	= "min_slab_ratio",
1528 		.data		= &sysctl_min_slab_ratio,
1529 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1530 		.mode		= 0644,
1531 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1532 		.extra1		= &zero,
1533 		.extra2		= &one_hundred,
1534 	},
1535 #endif
1536 #ifdef CONFIG_SMP
1537 	{
1538 		.procname	= "stat_interval",
1539 		.data		= &sysctl_stat_interval,
1540 		.maxlen		= sizeof(sysctl_stat_interval),
1541 		.mode		= 0644,
1542 		.proc_handler	= proc_dointvec_jiffies,
1543 	},
1544 	{
1545 		.procname	= "stat_refresh",
1546 		.data		= NULL,
1547 		.maxlen		= 0,
1548 		.mode		= 0600,
1549 		.proc_handler	= vmstat_refresh,
1550 	},
1551 #endif
1552 #ifdef CONFIG_MMU
1553 	{
1554 		.procname	= "mmap_min_addr",
1555 		.data		= &dac_mmap_min_addr,
1556 		.maxlen		= sizeof(unsigned long),
1557 		.mode		= 0644,
1558 		.proc_handler	= mmap_min_addr_handler,
1559 	},
1560 #endif
1561 #ifdef CONFIG_NUMA
1562 	{
1563 		.procname	= "numa_zonelist_order",
1564 		.data		= &numa_zonelist_order,
1565 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1566 		.mode		= 0644,
1567 		.proc_handler	= numa_zonelist_order_handler,
1568 	},
1569 #endif
1570 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1571    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1572 	{
1573 		.procname	= "vdso_enabled",
1574 #ifdef CONFIG_X86_32
1575 		.data		= &vdso32_enabled,
1576 		.maxlen		= sizeof(vdso32_enabled),
1577 #else
1578 		.data		= &vdso_enabled,
1579 		.maxlen		= sizeof(vdso_enabled),
1580 #endif
1581 		.mode		= 0644,
1582 		.proc_handler	= proc_dointvec,
1583 		.extra1		= &zero,
1584 	},
1585 #endif
1586 #ifdef CONFIG_HIGHMEM
1587 	{
1588 		.procname	= "highmem_is_dirtyable",
1589 		.data		= &vm_highmem_is_dirtyable,
1590 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1591 		.mode		= 0644,
1592 		.proc_handler	= proc_dointvec_minmax,
1593 		.extra1		= &zero,
1594 		.extra2		= &one,
1595 	},
1596 #endif
1597 #ifdef CONFIG_MEMORY_FAILURE
1598 	{
1599 		.procname	= "memory_failure_early_kill",
1600 		.data		= &sysctl_memory_failure_early_kill,
1601 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1602 		.mode		= 0644,
1603 		.proc_handler	= proc_dointvec_minmax,
1604 		.extra1		= &zero,
1605 		.extra2		= &one,
1606 	},
1607 	{
1608 		.procname	= "memory_failure_recovery",
1609 		.data		= &sysctl_memory_failure_recovery,
1610 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1611 		.mode		= 0644,
1612 		.proc_handler	= proc_dointvec_minmax,
1613 		.extra1		= &zero,
1614 		.extra2		= &one,
1615 	},
1616 #endif
1617 	{
1618 		.procname	= "user_reserve_kbytes",
1619 		.data		= &sysctl_user_reserve_kbytes,
1620 		.maxlen		= sizeof(sysctl_user_reserve_kbytes),
1621 		.mode		= 0644,
1622 		.proc_handler	= proc_doulongvec_minmax,
1623 	},
1624 	{
1625 		.procname	= "admin_reserve_kbytes",
1626 		.data		= &sysctl_admin_reserve_kbytes,
1627 		.maxlen		= sizeof(sysctl_admin_reserve_kbytes),
1628 		.mode		= 0644,
1629 		.proc_handler	= proc_doulongvec_minmax,
1630 	},
1631 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1632 	{
1633 		.procname	= "mmap_rnd_bits",
1634 		.data		= &mmap_rnd_bits,
1635 		.maxlen		= sizeof(mmap_rnd_bits),
1636 		.mode		= 0600,
1637 		.proc_handler	= proc_dointvec_minmax,
1638 		.extra1		= (void *)&mmap_rnd_bits_min,
1639 		.extra2		= (void *)&mmap_rnd_bits_max,
1640 	},
1641 #endif
1642 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1643 	{
1644 		.procname	= "mmap_rnd_compat_bits",
1645 		.data		= &mmap_rnd_compat_bits,
1646 		.maxlen		= sizeof(mmap_rnd_compat_bits),
1647 		.mode		= 0600,
1648 		.proc_handler	= proc_dointvec_minmax,
1649 		.extra1		= (void *)&mmap_rnd_compat_bits_min,
1650 		.extra2		= (void *)&mmap_rnd_compat_bits_max,
1651 	},
1652 #endif
1653 	{ }
1654 };
1655 
1656 static struct ctl_table fs_table[] = {
1657 	{
1658 		.procname	= "inode-nr",
1659 		.data		= &inodes_stat,
1660 		.maxlen		= 2*sizeof(long),
1661 		.mode		= 0444,
1662 		.proc_handler	= proc_nr_inodes,
