xref: /linux-6.15/kernel/sysctl.c (revision 258e4bfc)
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 	{ }
1209 };
1210 
1211 static struct ctl_table vm_table[] = {
1212 	{
1213 		.procname	= "overcommit_memory",
1214 		.data		= &sysctl_overcommit_memory,
1215 		.maxlen		= sizeof(sysctl_overcommit_memory),
1216 		.mode		= 0644,
1217 		.proc_handler	= proc_dointvec_minmax,
1218 		.extra1		= &zero,
1219 		.extra2		= &two,
1220 	},
1221 	{
1222 		.procname	= "panic_on_oom",
1223 		.data		= &sysctl_panic_on_oom,
1224 		.maxlen		= sizeof(sysctl_panic_on_oom),
1225 		.mode		= 0644,
1226 		.proc_handler	= proc_dointvec_minmax,
1227 		.extra1		= &zero,
1228 		.extra2		= &two,
1229 	},
1230 	{
1231 		.procname	= "oom_kill_allocating_task",
1232 		.data		= &sysctl_oom_kill_allocating_task,
1233 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1234 		.mode		= 0644,
1235 		.proc_handler	= proc_dointvec,
1236 	},
1237 	{
1238 		.procname	= "oom_dump_tasks",
1239 		.data		= &sysctl_oom_dump_tasks,
1240 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1241 		.mode		= 0644,
1242 		.proc_handler	= proc_dointvec,
1243 	},
1244 	{
1245 		.procname	= "overcommit_ratio",
1246 		.data		= &sysctl_overcommit_ratio,
1247 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1248 		.mode		= 0644,
1249 		.proc_handler	= overcommit_ratio_handler,
1250 	},
1251 	{
1252 		.procname	= "overcommit_kbytes",
1253 		.data		= &sysctl_overcommit_kbytes,
1254 		.maxlen		= sizeof(sysctl_overcommit_kbytes),
1255 		.mode		= 0644,
1256 		.proc_handler	= overcommit_kbytes_handler,
1257 	},
1258 	{
1259 		.procname	= "page-cluster",
1260 		.data		= &page_cluster,
1261 		.maxlen		= sizeof(int),
1262 		.mode		= 0644,
1263 		.proc_handler	= proc_dointvec_minmax,
1264 		.extra1		= &zero,
1265 	},
1266 	{
1267 		.procname	= "dirty_background_ratio",
1268 		.data		= &dirty_background_ratio,
1269 		.maxlen		= sizeof(dirty_background_ratio),
1270 		.mode		= 0644,
1271 		.proc_handler	= dirty_background_ratio_handler,
1272 		.extra1		= &zero,
1273 		.extra2		= &one_hundred,
1274 	},
1275 	{
1276 		.procname	= "dirty_background_bytes",
1277 		.data		= &dirty_background_bytes,
1278 		.maxlen		= sizeof(dirty_background_bytes),
1279 		.mode		= 0644,
1280 		.proc_handler	= dirty_background_bytes_handler,
1281 		.extra1		= &one_ul,
1282 	},
1283 	{
1284 		.procname	= "dirty_ratio",
1285 		.data		= &vm_dirty_ratio,
1286 		.maxlen		= sizeof(vm_dirty_ratio),
1287 		.mode		= 0644,
1288 		.proc_handler	= dirty_ratio_handler,
1289 		.extra1		= &zero,
1290 		.extra2		= &one_hundred,
1291 	},
1292 	{
1293 		.procname	= "dirty_bytes",
1294 		.data		= &vm_dirty_bytes,
1295 		.maxlen		= sizeof(vm_dirty_bytes),
1296 		.mode		= 0644,
1297 		.proc_handler	= dirty_bytes_handler,
1298 		.extra1		= &dirty_bytes_min,
1299 	},
1300 	{
1301 		.procname	= "dirty_writeback_centisecs",
1302 		.data		= &dirty_writeback_interval,
1303 		.maxlen		= sizeof(dirty_writeback_interval),
1304 		.mode		= 0644,
1305 		.proc_handler	= dirty_writeback_centisecs_handler,
1306 	},
1307 	{
1308 		.procname	= "dirty_expire_centisecs",
1309 		.data		= &dirty_expire_interval,
1310 		.maxlen		= sizeof(dirty_expire_interval),
1311 		.mode		= 0644,
1312 		.proc_handler	= proc_dointvec_minmax,
1313 		.extra1		= &zero,
1314 	},
1315 	{
1316 		.procname	= "dirtytime_expire_seconds",
1317 		.data		= &dirtytime_expire_interval,
1318 		.maxlen		= sizeof(dirty_expire_interval),
1319 		.mode		= 0644,
1320 		.proc_handler	= dirtytime_interval_handler,
1321 		.extra1		= &zero,
1322 	},
1323 	{
1324 		.procname       = "nr_pdflush_threads",
1325 		.mode           = 0444 /* read-only */,
1326 		.proc_handler   = pdflush_proc_obsolete,
1327 	},
1328 	{
1329 		.procname	= "swappiness",
1330 		.data		= &vm_swappiness,
1331 		.maxlen		= sizeof(vm_swappiness),
1332 		.mode		= 0644,
1333 		.proc_handler	= proc_dointvec_minmax,
1334 		.extra1		= &zero,
1335 		.extra2		= &one_hundred,
1336 	},
1337 #ifdef CONFIG_HUGETLB_PAGE
1338 	{
1339 		.procname	= "nr_hugepages",
1340 		.data		= NULL,
1341 		.maxlen		= sizeof(unsigned long),
1342 		.mode		= 0644,
1343 		.proc_handler	= hugetlb_sysctl_handler,
1344 	},
1345 #ifdef CONFIG_NUMA
1346 	{
1347 		.procname       = "nr_hugepages_mempolicy",
1348 		.data           = NULL,
1349 		.maxlen         = sizeof(unsigned long),
1350 		.mode           = 0644,
1351 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1352 	},
1353 #endif
1354 	 {
1355 		.procname	= "hugetlb_shm_group",
1356 		.data		= &sysctl_hugetlb_shm_group,
1357 		.maxlen		= sizeof(gid_t),
1358 		.mode		= 0644,
1359 		.proc_handler	= proc_dointvec,
1360 	 },
1361 	 {
1362 		.procname	= "hugepages_treat_as_movable",
1363 		.data		= &hugepages_treat_as_movable,
1364 		.maxlen		= sizeof(int),
1365 		.mode		= 0644,
1366 		.proc_handler	= proc_dointvec,
1367 	},
1368 	{
1369 		.procname	= "nr_overcommit_hugepages",
1370 		.data		= NULL,
1371 		.maxlen		= sizeof(unsigned long),
1372 		.mode		= 0644,
1373 		.proc_handler	= hugetlb_overcommit_handler,
1374 	},
1375 #endif
1376 	{
1377 		.procname	= "lowmem_reserve_ratio",
1378 		.data		= &sysctl_lowmem_reserve_ratio,
1379 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1380 		.mode		= 0644,
