xref: /openbmc/linux/kernel/sysctl.c (revision d8eabc37)
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/kmemleak.h>
34 #include <linux/fs.h>
35 #include <linux/init.h>
36 #include <linux/kernel.h>
37 #include <linux/kobject.h>
38 #include <linux/net.h>
39 #include <linux/sysrq.h>
40 #include <linux/highuid.h>
41 #include <linux/writeback.h>
42 #include <linux/ratelimit.h>
43 #include <linux/compaction.h>
44 #include <linux/hugetlb.h>
45 #include <linux/initrd.h>
46 #include <linux/key.h>
47 #include <linux/times.h>
48 #include <linux/limits.h>
49 #include <linux/dcache.h>
50 #include <linux/dnotify.h>
51 #include <linux/syscalls.h>
52 #include <linux/vmstat.h>
53 #include <linux/nfs_fs.h>
54 #include <linux/acpi.h>
55 #include <linux/reboot.h>
56 #include <linux/ftrace.h>
57 #include <linux/perf_event.h>
58 #include <linux/kprobes.h>
59 #include <linux/pipe_fs_i.h>
60 #include <linux/oom.h>
61 #include <linux/kmod.h>
62 #include <linux/capability.h>
63 #include <linux/binfmts.h>
64 #include <linux/sched/sysctl.h>
65 #include <linux/sched/coredump.h>
66 #include <linux/kexec.h>
67 #include <linux/bpf.h>
68 #include <linux/mount.h>
69 
70 #include <linux/uaccess.h>
71 #include <asm/processor.h>
72 
73 #ifdef CONFIG_X86
74 #include <asm/nmi.h>
75 #include <asm/stacktrace.h>
76 #include <asm/io.h>
77 #endif
78 #ifdef CONFIG_SPARC
79 #include <asm/setup.h>
80 #endif
81 #ifdef CONFIG_BSD_PROCESS_ACCT
82 #include <linux/acct.h>
83 #endif
84 #ifdef CONFIG_RT_MUTEXES
85 #include <linux/rtmutex.h>
86 #endif
87 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
88 #include <linux/lockdep.h>
89 #endif
90 #ifdef CONFIG_CHR_DEV_SG
91 #include <scsi/sg.h>
92 #endif
93 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
94 #include <linux/stackleak.h>
95 #endif
96 #ifdef CONFIG_LOCKUP_DETECTOR
97 #include <linux/nmi.h>
98 #endif
99 
100 #if defined(CONFIG_SYSCTL)
101 
102 /* External variables not in a header file. */
103 extern int suid_dumpable;
104 #ifdef CONFIG_COREDUMP
105 extern int core_uses_pid;
106 extern char core_pattern[];
107 extern unsigned int core_pipe_limit;
108 #endif
109 extern int pid_max;
110 extern int pid_max_min, pid_max_max;
111 extern int percpu_pagelist_fraction;
112 extern int latencytop_enabled;
113 extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
114 #ifndef CONFIG_MMU
115 extern int sysctl_nr_trim_pages;
116 #endif
117 
118 /* Constants used for minimum and  maximum */
119 #ifdef CONFIG_LOCKUP_DETECTOR
120 static int sixty = 60;
121 #endif
122 
123 static int __maybe_unused neg_one = -1;
124 
125 static int zero;
126 static int __maybe_unused one = 1;
127 static int __maybe_unused two = 2;
128 static int __maybe_unused four = 4;
129 static unsigned long one_ul = 1;
130 static int one_hundred = 100;
131 static int one_thousand = 1000;
132 #ifdef CONFIG_PRINTK
133 static int ten_thousand = 10000;
134 #endif
135 #ifdef CONFIG_PERF_EVENTS
136 static int six_hundred_forty_kb = 640 * 1024;
137 #endif
138 
139 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
140 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
141 
142 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
143 static int maxolduid = 65535;
144 static int minolduid;
145 
146 static int ngroups_max = NGROUPS_MAX;
147 static const int cap_last_cap = CAP_LAST_CAP;
148 
149 /*
150  * This is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs
151  * and hung_task_check_interval_secs
152  */
153 #ifdef CONFIG_DETECT_HUNG_TASK
154 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
155 #endif
156 
157 #ifdef CONFIG_INOTIFY_USER
158 #include <linux/inotify.h>
159 #endif
160 #ifdef CONFIG_SPARC
161 #endif
162 
163 #ifdef __hppa__
164 extern int pwrsw_enabled;
165 #endif
166 
167 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
168 extern int unaligned_enabled;
169 #endif
170 
171 #ifdef CONFIG_IA64
172 extern int unaligned_dump_stack;
173 #endif
174 
175 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
176 extern int no_unaligned_warning;
177 #endif
178 
179 #ifdef CONFIG_PROC_SYSCTL
180 
181 /**
182  * enum sysctl_writes_mode - supported sysctl write modes
183  *
184  * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
185  * 	to be written, and multiple writes on the same sysctl file descriptor
186  * 	will rewrite the sysctl value, regardless of file position. No warning
187  * 	is issued when the initial position is not 0.
188  * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
189  * 	not 0.
190  * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
191  * 	file position 0 and the value must be fully contained in the buffer
192  * 	sent to the write syscall. If dealing with strings respect the file
193  * 	position, but restrict this to the max length of the buffer, anything
194  * 	passed the max lenght will be ignored. Multiple writes will append
195  * 	to the buffer.
196  *
197  * These write modes control how current file position affects the behavior of
198  * updating sysctl values through the proc interface on each write.
199  */
200 enum sysctl_writes_mode {
201 	SYSCTL_WRITES_LEGACY		= -1,
202 	SYSCTL_WRITES_WARN		= 0,
203 	SYSCTL_WRITES_STRICT		= 1,
204 };
205 
206 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
207 
208 static int proc_do_cad_pid(struct ctl_table *table, int write,
209 		  void __user *buffer, size_t *lenp, loff_t *ppos);
210 static int proc_taint(struct ctl_table *table, int write,
211 			       void __user *buffer, size_t *lenp, loff_t *ppos);
212 #endif
213 
214 #ifdef CONFIG_PRINTK
215 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
216 				void __user *buffer, size_t *lenp, loff_t *ppos);
217 #endif
218 
219 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
220 		void __user *buffer, size_t *lenp, loff_t *ppos);
221 #ifdef CONFIG_COREDUMP
222 static int proc_dostring_coredump(struct ctl_table *table, int write,
223 		void __user *buffer, size_t *lenp, loff_t *ppos);
224 #endif
225 static int proc_dopipe_max_size(struct ctl_table *table, int write,
226 		void __user *buffer, size_t *lenp, loff_t *ppos);
227 
228 #ifdef CONFIG_MAGIC_SYSRQ
229 /* Note: sysrq code uses its own private copy */
230 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
231 
232 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
233 				void __user *buffer, size_t *lenp,
234 				loff_t *ppos)
235 {
236 	int error;
237 
238 	error = proc_dointvec(table, write, buffer, lenp, ppos);
239 	if (error)
240 		return error;
241 
242 	if (write)
243 		sysrq_toggle_support(__sysrq_enabled);
244 
245 	return 0;
246 }
247 
248 #endif
249 
250 static struct ctl_table kern_table[];
251 static struct ctl_table vm_table[];
252 static struct ctl_table fs_table[];
253 static struct ctl_table debug_table[];
254 static struct ctl_table dev_table[];
255 extern struct ctl_table random_table[];
256 #ifdef CONFIG_EPOLL
257 extern struct ctl_table epoll_table[];
258 #endif
259 
260 #ifdef CONFIG_FW_LOADER_USER_HELPER
261 extern struct ctl_table firmware_config_table[];
262 #endif
263 
264 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
265 int sysctl_legacy_va_layout;
266 #endif
267 
268 /* The default sysctl tables: */
269 
270 static struct ctl_table sysctl_base_table[] = {
271 	{
272 		.procname	= "kernel",
273 		.mode		= 0555,
274 		.child		= kern_table,
275 	},
276 	{
277 		.procname	= "vm",
278 		.mode		= 0555,
279 		.child		= vm_table,
280 	},
281 	{
282 		.procname	= "fs",
283 		.mode		= 0555,
284 		.child		= fs_table,
285 	},
286 	{
287 		.procname	= "debug",
288 		.mode		= 0555,
289 		.child		= debug_table,
290 	},
291 	{
292 		.procname	= "dev",
293 		.mode		= 0555,
294 		.child		= dev_table,
295 	},
296 	{ }
297 };
298 
299 #ifdef CONFIG_SCHED_DEBUG
300 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
301 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
302 static int min_wakeup_granularity_ns;			/* 0 usecs */
303 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
304 #ifdef CONFIG_SMP
305 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
306 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
307 #endif /* CONFIG_SMP */
308 #endif /* CONFIG_SCHED_DEBUG */
309 
310 #ifdef CONFIG_COMPACTION
311 static int min_extfrag_threshold;
312 static int max_extfrag_threshold = 1000;
313 #endif
314 
315 static struct ctl_table kern_table[] = {
316 	{
317 		.procname	= "sched_child_runs_first",
318 		.data		= &sysctl_sched_child_runs_first,
319 		.maxlen		= sizeof(unsigned int),
320 		.mode		= 0644,
321 		.proc_handler	= proc_dointvec,
322 	},
323 #ifdef CONFIG_SCHED_DEBUG
324 	{
325 		.procname	= "sched_min_granularity_ns",
326 		.data		= &sysctl_sched_min_granularity,
327 		.maxlen		= sizeof(unsigned int),
328 		.mode		= 0644,
329 		.proc_handler	= sched_proc_update_handler,
330 		.extra1		= &min_sched_granularity_ns,
331 		.extra2		= &max_sched_granularity_ns,
332 	},
333 	{
334 		.procname	= "sched_latency_ns",
335 		.data		= &sysctl_sched_latency,
336 		.maxlen		= sizeof(unsigned int),
337 		.mode		= 0644,
338 		.proc_handler	= sched_proc_update_handler,
339 		.extra1		= &min_sched_granularity_ns,
340 		.extra2		= &max_sched_granularity_ns,
341 	},
342 	{
343 		.procname	= "sched_wakeup_granularity_ns",
344 		.data		= &sysctl_sched_wakeup_granularity,
345 		.maxlen		= sizeof(unsigned int),
346 		.mode		= 0644,
347 		.proc_handler	= sched_proc_update_handler,
348 		.extra1		= &min_wakeup_granularity_ns,
349 		.extra2		= &max_wakeup_granularity_ns,
350 	},
351 #ifdef CONFIG_SMP
352 	{
353 		.procname	= "sched_tunable_scaling",
354 		.data		= &sysctl_sched_tunable_scaling,
355 		.maxlen		= sizeof(enum sched_tunable_scaling),
356 		.mode		= 0644,
357 		.proc_handler	= sched_proc_update_handler,
358 		.extra1		= &min_sched_tunable_scaling,
359 		.extra2		= &max_sched_tunable_scaling,
360 	},
361 	{
362 		.procname	= "sched_migration_cost_ns",
363 		.data		= &sysctl_sched_migration_cost,
364 		.maxlen		= sizeof(unsigned int),
365 		.mode		= 0644,
366 		.proc_handler	= proc_dointvec,
367 	},
368 	{
369 		.procname	= "sched_nr_migrate",
370 		.data		= &sysctl_sched_nr_migrate,