1663 	},
1664 	{
1665 		.procname	= "inode-state",
1666 		.data		= &inodes_stat,
1667 		.maxlen		= 7*sizeof(long),
1668 		.mode		= 0444,
1669 		.proc_handler	= proc_nr_inodes,
1670 	},
1671 	{
1672 		.procname	= "file-nr",
1673 		.data		= &files_stat,
1674 		.maxlen		= sizeof(files_stat),
1675 		.mode		= 0444,
1676 		.proc_handler	= proc_nr_files,
1677 	},
1678 	{
1679 		.procname	= "file-max",
1680 		.data		= &files_stat.max_files,
1681 		.maxlen		= sizeof(files_stat.max_files),
1682 		.mode		= 0644,
1683 		.proc_handler	= proc_doulongvec_minmax,
1684 	},
1685 	{
1686 		.procname	= "nr_open",
1687 		.data		= &sysctl_nr_open,
1688 		.maxlen		= sizeof(int),
1689 		.mode		= 0644,
1690 		.proc_handler	= proc_dointvec_minmax,
1691 		.extra1		= &sysctl_nr_open_min,
1692 		.extra2		= &sysctl_nr_open_max,
1693 	},
1694 	{
1695 		.procname	= "dentry-state",
1696 		.data		= &dentry_stat,
1697 		.maxlen		= 6*sizeof(long),
1698 		.mode		= 0444,
1699 		.proc_handler	= proc_nr_dentry,
1700 	},
1701 	{
1702 		.procname	= "overflowuid",
1703 		.data		= &fs_overflowuid,
1704 		.maxlen		= sizeof(int),
1705 		.mode		= 0644,
1706 		.proc_handler	= proc_dointvec_minmax,
1707 		.extra1		= &minolduid,
1708 		.extra2		= &maxolduid,
1709 	},
1710 	{
1711 		.procname	= "overflowgid",
1712 		.data		= &fs_overflowgid,
1713 		.maxlen		= sizeof(int),
1714 		.mode		= 0644,
1715 		.proc_handler	= proc_dointvec_minmax,
1716 		.extra1		= &minolduid,
1717 		.extra2		= &maxolduid,
1718 	},
1719 #ifdef CONFIG_FILE_LOCKING
1720 	{
1721 		.procname	= "leases-enable",
1722 		.data		= &leases_enable,
1723 		.maxlen		= sizeof(int),
1724 		.mode		= 0644,
1725 		.proc_handler	= proc_dointvec,
1726 	},
1727 #endif
1728 #ifdef CONFIG_DNOTIFY
1729 	{
1730 		.procname	= "dir-notify-enable",
1731 		.data		= &dir_notify_enable,
1732 		.maxlen		= sizeof(int),
1733 		.mode		= 0644,
1734 		.proc_handler	= proc_dointvec,
1735 	},
1736 #endif
1737 #ifdef CONFIG_MMU
1738 #ifdef CONFIG_FILE_LOCKING
1739 	{
1740 		.procname	= "lease-break-time",
1741 		.data		= &lease_break_time,
1742 		.maxlen		= sizeof(int),
1743 		.mode		= 0644,
1744 		.proc_handler	= proc_dointvec,
1745 	},
1746 #endif
1747 #ifdef CONFIG_AIO
1748 	{
1749 		.procname	= "aio-nr",
1750 		.data		= &aio_nr,
1751 		.maxlen		= sizeof(aio_nr),
1752 		.mode		= 0444,
1753 		.proc_handler	= proc_doulongvec_minmax,
1754 	},
1755 	{
1756 		.procname	= "aio-max-nr",
1757 		.data		= &aio_max_nr,
1758 		.maxlen		= sizeof(aio_max_nr),
1759 		.mode		= 0644,
1760 		.proc_handler	= proc_doulongvec_minmax,
1761 	},
1762 #endif /* CONFIG_AIO */
1763 #ifdef CONFIG_INOTIFY_USER
1764 	{
1765 		.procname	= "inotify",
1766 		.mode		= 0555,
1767 		.child		= inotify_table,
1768 	},
1769 #endif
1770 #ifdef CONFIG_EPOLL
1771 	{
1772 		.procname	= "epoll",
1773 		.mode		= 0555,
1774 		.child		= epoll_table,
1775 	},
1776 #endif
1777 #endif
1778 	{
1779 		.procname	= "protected_symlinks",
1780 		.data		= &sysctl_protected_symlinks,
1781 		.maxlen		= sizeof(int),
1782 		.mode		= 0600,
1783 		.proc_handler	= proc_dointvec_minmax,
1784 		.extra1		= &zero,
1785 		.extra2		= &one,
1786 	},
1787 	{
1788 		.procname	= "protected_hardlinks",
1789 		.data		= &sysctl_protected_hardlinks,
1790 		.maxlen		= sizeof(int),
1791 		.mode		= 0600,
1792 		.proc_handler	= proc_dointvec_minmax,
1793 		.extra1		= &zero,
1794 		.extra2		= &one,
1795 	},
1796 	{
1797 		.procname	= "suid_dumpable",
1798 		.data		= &suid_dumpable,
1799 		.maxlen		= sizeof(int),
1800 		.mode		= 0644,
1801 		.proc_handler	= proc_dointvec_minmax_coredump,
1802 		.extra1		= &zero,
1803 		.extra2		= &two,
1804 	},
1805 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1806 	{
1807 		.procname	= "binfmt_misc",
1808 		.mode		= 0555,
1809 		.child		= sysctl_mount_point,
1810 	},
1811 #endif
1812 	{
1813 		.procname	= "pipe-max-size",
1814 		.data		= &pipe_max_size,
1815 		.maxlen		= sizeof(int),
1816 		.mode		= 0644,
1817 		.proc_handler	= &pipe_proc_fn,
1818 		.extra1		= &pipe_min_size,