1381 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1382 	},
1383 	{
1384 		.procname	= "drop_caches",
1385 		.data		= &sysctl_drop_caches,
1386 		.maxlen		= sizeof(int),
1387 		.mode		= 0644,
1388 		.proc_handler	= drop_caches_sysctl_handler,
1389 		.extra1		= &one,
1390 		.extra2		= &four,
1391 	},
1392 #ifdef CONFIG_COMPACTION
1393 	{
1394 		.procname	= "compact_memory",
1395 		.data		= &sysctl_compact_memory,
1396 		.maxlen		= sizeof(int),
1397 		.mode		= 0200,
1398 		.proc_handler	= sysctl_compaction_handler,
1399 	},
1400 	{
1401 		.procname	= "extfrag_threshold",
1402 		.data		= &sysctl_extfrag_threshold,
1403 		.maxlen		= sizeof(int),
1404 		.mode		= 0644,
1405 		.proc_handler	= sysctl_extfrag_handler,
1406 		.extra1		= &min_extfrag_threshold,
1407 		.extra2		= &max_extfrag_threshold,
1408 	},
1409 	{
1410 		.procname	= "compact_unevictable_allowed",
1411 		.data		= &sysctl_compact_unevictable_allowed,
1412 		.maxlen		= sizeof(int),
1413 		.mode		= 0644,
1414 		.proc_handler	= proc_dointvec,
1415 		.extra1		= &zero,
1416 		.extra2		= &one,
1417 	},
1418 
1419 #endif /* CONFIG_COMPACTION */
1420 	{
1421 		.procname	= "min_free_kbytes",
1422 		.data		= &min_free_kbytes,
1423 		.maxlen		= sizeof(min_free_kbytes),
1424 		.mode		= 0644,
1425 		.proc_handler	= min_free_kbytes_sysctl_handler,
1426 		.extra1		= &zero,
1427 	},
1428 	{
1429 		.procname	= "watermark_scale_factor",
1430 		.data		= &watermark_scale_factor,
1431 		.maxlen		= sizeof(watermark_scale_factor),
1432 		.mode		= 0644,
1433 		.proc_handler	= watermark_scale_factor_sysctl_handler,
1434 		.extra1		= &one,
1435 		.extra2		= &one_thousand,
1436 	},
1437 	{
1438 		.procname	= "percpu_pagelist_fraction",
1439 		.data		= &percpu_pagelist_fraction,
1440 		.maxlen		= sizeof(percpu_pagelist_fraction),
1441 		.mode		= 0644,
1442 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1443 		.extra1		= &zero,
1444 	},
1445 #ifdef CONFIG_MMU
1446 	{
1447 		.procname	= "max_map_count",
1448 		.data		= &sysctl_max_map_count,
1449 		.maxlen		= sizeof(sysctl_max_map_count),
1450 		.mode		= 0644,
1451 		.proc_handler	= proc_dointvec_minmax,
1452 		.extra1		= &zero,
1453 	},
1454 #else
1455 	{
1456 		.procname	= "nr_trim_pages",
1457 		.data		= &sysctl_nr_trim_pages,
1458 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1459 		.mode		= 0644,
1460 		.proc_handler	= proc_dointvec_minmax,
1461 		.extra1		= &zero,
1462 	},
1463 #endif
1464 	{
1465 		.procname	= "laptop_mode",
1466 		.data		= &laptop_mode,
1467 		.maxlen		= sizeof(laptop_mode),
1468 		.mode		= 0644,
1469 		.proc_handler	= proc_dointvec_jiffies,
1470 	},
1471 	{
1472 		.procname	= "block_dump",
1473 		.data		= &block_dump,
1474 		.maxlen		= sizeof(block_dump),
1475 		.mode		= 0644,
1476 		.proc_handler	= proc_dointvec,
1477 		.extra1		= &zero,
1478 	},
1479 	{
1480 		.procname	= "vfs_cache_pressure",
1481 		.data		= &sysctl_vfs_cache_pressure,
1482 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1483 		.mode		= 0644,
1484 		.proc_handler	= proc_dointvec,
1485 		.extra1		= &zero,
1486 	},
1487 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1488 	{
1489 		.procname	= "legacy_va_layout",
1490 		.data		= &sysctl_legacy_va_layout,
1491 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1492 		.mode		= 0644,
1493 		.proc_handler	= proc_dointvec,
1494 		.extra1		= &zero,
1495 	},
1496 #endif
1497 #ifdef CONFIG_NUMA
1498 	{
1499 		.procname	= "zone_reclaim_mode",
1500 		.data		= &zone_reclaim_mode,
1501 		.maxlen		= sizeof(zone_reclaim_mode),
1502 		.mode		= 0644,
1503 		.proc_handler	= proc_dointvec,
1504 		.extra1		= &zero,
1505 	},
1506 	{
1507 		.procname	= "min_unmapped_ratio",
1508 		.data		= &sysctl_min_unmapped_ratio,
1509 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1510 		.mode		= 0644,
1511 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1512 		.extra1		= &zero,
1513 		.extra2		= &one_hundred,
1514 	},
1515 	{
1516 		.procname	= "min_slab_ratio",
1517 		.data		= &sysctl_min_slab_ratio,
1518 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1519 		.mode		= 0644,
1520 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1521 		.extra1		= &zero,
1522 		.extra2		= &one_hundred,
1523 	},
1524 #endif
1525 #ifdef CONFIG_SMP
1526 	{
1527 		.procname	= "stat_interval",
1528 		.data		= &sysctl_stat_interval,
1529 		.maxlen		= sizeof(sysctl_stat_interval),
1530 		.mode		= 0644,
1531 		.proc_handler	= proc_dointvec_jiffies,
1532 	},
1533 #endif
1534 #ifdef CONFIG_MMU
1535 	{
1536 		.procname	= "mmap_min_addr",
1537 		.data		= &dac_mmap_min_addr,
1538 		.maxlen		= sizeof(unsigned long),
1539 		.mode		= 0644,
1540 		.proc_handler	= mmap_min_addr_handler,
1541 	},
1542 #endif
1543 #ifdef CONFIG_NUMA
1544 	{
1545 		.procname	= "numa_zonelist_order",
1546 		.data		= &numa_zonelist_order,
1547 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1548 		.mode		= 0644,
1549 		.proc_handler	= numa_zonelist_order_handler,
1550 	},
1551 #endif
1552 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1553    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1554 	{
1555 		.procname	= "vdso_enabled",
1556 #ifdef CONFIG_X86_32
1557 		.data		= &vdso32_enabled,
1558 		.maxlen		= sizeof(vdso32_enabled),
1559 #else