371 		.maxlen		= sizeof(unsigned int),
372 		.mode		= 0644,
373 		.proc_handler	= proc_dointvec,
374 	},
375 #ifdef CONFIG_SCHEDSTATS
376 	{
377 		.procname	= "sched_schedstats",
378 		.data		= NULL,
379 		.maxlen		= sizeof(unsigned int),
380 		.mode		= 0644,
381 		.proc_handler	= sysctl_schedstats,
382 		.extra1		= &zero,
383 		.extra2		= &one,
384 	},
385 #endif /* CONFIG_SCHEDSTATS */
386 #endif /* CONFIG_SMP */
387 #ifdef CONFIG_NUMA_BALANCING
388 	{
389 		.procname	= "numa_balancing_scan_delay_ms",
390 		.data		= &sysctl_numa_balancing_scan_delay,
391 		.maxlen		= sizeof(unsigned int),
392 		.mode		= 0644,
393 		.proc_handler	= proc_dointvec,
394 	},
395 	{
396 		.procname	= "numa_balancing_scan_period_min_ms",
397 		.data		= &sysctl_numa_balancing_scan_period_min,
398 		.maxlen		= sizeof(unsigned int),
399 		.mode		= 0644,
400 		.proc_handler	= proc_dointvec,
401 	},
402 	{
403 		.procname	= "numa_balancing_scan_period_max_ms",
404 		.data		= &sysctl_numa_balancing_scan_period_max,
405 		.maxlen		= sizeof(unsigned int),
406 		.mode		= 0644,
407 		.proc_handler	= proc_dointvec,
408 	},
409 	{
410 		.procname	= "numa_balancing_scan_size_mb",
411 		.data		= &sysctl_numa_balancing_scan_size,
412 		.maxlen		= sizeof(unsigned int),
413 		.mode		= 0644,
414 		.proc_handler	= proc_dointvec_minmax,
415 		.extra1		= &one,
416 	},
417 	{
418 		.procname	= "numa_balancing",
419 		.data		= NULL, /* filled in by handler */
420 		.maxlen		= sizeof(unsigned int),
421 		.mode		= 0644,
422 		.proc_handler	= sysctl_numa_balancing,
423 		.extra1		= &zero,
424 		.extra2		= &one,
425 	},
426 #endif /* CONFIG_NUMA_BALANCING */
427 #endif /* CONFIG_SCHED_DEBUG */
428 	{
429 		.procname	= "sched_rt_period_us",
430 		.data		= &sysctl_sched_rt_period,
431 		.maxlen		= sizeof(unsigned int),
432 		.mode		= 0644,
433 		.proc_handler	= sched_rt_handler,
434 	},
435 	{
436 		.procname	= "sched_rt_runtime_us",
437 		.data		= &sysctl_sched_rt_runtime,
438 		.maxlen		= sizeof(int),
439 		.mode		= 0644,
440 		.proc_handler	= sched_rt_handler,
441 	},
442 	{
443 		.procname	= "sched_rr_timeslice_ms",
444 		.data		= &sysctl_sched_rr_timeslice,
445 		.maxlen		= sizeof(int),
446 		.mode		= 0644,
447 		.proc_handler	= sched_rr_handler,
448 	},
449 #ifdef CONFIG_SCHED_AUTOGROUP
450 	{
451 		.procname	= "sched_autogroup_enabled",
452 		.data		= &sysctl_sched_autogroup_enabled,
453 		.maxlen		= sizeof(unsigned int),
454 		.mode		= 0644,
455 		.proc_handler	= proc_dointvec_minmax,
456 		.extra1		= &zero,
457 		.extra2		= &one,
458 	},
459 #endif
460 #ifdef CONFIG_CFS_BANDWIDTH
461 	{
462 		.procname	= "sched_cfs_bandwidth_slice_us",
463 		.data		= &sysctl_sched_cfs_bandwidth_slice,
464 		.maxlen		= sizeof(unsigned int),
465 		.mode		= 0644,
466 		.proc_handler	= proc_dointvec_minmax,
467 		.extra1		= &one,
468 	},
469 #endif
470 #ifdef CONFIG_PROVE_LOCKING
471 	{
472 		.procname	= "prove_locking",
473 		.data		= &prove_locking,
474 		.maxlen		= sizeof(int),
475 		.mode		= 0644,
476 		.proc_handler	= proc_dointvec,
477 	},
478 #endif
479 #ifdef CONFIG_LOCK_STAT
480 	{
481 		.procname	= "lock_stat",
482 		.data		= &lock_stat,
483 		.maxlen		= sizeof(int),
484 		.mode		= 0644,
485 		.proc_handler	= proc_dointvec,
486 	},
487 #endif
488 	{
489 		.procname	= "panic",
490 		.data		= &panic_timeout,
491 		.maxlen		= sizeof(int),
492 		.mode		= 0644,
493 		.proc_handler	= proc_dointvec,
494 	},
495 #ifdef CONFIG_COREDUMP
496 	{
497 		.procname	= "core_uses_pid",
498 		.data		= &core_uses_pid,
499 		.maxlen		= sizeof(int),
500 		.mode		= 0644,
501 		.proc_handler	= proc_dointvec,
502 	},
503 	{
504 		.procname	= "core_pattern",
505 		.data		= core_pattern,
506 		.maxlen		= CORENAME_MAX_SIZE,
507 		.mode		= 0644,
508 		.proc_handler	= proc_dostring_coredump,
509 	},
510 	{
511 		.procname	= "core_pipe_limit",
512 		.data		= &core_pipe_limit,
513 		.maxlen		= sizeof(unsigned int),
514 		.mode		= 0644,
515 		.proc_handler	= proc_dointvec,
516 	},
517 #endif
518 #ifdef CONFIG_PROC_SYSCTL
519 	{
520 		.procname	= "tainted",
521 		.maxlen 	= sizeof(long),
522 		.mode		= 0644,
523 		.proc_handler	= proc_taint,
524 	},
525 	{
526 		.procname	= "sysctl_writes_strict",
527 		.data		= &sysctl_writes_strict,
528 		.maxlen		= sizeof(int),
529 		.mode		= 0644,
530 		.proc_handler	= proc_dointvec_minmax,
531 		.extra1		= &neg_one,
532 		.extra2		= &one,
533 	},
534 #endif
535 #ifdef CONFIG_LATENCYTOP
536 	{
537 		.procname	= "latencytop",
538 		.data		= &latencytop_enabled,
539 		.maxlen		= sizeof(int),
540 		.mode		= 0644,
541 		.proc_handler	= sysctl_latencytop,
542 	},
543 #endif
544 #ifdef CONFIG_BLK_DEV_INITRD
545 	{
546 		.procname	= "real-root-dev",
547 		.data		= &real_root_dev,
548 		.maxlen		= sizeof(int),
549 		.mode		= 0644,
550 		.proc_handler	= proc_dointvec,
551 	},
552 #endif
553 	{
554 		.procname	= "print-fatal-signals",
555 		.data		= &print_fatal_signals,
556 		.maxlen		= sizeof(int),
557 		.mode		= 0644,
558 		.proc_handler	= proc_dointvec,
559 	},
560 #ifdef CONFIG_SPARC
561 	{
562 		.procname	= "reboot-cmd",
563 		.data		= reboot_command,
564 		.maxlen		= 256,
565 		.mode		= 0644,
566 		.proc_handler	= proc_dostring,
567 	},
568 	{
569 		.procname	= "stop-a",
570 		.data		= &stop_a_enabled,
571 		.maxlen		= sizeof (int),
572 		.mode		= 0644,
573 		.proc_handler	= proc_dointvec,
574 	},
575 	{
576 		.procname	= "scons-poweroff",
577 		.data		= &scons_pwroff,
578 		.maxlen		= sizeof (int),
579 		.mode		= 0644,
580 		.proc_handler	= proc_dointvec,
581 	},
582 #endif
583 #ifdef CONFIG_SPARC64
584 	{
585 		.procname	= "tsb-ratio",
586 		.data		= &sysctl_tsb_ratio,
587 		.maxlen		= sizeof (int),
588 		.mode		= 0644,
589 		.proc_handler	= proc_dointvec,
590 	},
591 #endif
592 #ifdef __hppa__
593 	{
594 		.procname	= "soft-power",
595 		.data		= &pwrsw_enabled,
596 		.maxlen		= sizeof (int),
597 	 	.mode		= 0644,
598 		.proc_handler	= proc_dointvec,
599 	},
600 #endif
601 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
602 	{
603 		.procname	= "unaligned-trap",
604 		.data		= &unaligned_enabled,
605 		.maxlen		= sizeof (int),
606 		.mode		= 0644,
607 		.proc_handler	= proc_dointvec,
608 	},
609 #endif
610 	{
611 		.procname	= "ctrl-alt-del",
612 		.data		= &C_A_D,
613 		.maxlen		= sizeof(int),
614 		.mode		= 0644,
615 		.proc_handler	= proc_dointvec,
616 	},
617 #ifdef CONFIG_FUNCTION_TRACER
618 	{
619 		.procname	= "ftrace_enabled",
620 		.data		= &ftrace_enabled,
621 		.maxlen		= sizeof(int),
622 		.mode		= 0644,
623 		.proc_handler	= ftrace_enable_sysctl,
624 	},
625 #endif
626 #ifdef CONFIG_STACK_TRACER
627 	{
628 		.procname	= "stack_tracer_enabled",
629 		.data		= &stack_tracer_enabled,
630 		.maxlen		= sizeof(int),
631 		.mode		= 0644,
632 		.proc_handler	= stack_trace_sysctl,
633 	},
634 #endif
635 #ifdef CONFIG_TRACING
636 	{
637 		.procname	= "ftrace_dump_on_oops",
638 		.data		= &ftrace_dump_on_oops,
639 		.maxlen		= sizeof(int),
640 		.mode		= 0644,
641 		.proc_handler	= proc_dointvec,
642 	},
643 	{
644 		.procname	= "traceoff_on_warning",
645 		.data		= &__disable_trace_on_warning,
646 		.maxlen		= sizeof(__disable_trace_on_warning),
647 		.mode		= 0644,
648 		.proc_handler	= proc_dointvec,
649 	},
650 	{
651 		.procname	= "tracepoint_printk",
652 		.data		= &tracepoint_printk,
653 		.maxlen		= sizeof(tracepoint_printk),
654 		.mode		= 0644,
655 		.proc_handler	= tracepoint_printk_sysctl,
656 	},
657 #endif
658 #ifdef CONFIG_KEXEC_CORE
659 	{
660 		.procname	= "kexec_load_disabled",
661 		.data		= &kexec_load_disabled,
662 		.maxlen		= sizeof(int),
663 		.mode		= 0644,
664 		/* only handle a transition from default "0" to "1" */
665 		.proc_handler	= proc_dointvec_minmax,
666 		.extra1		= &one,
667 		.extra2		= &one,
668 	},
669 #endif
670 #ifdef CONFIG_MODULES
671 	{
672 		.procname	= "modprobe",
673 		.data		= &modprobe_path,
674 		.maxlen		= KMOD_PATH_LEN,
675 		.mode		= 0644,
676 		.proc_handler	= proc_dostring,
677 	},
678 	{
679 		.procname	= "modules_disabled",
680 		.data		= &modules_disabled,
681 		.maxlen		= sizeof(int),
682 		.mode		= 0644,
683 		/* only handle a transition from default "0" to "1" */
684 		.proc_handler	= proc_dointvec_minmax,
685 		.extra1		= &one,
686 		.extra2		= &one,
687 	},
688 #endif
689 #ifdef CONFIG_UEVENT_HELPER
690 	{
691 		.procname	= "hotplug",
692 		.data		= &uevent_helper,
693 		.maxlen		= UEVENT_HELPER_PATH_LEN,
694 		.mode		= 0644,
695 		.proc_handler	= proc_dostring,
696 	},
697 #endif
698 #ifdef CONFIG_CHR_DEV_SG
699 	{
700 		.procname	= "sg-big-buff",
701 		.data		= &sg_big_buff,
702 		.maxlen		= sizeof (int),
703 		.mode		= 0444,
704 		.proc_handler	= proc_dointvec,
705 	},
706 #endif
707 #ifdef CONFIG_BSD_PROCESS_ACCT
708 	{
709 		.procname	= "acct",
710 		.data		= &acct_parm,
711 		.maxlen		= 3*sizeof(int),
712 		.mode		= 0644,
713 		.proc_handler	= proc_dointvec,
714 	},
715 #endif
716 #ifdef CONFIG_MAGIC_SYSRQ
717 	{
718 		.procname	= "sysrq",
719 		.data		= &__sysrq_enabled,
720 		.maxlen		= sizeof (int),
721 		.mode		= 0644,
722 		.proc_handler	= sysrq_sysctl_handler,
723 	},
724 #endif
725 #ifdef CONFIG_PROC_SYSCTL
726 	{
727 		.procname	= "cad_pid",
728 		.data		= NULL,
729 		.maxlen		= sizeof (int),
730 		.mode		= 0600,
731 		.proc_handler	= proc_do_cad_pid,
732 	},
733 #endif
734 	{
735 		.procname	= "threads-max",
736 		.data		= NULL,
737 		.maxlen		= sizeof(int),
738 		.mode		= 0644,
739 		.proc_handler	= sysctl_max_threads,
740 	},
741 	{
742 		.procname	= "random",
743 		.mode		= 0555,
744 		.child		= random_table,
745 	},
746 	{
747 		.procname	= "usermodehelper",
748 		.mode		= 0555,
749 		.child		= usermodehelper_table,
750 	},
751 #ifdef CONFIG_FW_LOADER_USER_HELPER
752 	{
753 		.procname	= "firmware_config",
754 		.mode		= 0555,
755 		.child		= firmware_config_table,
756 	},
757 #endif
758 	{
759 		.procname	= "overflowuid",
760 		.data		= &overflowuid,
761 		.maxlen		= sizeof(int),
762 		.mode		= 0644,
763 		.proc_handler	= proc_dointvec_minmax,
764 		.extra1		= &minolduid,
765 		.extra2		= &maxolduid,
766 	},
767 	{
768 		.procname	= "overflowgid",
769 		.data		= &overflowgid,
770 		.maxlen		= sizeof(int),