1819 	},
1820 	{
1821 		.procname	= "pipe-user-pages-hard",
1822 		.data		= &pipe_user_pages_hard,
1823 		.maxlen		= sizeof(pipe_user_pages_hard),
1824 		.mode		= 0644,
1825 		.proc_handler	= proc_doulongvec_minmax,
1826 	},
1827 	{
1828 		.procname	= "pipe-user-pages-soft",
1829 		.data		= &pipe_user_pages_soft,
1830 		.maxlen		= sizeof(pipe_user_pages_soft),
1831 		.mode		= 0644,
1832 		.proc_handler	= proc_doulongvec_minmax,
1833 	},
1834 	{ }
1835 };
1836 
1837 static struct ctl_table debug_table[] = {
1838 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1839 	{
1840 		.procname	= "exception-trace",
1841 		.data		= &show_unhandled_signals,
1842 		.maxlen		= sizeof(int),
1843 		.mode		= 0644,
1844 		.proc_handler	= proc_dointvec
1845 	},
1846 #endif
1847 #if defined(CONFIG_OPTPROBES)
1848 	{
1849 		.procname	= "kprobes-optimization",
1850 		.data		= &sysctl_kprobes_optimization,
1851 		.maxlen		= sizeof(int),
1852 		.mode		= 0644,
1853 		.proc_handler	= proc_kprobes_optimization_handler,
1854 		.extra1		= &zero,
1855 		.extra2		= &one,
1856 	},
1857 #endif
1858 	{ }
1859 };
1860 
1861 static struct ctl_table dev_table[] = {
1862 	{ }
1863 };
1864 
1865 int __init sysctl_init(void)
1866 {
1867 	struct ctl_table_header *hdr;
1868 
1869 	hdr = register_sysctl_table(sysctl_base_table);
1870 	kmemleak_not_leak(hdr);
1871 	return 0;
1872 }
1873 
1874 #endif /* CONFIG_SYSCTL */
1875 
1876 /*
1877  * /proc/sys support
1878  */
1879 
1880 #ifdef CONFIG_PROC_SYSCTL
1881 
1882 static int _proc_do_string(char *data, int maxlen, int write,
1883 			   char __user *buffer,
1884 			   size_t *lenp, loff_t *ppos)
1885 {
1886 	size_t len;
1887 	char __user *p;
1888 	char c;
1889 
1890 	if (!data || !maxlen || !*lenp) {
1891 		*lenp = 0;
1892 		return 0;
1893 	}
1894 
1895 	if (write) {
1896 		if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1897 			/* Only continue writes not past the end of buffer. */
1898 			len = strlen(data);
1899 			if (len > maxlen - 1)
1900 				len = maxlen - 1;
1901 
1902 			if (*ppos > len)
1903 				return 0;
1904 			len = *ppos;
1905 		} else {
1906 			/* Start writing from beginning of buffer. */
1907 			len = 0;
1908 		}
1909 
1910 		*ppos += *lenp;
1911 		p = buffer;
1912 		while ((p - buffer) < *lenp && len < maxlen - 1) {
1913 			if (get_user(c, p++))
1914 				return -EFAULT;
1915 			if (c == 0 || c == '\n')
1916 				break;
1917 			data[len++] = c;
1918 		}
1919 		data[len] = 0;
1920 	} else {
1921 		len = strlen(data);
1922 		if (len > maxlen)
1923 			len = maxlen;
1924 
1925 		if (*ppos > len) {
1926 			*lenp = 0;
1927 			return 0;
1928 		}
1929 
1930 		data += *ppos;
1931 		len  -= *ppos;
1932 
1933 		if (len > *lenp)
1934 			len = *lenp;
1935 		if (len)
1936 			if (copy_to_user(buffer, data, len))
1937 				return -EFAULT;
1938 		if (len < *lenp) {
1939 			if (put_user('\n', buffer + len))
1940 				return -EFAULT;
1941 			len++;
1942 		}
1943 		*lenp = len;
1944 		*ppos += len;
1945 	}
1946 	return 0;
1947 }
1948 
1949 static void warn_sysctl_write(struct ctl_table *table)
1950 {
1951 	pr_warn_once("%s wrote to %s when file position was not 0!\n"
1952 		"This will not be supported in the future. To silence this\n"
1953 		"warning, set kernel.sysctl_writes_strict = -1\n",
1954 		current->comm, table->procname);
1955 }
1956 
1957 /**
1958  * proc_dostring - read a string sysctl
1959  * @table: the sysctl table
1960  * @write: %TRUE if this is a write to the sysctl file
1961  * @buffer: the user buffer
1962  * @lenp: the size of the user buffer
1963  * @ppos: file position
1964  *
1965  * Reads/writes a string from/to the user buffer. If the kernel
1966  * buffer provided is not large enough to hold the string, the
1967  * string is truncated. The copied string is %NULL-terminated.
1968  * If the string is being read by the user process, it is copied
1969  * and a newline '\n' is added. It is truncated if the buffer is
1970  * not large enough.
1971  *
1972  * Returns 0 on success.