1560 		.data		= &vdso_enabled,
1561 		.maxlen		= sizeof(vdso_enabled),
1562 #endif
1563 		.mode		= 0644,
1564 		.proc_handler	= proc_dointvec,
1565 		.extra1		= &zero,
1566 	},
1567 #endif
1568 #ifdef CONFIG_HIGHMEM
1569 	{
1570 		.procname	= "highmem_is_dirtyable",
1571 		.data		= &vm_highmem_is_dirtyable,
1572 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1573 		.mode		= 0644,
1574 		.proc_handler	= proc_dointvec_minmax,
1575 		.extra1		= &zero,
1576 		.extra2		= &one,
1577 	},
1578 #endif
1579 #ifdef CONFIG_MEMORY_FAILURE
1580 	{
1581 		.procname	= "memory_failure_early_kill",
1582 		.data		= &sysctl_memory_failure_early_kill,
1583 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1584 		.mode		= 0644,
1585 		.proc_handler	= proc_dointvec_minmax,
1586 		.extra1		= &zero,
1587 		.extra2		= &one,
1588 	},
1589 	{
1590 		.procname	= "memory_failure_recovery",
1591 		.data		= &sysctl_memory_failure_recovery,
1592 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1593 		.mode		= 0644,
1594 		.proc_handler	= proc_dointvec_minmax,
1595 		.extra1		= &zero,
1596 		.extra2		= &one,
1597 	},
1598 #endif
1599 	{
1600 		.procname	= "user_reserve_kbytes",
1601 		.data		= &sysctl_user_reserve_kbytes,
1602 		.maxlen		= sizeof(sysctl_user_reserve_kbytes),
1603 		.mode		= 0644,
1604 		.proc_handler	= proc_doulongvec_minmax,
1605 	},
1606 	{
1607 		.procname	= "admin_reserve_kbytes",
1608 		.data		= &sysctl_admin_reserve_kbytes,
1609 		.maxlen		= sizeof(sysctl_admin_reserve_kbytes),
1610 		.mode		= 0644,
1611 		.proc_handler	= proc_doulongvec_minmax,
1612 	},
1613 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1614 	{
1615 		.procname	= "mmap_rnd_bits",
1616 		.data		= &mmap_rnd_bits,
1617 		.maxlen		= sizeof(mmap_rnd_bits),
1618 		.mode		= 0600,
1619 		.proc_handler	= proc_dointvec_minmax,
1620 		.extra1		= (void *)&mmap_rnd_bits_min,
1621 		.extra2		= (void *)&mmap_rnd_bits_max,
1622 	},
1623 #endif
1624 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1625 	{
1626 		.procname	= "mmap_rnd_compat_bits",
1627 		.data		= &mmap_rnd_compat_bits,
1628 		.maxlen		= sizeof(mmap_rnd_compat_bits),
1629 		.mode		= 0600,
1630 		.proc_handler	= proc_dointvec_minmax,
1631 		.extra1		= (void *)&mmap_rnd_compat_bits_min,
1632 		.extra2		= (void *)&mmap_rnd_compat_bits_max,
1633 	},
1634 #endif
1635 	{ }
1636 };
1637 
1638 static struct ctl_table fs_table[] = {
1639 	{
1640 		.procname	= "inode-nr",
1641 		.data		= &inodes_stat,
1642 		.maxlen		= 2*sizeof(long),
1643 		.mode		= 0444,
1644 		.proc_handler	= proc_nr_inodes,
1645 	},
1646 	{
1647 		.procname	= "inode-state",
1648 		.data		= &inodes_stat,
1649 		.maxlen		= 7*sizeof(long),
1650 		.mode		= 0444,
1651 		.proc_handler	= proc_nr_inodes,
1652 	},
1653 	{
1654 		.procname	= "file-nr",
1655 		.data		= &files_stat,
1656 		.maxlen		= sizeof(files_stat),
1657 		.mode		= 0444,
1658 		.proc_handler	= proc_nr_files,
1659 	},
1660 	{
1661 		.procname	= "file-max",
1662 		.data		= &files_stat.max_files,
1663 		.maxlen		= sizeof(files_stat.max_files),
1664 		.mode		= 0644,
1665 		.proc_handler	= proc_doulongvec_minmax,
1666 	},
1667 	{
1668 		.procname	= "nr_open",
1669 		.data		= &sysctl_nr_open,
1670 		.maxlen		= sizeof(int),
1671 		.mode		= 0644,
1672 		.proc_handler	= proc_dointvec_minmax,
1673 		.extra1		= &sysctl_nr_open_min,
1674 		.extra2		= &sysctl_nr_open_max,
1675 	},
1676 	{
1677 		.procname	= "dentry-state",
1678 		.data		= &dentry_stat,
1679 		.maxlen		= 6*sizeof(long),
1680 		.mode		= 0444,
1681 		.proc_handler	= proc_nr_dentry,
1682 	},
1683 	{
1684 		.procname	= "overflowuid",
1685 		.data		= &fs_overflowuid,
1686 		.maxlen		= sizeof(int),
1687 		.mode		= 0644,
1688 		.proc_handler	= proc_dointvec_minmax,
1689 		.extra1		= &minolduid,
1690 		.extra2		= &maxolduid,
1691 	},
1692 	{
1693 		.procname	= "overflowgid",
1694 		.data		= &fs_overflowgid,
1695 		.maxlen		= sizeof(int),
1696 		.mode		= 0644,
1697 		.proc_handler	= proc_dointvec_minmax,
1698 		.extra1		= &minolduid,
1699 		.extra2		= &maxolduid,
1700 	},
1701 #ifdef CONFIG_FILE_LOCKING
1702 	{
1703 		.procname	= "leases-enable",
1704 		.data		= &leases_enable,
1705 		.maxlen		= sizeof(int),
1706 		.mode		= 0644,
1707 		.proc_handler	= proc_dointvec,
1708 	},
1709 #endif
1710 #ifdef CONFIG_DNOTIFY
1711 	{
1712 		.procname	= "dir-notify-enable",
1713 		.data		= &dir_notify_enable,
1714 		.maxlen		= sizeof(int),
1715 		.mode		= 0644,
1716 		.proc_handler	= proc_dointvec,
1717 	},
1718 #endif
1719 #ifdef CONFIG_MMU
1720 #ifdef CONFIG_FILE_LOCKING
1721 	{
1722 		.procname	= "lease-break-time",
1723 		.data		= &lease_break_time,
1724 		.maxlen		= sizeof(int),
1725 		.mode		= 0644,
1726 		.proc_handler	= proc_dointvec,
1727 	},
1728 #endif
1729 #ifdef CONFIG_AIO
1730 	{
1731 		.procname	= "aio-nr",
1732 		.data		= &aio_nr,
1733 		.maxlen		= sizeof(aio_nr),
1734 		.mode		= 0444,
1735 		.proc_handler	= proc_doulongvec_minmax,
1736 	},
1737 	{
1738 		.procname	= "aio-max-nr",
1739 		.data		= &aio_max_nr,
1740 		.maxlen		= sizeof(aio_max_nr),
1741 		.mode		= 0644,
1742 		.proc_handler	= proc_doulongvec_minmax,
1743 	},
1744 #endif /* CONFIG_AIO */
1745 #ifdef CONFIG_INOTIFY_USER
1746 	{
1747 		.procname	= "inotify",
1748 		.mode		= 0555,
1749 		.child		= inotify_table,