771 		.mode		= 0644,
772 		.proc_handler	= proc_dointvec_minmax,
773 		.extra1		= &minolduid,
774 		.extra2		= &maxolduid,
775 	},
776 #ifdef CONFIG_S390
777 #ifdef CONFIG_MATHEMU
778 	{
779 		.procname	= "ieee_emulation_warnings",
780 		.data		= &sysctl_ieee_emulation_warnings,
781 		.maxlen		= sizeof(int),
782 		.mode		= 0644,
783 		.proc_handler	= proc_dointvec,
784 	},
785 #endif
786 	{
787 		.procname	= "userprocess_debug",
788 		.data		= &show_unhandled_signals,
789 		.maxlen		= sizeof(int),
790 		.mode		= 0644,
791 		.proc_handler	= proc_dointvec,
792 	},
793 #endif
794 	{
795 		.procname	= "pid_max",
796 		.data		= &pid_max,
797 		.maxlen		= sizeof (int),
798 		.mode		= 0644,
799 		.proc_handler	= proc_dointvec_minmax,
800 		.extra1		= &pid_max_min,
801 		.extra2		= &pid_max_max,
802 	},
803 	{
804 		.procname	= "panic_on_oops",
805 		.data		= &panic_on_oops,
806 		.maxlen		= sizeof(int),
807 		.mode		= 0644,
808 		.proc_handler	= proc_dointvec,
809 	},
810 	{
811 		.procname	= "panic_print",
812 		.data		= &panic_print,
813 		.maxlen		= sizeof(unsigned long),
814 		.mode		= 0644,
815 		.proc_handler	= proc_doulongvec_minmax,
816 	},
817 #if defined CONFIG_PRINTK
818 	{
819 		.procname	= "printk",
820 		.data		= &console_loglevel,
821 		.maxlen		= 4*sizeof(int),
822 		.mode		= 0644,
823 		.proc_handler	= proc_dointvec,
824 	},
825 	{
826 		.procname	= "printk_ratelimit",
827 		.data		= &printk_ratelimit_state.interval,
828 		.maxlen		= sizeof(int),
829 		.mode		= 0644,
830 		.proc_handler	= proc_dointvec_jiffies,
831 	},
832 	{
833 		.procname	= "printk_ratelimit_burst",
834 		.data		= &printk_ratelimit_state.burst,
835 		.maxlen		= sizeof(int),
836 		.mode		= 0644,
837 		.proc_handler	= proc_dointvec,
838 	},
839 	{
840 		.procname	= "printk_delay",
841 		.data		= &printk_delay_msec,
842 		.maxlen		= sizeof(int),
843 		.mode		= 0644,
844 		.proc_handler	= proc_dointvec_minmax,
845 		.extra1		= &zero,
846 		.extra2		= &ten_thousand,
847 	},
848 	{
849 		.procname	= "printk_devkmsg",
850 		.data		= devkmsg_log_str,
851 		.maxlen		= DEVKMSG_STR_MAX_SIZE,
852 		.mode		= 0644,
853 		.proc_handler	= devkmsg_sysctl_set_loglvl,
854 	},
855 	{
856 		.procname	= "dmesg_restrict",
857 		.data		= &dmesg_restrict,
858 		.maxlen		= sizeof(int),
859 		.mode		= 0644,
860 		.proc_handler	= proc_dointvec_minmax_sysadmin,
861 		.extra1		= &zero,
862 		.extra2		= &one,
863 	},
864 	{
865 		.procname	= "kptr_restrict",
866 		.data		= &kptr_restrict,
867 		.maxlen		= sizeof(int),
868 		.mode		= 0644,
869 		.proc_handler	= proc_dointvec_minmax_sysadmin,
870 		.extra1		= &zero,
871 		.extra2		= &two,
872 	},
873 #endif
874 	{
875 		.procname	= "ngroups_max",
876 		.data		= &ngroups_max,
877 		.maxlen		= sizeof (int),
878 		.mode		= 0444,
879 		.proc_handler	= proc_dointvec,
880 	},
881 	{
882 		.procname	= "cap_last_cap",
883 		.data		= (void *)&cap_last_cap,
884 		.maxlen		= sizeof(int),
885 		.mode		= 0444,
886 		.proc_handler	= proc_dointvec,
887 	},
888 #if defined(CONFIG_LOCKUP_DETECTOR)
889 	{
890 		.procname       = "watchdog",
891 		.data		= &watchdog_user_enabled,
892 		.maxlen		= sizeof(int),
893 		.mode		= 0644,
894 		.proc_handler   = proc_watchdog,
895 		.extra1		= &zero,
896 		.extra2		= &one,
897 	},
898 	{
899 		.procname	= "watchdog_thresh",
900 		.data		= &watchdog_thresh,
901 		.maxlen		= sizeof(int),
902 		.mode		= 0644,
903 		.proc_handler	= proc_watchdog_thresh,
904 		.extra1		= &zero,
905 		.extra2		= &sixty,
906 	},
907 	{
908 		.procname       = "nmi_watchdog",
909 		.data		= &nmi_watchdog_user_enabled,
910 		.maxlen		= sizeof(int),
911 		.mode		= NMI_WATCHDOG_SYSCTL_PERM,
912 		.proc_handler   = proc_nmi_watchdog,
913 		.extra1		= &zero,
914 		.extra2		= &one,
915 	},
916 	{
917 		.procname	= "watchdog_cpumask",
918 		.data		= &watchdog_cpumask_bits,
919 		.maxlen		= NR_CPUS,
920 		.mode		= 0644,
921 		.proc_handler	= proc_watchdog_cpumask,
922 	},
923 #ifdef CONFIG_SOFTLOCKUP_DETECTOR
924 	{
925 		.procname       = "soft_watchdog",
926 		.data		= &soft_watchdog_user_enabled,
927 		.maxlen		= sizeof(int),
928 		.mode		= 0644,
929 		.proc_handler   = proc_soft_watchdog,
930 		.extra1		= &zero,
931 		.extra2		= &one,
932 	},
933 	{
934 		.procname	= "softlockup_panic",
935 		.data		= &softlockup_panic,
936 		.maxlen		= sizeof(int),
937 		.mode		= 0644,
938 		.proc_handler	= proc_dointvec_minmax,
939 		.extra1		= &zero,
940 		.extra2		= &one,
941 	},
942 #ifdef CONFIG_SMP
943 	{
944 		.procname	= "softlockup_all_cpu_backtrace",
945 		.data		= &sysctl_softlockup_all_cpu_backtrace,
946 		.maxlen		= sizeof(int),
947 		.mode		= 0644,
948 		.proc_handler	= proc_dointvec_minmax,
949 		.extra1		= &zero,
950 		.extra2		= &one,
951 	},
952 #endif /* CONFIG_SMP */
953 #endif
954 #ifdef CONFIG_HARDLOCKUP_DETECTOR
955 	{
956 		.procname	= "hardlockup_panic",
957 		.data		= &hardlockup_panic,
958 		.maxlen		= sizeof(int),
959 		.mode		= 0644,
960 		.proc_handler	= proc_dointvec_minmax,
961 		.extra1		= &zero,
962 		.extra2		= &one,
963 	},
964 #ifdef CONFIG_SMP
965 	{
966 		.procname	= "hardlockup_all_cpu_backtrace",
967 		.data		= &sysctl_hardlockup_all_cpu_backtrace,
968 		.maxlen		= sizeof(int),
969 		.mode		= 0644,
970 		.proc_handler	= proc_dointvec_minmax,
971 		.extra1		= &zero,
972 		.extra2		= &one,
973 	},
974 #endif /* CONFIG_SMP */
975 #endif
976 #endif
977 
978 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
979 	{
980 		.procname       = "unknown_nmi_panic",
981 		.data           = &unknown_nmi_panic,
982 		.maxlen         = sizeof (int),
983 		.mode           = 0644,
984 		.proc_handler   = proc_dointvec,
985 	},
986 #endif
987 #if defined(CONFIG_X86)
988 	{
989 		.procname	= "panic_on_unrecovered_nmi",
990 		.data		= &panic_on_unrecovered_nmi,
991 		.maxlen		= sizeof(int),
992 		.mode		= 0644,
993 		.proc_handler	= proc_dointvec,
994 	},
995 	{
996 		.procname	= "panic_on_io_nmi",
997 		.data		= &panic_on_io_nmi,
998 		.maxlen		= sizeof(int),
999 		.mode		= 0644,
1000 		.proc_handler	= proc_dointvec,
1001 	},
1002 #ifdef CONFIG_DEBUG_STACKOVERFLOW
1003 	{
1004 		.procname	= "panic_on_stackoverflow",
1005 		.data		= &sysctl_panic_on_stackoverflow,
1006 		.maxlen		= sizeof(int),
1007 		.mode		= 0644,
1008 		.proc_handler	= proc_dointvec,
1009 	},
1010 #endif
1011 	{
1012 		.procname	= "bootloader_type",
1013 		.data		= &bootloader_type,
1014 		.maxlen		= sizeof (int),
1015 		.mode		= 0444,
1016 		.proc_handler	= proc_dointvec,
1017 	},
1018 	{
1019 		.procname	= "bootloader_version",
1020 		.data		= &bootloader_version,
1021 		.maxlen		= sizeof (int),
1022 		.mode		= 0444,
1023 		.proc_handler	= proc_dointvec,
1024 	},
1025 	{
1026 		.procname	= "io_delay_type",
1027 		.data		= &io_delay_type,
1028 		.maxlen		= sizeof(int),
1029 		.mode		= 0644,
1030 		.proc_handler	= proc_dointvec,
1031 	},
1032 #endif
1033 #if defined(CONFIG_MMU)
1034 	{
1035 		.procname	= "randomize_va_space",
1036 		.data		= &randomize_va_space,
1037 		.maxlen		= sizeof(int),
1038 		.mode		= 0644,
1039 		.proc_handler	= proc_dointvec,
1040 	},
1041 #endif
1042 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1043 	{
1044 		.procname	= "spin_retry",
1045 		.data		= &spin_retry,
1046 		.maxlen		= sizeof (int),
1047 		.mode		= 0644,
1048 		.proc_handler	= proc_dointvec,
1049 	},
1050 #endif
1051 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1052 	{
1053 		.procname	= "acpi_video_flags",
1054 		.data		= &acpi_realmode_flags,
1055 		.maxlen		= sizeof (unsigned long),
1056 		.mode		= 0644,
1057 		.proc_handler	= proc_doulongvec_minmax,
1058 	},
1059 #endif
1060 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1061 	{
1062 		.procname	= "ignore-unaligned-usertrap",
1063 		.data		= &no_unaligned_warning,
1064 		.maxlen		= sizeof (int),
1065 	 	.mode		= 0644,
1066 		.proc_handler	= proc_dointvec,
1067 	},
1068 #endif
1069 #ifdef CONFIG_IA64
1070 	{
1071 		.procname	= "unaligned-dump-stack",
1072 		.data		= &unaligned_dump_stack,
1073 		.maxlen		= sizeof (int),
1074 		.mode		= 0644,
1075 		.proc_handler	= proc_dointvec,
1076 	},
1077 #endif
1078 #ifdef CONFIG_DETECT_HUNG_TASK
1079 	{
1080 		.procname	= "hung_task_panic",
1081 		.data		= &sysctl_hung_task_panic,
1082 		.maxlen		= sizeof(int),
1083 		.mode		= 0644,
1084 		.proc_handler	= proc_dointvec_minmax,
1085 		.extra1		= &zero,
1086 		.extra2		= &one,
1087 	},
1088 	{
1089 		.procname	= "hung_task_check_count",
1090 		.data		= &sysctl_hung_task_check_count,
1091 		.maxlen		= sizeof(int),
1092 		.mode		= 0644,
1093 		.proc_handler	= proc_dointvec_minmax,
1094 		.extra1		= &zero,
1095 	},
1096 	{
1097 		.procname	= "hung_task_timeout_secs",
1098 		.data		= &sysctl_hung_task_timeout_secs,
1099 		.maxlen		= sizeof(unsigned long),
1100 		.mode		= 0644,
1101 		.proc_handler	= proc_dohung_task_timeout_secs,
1102 		.extra2		= &hung_task_timeout_max,
1103 	},
1104 	{
1105 		.procname	= "hung_task_check_interval_secs",
1106 		.data		= &sysctl_hung_task_check_interval_secs,
1107 		.maxlen		= sizeof(unsigned long),
1108 		.mode		= 0644,
1109 		.proc_handler	= proc_dohung_task_timeout_secs,
1110 		.extra2		= &hung_task_timeout_max,
1111 	},
1112 	{
1113 		.procname	= "hung_task_warnings",
1114 		.data		= &sysctl_hung_task_warnings,
1115 		.maxlen		= sizeof(int),
1116 		.mode		= 0644,
1117 		.proc_handler	= proc_dointvec_minmax,
1118 		.extra1		= &neg_one,
1119 	},
1120 #endif
1121 #ifdef CONFIG_RT_MUTEXES
1122 	{
1123 		.procname	= "max_lock_depth",
1124 		.data		= &max_lock_depth,
1125 		.maxlen		= sizeof(int),
1126 		.mode		= 0644,
1127 		.proc_handler	= proc_dointvec,
1128 	},
1129 #endif
1130 	{
1131 		.procname	= "poweroff_cmd",
1132 		.data		= &poweroff_cmd,
1133 		.maxlen		= POWEROFF_CMD_PATH_LEN,
1134 		.mode		= 0644,
1135 		.proc_handler	= proc_dostring,
1136 	},
1137 #ifdef CONFIG_KEYS
1138 	{
1139 		.procname	= "keys",
1140 		.mode		= 0555,
1141 		.child		= key_sysctls,
1142 	},
1143 #endif
1144 #ifdef CONFIG_PERF_EVENTS
1145 	/*
1146 	 * User-space scripts rely on the existence of this file
1147 	 * as a feature check for perf_events being enabled.
1148 	 *
1149 	 * So it's an ABI, do not remove!