1973  */
1974 int proc_dostring(struct ctl_table *table, int write,
1975 		  void __user *buffer, size_t *lenp, loff_t *ppos)
1976 {
1977 	if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN)
1978 		warn_sysctl_write(table);
1979 
1980 	return _proc_do_string((char *)(table->data), table->maxlen, write,
1981 			       (char __user *)buffer, lenp, ppos);
1982 }
1983 
1984 static size_t proc_skip_spaces(char **buf)
1985 {
1986 	size_t ret;
1987 	char *tmp = skip_spaces(*buf);
1988 	ret = tmp - *buf;
1989 	*buf = tmp;
1990 	return ret;
1991 }
1992 
1993 static void proc_skip_char(char **buf, size_t *size, const char v)
1994 {
1995 	while (*size) {
1996 		if (**buf != v)
1997 			break;
1998 		(*size)--;
1999 		(*buf)++;
2000 	}
2001 }
2002 
2003 #define TMPBUFLEN 22
2004 /**
2005  * proc_get_long - reads an ASCII formatted integer from a user buffer
2006  *
2007  * @buf: a kernel buffer
2008  * @size: size of the kernel buffer
2009  * @val: this is where the number will be stored
2010  * @neg: set to %TRUE if number is negative
2011  * @perm_tr: a vector which contains the allowed trailers
2012  * @perm_tr_len: size of the perm_tr vector
2013  * @tr: pointer to store the trailer character
2014  *
2015  * In case of success %0 is returned and @buf and @size are updated with
2016  * the amount of bytes read. If @tr is non-NULL and a trailing
2017  * character exists (size is non-zero after returning from this
2018  * function), @tr is updated with the trailing character.
2019  */
2020 static int proc_get_long(char **buf, size_t *size,
2021 			  unsigned long *val, bool *neg,
2022 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
2023 {
2024 	int len;
2025 	char *p, tmp[TMPBUFLEN];
2026 
2027 	if (!*size)
2028 		return -EINVAL;
2029 
2030 	len = *size;
2031 	if (len > TMPBUFLEN - 1)
2032 		len = TMPBUFLEN - 1;
2033 
2034 	memcpy(tmp, *buf, len);
2035 
2036 	tmp[len] = 0;
2037 	p = tmp;
2038 	if (*p == '-' && *size > 1) {
2039 		*neg = true;
2040 		p++;
2041 	} else
2042 		*neg = false;
2043 	if (!isdigit(*p))
2044 		return -EINVAL;
2045 
2046 	*val = simple_strtoul(p, &p, 0);
2047 
2048 	len = p - tmp;
2049 
2050 	/* We don't know if the next char is whitespace thus we may accept
2051 	 * invalid integers (e.g. 1234...a) or two integers instead of one
2052 	 * (e.g. 123...1). So lets not allow such large numbers. */
2053 	if (len == TMPBUFLEN - 1)
2054 		return -EINVAL;
2055 
2056 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2057 		return -EINVAL;
2058 
2059 	if (tr && (len < *size))
2060 		*tr = *p;
2061 
2062 	*buf += len;
2063 	*size -= len;
2064 
2065 	return 0;
2066 }
2067 
2068 /**
2069  * proc_put_long - converts an integer to a decimal ASCII formatted string
2070  *
2071  * @buf: the user buffer
2072  * @size: the size of the user buffer
2073  * @val: the integer to be converted
2074  * @neg: sign of the number, %TRUE for negative
2075  *
2076  * In case of success %0 is returned and @buf and @size are updated with
2077  * the amount of bytes written.
2078  */
2079 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2080 			  bool neg)
2081 {
2082 	int len;
2083 	char tmp[TMPBUFLEN], *p = tmp;
2084 
2085 	sprintf(p, "%s%lu", neg ? "-" : "", val);
2086 	len = strlen(tmp);
2087 	if (len > *size)
2088 		len = *size;
2089 	if (copy_to_user(*buf, tmp, len))
2090 		return -EFAULT;
2091 	*size -= len;
2092 	*buf += len;
2093 	return 0;
2094 }
2095 #undef TMPBUFLEN
2096 
2097 static int proc_put_char(void __user **buf, size_t *size, char c)
2098 {
2099 	if (*size) {
2100 		char __user **buffer = (char __user **)buf;
2101 		if (put_user(c, *buffer))
2102 			return -EFAULT;
2103 		(*size)--, (*buffer)++;
2104 		*buf = *buffer;
2105 	}
2106 	return 0;
2107 }
2108 
2109 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2110 				 int *valp,
2111 				 int write, void *data)
2112 {
2113 	if (write) {
2114 		if (*negp) {
2115 			if (*lvalp > (unsigned long) INT_MAX + 1)
2116 				return -EINVAL;
2117 			*valp = -*lvalp;
2118 		} else {
2119 			if (*lvalp > (unsigned long) INT_MAX)
2120 				return -EINVAL;
2121 			*valp = *lvalp;
2122 		}
2123 	} else {
2124 		int val = *valp;
2125 		if (val < 0) {
2126 			*negp = true;
2127 			*lvalp = -(unsigned long)val;
2128 		} else {
2129 			*negp = false;
2130 			*lvalp = (unsigned long)val;
2131 		}
2132 	}
2133 	return 0;
2134 }
2135 
2136 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2137 
2138 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2139 		  int write, void __user *buffer,
2140 		  size_t *lenp, loff_t *ppos,
2141 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2142 			      int write, void *data),
2143 		  void *data)
2144 {
2145 	int *i, vleft, first = 1, err = 0;
2146 	size_t left;
2147 	char *kbuf = NULL, *p;
2148 
2149 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2150 		*lenp = 0;
2151 		return 0;
2152 	}
2153 
2154 	i = (int *) tbl_data;
2155 	vleft = table->maxlen / sizeof(*i);
2156 	left = *lenp;
2157 
2158 	if (!conv)