1750 	},
1751 #endif
1752 #ifdef CONFIG_EPOLL
1753 	{
1754 		.procname	= "epoll",
1755 		.mode		= 0555,
1756 		.child		= epoll_table,
1757 	},
1758 #endif
1759 #endif
1760 	{
1761 		.procname	= "protected_symlinks",
1762 		.data		= &sysctl_protected_symlinks,
1763 		.maxlen		= sizeof(int),
1764 		.mode		= 0600,
1765 		.proc_handler	= proc_dointvec_minmax,
1766 		.extra1		= &zero,
1767 		.extra2		= &one,
1768 	},
1769 	{
1770 		.procname	= "protected_hardlinks",
1771 		.data		= &sysctl_protected_hardlinks,
1772 		.maxlen		= sizeof(int),
1773 		.mode		= 0600,
1774 		.proc_handler	= proc_dointvec_minmax,
1775 		.extra1		= &zero,
1776 		.extra2		= &one,
1777 	},
1778 	{
1779 		.procname	= "suid_dumpable",
1780 		.data		= &suid_dumpable,
1781 		.maxlen		= sizeof(int),
1782 		.mode		= 0644,
1783 		.proc_handler	= proc_dointvec_minmax_coredump,
1784 		.extra1		= &zero,
1785 		.extra2		= &two,
1786 	},
1787 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1788 	{
1789 		.procname	= "binfmt_misc",
1790 		.mode		= 0555,
1791 		.child		= sysctl_mount_point,
1792 	},
1793 #endif
1794 	{
1795 		.procname	= "pipe-max-size",
1796 		.data		= &pipe_max_size,
1797 		.maxlen		= sizeof(int),
1798 		.mode		= 0644,
1799 		.proc_handler	= &pipe_proc_fn,
1800 		.extra1		= &pipe_min_size,
1801 	},
1802 	{
1803 		.procname	= "pipe-user-pages-hard",
1804 		.data		= &pipe_user_pages_hard,
1805 		.maxlen		= sizeof(pipe_user_pages_hard),
1806 		.mode		= 0644,
1807 		.proc_handler	= proc_doulongvec_minmax,
1808 	},
1809 	{
1810 		.procname	= "pipe-user-pages-soft",
1811 		.data		= &pipe_user_pages_soft,
1812 		.maxlen		= sizeof(pipe_user_pages_soft),
1813 		.mode		= 0644,
1814 		.proc_handler	= proc_doulongvec_minmax,
1815 	},
1816 	{ }
1817 };
1818 
1819 static struct ctl_table debug_table[] = {
1820 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1821 	{
1822 		.procname	= "exception-trace",
1823 		.data		= &show_unhandled_signals,
1824 		.maxlen		= sizeof(int),
1825 		.mode		= 0644,
1826 		.proc_handler	= proc_dointvec
1827 	},
1828 #endif
1829 #if defined(CONFIG_OPTPROBES)
1830 	{
1831 		.procname	= "kprobes-optimization",
1832 		.data		= &sysctl_kprobes_optimization,
1833 		.maxlen		= sizeof(int),
1834 		.mode		= 0644,
1835 		.proc_handler	= proc_kprobes_optimization_handler,
1836 		.extra1		= &zero,
1837 		.extra2		= &one,
1838 	},
1839 #endif
1840 	{ }
1841 };
1842 
1843 static struct ctl_table dev_table[] = {
1844 	{ }
1845 };
1846 
1847 int __init sysctl_init(void)
1848 {
1849 	struct ctl_table_header *hdr;
1850 
1851 	hdr = register_sysctl_table(sysctl_base_table);
1852 	kmemleak_not_leak(hdr);
1853 	return 0;
1854 }
1855 
1856 #endif /* CONFIG_SYSCTL */
1857 
1858 /*
1859  * /proc/sys support
1860  */
1861 
1862 #ifdef CONFIG_PROC_SYSCTL
1863 
1864 static int _proc_do_string(char *data, int maxlen, int write,
1865 			   char __user *buffer,
1866 			   size_t *lenp, loff_t *ppos)
1867 {
1868 	size_t len;
1869 	char __user *p;
1870 	char c;
1871 
1872 	if (!data || !maxlen || !*lenp) {
1873 		*lenp = 0;
1874 		return 0;
1875 	}
1876 
1877 	if (write) {
1878 		if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1879 			/* Only continue writes not past the end of buffer. */
1880 			len = strlen(data);
1881 			if (len > maxlen - 1)
1882 				len = maxlen - 1;
1883 
1884 			if (*ppos > len)
1885 				return 0;
1886 			len = *ppos;
1887 		} else {
1888 			/* Start writing from beginning of buffer. */
1889 			len = 0;
1890 		}
1891 
1892 		*ppos += *lenp;
1893 		p = buffer;
1894 		while ((p - buffer) < *lenp && len < maxlen - 1) {
1895 			if (get_user(c, p++))
1896 				return -EFAULT;
1897 			if (c == 0 || c == '\n')
1898 				break;
1899 			data[len++] = c;
1900 		}
1901 		data[len] = 0;
1902 	} else {
1903 		len = strlen(data);
1904 		if (len > maxlen)
1905 			len = maxlen;
1906 
1907 		if (*ppos > len) {
1908 			*lenp = 0;
1909 			return 0;
1910 		}
1911 
1912 		data += *ppos;
1913 		len  -= *ppos;
1914 
1915 		if (len > *lenp)
1916 			len = *lenp;
1917 		if (len)
1918 			if (copy_to_user(buffer, data, len))
1919 				return -EFAULT;
1920 		if (len < *lenp) {
1921 			if (put_user('\n', buffer + len))
1922 				return -EFAULT;
1923 			len++;
1924 		}
1925 		*lenp = len;
1926 		*ppos += len;
1927 	}
1928 	return 0;
1929 }
1930 
1931 static void warn_sysctl_write(struct ctl_table *table)
1932 {
1933 	pr_warn_once("%s wrote to %s when file position was not 0!\n"
1934 		"This will not be supported in the future. To silence this\n"
1935 		"warning, set kernel.sysctl_writes_strict = -1\n",
1936 		current->comm, table->procname);
1937 }
1938 
1939 /**
1940  * proc_dostring - read a string sysctl
1941  * @table: the sysctl table
1942  * @write: %TRUE if this is a write to the sysctl file
1943  * @buffer: the user buffer
1944  * @lenp: the size of the user buffer
1945  * @ppos: file position
1946  *
1947  * Reads/writes a string from/to the user buffer. If the kernel
1948  * buffer provided is not large enough to hold the string, the
1949  * string is truncated. The copied string is %NULL-terminated.
1950  * If the string is being read by the user process, it is copied
1951  * and a newline '\n' is added. It is truncated if the buffer is
1952  * not large enough.