1150 	 */
1151 	{
1152 		.procname	= "perf_event_paranoid",
1153 		.data		= &sysctl_perf_event_paranoid,
1154 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
1155 		.mode		= 0644,
1156 		.proc_handler	= proc_dointvec,
1157 	},
1158 	{
1159 		.procname	= "perf_event_mlock_kb",
1160 		.data		= &sysctl_perf_event_mlock,
1161 		.maxlen		= sizeof(sysctl_perf_event_mlock),
1162 		.mode		= 0644,
1163 		.proc_handler	= proc_dointvec,
1164 	},
1165 	{
1166 		.procname	= "perf_event_max_sample_rate",
1167 		.data		= &sysctl_perf_event_sample_rate,
1168 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
1169 		.mode		= 0644,
1170 		.proc_handler	= perf_proc_update_handler,
1171 		.extra1		= &one,
1172 	},
1173 	{
1174 		.procname	= "perf_cpu_time_max_percent",
1175 		.data		= &sysctl_perf_cpu_time_max_percent,
1176 		.maxlen		= sizeof(sysctl_perf_cpu_time_max_percent),
1177 		.mode		= 0644,
1178 		.proc_handler	= perf_cpu_time_max_percent_handler,
1179 		.extra1		= &zero,
1180 		.extra2		= &one_hundred,
1181 	},
1182 	{
1183 		.procname	= "perf_event_max_stack",
1184 		.data		= &sysctl_perf_event_max_stack,
1185 		.maxlen		= sizeof(sysctl_perf_event_max_stack),
1186 		.mode		= 0644,
1187 		.proc_handler	= perf_event_max_stack_handler,
1188 		.extra1		= &zero,
1189 		.extra2		= &six_hundred_forty_kb,
1190 	},
1191 	{
1192 		.procname	= "perf_event_max_contexts_per_stack",
1193 		.data		= &sysctl_perf_event_max_contexts_per_stack,
1194 		.maxlen		= sizeof(sysctl_perf_event_max_contexts_per_stack),
1195 		.mode		= 0644,
1196 		.proc_handler	= perf_event_max_stack_handler,
1197 		.extra1		= &zero,
1198 		.extra2		= &one_thousand,
1199 	},
1200 #endif
1201 	{
1202 		.procname	= "panic_on_warn",
1203 		.data		= &panic_on_warn,
1204 		.maxlen		= sizeof(int),
1205 		.mode		= 0644,
1206 		.proc_handler	= proc_dointvec_minmax,
1207 		.extra1		= &zero,
1208 		.extra2		= &one,
1209 	},
1210 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1211 	{
1212 		.procname	= "timer_migration",
1213 		.data		= &sysctl_timer_migration,
1214 		.maxlen		= sizeof(unsigned int),
1215 		.mode		= 0644,
1216 		.proc_handler	= timer_migration_handler,
1217 		.extra1		= &zero,
1218 		.extra2		= &one,
1219 	},
1220 #endif
1221 #ifdef CONFIG_BPF_SYSCALL
1222 	{
1223 		.procname	= "unprivileged_bpf_disabled",
1224 		.data		= &sysctl_unprivileged_bpf_disabled,
1225 		.maxlen		= sizeof(sysctl_unprivileged_bpf_disabled),
1226 		.mode		= 0644,
1227 		/* only handle a transition from default "0" to "1" */
1228 		.proc_handler	= proc_dointvec_minmax,
1229 		.extra1		= &one,
1230 		.extra2		= &one,
1231 	},
1232 #endif
1233 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1234 	{
1235 		.procname	= "panic_on_rcu_stall",
1236 		.data		= &sysctl_panic_on_rcu_stall,
1237 		.maxlen		= sizeof(sysctl_panic_on_rcu_stall),
1238 		.mode		= 0644,
1239 		.proc_handler	= proc_dointvec_minmax,
1240 		.extra1		= &zero,
1241 		.extra2		= &one,
1242 	},
1243 #endif
1244 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
1245 	{
1246 		.procname	= "stack_erasing",
1247 		.data		= NULL,
1248 		.maxlen		= sizeof(int),
1249 		.mode		= 0600,
1250 		.proc_handler	= stack_erasing_sysctl,
1251 		.extra1		= &zero,
1252 		.extra2		= &one,
1253 	},
1254 #endif
1255 	{ }
1256 };
1257 
1258 static struct ctl_table vm_table[] = {
1259 	{
1260 		.procname	= "overcommit_memory",
1261 		.data		= &sysctl_overcommit_memory,
1262 		.maxlen		= sizeof(sysctl_overcommit_memory),
1263 		.mode		= 0644,
1264 		.proc_handler	= proc_dointvec_minmax,
1265 		.extra1		= &zero,
1266 		.extra2		= &two,
1267 	},
1268 	{
1269 		.procname	= "panic_on_oom",
1270 		.data		= &sysctl_panic_on_oom,
1271 		.maxlen		= sizeof(sysctl_panic_on_oom),
1272 		.mode		= 0644,
1273 		.proc_handler	= proc_dointvec_minmax,
1274 		.extra1		= &zero,
1275 		.extra2		= &two,
1276 	},
1277 	{
1278 		.procname	= "oom_kill_allocating_task",
1279 		.data		= &sysctl_oom_kill_allocating_task,
1280 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1281 		.mode		= 0644,
1282 		.proc_handler	= proc_dointvec,
1283 	},
1284 	{
1285 		.procname	= "oom_dump_tasks",
1286 		.data		= &sysctl_oom_dump_tasks,
1287 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1288 		.mode		= 0644,
1289 		.proc_handler	= proc_dointvec,
1290 	},
1291 	{
1292 		.procname	= "overcommit_ratio",
1293 		.data		= &sysctl_overcommit_ratio,
1294 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1295 		.mode		= 0644,
1296 		.proc_handler	= overcommit_ratio_handler,
1297 	},
1298 	{
1299 		.procname	= "overcommit_kbytes",
1300 		.data		= &sysctl_overcommit_kbytes,
1301 		.maxlen		= sizeof(sysctl_overcommit_kbytes),
1302 		.mode		= 0644,
1303 		.proc_handler	= overcommit_kbytes_handler,
1304 	},
1305 	{
1306 		.procname	= "page-cluster",
1307 		.data		= &page_cluster,
1308 		.maxlen		= sizeof(int),
1309 		.mode		= 0644,
1310 		.proc_handler	= proc_dointvec_minmax,
1311 		.extra1		= &zero,
1312 	},
1313 	{
1314 		.procname	= "dirty_background_ratio",
1315 		.data		= &dirty_background_ratio,
1316 		.maxlen		= sizeof(dirty_background_ratio),
1317 		.mode		= 0644,
1318 		.proc_handler	= dirty_background_ratio_handler,
1319 		.extra1		= &zero,
1320 		.extra2		= &one_hundred,
1321 	},
1322 	{
1323 		.procname	= "dirty_background_bytes",
1324 		.data		= &dirty_background_bytes,
1325 		.maxlen		= sizeof(dirty_background_bytes),
1326 		.mode		= 0644,
1327 		.proc_handler	= dirty_background_bytes_handler,
1328 		.extra1		= &one_ul,
1329 	},
1330 	{
1331 		.procname	= "dirty_ratio",
1332 		.data		= &vm_dirty_ratio,
1333 		.maxlen		= sizeof(vm_dirty_ratio),
1334 		.mode		= 0644,
1335 		.proc_handler	= dirty_ratio_handler,
1336 		.extra1		= &zero,
1337 		.extra2		= &one_hundred,
1338 	},
1339 	{
1340 		.procname	= "dirty_bytes",
1341 		.data		= &vm_dirty_bytes,
1342 		.maxlen		= sizeof(vm_dirty_bytes),
1343 		.mode		= 0644,
1344 		.proc_handler	= dirty_bytes_handler,
1345 		.extra1		= &dirty_bytes_min,
1346 	},
1347 	{
1348 		.procname	= "dirty_writeback_centisecs",
1349 		.data		= &dirty_writeback_interval,
1350 		.maxlen		= sizeof(dirty_writeback_interval),
1351 		.mode		= 0644,
1352 		.proc_handler	= dirty_writeback_centisecs_handler,
1353 	},
1354 	{
1355 		.procname	= "dirty_expire_centisecs",
1356 		.data		= &dirty_expire_interval,
1357 		.maxlen		= sizeof(dirty_expire_interval),
1358 		.mode		= 0644,
1359 		.proc_handler	= proc_dointvec_minmax,
1360 		.extra1		= &zero,
1361 	},
1362 	{
1363 		.procname	= "dirtytime_expire_seconds",
1364 		.data		= &dirtytime_expire_interval,
1365 		.maxlen		= sizeof(dirtytime_expire_interval),
1366 		.mode		= 0644,
1367 		.proc_handler	= dirtytime_interval_handler,
1368 		.extra1		= &zero,
1369 	},
1370 	{
1371 		.procname	= "swappiness",
1372 		.data		= &vm_swappiness,
1373 		.maxlen		= sizeof(vm_swappiness),
1374 		.mode		= 0644,
1375 		.proc_handler	= proc_dointvec_minmax,
1376 		.extra1		= &zero,
1377 		.extra2		= &one_hundred,
1378 	},
1379 #ifdef CONFIG_HUGETLB_PAGE
1380 	{
1381 		.procname	= "nr_hugepages",
1382 		.data		= NULL,
1383 		.maxlen		= sizeof(unsigned long),
1384 		.mode		= 0644,
1385 		.proc_handler	= hugetlb_sysctl_handler,
1386 	},
1387 #ifdef CONFIG_NUMA
1388 	{
1389 		.procname       = "nr_hugepages_mempolicy",
1390 		.data           = NULL,
1391 		.maxlen         = sizeof(unsigned long),
1392 		.mode           = 0644,
1393 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1394 	},
1395 	{
1396 		.procname		= "numa_stat",
1397 		.data			= &sysctl_vm_numa_stat,
1398 		.maxlen			= sizeof(int),
1399 		.mode			= 0644,
1400 		.proc_handler	= sysctl_vm_numa_stat_handler,
1401 		.extra1			= &zero,
1402 		.extra2			= &one,
1403 	},
1404 #endif
1405 	 {
1406 		.procname	= "hugetlb_shm_group",
1407 		.data		= &sysctl_hugetlb_shm_group,
1408 		.maxlen		= sizeof(gid_t),
1409 		.mode		= 0644,
1410 		.proc_handler	= proc_dointvec,
1411 	 },
1412 	{
1413 		.procname	= "nr_overcommit_hugepages",
1414 		.data		= NULL,
1415 		.maxlen		= sizeof(unsigned long),
1416 		.mode		= 0644,
1417 		.proc_handler	= hugetlb_overcommit_handler,
1418 	},
1419 #endif
1420 	{
1421 		.procname	= "lowmem_reserve_ratio",
1422 		.data		= &sysctl_lowmem_reserve_ratio,
1423 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1424 		.mode		= 0644,
1425 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1426 	},
1427 	{
1428 		.procname	= "drop_caches",
1429 		.data		= &sysctl_drop_caches,
1430 		.maxlen		= sizeof(int),
1431 		.mode		= 0644,
1432 		.proc_handler	= drop_caches_sysctl_handler,
1433 		.extra1		= &one,
1434 		.extra2		= &four,
1435 	},
1436 #ifdef CONFIG_COMPACTION
1437 	{
1438 		.procname	= "compact_memory",
1439 		.data		= &sysctl_compact_memory,
1440 		.maxlen		= sizeof(int),
1441 		.mode		= 0200,
1442 		.proc_handler	= sysctl_compaction_handler,
1443 	},
1444 	{
1445 		.procname	= "extfrag_threshold",
1446 		.data		= &sysctl_extfrag_threshold,
1447 		.maxlen		= sizeof(int),
1448 		.mode		= 0644,
1449 		.proc_handler	= sysctl_extfrag_handler,
1450 		.extra1		= &min_extfrag_threshold,
1451 		.extra2		= &max_extfrag_threshold,
1452 	},
1453 	{
1454 		.procname	= "compact_unevictable_allowed",
1455 		.data		= &sysctl_compact_unevictable_allowed,
1456 		.maxlen		= sizeof(int),
1457 		.mode		= 0644,
1458 		.proc_handler	= proc_dointvec,
1459 		.extra1		= &zero,
1460 		.extra2		= &one,
1461 	},
1462 
1463 #endif /* CONFIG_COMPACTION */
1464 	{
1465 		.procname	= "min_free_kbytes",
1466 		.data		= &min_free_kbytes,
1467 		.maxlen		= sizeof(min_free_kbytes),
1468 		.mode		= 0644,
1469 		.proc_handler	= min_free_kbytes_sysctl_handler,
1470 		.extra1		= &zero,
1471 	},
1472 	{
1473 		.procname	= "watermark_boost_factor",
1474 		.data		= &watermark_boost_factor,
1475 		.maxlen		= sizeof(watermark_boost_factor),
1476 		.mode		= 0644,
1477 		.proc_handler	= watermark_boost_factor_sysctl_handler,
1478 		.extra1		= &zero,
1479 	},
1480 	{
1481 		.procname	= "watermark_scale_factor",
1482 		.data		= &watermark_scale_factor,
1483 		.maxlen		= sizeof(watermark_scale_factor),
1484 		.mode		= 0644,
1485 		.proc_handler	= watermark_scale_factor_sysctl_handler,
1486 		.extra1		= &one,
1487 		.extra2		= &one_thousand,
1488 	},
1489 	{
1490 		.procname	= "percpu_pagelist_fraction",
1491 		.data		= &percpu_pagelist_fraction,
1492 		.maxlen		= sizeof(percpu_pagelist_fraction),
1493 		.mode		= 0644,
1494 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1495 		.extra1		= &zero,
1496 	},
1497 #ifdef CONFIG_MMU
1498 	{
1499 		.procname	= "max_map_count",
1500 		.data		= &sysctl_max_map_count,
1501 		.maxlen		= sizeof(sysctl_max_map_count),
1502 		.mode		= 0644,
1503 		.proc_handler	= proc_dointvec_minmax,
1504 		.extra1		= &zero,
1505 	},
1506 #else
1507 	{
1508 		.procname	= "nr_trim_pages",
1509 		.data		= &sysctl_nr_trim_pages,
1510 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1511 		.mode		= 0644,
1512 		.proc_handler	= proc_dointvec_minmax,
1513 		.extra1		= &zero,
1514 	},
1515 #endif
1516 	{
1517 		.procname	= "laptop_mode",
1518 		.data		= &laptop_mode,
1519 		.maxlen		= sizeof(laptop_mode),
1520 		.mode		= 0644,
1521 		.proc_handler	= proc_dointvec_jiffies,
1522 	},
1523 	{
1524 		.procname	= "block_dump",
1525 		.data		= &block_dump,
1526 		.maxlen		= sizeof(block_dump),
1527 		.mode		= 0644,
1528 		.proc_handler	= proc_dointvec,
1529 		.extra1		= &zero,
1530 	},
1531 	{
1532 		.procname	= "vfs_cache_pressure",
1533 		.data		= &sysctl_vfs_cache_pressure,
1534 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1535 		.mode		= 0644,
1536 		.proc_handler	= proc_dointvec,
1537 		.extra1		= &zero,
1538 	},
1539 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1540 	{
1541 		.procname	= "legacy_va_layout",
1542 		.data		= &sysctl_legacy_va_layout,
1543 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1544 		.mode		= 0644,
1545 		.proc_handler	= proc_dointvec,
1546 		.extra1		= &zero,
1547 	},
1548 #endif
1549 #ifdef CONFIG_NUMA