2159 		conv = do_proc_dointvec_conv;
2160 
2161 	if (write) {
2162 		if (*ppos) {
2163 			switch (sysctl_writes_strict) {
2164 			case SYSCTL_WRITES_STRICT:
2165 				goto out;
2166 			case SYSCTL_WRITES_WARN:
2167 				warn_sysctl_write(table);
2168 				break;
2169 			default:
2170 				break;
2171 			}
2172 		}
2173 
2174 		if (left > PAGE_SIZE - 1)
2175 			left = PAGE_SIZE - 1;
2176 		p = kbuf = memdup_user_nul(buffer, left);
2177 		if (IS_ERR(kbuf))
2178 			return PTR_ERR(kbuf);
2179 	}
2180 
2181 	for (; left && vleft--; i++, first=0) {
2182 		unsigned long lval;
2183 		bool neg;
2184 
2185 		if (write) {
2186 			left -= proc_skip_spaces(&p);
2187 
2188 			if (!left)
2189 				break;
2190 			err = proc_get_long(&p, &left, &lval, &neg,
2191 					     proc_wspace_sep,
2192 					     sizeof(proc_wspace_sep), NULL);
2193 			if (err)
2194 				break;
2195 			if (conv(&neg, &lval, i, 1, data)) {
2196 				err = -EINVAL;
2197 				break;
2198 			}
2199 		} else {
2200 			if (conv(&neg, &lval, i, 0, data)) {
2201 				err = -EINVAL;
2202 				break;
2203 			}
2204 			if (!first)
2205 				err = proc_put_char(&buffer, &left, '\t');
2206 			if (err)
2207 				break;
2208 			err = proc_put_long(&buffer, &left, lval, neg);
2209 			if (err)
2210 				break;
2211 		}
2212 	}
2213 
2214 	if (!write && !first && left && !err)
2215 		err = proc_put_char(&buffer, &left, '\n');
2216 	if (write && !err && left)
2217 		left -= proc_skip_spaces(&p);
2218 	if (write) {
2219 		kfree(kbuf);
2220 		if (first)
2221 			return err ? : -EINVAL;
2222 	}
2223 	*lenp -= left;
2224 out:
2225 	*ppos += *lenp;
2226 	return err;
2227 }
2228 
2229 static int do_proc_dointvec(struct ctl_table *table, int write,
2230 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2231 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2232 			      int write, void *data),
2233 		  void *data)
2234 {
2235 	return __do_proc_dointvec(table->data, table, write,
2236 			buffer, lenp, ppos, conv, data);
2237 }
2238 
2239 /**
2240  * proc_dointvec - read a vector of integers
2241  * @table: the sysctl table
2242  * @write: %TRUE if this is a write to the sysctl file
2243  * @buffer: the user buffer
2244  * @lenp: the size of the user buffer
2245  * @ppos: file position
2246  *
2247  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2248  * values from/to the user buffer, treated as an ASCII string.
2249  *
2250  * Returns 0 on success.
2251  */
2252 int proc_dointvec(struct ctl_table *table, int write,
2253 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2254 {
2255     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2256 		    	    NULL,NULL);
2257 }
2258 
2259 /*
2260  * Taint values can only be increased
2261  * This means we can safely use a temporary.
2262  */
2263 static int proc_taint(struct ctl_table *table, int write,
2264 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2265 {
2266 	struct ctl_table t;
2267 	unsigned long tmptaint = get_taint();
2268 	int err;
2269 
2270 	if (write && !capable(CAP_SYS_ADMIN))
2271 		return -EPERM;
2272 
2273 	t = *table;
2274 	t.data = &tmptaint;
2275 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2276 	if (err < 0)
2277 		return err;
2278 
2279 	if (write) {
2280 		/*
2281 		 * Poor man's atomic or. Not worth adding a primitive
2282 		 * to everyone's atomic.h for this
2283 		 */
2284 		int i;
2285 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2286 			if ((tmptaint >> i) & 1)
2287 				add_taint(i, LOCKDEP_STILL_OK);
2288 		}
2289 	}
2290 
2291 	return err;
2292 }
2293 
2294 #ifdef CONFIG_PRINTK
2295 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2296 				void __user *buffer, size_t *lenp, loff_t *ppos)
2297 {
2298 	if (write && !capable(CAP_SYS_ADMIN))
2299 		return -EPERM;
2300 
2301 	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2302 }
2303 #endif
2304 
2305 struct do_proc_dointvec_minmax_conv_param {
2306 	int *min;
2307 	int *max;
2308 };
2309 
2310 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2311 					int *valp,
2312 					int write, void *data)
2313 {
2314 	struct do_proc_dointvec_minmax_conv_param *param = data;
2315 	if (write) {
2316 		int val = *negp ? -*lvalp : *lvalp;
2317 		if ((param->min && *param->min > val) ||
2318 		    (param->max && *param->max < val))
2319 			return -EINVAL;
2320 		*valp = val;
2321 	} else {
2322 		int val = *valp;
2323 		if (val < 0) {
2324 			*negp = true;
2325 			*lvalp = -(unsigned long)val;
2326 		} else {
2327 			*negp = false;
2328 			*lvalp = (unsigned long)val;
2329 		}
2330 	}
2331 	return 0;
2332 }
2333 
2334 /**
2335  * proc_dointvec_minmax - read a vector of integers with min/max values
2336  * @table: the sysctl table
2337  * @write: %TRUE if this is a write to the sysctl file
2338  * @buffer: the user buffer
2339  * @lenp: the size of the user buffer
2340  * @ppos: file position
2341  *
2342  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2343  * values from/to the user buffer, treated as an ASCII string.