1953  *
1954  * Returns 0 on success.
1955  */
1956 int proc_dostring(struct ctl_table *table, int write,
1957 		  void __user *buffer, size_t *lenp, loff_t *ppos)
1958 {
1959 	if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN)
1960 		warn_sysctl_write(table);
1961 
1962 	return _proc_do_string((char *)(table->data), table->maxlen, write,
1963 			       (char __user *)buffer, lenp, ppos);
1964 }
1965 
1966 static size_t proc_skip_spaces(char **buf)
1967 {
1968 	size_t ret;
1969 	char *tmp = skip_spaces(*buf);
1970 	ret = tmp - *buf;
1971 	*buf = tmp;
1972 	return ret;
1973 }
1974 
1975 static void proc_skip_char(char **buf, size_t *size, const char v)
1976 {
1977 	while (*size) {
1978 		if (**buf != v)
1979 			break;
1980 		(*size)--;
1981 		(*buf)++;
1982 	}
1983 }
1984 
1985 #define TMPBUFLEN 22
1986 /**
1987  * proc_get_long - reads an ASCII formatted integer from a user buffer
1988  *
1989  * @buf: a kernel buffer
1990  * @size: size of the kernel buffer
1991  * @val: this is where the number will be stored
1992  * @neg: set to %TRUE if number is negative
1993  * @perm_tr: a vector which contains the allowed trailers
1994  * @perm_tr_len: size of the perm_tr vector
1995  * @tr: pointer to store the trailer character
1996  *
1997  * In case of success %0 is returned and @buf and @size are updated with
1998  * the amount of bytes read. If @tr is non-NULL and a trailing
1999  * character exists (size is non-zero after returning from this
2000  * function), @tr is updated with the trailing character.
2001  */
2002 static int proc_get_long(char **buf, size_t *size,
2003 			  unsigned long *val, bool *neg,
2004 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
2005 {
2006 	int len;
2007 	char *p, tmp[TMPBUFLEN];
2008 
2009 	if (!*size)
2010 		return -EINVAL;
2011 
2012 	len = *size;
2013 	if (len > TMPBUFLEN - 1)
2014 		len = TMPBUFLEN - 1;
2015 
2016 	memcpy(tmp, *buf, len);
2017 
2018 	tmp[len] = 0;
2019 	p = tmp;
2020 	if (*p == '-' && *size > 1) {
2021 		*neg = true;
2022 		p++;
2023 	} else
2024 		*neg = false;
2025 	if (!isdigit(*p))
2026 		return -EINVAL;
2027 
2028 	*val = simple_strtoul(p, &p, 0);
2029 
2030 	len = p - tmp;
2031 
2032 	/* We don't know if the next char is whitespace thus we may accept
2033 	 * invalid integers (e.g. 1234...a) or two integers instead of one
2034 	 * (e.g. 123...1). So lets not allow such large numbers. */
2035 	if (len == TMPBUFLEN - 1)
2036 		return -EINVAL;
2037 
2038 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2039 		return -EINVAL;
2040 
2041 	if (tr && (len < *size))
2042 		*tr = *p;
2043 
2044 	*buf += len;
2045 	*size -= len;
2046 
2047 	return 0;
2048 }
2049 
2050 /**
2051  * proc_put_long - converts an integer to a decimal ASCII formatted string
2052  *
2053  * @buf: the user buffer
2054  * @size: the size of the user buffer
2055  * @val: the integer to be converted
2056  * @neg: sign of the number, %TRUE for negative
2057  *
2058  * In case of success %0 is returned and @buf and @size are updated with
2059  * the amount of bytes written.
2060  */
2061 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2062 			  bool neg)
2063 {
2064 	int len;
2065 	char tmp[TMPBUFLEN], *p = tmp;
2066 
2067 	sprintf(p, "%s%lu", neg ? "-" : "", val);
2068 	len = strlen(tmp);
2069 	if (len > *size)
2070 		len = *size;
2071 	if (copy_to_user(*buf, tmp, len))
2072 		return -EFAULT;
2073 	*size -= len;
2074 	*buf += len;
2075 	return 0;
2076 }
2077 #undef TMPBUFLEN
2078 
2079 static int proc_put_char(void __user **buf, size_t *size, char c)
2080 {
2081 	if (*size) {
2082 		char __user **buffer = (char __user **)buf;
2083 		if (put_user(c, *buffer))
2084 			return -EFAULT;
2085 		(*size)--, (*buffer)++;
2086 		*buf = *buffer;
2087 	}
2088 	return 0;
2089 }
2090 
2091 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2092 				 int *valp,
2093 				 int write, void *data)
2094 {
2095 	if (write) {
2096 		if (*negp) {
2097 			if (*lvalp > (unsigned long) INT_MAX + 1)
2098 				return -EINVAL;
2099 			*valp = -*lvalp;
2100 		} else {
2101 			if (*lvalp > (unsigned long) INT_MAX)
2102 				return -EINVAL;
2103 			*valp = *lvalp;
2104 		}
2105 	} else {
2106 		int val = *valp;
2107 		if (val < 0) {
2108 			*negp = true;
2109 			*lvalp = -(unsigned long)val;
2110 		} else {
2111 			*negp = false;
2112 			*lvalp = (unsigned long)val;
2113 		}
2114 	}
2115 	return 0;
2116 }
2117 
2118 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2119 
2120 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2121 		  int write, void __user *buffer,
2122 		  size_t *lenp, loff_t *ppos,
2123 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2124 			      int write, void *data),
2125 		  void *data)
2126 {
2127 	int *i, vleft, first = 1, err = 0;
2128 	size_t left;
2129 	char *kbuf = NULL, *p;
2130 
2131 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2132 		*lenp = 0;
2133 		return 0;
2134 	}
2135 
2136 	i = (int *) tbl_data;
2137 	vleft = table->maxlen / sizeof(*i);
2138 	left = *lenp;
2139 
2140 	if (!conv)
2141 		conv = do_proc_dointvec_conv;
2142 
2143 	if (write) {
2144 		if (*ppos) {
2145 			switch (sysctl_writes_strict) {
2146 			case SYSCTL_WRITES_STRICT:
2147 				goto out;
2148 			case SYSCTL_WRITES_WARN:
2149 				warn_sysctl_write(table);
2150 				break;
2151 			default:
2152 				break;
2153 			}
2154 		}
2155 
2156 		if (left > PAGE_SIZE - 1)
2157 			left = PAGE_SIZE - 1;
2158 		p = kbuf = memdup_user_nul(buffer, left);
2159 		if (IS_ERR(kbuf))
2160 			return PTR_ERR(kbuf);
2161 	}
2162 
2163 	for (; left && vleft--; i++, first=0) {
2164 		unsigned long lval;
2165 		bool neg;
2166 
2167 		if (write) {
2168 			left -= proc_skip_spaces(&p);
2169 
2170 			if (!left)
2171 				break;
2172 			err = proc_get_long(&p, &left, &lval, &neg,
2173 					     proc_wspace_sep,
2174 					     sizeof(proc_wspace_sep), NULL);
2175 			if (err)
2176 				break;
2177 			if (conv(&neg, &lval, i, 1, data)) {
2178 				err = -EINVAL;
2179 				break;
2180 			}
2181 		} else {
2182 			if (conv(&neg, &lval, i, 0, data)) {
2183 				err = -EINVAL;
2184 				break;
2185 			}
2186 			if (!first)
2187 				err = proc_put_char(&buffer, &left, '\t');
2188 			if (err)
2189 				break;
2190 			err = proc_put_long(&buffer, &left, lval, neg);
2191 			if (err)
2192 				break;
2193 		}
2194 	}
2195 
2196 	if (!write && !first && left && !err)
2197 		err = proc_put_char(&buffer, &left, '\n');
2198 	if (write && !err && left)
2199 		left -= proc_skip_spaces(&p);
2200 	if (write) {
2201 		kfree(kbuf);
2202 		if (first)
2203 			return err ? : -EINVAL;
2204 	}
2205 	*lenp -= left;
2206 out:
2207 	*ppos += *lenp;
2208 	return err;
2209 }
2210 
2211 static int do_proc_dointvec(struct ctl_table *table, int write,
2212 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2213 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2214 			      int write, void *data),
2215 		  void *data)
2216 {
2217 	return __do_proc_dointvec(table->data, table, write,
2218 			buffer, lenp, ppos, conv, data);
2219 }
2220 
2221 /**
2222  * proc_dointvec - read a vector of integers
2223  * @table: the sysctl table
2224  * @write: %TRUE if this is a write to the sysctl file
2225  * @buffer: the user buffer
2226  * @lenp: the size of the user buffer
2227  * @ppos: file position
2228  *
2229  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2230  * values from/to the user buffer, treated as an ASCII string.