1550 	{
1551 		.procname	= "zone_reclaim_mode",
1552 		.data		= &node_reclaim_mode,
1553 		.maxlen		= sizeof(node_reclaim_mode),
1554 		.mode		= 0644,
1555 		.proc_handler	= proc_dointvec,
1556 		.extra1		= &zero,
1557 	},
1558 	{
1559 		.procname	= "min_unmapped_ratio",
1560 		.data		= &sysctl_min_unmapped_ratio,
1561 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1562 		.mode		= 0644,
1563 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1564 		.extra1		= &zero,
1565 		.extra2		= &one_hundred,
1566 	},
1567 	{
1568 		.procname	= "min_slab_ratio",
1569 		.data		= &sysctl_min_slab_ratio,
1570 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1571 		.mode		= 0644,
1572 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1573 		.extra1		= &zero,
1574 		.extra2		= &one_hundred,
1575 	},
1576 #endif
1577 #ifdef CONFIG_SMP
1578 	{
1579 		.procname	= "stat_interval",
1580 		.data		= &sysctl_stat_interval,
1581 		.maxlen		= sizeof(sysctl_stat_interval),
1582 		.mode		= 0644,
1583 		.proc_handler	= proc_dointvec_jiffies,
1584 	},
1585 	{
1586 		.procname	= "stat_refresh",
1587 		.data		= NULL,
1588 		.maxlen		= 0,
1589 		.mode		= 0600,
1590 		.proc_handler	= vmstat_refresh,
1591 	},
1592 #endif
1593 #ifdef CONFIG_MMU
1594 	{
1595 		.procname	= "mmap_min_addr",
1596 		.data		= &dac_mmap_min_addr,
1597 		.maxlen		= sizeof(unsigned long),
1598 		.mode		= 0644,
1599 		.proc_handler	= mmap_min_addr_handler,
1600 	},
1601 #endif
1602 #ifdef CONFIG_NUMA
1603 	{
1604 		.procname	= "numa_zonelist_order",
1605 		.data		= &numa_zonelist_order,
1606 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1607 		.mode		= 0644,
1608 		.proc_handler	= numa_zonelist_order_handler,
1609 	},
1610 #endif
1611 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1612    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1613 	{
1614 		.procname	= "vdso_enabled",
1615 #ifdef CONFIG_X86_32
1616 		.data		= &vdso32_enabled,
1617 		.maxlen		= sizeof(vdso32_enabled),
1618 #else
1619 		.data		= &vdso_enabled,
1620 		.maxlen		= sizeof(vdso_enabled),
1621 #endif
1622 		.mode		= 0644,
1623 		.proc_handler	= proc_dointvec,
1624 		.extra1		= &zero,
1625 	},
1626 #endif
1627 #ifdef CONFIG_HIGHMEM
1628 	{
1629 		.procname	= "highmem_is_dirtyable",
1630 		.data		= &vm_highmem_is_dirtyable,
1631 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1632 		.mode		= 0644,
1633 		.proc_handler	= proc_dointvec_minmax,
1634 		.extra1		= &zero,
1635 		.extra2		= &one,
1636 	},
1637 #endif
1638 #ifdef CONFIG_MEMORY_FAILURE
1639 	{
1640 		.procname	= "memory_failure_early_kill",
1641 		.data		= &sysctl_memory_failure_early_kill,
1642 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1643 		.mode		= 0644,
1644 		.proc_handler	= proc_dointvec_minmax,
1645 		.extra1		= &zero,
1646 		.extra2		= &one,
1647 	},
1648 	{
1649 		.procname	= "memory_failure_recovery",
1650 		.data		= &sysctl_memory_failure_recovery,
1651 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1652 		.mode		= 0644,
1653 		.proc_handler	= proc_dointvec_minmax,
1654 		.extra1		= &zero,
1655 		.extra2		= &one,
1656 	},
1657 #endif
1658 	{
1659 		.procname	= "user_reserve_kbytes",
1660 		.data		= &sysctl_user_reserve_kbytes,
1661 		.maxlen		= sizeof(sysctl_user_reserve_kbytes),
1662 		.mode		= 0644,
1663 		.proc_handler	= proc_doulongvec_minmax,
1664 	},
1665 	{
1666 		.procname	= "admin_reserve_kbytes",
1667 		.data		= &sysctl_admin_reserve_kbytes,
1668 		.maxlen		= sizeof(sysctl_admin_reserve_kbytes),
1669 		.mode		= 0644,
1670 		.proc_handler	= proc_doulongvec_minmax,
1671 	},
1672 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1673 	{
1674 		.procname	= "mmap_rnd_bits",
1675 		.data		= &mmap_rnd_bits,
1676 		.maxlen		= sizeof(mmap_rnd_bits),
1677 		.mode		= 0600,
1678 		.proc_handler	= proc_dointvec_minmax,
1679 		.extra1		= (void *)&mmap_rnd_bits_min,
1680 		.extra2		= (void *)&mmap_rnd_bits_max,
1681 	},
1682 #endif
1683 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1684 	{
1685 		.procname	= "mmap_rnd_compat_bits",
1686 		.data		= &mmap_rnd_compat_bits,
1687 		.maxlen		= sizeof(mmap_rnd_compat_bits),
1688 		.mode		= 0600,
1689 		.proc_handler	= proc_dointvec_minmax,
1690 		.extra1		= (void *)&mmap_rnd_compat_bits_min,
1691 		.extra2		= (void *)&mmap_rnd_compat_bits_max,
1692 	},
1693 #endif
1694 	{ }
1695 };
1696 
1697 static struct ctl_table fs_table[] = {
1698 	{
1699 		.procname	= "inode-nr",
1700 		.data		= &inodes_stat,
1701 		.maxlen		= 2*sizeof(long),
1702 		.mode		= 0444,
1703 		.proc_handler	= proc_nr_inodes,
1704 	},
1705 	{
1706 		.procname	= "inode-state",
1707 		.data		= &inodes_stat,
1708 		.maxlen		= 7*sizeof(long),
1709 		.mode		= 0444,
1710 		.proc_handler	= proc_nr_inodes,
1711 	},
1712 	{
1713 		.procname	= "file-nr",
1714 		.data		= &files_stat,
1715 		.maxlen		= sizeof(files_stat),
1716 		.mode		= 0444,
1717 		.proc_handler	= proc_nr_files,
1718 	},
1719 	{
1720 		.procname	= "file-max",
1721 		.data		= &files_stat.max_files,
1722 		.maxlen		= sizeof(files_stat.max_files),
1723 		.mode		= 0644,
1724 		.proc_handler	= proc_doulongvec_minmax,
1725 	},
1726 	{
1727 		.procname	= "nr_open",
1728 		.data		= &sysctl_nr_open,
1729 		.maxlen		= sizeof(unsigned int),
1730 		.mode		= 0644,
1731 		.proc_handler	= proc_dointvec_minmax,
1732 		.extra1		= &sysctl_nr_open_min,
1733 		.extra2		= &sysctl_nr_open_max,
1734 	},
1735 	{
1736 		.procname	= "dentry-state",
1737 		.data		= &dentry_stat,
1738 		.maxlen		= 6*sizeof(long),
1739 		.mode		= 0444,
1740 		.proc_handler	= proc_nr_dentry,
1741 	},
1742 	{
1743 		.procname	= "overflowuid",
1744 		.data		= &fs_overflowuid,
1745 		.maxlen		= sizeof(int),
1746 		.mode		= 0644,
1747 		.proc_handler	= proc_dointvec_minmax,
1748 		.extra1		= &minolduid,
1749 		.extra2		= &maxolduid,
1750 	},
1751 	{
1752 		.procname	= "overflowgid",
1753 		.data		= &fs_overflowgid,
1754 		.maxlen		= sizeof(int),
1755 		.mode		= 0644,
1756 		.proc_handler	= proc_dointvec_minmax,
1757 		.extra1		= &minolduid,
1758 		.extra2		= &maxolduid,
1759 	},
1760 #ifdef CONFIG_FILE_LOCKING
1761 	{
1762 		.procname	= "leases-enable",
1763 		.data		= &leases_enable,
1764 		.maxlen		= sizeof(int),
1765 		.mode		= 0644,
1766 		.proc_handler	= proc_dointvec,
1767 	},
1768 #endif
1769 #ifdef CONFIG_DNOTIFY
1770 	{
1771 		.procname	= "dir-notify-enable",
1772 		.data		= &dir_notify_enable,
1773 		.maxlen		= sizeof(int),
1774 		.mode		= 0644,
1775 		.proc_handler	= proc_dointvec,
1776 	},
1777 #endif
1778 #ifdef CONFIG_MMU
1779 #ifdef CONFIG_FILE_LOCKING
1780 	{
1781 		.procname	= "lease-break-time",
1782 		.data		= &lease_break_time,
1783 		.maxlen		= sizeof(int),
1784 		.mode		= 0644,
1785 		.proc_handler	= proc_dointvec,
1786 	},
1787 #endif
1788 #ifdef CONFIG_AIO
1789 	{
1790 		.procname	= "aio-nr",
1791 		.data		= &aio_nr,
1792 		.maxlen		= sizeof(aio_nr),
1793 		.mode		= 0444,
1794 		.proc_handler	= proc_doulongvec_minmax,
1795 	},
1796 	{
1797 		.procname	= "aio-max-nr",
1798 		.data		= &aio_max_nr,
1799 		.maxlen		= sizeof(aio_max_nr),
1800 		.mode		= 0644,
1801 		.proc_handler	= proc_doulongvec_minmax,
1802 	},
1803 #endif /* CONFIG_AIO */
1804 #ifdef CONFIG_INOTIFY_USER
1805 	{
1806 		.procname	= "inotify",
1807 		.mode		= 0555,
1808 		.child		= inotify_table,
1809 	},
1810 #endif
1811 #ifdef CONFIG_EPOLL
1812 	{
1813 		.procname	= "epoll",
1814 		.mode		= 0555,
1815 		.child		= epoll_table,
1816 	},
1817 #endif
1818 #endif
1819 	{
1820 		.procname	= "protected_symlinks",
1821 		.data		= &sysctl_protected_symlinks,
1822 		.maxlen		= sizeof(int),
1823 		.mode		= 0600,
1824 		.proc_handler	= proc_dointvec_minmax,
1825 		.extra1		= &zero,
1826 		.extra2		= &one,
1827 	},
1828 	{
1829 		.procname	= "protected_hardlinks",
1830 		.data		= &sysctl_protected_hardlinks,
1831 		.maxlen		= sizeof(int),
1832 		.mode		= 0600,
1833 		.proc_handler	= proc_dointvec_minmax,
1834 		.extra1		= &zero,
1835 		.extra2		= &one,
1836 	},
1837 	{
1838 		.procname	= "protected_fifos",
1839 		.data		= &sysctl_protected_fifos,
1840 		.maxlen		= sizeof(int),
1841 		.mode		= 0600,
1842 		.proc_handler	= proc_dointvec_minmax,
1843 		.extra1		= &zero,
1844 		.extra2		= &two,
1845 	},
1846 	{
1847 		.procname	= "protected_regular",
1848 		.data		= &sysctl_protected_regular,
1849 		.maxlen		= sizeof(int),
1850 		.mode		= 0600,
1851 		.proc_handler	= proc_dointvec_minmax,
1852 		.extra1		= &zero,
1853 		.extra2		= &two,
1854 	},
1855 	{
1856 		.procname	= "suid_dumpable",
1857 		.data		= &suid_dumpable,
1858 		.maxlen		= sizeof(int),
1859 		.mode		= 0644,
1860 		.proc_handler	= proc_dointvec_minmax_coredump,
1861 		.extra1		= &zero,
1862 		.extra2		= &two,
1863 	},
1864 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1865 	{
1866 		.procname	= "binfmt_misc",
1867 		.mode		= 0555,
1868 		.child		= sysctl_mount_point,
1869 	},
1870 #endif
1871 	{
1872 		.procname	= "pipe-max-size",
1873 		.data		= &pipe_max_size,
1874 		.maxlen		= sizeof(pipe_max_size),
1875 		.mode		= 0644,
1876 		.proc_handler	= proc_dopipe_max_size,
1877 	},
1878 	{
1879 		.procname	= "pipe-user-pages-hard",
1880 		.data		= &pipe_user_pages_hard,
1881 		.maxlen		= sizeof(pipe_user_pages_hard),
1882 		.mode		= 0644,
1883 		.proc_handler	= proc_doulongvec_minmax,
1884 	},
1885 	{
1886 		.procname	= "pipe-user-pages-soft",
1887 		.data		= &pipe_user_pages_soft,
1888 		.maxlen		= sizeof(pipe_user_pages_soft),
1889 		.mode		= 0644,
1890 		.proc_handler	= proc_doulongvec_minmax,
1891 	},
1892 	{
1893 		.procname	= "mount-max",
1894 		.data		= &sysctl_mount_max,
1895 		.maxlen		= sizeof(unsigned int),
1896 		.mode		= 0644,
1897 		.proc_handler	= proc_dointvec_minmax,
1898 		.extra1		= &one,
1899 	},
1900 	{ }
1901 };
1902 
1903 static struct ctl_table debug_table[] = {
1904 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1905 	{
1906 		.procname	= "exception-trace",
1907 		.data		= &show_unhandled_signals,
1908 		.maxlen		= sizeof(int),
1909 		.mode		= 0644,
1910 		.proc_handler	= proc_dointvec
1911 	},
1912 #endif
1913 #if defined(CONFIG_OPTPROBES)
1914 	{
1915 		.procname	= "kprobes-optimization",
1916 		.data		= &sysctl_kprobes_optimization,
1917 		.maxlen		= sizeof(int),
1918 		.mode		= 0644,
1919 		.proc_handler	= proc_kprobes_optimization_handler,
1920 		.extra1		= &zero,
1921 		.extra2		= &one,
1922 	},
1923 #endif
1924 	{ }
1925 };
1926 
1927 static struct ctl_table dev_table[] = {
1928 	{ }
1929 };
1930 
1931 int __init sysctl_init(void)
1932 {
1933 	struct ctl_table_header *hdr;
1934 
1935 	hdr = register_sysctl_table(sysctl_base_table);
1936 	kmemleak_not_leak(hdr);
1937 	return 0;
1938 }
1939 
1940 #endif /* CONFIG_SYSCTL */
1941 
1942 /*
1943  * /proc/sys support
1944  */
1945 
1946 #ifdef CONFIG_PROC_SYSCTL
1947 
1948 static int _proc_do_string(char *data, int maxlen, int write,
1949 			   char __user *buffer,
1950 			   size_t *lenp, loff_t *ppos)
1951 {
1952 	size_t len;
1953 	char __user *p;
1954 	char c;
1955 
1956 	if (!data || !maxlen || !*lenp) {
1957 		*lenp = 0;
1958 		return 0;
1959 	}
1960 
1961 	if (write) {
1962 		if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1963 			/* Only continue writes not past the end of buffer. */
1964 			len = strlen(data);
1965 			if (len > maxlen - 1)
1966 				len = maxlen - 1;
1967 
1968 			if (*ppos > len)
1969 				return 0;
1970 			len = *ppos;
1971 		} else {
1972 			/* Start writing from beginning of buffer. */
1973 			len = 0;
1974 		}
1975 
1976 		*ppos += *lenp;
1977 		p = buffer;
1978 		while ((p - buffer) < *lenp && len < maxlen - 1) {
1979 			if (get_user(c, p++))
1980 				return -EFAULT;
1981 			if (c == 0 || c == '\n')
1982 				break;
1983 			data[len++] = c;
1984 		}
1985 		data[len] = 0;
1986 	} else {
1987 		len = strlen(data);
1988 		if (len > maxlen)
1989 			len = maxlen;
1990 
1991 		if (*ppos > len) {
1992 			*lenp = 0;
1993 			return 0;
1994 		}
1995 
1996 		data += *ppos;
1997 		len  -= *ppos;
1998 
1999 		if (len > *lenp)