2344  *
2345  * This routine will ensure the values are within the range specified by
2346  * table->extra1 (min) and table->extra2 (max).
2347  *
2348  * Returns 0 on success.
2349  */
2350 int proc_dointvec_minmax(struct ctl_table *table, int write,
2351 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2352 {
2353 	struct do_proc_dointvec_minmax_conv_param param = {
2354 		.min = (int *) table->extra1,
2355 		.max = (int *) table->extra2,
2356 	};
2357 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2358 				do_proc_dointvec_minmax_conv, &param);
2359 }
2360 
2361 static void validate_coredump_safety(void)
2362 {
2363 #ifdef CONFIG_COREDUMP
2364 	if (suid_dumpable == SUID_DUMP_ROOT &&
2365 	    core_pattern[0] != '/' && core_pattern[0] != '|') {
2366 		printk(KERN_WARNING "Unsafe core_pattern used with "\
2367 			"suid_dumpable=2. Pipe handler or fully qualified "\
2368 			"core dump path required.\n");
2369 	}
2370 #endif
2371 }
2372 
2373 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2374 		void __user *buffer, size_t *lenp, loff_t *ppos)
2375 {
2376 	int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2377 	if (!error)
2378 		validate_coredump_safety();
2379 	return error;
2380 }
2381 
2382 #ifdef CONFIG_COREDUMP
2383 static int proc_dostring_coredump(struct ctl_table *table, int write,
2384 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2385 {
2386 	int error = proc_dostring(table, write, buffer, lenp, ppos);
2387 	if (!error)
2388 		validate_coredump_safety();
2389 	return error;
2390 }
2391 #endif
2392 
2393 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2394 				     void __user *buffer,
2395 				     size_t *lenp, loff_t *ppos,
2396 				     unsigned long convmul,
2397 				     unsigned long convdiv)
2398 {
2399 	unsigned long *i, *min, *max;
2400 	int vleft, first = 1, err = 0;
2401 	size_t left;
2402 	char *kbuf = NULL, *p;
2403 
2404 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2405 		*lenp = 0;
2406 		return 0;
2407 	}
2408 
2409 	i = (unsigned long *) data;
2410 	min = (unsigned long *) table->extra1;
2411 	max = (unsigned long *) table->extra2;
2412 	vleft = table->maxlen / sizeof(unsigned long);
2413 	left = *lenp;
2414 
2415 	if (write) {
2416 		if (*ppos) {
2417 			switch (sysctl_writes_strict) {
2418 			case SYSCTL_WRITES_STRICT:
2419 				goto out;
2420 			case SYSCTL_WRITES_WARN:
2421 				warn_sysctl_write(table);
2422 				break;
2423 			default:
2424 				break;
2425 			}
2426 		}
2427 
2428 		if (left > PAGE_SIZE - 1)
2429 			left = PAGE_SIZE - 1;
2430 		p = kbuf = memdup_user_nul(buffer, left);
2431 		if (IS_ERR(kbuf))
2432 			return PTR_ERR(kbuf);
2433 	}
2434 
2435 	for (; left && vleft--; i++, first = 0) {
2436 		unsigned long val;
2437 
2438 		if (write) {
2439 			bool neg;
2440 
2441 			left -= proc_skip_spaces(&p);
2442 
2443 			err = proc_get_long(&p, &left, &val, &neg,
2444 					     proc_wspace_sep,
2445 					     sizeof(proc_wspace_sep), NULL);
2446 			if (err)
2447 				break;
2448 			if (neg)
2449 				continue;
2450 			if ((min && val < *min) || (max && val > *max))
2451 				continue;
2452 			*i = val;
2453 		} else {
2454 			val = convdiv * (*i) / convmul;
2455 			if (!first) {
2456 				err = proc_put_char(&buffer, &left, '\t');
2457 				if (err)
2458 					break;
2459 			}
2460 			err = proc_put_long(&buffer, &left, val, false);
2461 			if (err)
2462 				break;
2463 		}
2464 	}
2465 
2466 	if (!write && !first && left && !err)
2467 		err = proc_put_char(&buffer, &left, '\n');
2468 	if (write && !err)
2469 		left -= proc_skip_spaces(&p);
2470 	if (write) {
2471 		kfree(kbuf);
2472 		if (first)
2473 			return err ? : -EINVAL;
2474 	}
2475 	*lenp -= left;
2476 out:
2477 	*ppos += *lenp;
2478 	return err;
2479 }
2480 
2481 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2482 				     void __user *buffer,
2483 				     size_t *lenp, loff_t *ppos,
2484 				     unsigned long convmul,
2485 				     unsigned long convdiv)
2486 {
2487 	return __do_proc_doulongvec_minmax(table->data, table, write,
2488 			buffer, lenp, ppos, convmul, convdiv);
2489 }
2490 
2491 /**
2492  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2493  * @table: the sysctl table
2494  * @write: %TRUE if this is a write to the sysctl file
2495  * @buffer: the user buffer
2496  * @lenp: the size of the user buffer
2497  * @ppos: file position
2498  *
2499  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2500  * values from/to the user buffer, treated as an ASCII string.
2501  *
2502  * This routine will ensure the values are within the range specified by
2503  * table->extra1 (min) and table->extra2 (max).
2504  *
2505  * Returns 0 on success.