2231  *
2232  * Returns 0 on success.
2233  */
2234 int proc_dointvec(struct ctl_table *table, int write,
2235 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2236 {
2237     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2238 		    	    NULL,NULL);
2239 }
2240 
2241 /*
2242  * Taint values can only be increased
2243  * This means we can safely use a temporary.
2244  */
2245 static int proc_taint(struct ctl_table *table, int write,
2246 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2247 {
2248 	struct ctl_table t;
2249 	unsigned long tmptaint = get_taint();
2250 	int err;
2251 
2252 	if (write && !capable(CAP_SYS_ADMIN))
2253 		return -EPERM;
2254 
2255 	t = *table;
2256 	t.data = &tmptaint;
2257 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2258 	if (err < 0)
2259 		return err;
2260 
2261 	if (write) {
2262 		/*
2263 		 * Poor man's atomic or. Not worth adding a primitive
2264 		 * to everyone's atomic.h for this
2265 		 */
2266 		int i;
2267 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2268 			if ((tmptaint >> i) & 1)
2269 				add_taint(i, LOCKDEP_STILL_OK);
2270 		}
2271 	}
2272 
2273 	return err;
2274 }
2275 
2276 #ifdef CONFIG_PRINTK
2277 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2278 				void __user *buffer, size_t *lenp, loff_t *ppos)
2279 {
2280 	if (write && !capable(CAP_SYS_ADMIN))
2281 		return -EPERM;
2282 
2283 	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2284 }
2285 #endif
2286 
2287 struct do_proc_dointvec_minmax_conv_param {
2288 	int *min;
2289 	int *max;
2290 };
2291 
2292 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2293 					int *valp,
2294 					int write, void *data)
2295 {
2296 	struct do_proc_dointvec_minmax_conv_param *param = data;
2297 	if (write) {
2298 		int val = *negp ? -*lvalp : *lvalp;
2299 		if ((param->min && *param->min > val) ||
2300 		    (param->max && *param->max < val))
2301 			return -EINVAL;
2302 		*valp = val;
2303 	} else {
2304 		int val = *valp;
2305 		if (val < 0) {
2306 			*negp = true;
2307 			*lvalp = -(unsigned long)val;
2308 		} else {
2309 			*negp = false;
2310 			*lvalp = (unsigned long)val;
2311 		}
2312 	}
2313 	return 0;
2314 }
2315 
2316 /**
2317  * proc_dointvec_minmax - read a vector of integers with min/max values
2318  * @table: the sysctl table
2319  * @write: %TRUE if this is a write to the sysctl file
2320  * @buffer: the user buffer
2321  * @lenp: the size of the user buffer
2322  * @ppos: file position
2323  *
2324  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2325  * values from/to the user buffer, treated as an ASCII string.
2326  *
2327  * This routine will ensure the values are within the range specified by
2328  * table->extra1 (min) and table->extra2 (max).
2329  *
2330  * Returns 0 on success.
2331  */
2332 int proc_dointvec_minmax(struct ctl_table *table, int write,
2333 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2334 {
2335 	struct do_proc_dointvec_minmax_conv_param param = {
2336 		.min = (int *) table->extra1,
2337 		.max = (int *) table->extra2,
2338 	};
2339 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2340 				do_proc_dointvec_minmax_conv, &param);
2341 }
2342 
2343 static void validate_coredump_safety(void)
2344 {
2345 #ifdef CONFIG_COREDUMP
2346 	if (suid_dumpable == SUID_DUMP_ROOT &&
2347 	    core_pattern[0] != '/' && core_pattern[0] != '|') {
2348 		printk(KERN_WARNING "Unsafe core_pattern used with "\
2349 			"suid_dumpable=2. Pipe handler or fully qualified "\
2350 			"core dump path required.\n");
2351 	}
2352 #endif
2353 }
2354 
2355 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2356 		void __user *buffer, size_t *lenp, loff_t *ppos)
2357 {
2358 	int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2359 	if (!error)
2360 		validate_coredump_safety();
2361 	return error;
2362 }
2363 
2364 #ifdef CONFIG_COREDUMP
2365 static int proc_dostring_coredump(struct ctl_table *table, int write,
2366 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2367 {
2368 	int error = proc_dostring(table, write, buffer, lenp, ppos);
2369 	if (!error)
2370 		validate_coredump_safety();
2371 	return error;
2372 }
2373 #endif
2374 
2375 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2376 				     void __user *buffer,
2377 				     size_t *lenp, loff_t *ppos,
2378 				     unsigned long convmul,
2379 				     unsigned long convdiv)
2380 {
2381 	unsigned long *i, *min, *max;
2382 	int vleft, first = 1, err = 0;
2383 	size_t left;
2384 	char *kbuf = NULL, *p;
2385 
2386 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2387 		*lenp = 0;
2388 		return 0;
2389 	}
2390 
2391 	i = (unsigned long *) data;
2392 	min = (unsigned long *) table->extra1;
2393 	max = (unsigned long *) table->extra2;
2394 	vleft = table->maxlen / sizeof(unsigned long);
2395 	left = *lenp;
2396 
2397 	if (write) {
2398 		if (*ppos) {
2399 			switch (sysctl_writes_strict) {
2400 			case SYSCTL_WRITES_STRICT:
2401 				goto out;
2402 			case SYSCTL_WRITES_WARN:
2403 				warn_sysctl_write(table);
2404 				break;
2405 			default:
2406 				break;
2407 			}
2408 		}
2409 
2410 		if (left > PAGE_SIZE - 1)
2411 			left = PAGE_SIZE - 1;
2412 		p = kbuf = memdup_user_nul(buffer, left);
2413 		if (IS_ERR(kbuf))
2414 			return PTR_ERR(kbuf);
2415 	}
2416 
2417 	for (; left && vleft--; i++, first = 0) {
2418 		unsigned long val;
2419 
2420 		if (write) {
2421 			bool neg;
2422 
2423 			left -= proc_skip_spaces(&p);
2424 
2425 			err = proc_get_long(&p, &left, &val, &neg,
2426 					     proc_wspace_sep,
2427 					     sizeof(proc_wspace_sep), NULL);
2428 			if (err)
2429 				break;
2430 			if (neg)
2431 				continue;
2432 			if ((min && val < *min) || (max && val > *max))
2433 				continue;
2434 			*i = val;
2435 		} else {