2000 			len = *lenp;
2001 		if (len)
2002 			if (copy_to_user(buffer, data, len))
2003 				return -EFAULT;
2004 		if (len < *lenp) {
2005 			if (put_user('\n', buffer + len))
2006 				return -EFAULT;
2007 			len++;
2008 		}
2009 		*lenp = len;
2010 		*ppos += len;
2011 	}
2012 	return 0;
2013 }
2014 
2015 static void warn_sysctl_write(struct ctl_table *table)
2016 {
2017 	pr_warn_once("%s wrote to %s when file position was not 0!\n"
2018 		"This will not be supported in the future. To silence this\n"
2019 		"warning, set kernel.sysctl_writes_strict = -1\n",
2020 		current->comm, table->procname);
2021 }
2022 
2023 /**
2024  * proc_first_pos_non_zero_ignore - check if first position is allowed
2025  * @ppos: file position
2026  * @table: the sysctl table
2027  *
2028  * Returns true if the first position is non-zero and the sysctl_writes_strict
2029  * mode indicates this is not allowed for numeric input types. String proc
2030  * handlers can ignore the return value.
2031  */
2032 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
2033 					   struct ctl_table *table)
2034 {
2035 	if (!*ppos)
2036 		return false;
2037 
2038 	switch (sysctl_writes_strict) {
2039 	case SYSCTL_WRITES_STRICT:
2040 		return true;
2041 	case SYSCTL_WRITES_WARN:
2042 		warn_sysctl_write(table);
2043 		return false;
2044 	default:
2045 		return false;
2046 	}
2047 }
2048 
2049 /**
2050  * proc_dostring - read a string sysctl
2051  * @table: the sysctl table
2052  * @write: %TRUE if this is a write to the sysctl file
2053  * @buffer: the user buffer
2054  * @lenp: the size of the user buffer
2055  * @ppos: file position
2056  *
2057  * Reads/writes a string from/to the user buffer. If the kernel
2058  * buffer provided is not large enough to hold the string, the
2059  * string is truncated. The copied string is %NULL-terminated.
2060  * If the string is being read by the user process, it is copied
2061  * and a newline '\n' is added. It is truncated if the buffer is
2062  * not large enough.
2063  *
2064  * Returns 0 on success.
2065  */
2066 int proc_dostring(struct ctl_table *table, int write,
2067 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2068 {
2069 	if (write)
2070 		proc_first_pos_non_zero_ignore(ppos, table);
2071 
2072 	return _proc_do_string((char *)(table->data), table->maxlen, write,
2073 			       (char __user *)buffer, lenp, ppos);
2074 }
2075 
2076 static size_t proc_skip_spaces(char **buf)
2077 {
2078 	size_t ret;
2079 	char *tmp = skip_spaces(*buf);
2080 	ret = tmp - *buf;
2081 	*buf = tmp;
2082 	return ret;
2083 }
2084 
2085 static void proc_skip_char(char **buf, size_t *size, const char v)
2086 {
2087 	while (*size) {
2088 		if (**buf != v)
2089 			break;
2090 		(*size)--;
2091 		(*buf)++;
2092 	}
2093 }
2094 
2095 #define TMPBUFLEN 22
2096 /**
2097  * proc_get_long - reads an ASCII formatted integer from a user buffer
2098  *
2099  * @buf: a kernel buffer
2100  * @size: size of the kernel buffer
2101  * @val: this is where the number will be stored
2102  * @neg: set to %TRUE if number is negative
2103  * @perm_tr: a vector which contains the allowed trailers
2104  * @perm_tr_len: size of the perm_tr vector
2105  * @tr: pointer to store the trailer character
2106  *
2107  * In case of success %0 is returned and @buf and @size are updated with
2108  * the amount of bytes read. If @tr is non-NULL and a trailing
2109  * character exists (size is non-zero after returning from this
2110  * function), @tr is updated with the trailing character.
2111  */
2112 static int proc_get_long(char **buf, size_t *size,
2113 			  unsigned long *val, bool *neg,
2114 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
2115 {
2116 	int len;
2117 	char *p, tmp[TMPBUFLEN];
2118 
2119 	if (!*size)
2120 		return -EINVAL;
2121 
2122 	len = *size;
2123 	if (len > TMPBUFLEN - 1)
2124 		len = TMPBUFLEN - 1;
2125 
2126 	memcpy(tmp, *buf, len);
2127 
2128 	tmp[len] = 0;
2129 	p = tmp;
2130 	if (*p == '-' && *size > 1) {
2131 		*neg = true;
2132 		p++;
2133 	} else
2134 		*neg = false;
2135 	if (!isdigit(*p))
2136 		return -EINVAL;
2137 
2138 	*val = simple_strtoul(p, &p, 0);
2139 
2140 	len = p - tmp;
2141 
2142 	/* We don't know if the next char is whitespace thus we may accept
2143 	 * invalid integers (e.g. 1234...a) or two integers instead of one
2144 	 * (e.g. 123...1). So lets not allow such large numbers. */
2145 	if (len == TMPBUFLEN - 1)
2146 		return -EINVAL;
2147 
2148 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2149 		return -EINVAL;
2150 
2151 	if (tr && (len < *size))
2152 		*tr = *p;
2153 
2154 	*buf += len;
2155 	*size -= len;
2156 
2157 	return 0;
2158 }
2159 
2160 /**
2161  * proc_put_long - converts an integer to a decimal ASCII formatted string
2162  *
2163  * @buf: the user buffer
2164  * @size: the size of the user buffer
2165  * @val: the integer to be converted
2166  * @neg: sign of the number, %TRUE for negative
2167  *
2168  * In case of success %0 is returned and @buf and @size are updated with
2169  * the amount of bytes written.
2170  */
2171 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2172 			  bool neg)
2173 {
2174 	int len;
2175 	char tmp[TMPBUFLEN], *p = tmp;
2176 
2177 	sprintf(p, "%s%lu", neg ? "-" : "", val);
2178 	len = strlen(tmp);
2179 	if (len > *size)
2180 		len = *size;
2181 	if (copy_to_user(*buf, tmp, len))
2182 		return -EFAULT;
2183 	*size -= len;
2184 	*buf += len;
2185 	return 0;
2186 }
2187 #undef TMPBUFLEN
2188 
2189 static int proc_put_char(void __user **buf, size_t *size, char c)
2190 {
2191 	if (*size) {
2192 		char __user **buffer = (char __user **)buf;
2193 		if (put_user(c, *buffer))
2194 			return -EFAULT;
2195 		(*size)--, (*buffer)++;
2196 		*buf = *buffer;
2197 	}
2198 	return 0;
2199 }
2200 
2201 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2202 				 int *valp,
2203 				 int write, void *data)
2204 {
2205 	if (write) {
2206 		if (*negp) {
2207 			if (*lvalp > (unsigned long) INT_MAX + 1)
2208 				return -EINVAL;
2209 			*valp = -*lvalp;
2210 		} else {
2211 			if (*lvalp > (unsigned long) INT_MAX)
2212 				return -EINVAL;
2213 			*valp = *lvalp;
2214 		}
2215 	} else {
2216 		int val = *valp;
2217 		if (val < 0) {
2218 			*negp = true;
2219 			*lvalp = -(unsigned long)val;
2220 		} else {
2221 			*negp = false;
2222 			*lvalp = (unsigned long)val;
2223 		}
2224 	}
2225 	return 0;
2226 }
2227 
2228 static int do_proc_douintvec_conv(unsigned long *lvalp,
2229 				  unsigned int *valp,
2230 				  int write, void *data)
2231 {
2232 	if (write) {
2233 		if (*lvalp > UINT_MAX)
2234 			return -EINVAL;
2235 		*valp = *lvalp;
2236 	} else {
2237 		unsigned int val = *valp;
2238 		*lvalp = (unsigned long)val;
2239 	}
2240 	return 0;
2241 }
2242 
2243 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2244 
2245 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2246 		  int write, void __user *buffer,
2247 		  size_t *lenp, loff_t *ppos,
2248 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2249 			      int write, void *data),
2250 		  void *data)
2251 {
2252 	int *i, vleft, first = 1, err = 0;
2253 	size_t left;
2254 	char *kbuf = NULL, *p;
2255 
2256 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2257 		*lenp = 0;
2258 		return 0;
2259 	}
2260 
2261 	i = (int *) tbl_data;
2262 	vleft = table->maxlen / sizeof(*i);
2263 	left = *lenp;
2264 
2265 	if (!conv)
2266 		conv = do_proc_dointvec_conv;
2267 
2268 	if (write) {
2269 		if (proc_first_pos_non_zero_ignore(ppos, table))
2270 			goto out;
2271 
2272 		if (left > PAGE_SIZE - 1)
2273 			left = PAGE_SIZE - 1;
2274 		p = kbuf = memdup_user_nul(buffer, left);
2275 		if (IS_ERR(kbuf))
2276 			return PTR_ERR(kbuf);
2277 	}
2278 
2279 	for (; left && vleft--; i++, first=0) {
2280 		unsigned long lval;
2281 		bool neg;
2282 
2283 		if (write) {
2284 			left -= proc_skip_spaces(&p);
2285 
2286 			if (!left)
2287 				break;
2288 			err = proc_get_long(&p, &left, &lval, &neg,
2289 					     proc_wspace_sep,
2290 					     sizeof(proc_wspace_sep), NULL);
2291 			if (err)
2292 				break;
2293 			if (conv(&neg, &lval, i, 1, data)) {
2294 				err = -EINVAL;
2295 				break;
2296 			}
2297 		} else {
2298 			if (conv(&neg, &lval, i, 0, data)) {
2299 				err = -EINVAL;
2300 				break;
2301 			}
2302 			if (!first)
2303 				err = proc_put_char(&buffer, &left, '\t');
2304 			if (err)
2305 				break;
2306 			err = proc_put_long(&buffer, &left, lval, neg);
2307 			if (err)
2308 				break;
2309 		}
2310 	}
2311 
2312 	if (!write && !first && left && !err)
2313 		err = proc_put_char(&buffer, &left, '\n');
2314 	if (write && !err && left)
2315 		left -= proc_skip_spaces(&p);
2316 	if (write) {
2317 		kfree(kbuf);
2318 		if (first)
2319 			return err ? : -EINVAL;
2320 	}
2321 	*lenp -= left;
2322 out:
2323 	*ppos += *lenp;
2324 	return err;
2325 }
2326 
2327 static int do_proc_dointvec(struct ctl_table *table, int write,
2328 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2329 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2330 			      int write, void *data),
2331 		  void *data)
2332 {
2333 	return __do_proc_dointvec(table->data, table, write,
2334 			buffer, lenp, ppos, conv, data);
2335 }
2336 
2337 static int do_proc_douintvec_w(unsigned int *tbl_data,
2338 			       struct ctl_table *table,
2339 			       void __user *buffer,
2340 			       size_t *lenp, loff_t *ppos,
2341 			       int (*conv)(unsigned long *lvalp,
2342 					   unsigned int *valp,
2343 					   int write, void *data),
2344 			       void *data)
2345 {
2346 	unsigned long lval;
2347 	int err = 0;
2348 	size_t left;
2349 	bool neg;
2350 	char *kbuf = NULL, *p;
2351 
2352 	left = *lenp;
2353 
2354 	if (proc_first_pos_non_zero_ignore(ppos, table))
2355 		goto bail_early;
2356 
2357 	if (left > PAGE_SIZE - 1)
2358 		left = PAGE_SIZE - 1;
2359 
2360 	p = kbuf = memdup_user_nul(buffer, left);
2361 	if (IS_ERR(kbuf))
2362 		return -EINVAL;
2363 
2364 	left -= proc_skip_spaces(&p);
2365 	if (!left) {
2366 		err = -EINVAL;
2367 		goto out_free;
2368 	}
2369 
2370 	err = proc_get_long(&p, &left, &lval, &neg,
2371 			     proc_wspace_sep,
2372 			     sizeof(proc_wspace_sep), NULL);
2373 	if (err || neg) {
2374 		err = -EINVAL;
2375 		goto out_free;
2376 	}
2377 
2378 	if (conv(&lval, tbl_data, 1, data)) {
2379 		err = -EINVAL;
2380 		goto out_free;
2381 	}
2382 
2383 	if (!err && left)
2384 		left -= proc_skip_spaces(&p);
2385 
2386 out_free:
2387 	kfree(kbuf);
2388 	if (err)
2389 		return -EINVAL;
2390 
2391 	return 0;
2392 
2393 	/* This is in keeping with old __do_proc_dointvec() */
2394 bail_early:
2395 	*ppos += *lenp;
2396 	return err;
2397 }
2398 
2399 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2400 			       size_t *lenp, loff_t *ppos,
2401 			       int (*conv)(unsigned long *lvalp,
2402 					   unsigned int *valp,
2403 					   int write, void *data),
2404 			       void *data)
2405 {
2406 	unsigned long lval;
2407 	int err = 0;
2408 	size_t left;
2409 
2410 	left = *lenp;
2411 
2412 	if (conv(&lval, tbl_data, 0, data)) {
2413 		err = -EINVAL;
2414 		goto out;
2415 	}
2416 
2417 	err = proc_put_long(&buffer, &left, lval, false);
2418 	if (err || !left)
2419 		goto out;
2420 
2421 	err = proc_put_char(&buffer, &left, '\n');
2422 
2423 out:
2424 	*lenp -= left;
2425 	*ppos += *lenp;
2426 
2427 	return err;
2428 }
2429 
2430 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2431 			       int write, void __user *buffer,
2432 			       size_t *lenp, loff_t *ppos,
2433 			       int (*conv)(unsigned long *lvalp,
2434 					   unsigned int *valp,
2435 					   int write, void *data),
2436 			       void *data)
2437 {
2438 	unsigned int *i, vleft;
2439 
2440 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2441 		*lenp = 0;
2442 		return 0;
2443 	}
2444 
2445 	i = (unsigned int *) tbl_data;
2446 	vleft = table->maxlen / sizeof(*i);
2447 
2448 	/*
2449 	 * Arrays are not supported, keep this simple. *Do not* add
2450 	 * support for them.