2506  */
2507 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2508 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2509 {
2510     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2511 }
2512 
2513 /**
2514  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2515  * @table: the sysctl table
2516  * @write: %TRUE if this is a write to the sysctl file
2517  * @buffer: the user buffer
2518  * @lenp: the size of the user buffer
2519  * @ppos: file position
2520  *
2521  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2522  * values from/to the user buffer, treated as an ASCII string. The values
2523  * are treated as milliseconds, and converted to jiffies when they are stored.
2524  *
2525  * This routine will ensure the values are within the range specified by
2526  * table->extra1 (min) and table->extra2 (max).
2527  *
2528  * Returns 0 on success.
2529  */
2530 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2531 				      void __user *buffer,
2532 				      size_t *lenp, loff_t *ppos)
2533 {
2534     return do_proc_doulongvec_minmax(table, write, buffer,
2535 				     lenp, ppos, HZ, 1000l);
2536 }
2537 
2538 
2539 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2540 					 int *valp,
2541 					 int write, void *data)
2542 {
2543 	if (write) {
2544 		if (*lvalp > LONG_MAX / HZ)
2545 			return 1;
2546 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2547 	} else {
2548 		int val = *valp;
2549 		unsigned long lval;
2550 		if (val < 0) {
2551 			*negp = true;
2552 			lval = -(unsigned long)val;
2553 		} else {
2554 			*negp = false;
2555 			lval = (unsigned long)val;
2556 		}
2557 		*lvalp = lval / HZ;
2558 	}
2559 	return 0;
2560 }
2561 
2562 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2563 						int *valp,
2564 						int write, void *data)
2565 {
2566 	if (write) {
2567 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2568 			return 1;
2569 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2570 	} else {
2571 		int val = *valp;
2572 		unsigned long lval;
2573 		if (val < 0) {
2574 			*negp = true;
2575 			lval = -(unsigned long)val;
2576 		} else {
2577 			*negp = false;
2578 			lval = (unsigned long)val;
2579 		}
2580 		*lvalp = jiffies_to_clock_t(lval);
2581 	}
2582 	return 0;
2583 }
2584 
2585 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2586 					    int *valp,
2587 					    int write, void *data)
2588 {
2589 	if (write) {
2590 		unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2591 
2592 		if (jif > INT_MAX)
2593 			return 1;
2594 		*valp = (int)jif;
2595 	} else {
2596 		int val = *valp;
2597 		unsigned long lval;
2598 		if (val < 0) {
2599 			*negp = true;
2600 			lval = -(unsigned long)val;
2601 		} else {
2602 			*negp = false;
2603 			lval = (unsigned long)val;
2604 		}
2605 		*lvalp = jiffies_to_msecs(lval);
2606 	}
2607 	return 0;
2608 }
2609 
2610 /**
2611  * proc_dointvec_jiffies - read a vector of integers as seconds
2612  * @table: the sysctl table
2613  * @write: %TRUE if this is a write to the sysctl file
2614  * @buffer: the user buffer
2615  * @lenp: the size of the user buffer
2616  * @ppos: file position
2617  *
2618  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2619  * values from/to the user buffer, treated as an ASCII string.
2620  * The values read are assumed to be in seconds, and are converted into
2621  * jiffies.
2622  *
2623  * Returns 0 on success.
2624  */
2625 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2626 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2627 {
2628     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2629 		    	    do_proc_dointvec_jiffies_conv,NULL);
2630 }
2631 
2632 /**
2633  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2634  * @table: the sysctl table
2635  * @write: %TRUE if this is a write to the sysctl file
2636  * @buffer: the user buffer
2637  * @lenp: the size of the user buffer
2638  * @ppos: pointer to the file position
2639  *
2640  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2641  * values from/to the user buffer, treated as an ASCII string.
2642  * The values read are assumed to be in 1/USER_HZ seconds, and
2643  * are converted into jiffies.
2644  *
2645  * Returns 0 on success.
2646  */
2647 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2648 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2649 {
2650     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2651 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2652 }
2653 
2654 /**
2655  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2656  * @table: the sysctl table
2657  * @write: %TRUE if this is a write to the sysctl file
2658  * @buffer: the user buffer
2659  * @lenp: the size of the user buffer
2660  * @ppos: file position
2661  * @ppos: the current position in the file
2662  *
2663  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2664  * values from/to the user buffer, treated as an ASCII string.
2665  * The values read are assumed to be in 1/1000 seconds, and
2666  * are converted into jiffies.
2667  *
2668  * Returns 0 on success.
2669  */
2670 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2671 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2672 {
2673 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2674 				do_proc_dointvec_ms_jiffies_conv, NULL);
2675 }
2676 
2677 static int proc_do_cad_pid(struct ctl_table *table, int write,
2678 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2679 {
2680 	struct pid *new_pid;
2681 	pid_t tmp;
2682 	int r;
2683 
2684 	tmp = pid_vnr(cad_pid);
2685 
2686 	r = __do_proc_dointvec(&tmp, table, write, buffer,
2687 			       lenp, ppos, NULL, NULL);
2688 	if (r || !write)
2689 		return r;
2690 
2691 	new_pid = find_get_pid(tmp);
2692 	if (!new_pid)
2693 		return -ESRCH;
2694 
2695 	put_pid(xchg(&cad_pid, new_pid));
2696 	return 0;
2697 }
2698 
2699 /**
2700  * proc_do_large_bitmap - read/write from/to a large bitmap
2701  * @table: the sysctl table
2702  * @write: %TRUE if this is a write to the sysctl file
2703  * @buffer: the user buffer
2704  * @lenp: the size of the user buffer
2705  * @ppos: file position
2706  *
2707  * The bitmap is stored at table->data and the bitmap length (in bits)
2708  * in table->maxlen.