2436 			val = convdiv * (*i) / convmul;
2437 			if (!first) {
2438 				err = proc_put_char(&buffer, &left, '\t');
2439 				if (err)
2440 					break;
2441 			}
2442 			err = proc_put_long(&buffer, &left, val, false);
2443 			if (err)
2444 				break;
2445 		}
2446 	}
2447 
2448 	if (!write && !first && left && !err)
2449 		err = proc_put_char(&buffer, &left, '\n');
2450 	if (write && !err)
2451 		left -= proc_skip_spaces(&p);
2452 	if (write) {
2453 		kfree(kbuf);
2454 		if (first)
2455 			return err ? : -EINVAL;
2456 	}
2457 	*lenp -= left;
2458 out:
2459 	*ppos += *lenp;
2460 	return err;
2461 }
2462 
2463 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2464 				     void __user *buffer,
2465 				     size_t *lenp, loff_t *ppos,
2466 				     unsigned long convmul,
2467 				     unsigned long convdiv)
2468 {
2469 	return __do_proc_doulongvec_minmax(table->data, table, write,
2470 			buffer, lenp, ppos, convmul, convdiv);
2471 }
2472 
2473 /**
2474  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2475  * @table: the sysctl table
2476  * @write: %TRUE if this is a write to the sysctl file
2477  * @buffer: the user buffer
2478  * @lenp: the size of the user buffer
2479  * @ppos: file position
2480  *
2481  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2482  * values from/to the user buffer, treated as an ASCII string.
2483  *
2484  * This routine will ensure the values are within the range specified by
2485  * table->extra1 (min) and table->extra2 (max).
2486  *
2487  * Returns 0 on success.
2488  */
2489 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2490 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2491 {
2492     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2493 }
2494 
2495 /**
2496  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2497  * @table: the sysctl table
2498  * @write: %TRUE if this is a write to the sysctl file
2499  * @buffer: the user buffer
2500  * @lenp: the size of the user buffer
2501  * @ppos: file position
2502  *
2503  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2504  * values from/to the user buffer, treated as an ASCII string. The values
2505  * are treated as milliseconds, and converted to jiffies when they are stored.
2506  *
2507  * This routine will ensure the values are within the range specified by
2508  * table->extra1 (min) and table->extra2 (max).
2509  *
2510  * Returns 0 on success.
2511  */
2512 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2513 				      void __user *buffer,
2514 				      size_t *lenp, loff_t *ppos)
2515 {
2516     return do_proc_doulongvec_minmax(table, write, buffer,
2517 				     lenp, ppos, HZ, 1000l);
2518 }
2519 
2520 
2521 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2522 					 int *valp,
2523 					 int write, void *data)
2524 {
2525 	if (write) {
2526 		if (*lvalp > LONG_MAX / HZ)
2527 			return 1;
2528 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2529 	} else {
2530 		int val = *valp;
2531 		unsigned long lval;
2532 		if (val < 0) {
2533 			*negp = true;
2534 			lval = -(unsigned long)val;
2535 		} else {
2536 			*negp = false;
2537 			lval = (unsigned long)val;
2538 		}
2539 		*lvalp = lval / HZ;
2540 	}
2541 	return 0;
2542 }
2543 
2544 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2545 						int *valp,
2546 						int write, void *data)
2547 {
2548 	if (write) {
2549 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2550 			return 1;
2551 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2552 	} else {
2553 		int val = *valp;
2554 		unsigned long lval;
2555 		if (val < 0) {
2556 			*negp = true;
2557 			lval = -(unsigned long)val;
2558 		} else {
2559 			*negp = false;
2560 			lval = (unsigned long)val;
2561 		}
2562 		*lvalp = jiffies_to_clock_t(lval);
2563 	}
2564 	return 0;
2565 }
2566 
2567 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2568 					    int *valp,
2569 					    int write, void *data)
2570 {
2571 	if (write) {
2572 		unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2573 
2574 		if (jif > INT_MAX)
2575 			return 1;
2576 		*valp = (int)jif;
2577 	} else {
2578 		int val = *valp;
2579 		unsigned long lval;
2580 		if (val < 0) {
2581 			*negp = true;
2582 			lval = -(unsigned long)val;
2583 		} else {
2584 			*negp = false;
2585 			lval = (unsigned long)val;
2586 		}
2587 		*lvalp = jiffies_to_msecs(lval);
2588 	}
2589 	return 0;
2590 }
2591 
2592 /**
2593  * proc_dointvec_jiffies - read a vector of integers as seconds
2594  * @table: the sysctl table
2595  * @write: %TRUE if this is a write to the sysctl file
2596  * @buffer: the user buffer
2597  * @lenp: the size of the user buffer
2598  * @ppos: file position
2599  *
2600  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2601  * values from/to the user buffer, treated as an ASCII string.
2602  * The values read are assumed to be in seconds, and are converted into
2603  * jiffies.
2604  *
2605  * Returns 0 on success.
2606  */
2607 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2608 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2609 {
2610     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2611 		    	    do_proc_dointvec_jiffies_conv,NULL);
2612 }
2613 
2614 /**
2615  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2616  * @table: the sysctl table
2617  * @write: %TRUE if this is a write to the sysctl file
2618  * @buffer: the user buffer
2619  * @lenp: the size of the user buffer
2620  * @ppos: pointer to the file position
2621  *
2622  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2623  * values from/to the user buffer, treated as an ASCII string.
2624  * The values read are assumed to be in 1/USER_HZ seconds, and
2625  * are converted into jiffies.
2626  *
2627  * Returns 0 on success.