2451 	 */
2452 	if (vleft != 1) {
2453 		*lenp = 0;
2454 		return -EINVAL;
2455 	}
2456 
2457 	if (!conv)
2458 		conv = do_proc_douintvec_conv;
2459 
2460 	if (write)
2461 		return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2462 					   conv, data);
2463 	return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2464 }
2465 
2466 static int do_proc_douintvec(struct ctl_table *table, int write,
2467 			     void __user *buffer, size_t *lenp, loff_t *ppos,
2468 			     int (*conv)(unsigned long *lvalp,
2469 					 unsigned int *valp,
2470 					 int write, void *data),
2471 			     void *data)
2472 {
2473 	return __do_proc_douintvec(table->data, table, write,
2474 				   buffer, lenp, ppos, conv, data);
2475 }
2476 
2477 /**
2478  * proc_dointvec - read a vector of integers
2479  * @table: the sysctl table
2480  * @write: %TRUE if this is a write to the sysctl file
2481  * @buffer: the user buffer
2482  * @lenp: the size of the user buffer
2483  * @ppos: file position
2484  *
2485  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2486  * values from/to the user buffer, treated as an ASCII string.
2487  *
2488  * Returns 0 on success.
2489  */
2490 int proc_dointvec(struct ctl_table *table, int write,
2491 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2492 {
2493 	return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2494 }
2495 
2496 /**
2497  * proc_douintvec - read a vector of unsigned integers
2498  * @table: the sysctl table
2499  * @write: %TRUE if this is a write to the sysctl file
2500  * @buffer: the user buffer
2501  * @lenp: the size of the user buffer
2502  * @ppos: file position
2503  *
2504  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2505  * values from/to the user buffer, treated as an ASCII string.
2506  *
2507  * Returns 0 on success.
2508  */
2509 int proc_douintvec(struct ctl_table *table, int write,
2510 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2511 {
2512 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
2513 				 do_proc_douintvec_conv, NULL);
2514 }
2515 
2516 /*
2517  * Taint values can only be increased
2518  * This means we can safely use a temporary.
2519  */
2520 static int proc_taint(struct ctl_table *table, int write,
2521 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2522 {
2523 	struct ctl_table t;
2524 	unsigned long tmptaint = get_taint();
2525 	int err;
2526 
2527 	if (write && !capable(CAP_SYS_ADMIN))
2528 		return -EPERM;
2529 
2530 	t = *table;
2531 	t.data = &tmptaint;
2532 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2533 	if (err < 0)
2534 		return err;
2535 
2536 	if (write) {
2537 		/*
2538 		 * Poor man's atomic or. Not worth adding a primitive
2539 		 * to everyone's atomic.h for this
2540 		 */
2541 		int i;
2542 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2543 			if ((tmptaint >> i) & 1)
2544 				add_taint(i, LOCKDEP_STILL_OK);
2545 		}
2546 	}
2547 
2548 	return err;
2549 }
2550 
2551 #ifdef CONFIG_PRINTK
2552 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2553 				void __user *buffer, size_t *lenp, loff_t *ppos)
2554 {
2555 	if (write && !capable(CAP_SYS_ADMIN))
2556 		return -EPERM;
2557 
2558 	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2559 }
2560 #endif
2561 
2562 /**
2563  * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
2564  * @min: pointer to minimum allowable value
2565  * @max: pointer to maximum allowable value
2566  *
2567  * The do_proc_dointvec_minmax_conv_param structure provides the
2568  * minimum and maximum values for doing range checking for those sysctl
2569  * parameters that use the proc_dointvec_minmax() handler.
2570  */
2571 struct do_proc_dointvec_minmax_conv_param {
2572 	int *min;
2573 	int *max;
2574 };
2575 
2576 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2577 					int *valp,
2578 					int write, void *data)
2579 {
2580 	struct do_proc_dointvec_minmax_conv_param *param = data;
2581 	if (write) {
2582 		int val = *negp ? -*lvalp : *lvalp;
2583 		if ((param->min && *param->min > val) ||
2584 		    (param->max && *param->max < val))
2585 			return -EINVAL;
2586 		*valp = val;
2587 	} else {
2588 		int val = *valp;
2589 		if (val < 0) {
2590 			*negp = true;
2591 			*lvalp = -(unsigned long)val;
2592 		} else {
2593 			*negp = false;
2594 			*lvalp = (unsigned long)val;
2595 		}
2596 	}
2597 	return 0;
2598 }
2599 
2600 /**
2601  * proc_dointvec_minmax - read a vector of integers with min/max values
2602  * @table: the sysctl table
2603  * @write: %TRUE if this is a write to the sysctl file
2604  * @buffer: the user buffer
2605  * @lenp: the size of the user buffer
2606  * @ppos: file position
2607  *
2608  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2609  * values from/to the user buffer, treated as an ASCII string.
2610  *
2611  * This routine will ensure the values are within the range specified by
2612  * table->extra1 (min) and table->extra2 (max).
2613  *
2614  * Returns 0 on success or -EINVAL on write when the range check fails.
2615  */
2616 int proc_dointvec_minmax(struct ctl_table *table, int write,
2617 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2618 {
2619 	struct do_proc_dointvec_minmax_conv_param param = {
2620 		.min = (int *) table->extra1,
2621 		.max = (int *) table->extra2,
2622 	};
2623 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2624 				do_proc_dointvec_minmax_conv, &param);
2625 }
2626 
2627 /**
2628  * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
2629  * @min: pointer to minimum allowable value
2630  * @max: pointer to maximum allowable value
2631  *
2632  * The do_proc_douintvec_minmax_conv_param structure provides the
2633  * minimum and maximum values for doing range checking for those sysctl
2634  * parameters that use the proc_douintvec_minmax() handler.
2635  */
2636 struct do_proc_douintvec_minmax_conv_param {
2637 	unsigned int *min;
2638 	unsigned int *max;
2639 };
2640 
2641 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2642 					 unsigned int *valp,
2643 					 int write, void *data)
2644 {
2645 	struct do_proc_douintvec_minmax_conv_param *param = data;
2646 
2647 	if (write) {
2648 		unsigned int val = *lvalp;
2649 
2650 		if (*lvalp > UINT_MAX)
2651 			return -EINVAL;
2652 
2653 		if ((param->min && *param->min > val) ||
2654 		    (param->max && *param->max < val))
2655 			return -ERANGE;
2656 
2657 		*valp = val;
2658 	} else {
2659 		unsigned int val = *valp;
2660 		*lvalp = (unsigned long) val;
2661 	}
2662 
2663 	return 0;
2664 }
2665 
2666 /**
2667  * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2668  * @table: the sysctl table
2669  * @write: %TRUE if this is a write to the sysctl file
2670  * @buffer: the user buffer
2671  * @lenp: the size of the user buffer
2672  * @ppos: file position
2673  *
2674  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2675  * values from/to the user buffer, treated as an ASCII string. Negative
2676  * strings are not allowed.
2677  *
2678  * This routine will ensure the values are within the range specified by
2679  * table->extra1 (min) and table->extra2 (max). There is a final sanity
2680  * check for UINT_MAX to avoid having to support wrap around uses from
2681  * userspace.
2682  *
2683  * Returns 0 on success or -ERANGE on write when the range check fails.
2684  */
2685 int proc_douintvec_minmax(struct ctl_table *table, int write,
2686 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2687 {
2688 	struct do_proc_douintvec_minmax_conv_param param = {
2689 		.min = (unsigned int *) table->extra1,
2690 		.max = (unsigned int *) table->extra2,
2691 	};
2692 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
2693 				 do_proc_douintvec_minmax_conv, &param);
2694 }
2695 
2696 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
2697 					unsigned int *valp,
2698 					int write, void *data)
2699 {
2700 	if (write) {
2701 		unsigned int val;
2702 
2703 		val = round_pipe_size(*lvalp);
2704 		if (val == 0)
2705 			return -EINVAL;
2706 
2707 		*valp = val;
2708 	} else {
2709 		unsigned int val = *valp;
2710 		*lvalp = (unsigned long) val;
2711 	}
2712 
2713 	return 0;
2714 }
2715 
2716 static int proc_dopipe_max_size(struct ctl_table *table, int write,
2717 				void __user *buffer, size_t *lenp, loff_t *ppos)
2718 {
2719 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
2720 				 do_proc_dopipe_max_size_conv, NULL);
2721 }
2722 
2723 static void validate_coredump_safety(void)
2724 {
2725 #ifdef CONFIG_COREDUMP
2726 	if (suid_dumpable == SUID_DUMP_ROOT &&
2727 	    core_pattern[0] != '/' && core_pattern[0] != '|') {
2728 		printk(KERN_WARNING
2729 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2730 "Pipe handler or fully qualified core dump path required.\n"
2731 "Set kernel.core_pattern before fs.suid_dumpable.\n"
2732 		);
2733 	}
2734 #endif
2735 }
2736 
2737 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2738 		void __user *buffer, size_t *lenp, loff_t *ppos)
2739 {
2740 	int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2741 	if (!error)
2742 		validate_coredump_safety();
2743 	return error;
2744 }
2745 
2746 #ifdef CONFIG_COREDUMP
2747 static int proc_dostring_coredump(struct ctl_table *table, int write,
2748 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2749 {
2750 	int error = proc_dostring(table, write, buffer, lenp, ppos);
2751 	if (!error)
2752 		validate_coredump_safety();
2753 	return error;
2754 }
2755 #endif
2756 
2757 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2758 				     void __user *buffer,
2759 				     size_t *lenp, loff_t *ppos,
2760 				     unsigned long convmul,
2761 				     unsigned long convdiv)
2762 {
2763 	unsigned long *i, *min, *max;
2764 	int vleft, first = 1, err = 0;
2765 	size_t left;
2766 	char *kbuf = NULL, *p;
2767 
2768 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2769 		*lenp = 0;
2770 		return 0;
2771 	}
2772 
2773 	i = (unsigned long *) data;
2774 	min = (unsigned long *) table->extra1;
2775 	max = (unsigned long *) table->extra2;
2776 	vleft = table->maxlen / sizeof(unsigned long);
2777 	left = *lenp;
2778 
2779 	if (write) {
2780 		if (proc_first_pos_non_zero_ignore(ppos, table))
2781 			goto out;
2782 
2783 		if (left > PAGE_SIZE - 1)
2784 			left = PAGE_SIZE - 1;
2785 		p = kbuf = memdup_user_nul(buffer, left);
2786 		if (IS_ERR(kbuf))
2787 			return PTR_ERR(kbuf);
2788 	}
2789 
2790 	for (; left && vleft--; i++, first = 0) {
2791 		unsigned long val;
2792 
2793 		if (write) {
2794 			bool neg;
2795 
2796 			left -= proc_skip_spaces(&p);
2797 			if (!left)
2798 				break;
2799 
2800 			err = proc_get_long(&p, &left, &val, &neg,
2801 					     proc_wspace_sep,
2802 					     sizeof(proc_wspace_sep), NULL);
2803 			if (err)
2804 				break;
2805 			if (neg)
2806 				continue;
2807 			val = convmul * val / convdiv;
2808 			if ((min && val < *min) || (max && val > *max))
2809 				continue;
2810 			*i = val;
2811 		} else {
2812 			val = convdiv * (*i) / convmul;
2813 			if (!first) {
2814 				err = proc_put_char(&buffer, &left, '\t');
2815 				if (err)
2816 					break;
2817 			}
2818 			err = proc_put_long(&buffer, &left, val, false);
2819 			if (err)
2820 				break;
2821 		}
2822 	}
2823 
2824 	if (!write && !first && left && !err)
2825 		err = proc_put_char(&buffer, &left, '\n');
2826 	if (write && !err)
2827 		left -= proc_skip_spaces(&p);
2828 	if (write) {
2829 		kfree(kbuf);
2830 		if (first)
2831 			return err ? : -EINVAL;
2832 	}
2833 	*lenp -= left;
2834 out:
2835 	*ppos += *lenp;
2836 	return err;
2837 }
2838 
2839 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2840 				     void __user *buffer,
2841 				     size_t *lenp, loff_t *ppos,
2842 				     unsigned long convmul,
2843 				     unsigned long convdiv)
2844 {
2845 	return __do_proc_doulongvec_minmax(table->data, table, write,
2846 			buffer, lenp, ppos, convmul, convdiv);
2847 }
2848 
2849 /**
2850  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2851  * @table: the sysctl table
2852  * @write: %TRUE if this is a write to the sysctl file
2853  * @buffer: the user buffer
2854  * @lenp: the size of the user buffer
2855  * @ppos: file position
2856  *
2857  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2858  * values from/to the user buffer, treated as an ASCII string.