2709  *
2710  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2711  * large bitmaps may be represented in a compact manner. Writing into
2712  * the file will clear the bitmap then update it with the given input.
2713  *
2714  * Returns 0 on success.
2715  */
2716 int proc_do_large_bitmap(struct ctl_table *table, int write,
2717 			 void __user *buffer, size_t *lenp, loff_t *ppos)
2718 {
2719 	int err = 0;
2720 	bool first = 1;
2721 	size_t left = *lenp;
2722 	unsigned long bitmap_len = table->maxlen;
2723 	unsigned long *bitmap = *(unsigned long **) table->data;
2724 	unsigned long *tmp_bitmap = NULL;
2725 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2726 
2727 	if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2728 		*lenp = 0;
2729 		return 0;
2730 	}
2731 
2732 	if (write) {
2733 		char *kbuf, *p;
2734 
2735 		if (left > PAGE_SIZE - 1)
2736 			left = PAGE_SIZE - 1;
2737 
2738 		p = kbuf = memdup_user_nul(buffer, left);
2739 		if (IS_ERR(kbuf))
2740 			return PTR_ERR(kbuf);
2741 
2742 		tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2743 				     GFP_KERNEL);
2744 		if (!tmp_bitmap) {
2745 			kfree(kbuf);
2746 			return -ENOMEM;
2747 		}
2748 		proc_skip_char(&p, &left, '\n');
2749 		while (!err && left) {
2750 			unsigned long val_a, val_b;
2751 			bool neg;
2752 
2753 			err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
2754 					     sizeof(tr_a), &c);
2755 			if (err)
2756 				break;
2757 			if (val_a >= bitmap_len || neg) {
2758 				err = -EINVAL;
2759 				break;
2760 			}
2761 
2762 			val_b = val_a;
2763 			if (left) {
2764 				p++;
2765 				left--;
2766 			}
2767 
2768 			if (c == '-') {
2769 				err = proc_get_long(&p, &left, &val_b,
2770 						     &neg, tr_b, sizeof(tr_b),
2771 						     &c);
2772 				if (err)
2773 					break;
2774 				if (val_b >= bitmap_len || neg ||
2775 				    val_a > val_b) {
2776 					err = -EINVAL;
2777 					break;
2778 				}
2779 				if (left) {
2780 					p++;
2781 					left--;
2782 				}
2783 			}
2784 
2785 			bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2786 			first = 0;
2787 			proc_skip_char(&p, &left, '\n');
2788 		}
2789 		kfree(kbuf);
2790 	} else {
2791 		unsigned long bit_a, bit_b = 0;
2792 
2793 		while (left) {
2794 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2795 			if (bit_a >= bitmap_len)
2796 				break;
2797 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
2798 						   bit_a + 1) - 1;
2799 
2800 			if (!first) {
2801 				err = proc_put_char(&buffer, &left, ',');
2802 				if (err)
2803 					break;
2804 			}
2805 			err = proc_put_long(&buffer, &left, bit_a, false);
2806 			if (err)
2807 				break;
2808 			if (bit_a != bit_b) {
2809 				err = proc_put_char(&buffer, &left, '-');
2810 				if (err)
2811 					break;
2812 				err = proc_put_long(&buffer, &left, bit_b, false);
2813 				if (err)
2814 					break;
2815 			}
2816 
2817 			first = 0; bit_b++;
2818 		}
2819 		if (!err)
2820 			err = proc_put_char(&buffer, &left, '\n');
2821 	}
2822 
2823 	if (!err) {
2824 		if (write) {
2825 			if (*ppos)
2826 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2827 			else
2828 				bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2829 		}
2830 		kfree(tmp_bitmap);
2831 		*lenp -= left;
2832 		*ppos += *lenp;
2833 		return 0;
2834 	} else {
2835 		kfree(tmp_bitmap);
2836 		return err;
2837 	}
2838 }
2839 
2840 #else /* CONFIG_PROC_SYSCTL */
2841 
2842 int proc_dostring(struct ctl_table *table, int write,
2843 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2844 {
2845 	return -ENOSYS;
2846 }
2847 
2848 int proc_dointvec(struct ctl_table *table, int write,
2849 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2850 {
2851 	return -ENOSYS;
2852 }
2853 
2854 int proc_dointvec_minmax(struct ctl_table *table, int write,
2855 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2856 {
2857 	return -ENOSYS;
2858 }
2859 
2860 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2861 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2862 {
2863 	return -ENOSYS;
2864 }
2865 
2866 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2867 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2868 {
2869 	return -ENOSYS;
2870 }
2871 
2872 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2873 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2874 {
2875 	return -ENOSYS;
2876 }
2877 
2878 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2879 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2880 {
2881 	return -ENOSYS;
2882 }
2883 
2884 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2885 				      void __user *buffer,
2886 				      size_t *lenp, loff_t *ppos)
2887 {
2888     return -ENOSYS;
2889 }
2890 
2891 
2892 #endif /* CONFIG_PROC_SYSCTL */
2893 
2894 /*
2895  * No sense putting this after each symbol definition, twice,
2896  * exception granted :-)
2897  */
2898 EXPORT_SYMBOL(proc_dointvec);
2899 EXPORT_SYMBOL(proc_dointvec_jiffies);
2900 EXPORT_SYMBOL(proc_dointvec_minmax);
2901 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2902 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2903 EXPORT_SYMBOL(proc_dostring);
2904 EXPORT_SYMBOL(proc_doulongvec_minmax);
2905 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2906