2628  */
2629 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2630 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2631 {
2632     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2633 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2634 }
2635 
2636 /**
2637  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2638  * @table: the sysctl table
2639  * @write: %TRUE if this is a write to the sysctl file
2640  * @buffer: the user buffer
2641  * @lenp: the size of the user buffer
2642  * @ppos: file position
2643  * @ppos: the current position in the file
2644  *
2645  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2646  * values from/to the user buffer, treated as an ASCII string.
2647  * The values read are assumed to be in 1/1000 seconds, and
2648  * are converted into jiffies.
2649  *
2650  * Returns 0 on success.
2651  */
2652 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2653 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2654 {
2655 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2656 				do_proc_dointvec_ms_jiffies_conv, NULL);
2657 }
2658 
2659 static int proc_do_cad_pid(struct ctl_table *table, int write,
2660 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2661 {
2662 	struct pid *new_pid;
2663 	pid_t tmp;
2664 	int r;
2665 
2666 	tmp = pid_vnr(cad_pid);
2667 
2668 	r = __do_proc_dointvec(&tmp, table, write, buffer,
2669 			       lenp, ppos, NULL, NULL);
2670 	if (r || !write)
2671 		return r;
2672 
2673 	new_pid = find_get_pid(tmp);
2674 	if (!new_pid)
2675 		return -ESRCH;
2676 
2677 	put_pid(xchg(&cad_pid, new_pid));
2678 	return 0;
2679 }
2680 
2681 /**
2682  * proc_do_large_bitmap - read/write from/to a large bitmap
2683  * @table: the sysctl table
2684  * @write: %TRUE if this is a write to the sysctl file
2685  * @buffer: the user buffer
2686  * @lenp: the size of the user buffer
2687  * @ppos: file position
2688  *
2689  * The bitmap is stored at table->data and the bitmap length (in bits)
2690  * in table->maxlen.
2691  *
2692  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2693  * large bitmaps may be represented in a compact manner. Writing into
2694  * the file will clear the bitmap then update it with the given input.
2695  *
2696  * Returns 0 on success.
2697  */
2698 int proc_do_large_bitmap(struct ctl_table *table, int write,
2699 			 void __user *buffer, size_t *lenp, loff_t *ppos)
2700 {
2701 	int err = 0;
2702 	bool first = 1;
2703 	size_t left = *lenp;
2704 	unsigned long bitmap_len = table->maxlen;
2705 	unsigned long *bitmap = *(unsigned long **) table->data;
2706 	unsigned long *tmp_bitmap = NULL;
2707 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2708 
2709 	if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2710 		*lenp = 0;
2711 		return 0;
2712 	}
2713 
2714 	if (write) {
2715 		char *kbuf, *p;
2716 
2717 		if (left > PAGE_SIZE - 1)
2718 			left = PAGE_SIZE - 1;
2719 
2720 		p = kbuf = memdup_user_nul(buffer, left);
2721 		if (IS_ERR(kbuf))
2722 			return PTR_ERR(kbuf);
2723 
2724 		tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2725 				     GFP_KERNEL);
2726 		if (!tmp_bitmap) {
2727 			kfree(kbuf);
2728 			return -ENOMEM;
2729 		}
2730 		proc_skip_char(&p, &left, '\n');
2731 		while (!err && left) {
2732 			unsigned long val_a, val_b;
2733 			bool neg;
2734 
2735 			err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
2736 					     sizeof(tr_a), &c);
2737 			if (err)
2738 				break;
2739 			if (val_a >= bitmap_len || neg) {
2740 				err = -EINVAL;
2741 				break;
2742 			}
2743 
2744 			val_b = val_a;
2745 			if (left) {
2746 				p++;
2747 				left--;
2748 			}
2749 
2750 			if (c == '-') {
2751 				err = proc_get_long(&p, &left, &val_b,
2752 						     &neg, tr_b, sizeof(tr_b),
2753 						     &c);
2754 				if (err)
2755 					break;
2756 				if (val_b >= bitmap_len || neg ||
2757 				    val_a > val_b) {
2758 					err = -EINVAL;
2759 					break;
2760 				}
2761 				if (left) {
2762 					p++;
2763 					left--;
2764 				}
2765 			}
2766 
2767 			bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2768 			first = 0;
2769 			proc_skip_char(&p, &left, '\n');
2770 		}
2771 		kfree(kbuf);
2772 	} else {
2773 		unsigned long bit_a, bit_b = 0;
2774 
2775 		while (left) {
2776 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2777 			if (bit_a >= bitmap_len)
2778 				break;
2779 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
2780 						   bit_a + 1) - 1;
2781 
2782 			if (!first) {
2783 				err = proc_put_char(&buffer, &left, ',');
2784 				if (err)
2785 					break;
2786 			}
2787 			err = proc_put_long(&buffer, &left, bit_a, false);
2788 			if (err)
2789 				break;
2790 			if (bit_a != bit_b) {
2791 				err = proc_put_char(&buffer, &left, '-');
2792 				if (err)
2793 					break;
2794 				err = proc_put_long(&buffer, &left, bit_b, false);
2795 				if (err)
2796 					break;
2797 			}
2798 
2799 			first = 0; bit_b++;
2800 		}
2801 		if (!err)
2802 			err = proc_put_char(&buffer, &left, '\n');
2803 	}
2804 
2805 	if (!err) {
2806 		if (write) {
2807 			if (*ppos)
2808 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2809 			else
2810 				bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2811 		}
2812 		kfree(tmp_bitmap);
2813 		*lenp -= left;
2814 		*ppos += *lenp;
2815 		return 0;
2816 	} else {
2817 		kfree(tmp_bitmap);
2818 		return err;
2819 	}
2820 }
2821 
2822 #else /* CONFIG_PROC_SYSCTL */
2823 
2824 int proc_dostring(struct ctl_table *table, int write,
2825 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2826 {
2827 	return -ENOSYS;
2828 }
2829 
2830 int proc_dointvec(struct ctl_table *table, int write,
2831 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2832 {
2833 	return -ENOSYS;
2834 }
2835 
2836 int proc_dointvec_minmax(struct ctl_table *table, int write,
2837 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2838 {
2839 	return -ENOSYS;
2840 }
2841 
2842 int proc_dointvec_jiffies(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_userhz_jiffies(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_ms_jiffies(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_doulongvec_minmax(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_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2867 				      void __user *buffer,
2868 				      size_t *lenp, loff_t *ppos)
2869 {
2870     return -ENOSYS;
2871 }
2872 
2873 
2874 #endif /* CONFIG_PROC_SYSCTL */
2875 
2876 /*
2877  * No sense putting this after each symbol definition, twice,
2878  * exception granted :-)
2879  */
2880 EXPORT_SYMBOL(proc_dointvec);
2881 EXPORT_SYMBOL(proc_dointvec_jiffies);
2882 EXPORT_SYMBOL(proc_dointvec_minmax);
2883 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2884 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2885 EXPORT_SYMBOL(proc_dostring);
2886 EXPORT_SYMBOL(proc_doulongvec_minmax);
2887 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2888