2859  *
2860  * This routine will ensure the values are within the range specified by
2861  * table->extra1 (min) and table->extra2 (max).
2862  *
2863  * Returns 0 on success.
2864  */
2865 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2866 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2867 {
2868     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2869 }
2870 
2871 /**
2872  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2873  * @table: the sysctl table
2874  * @write: %TRUE if this is a write to the sysctl file
2875  * @buffer: the user buffer
2876  * @lenp: the size of the user buffer
2877  * @ppos: file position
2878  *
2879  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2880  * values from/to the user buffer, treated as an ASCII string. The values
2881  * are treated as milliseconds, and converted to jiffies when they are stored.
2882  *
2883  * This routine will ensure the values are within the range specified by
2884  * table->extra1 (min) and table->extra2 (max).
2885  *
2886  * Returns 0 on success.
2887  */
2888 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2889 				      void __user *buffer,
2890 				      size_t *lenp, loff_t *ppos)
2891 {
2892     return do_proc_doulongvec_minmax(table, write, buffer,
2893 				     lenp, ppos, HZ, 1000l);
2894 }
2895 
2896 
2897 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2898 					 int *valp,
2899 					 int write, void *data)
2900 {
2901 	if (write) {
2902 		if (*lvalp > INT_MAX / HZ)
2903 			return 1;
2904 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2905 	} else {
2906 		int val = *valp;
2907 		unsigned long lval;
2908 		if (val < 0) {
2909 			*negp = true;
2910 			lval = -(unsigned long)val;
2911 		} else {
2912 			*negp = false;
2913 			lval = (unsigned long)val;
2914 		}
2915 		*lvalp = lval / HZ;
2916 	}
2917 	return 0;
2918 }
2919 
2920 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2921 						int *valp,
2922 						int write, void *data)
2923 {
2924 	if (write) {
2925 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2926 			return 1;
2927 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2928 	} else {
2929 		int val = *valp;
2930 		unsigned long lval;
2931 		if (val < 0) {
2932 			*negp = true;
2933 			lval = -(unsigned long)val;
2934 		} else {
2935 			*negp = false;
2936 			lval = (unsigned long)val;
2937 		}
2938 		*lvalp = jiffies_to_clock_t(lval);
2939 	}
2940 	return 0;
2941 }
2942 
2943 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2944 					    int *valp,
2945 					    int write, void *data)
2946 {
2947 	if (write) {
2948 		unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2949 
2950 		if (jif > INT_MAX)
2951 			return 1;
2952 		*valp = (int)jif;
2953 	} else {
2954 		int val = *valp;
2955 		unsigned long lval;
2956 		if (val < 0) {
2957 			*negp = true;
2958 			lval = -(unsigned long)val;
2959 		} else {
2960 			*negp = false;
2961 			lval = (unsigned long)val;
2962 		}
2963 		*lvalp = jiffies_to_msecs(lval);
2964 	}
2965 	return 0;
2966 }
2967 
2968 /**
2969  * proc_dointvec_jiffies - read a vector of integers as seconds
2970  * @table: the sysctl table
2971  * @write: %TRUE if this is a write to the sysctl file
2972  * @buffer: the user buffer
2973  * @lenp: the size of the user buffer
2974  * @ppos: file position
2975  *
2976  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2977  * values from/to the user buffer, treated as an ASCII string.
2978  * The values read are assumed to be in seconds, and are converted into
2979  * jiffies.
2980  *
2981  * Returns 0 on success.
2982  */
2983 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2984 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2985 {
2986     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2987 		    	    do_proc_dointvec_jiffies_conv,NULL);
2988 }
2989 
2990 /**
2991  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2992  * @table: the sysctl table
2993  * @write: %TRUE if this is a write to the sysctl file
2994  * @buffer: the user buffer
2995  * @lenp: the size of the user buffer
2996  * @ppos: pointer to the file position
2997  *
2998  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2999  * values from/to the user buffer, treated as an ASCII string.
3000  * The values read are assumed to be in 1/USER_HZ seconds, and
3001  * are converted into jiffies.
3002  *
3003  * Returns 0 on success.
3004  */
3005 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3006 				 void __user *buffer, size_t *lenp, loff_t *ppos)
3007 {
3008     return do_proc_dointvec(table,write,buffer,lenp,ppos,
3009 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
3010 }
3011 
3012 /**
3013  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
3014  * @table: the sysctl table
3015  * @write: %TRUE if this is a write to the sysctl file
3016  * @buffer: the user buffer
3017  * @lenp: the size of the user buffer
3018  * @ppos: file position
3019  * @ppos: the current position in the file
3020  *
3021  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3022  * values from/to the user buffer, treated as an ASCII string.
3023  * The values read are assumed to be in 1/1000 seconds, and
3024  * are converted into jiffies.
3025  *
3026  * Returns 0 on success.
3027  */
3028 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3029 			     void __user *buffer, size_t *lenp, loff_t *ppos)
3030 {
3031 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
3032 				do_proc_dointvec_ms_jiffies_conv, NULL);
3033 }
3034 
3035 static int proc_do_cad_pid(struct ctl_table *table, int write,
3036 			   void __user *buffer, size_t *lenp, loff_t *ppos)
3037 {
3038 	struct pid *new_pid;
3039 	pid_t tmp;
3040 	int r;
3041 
3042 	tmp = pid_vnr(cad_pid);
3043 
3044 	r = __do_proc_dointvec(&tmp, table, write, buffer,
3045 			       lenp, ppos, NULL, NULL);
3046 	if (r || !write)
3047 		return r;
3048 
3049 	new_pid = find_get_pid(tmp);
3050 	if (!new_pid)
3051 		return -ESRCH;
3052 
3053 	put_pid(xchg(&cad_pid, new_pid));
3054 	return 0;
3055 }
3056 
3057 /**
3058  * proc_do_large_bitmap - read/write from/to a large bitmap
3059  * @table: the sysctl table
3060  * @write: %TRUE if this is a write to the sysctl file
3061  * @buffer: the user buffer
3062  * @lenp: the size of the user buffer
3063  * @ppos: file position
3064  *
3065  * The bitmap is stored at table->data and the bitmap length (in bits)
3066  * in table->maxlen.
3067  *
3068  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
3069  * large bitmaps may be represented in a compact manner. Writing into
3070  * the file will clear the bitmap then update it with the given input.
3071  *
3072  * Returns 0 on success.
3073  */
3074 int proc_do_large_bitmap(struct ctl_table *table, int write,
3075 			 void __user *buffer, size_t *lenp, loff_t *ppos)
3076 {
3077 	int err = 0;
3078 	bool first = 1;
3079 	size_t left = *lenp;
3080 	unsigned long bitmap_len = table->maxlen;
3081 	unsigned long *bitmap = *(unsigned long **) table->data;
3082 	unsigned long *tmp_bitmap = NULL;
3083 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
3084 
3085 	if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
3086 		*lenp = 0;
3087 		return 0;
3088 	}
3089 
3090 	if (write) {
3091 		char *kbuf, *p;
3092 
3093 		if (left > PAGE_SIZE - 1)
3094 			left = PAGE_SIZE - 1;
3095 
3096 		p = kbuf = memdup_user_nul(buffer, left);
3097 		if (IS_ERR(kbuf))
3098 			return PTR_ERR(kbuf);
3099 
3100 		tmp_bitmap = kcalloc(BITS_TO_LONGS(bitmap_len),
3101 				     sizeof(unsigned long),
3102 				     GFP_KERNEL);
3103 		if (!tmp_bitmap) {
3104 			kfree(kbuf);
3105 			return -ENOMEM;
3106 		}
3107 		proc_skip_char(&p, &left, '\n');
3108 		while (!err && left) {
3109 			unsigned long val_a, val_b;
3110 			bool neg;
3111 
3112 			err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3113 					     sizeof(tr_a), &c);
3114 			if (err)
3115 				break;
3116 			if (val_a >= bitmap_len || neg) {
3117 				err = -EINVAL;
3118 				break;
3119 			}
3120 
3121 			val_b = val_a;
3122 			if (left) {
3123 				p++;
3124 				left--;
3125 			}
3126 
3127 			if (c == '-') {
3128 				err = proc_get_long(&p, &left, &val_b,
3129 						     &neg, tr_b, sizeof(tr_b),
3130 						     &c);
3131 				if (err)
3132 					break;
3133 				if (val_b >= bitmap_len || neg ||
3134 				    val_a > val_b) {
3135 					err = -EINVAL;
3136 					break;
3137 				}
3138 				if (left) {
3139 					p++;
3140 					left--;
3141 				}
3142 			}
3143 
3144 			bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3145 			first = 0;
3146 			proc_skip_char(&p, &left, '\n');
3147 		}
3148 		kfree(kbuf);
3149 	} else {
3150 		unsigned long bit_a, bit_b = 0;
3151 
3152 		while (left) {
3153 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3154 			if (bit_a >= bitmap_len)
3155 				break;
3156 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
3157 						   bit_a + 1) - 1;
3158 
3159 			if (!first) {
3160 				err = proc_put_char(&buffer, &left, ',');
3161 				if (err)
3162 					break;
3163 			}
3164 			err = proc_put_long(&buffer, &left, bit_a, false);
3165 			if (err)
3166 				break;
3167 			if (bit_a != bit_b) {
3168 				err = proc_put_char(&buffer, &left, '-');
3169 				if (err)
3170 					break;
3171 				err = proc_put_long(&buffer, &left, bit_b, false);
3172 				if (err)
3173 					break;
3174 			}
3175 
3176 			first = 0; bit_b++;
3177 		}
3178 		if (!err)
3179 			err = proc_put_char(&buffer, &left, '\n');
3180 	}
3181 
3182 	if (!err) {
3183 		if (write) {
3184 			if (*ppos)
3185 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3186 			else
3187 				bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3188 		}
3189 		*lenp -= left;
3190 		*ppos += *lenp;
3191 	}
3192 
3193 	kfree(tmp_bitmap);
3194 	return err;
3195 }
3196 
3197 #else /* CONFIG_PROC_SYSCTL */
3198 
3199 int proc_dostring(struct ctl_table *table, int write,
3200 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3201 {
3202 	return -ENOSYS;
3203 }
3204 
3205 int proc_dointvec(struct ctl_table *table, int write,
3206 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3207 {
3208 	return -ENOSYS;
3209 }
3210 
3211 int proc_douintvec(struct ctl_table *table, int write,
3212 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3213 {
3214 	return -ENOSYS;
3215 }
3216 
3217 int proc_dointvec_minmax(struct ctl_table *table, int write,
3218 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3219 {
3220 	return -ENOSYS;
3221 }
3222 
3223 int proc_douintvec_minmax(struct ctl_table *table, int write,
3224 			  void __user *buffer, size_t *lenp, loff_t *ppos)
3225 {
3226 	return -ENOSYS;
3227 }
3228 
3229 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3230 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3231 {
3232 	return -ENOSYS;
3233 }
3234 
3235 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3236 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3237 {
3238 	return -ENOSYS;
3239 }
3240 
3241 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3242 			     void __user *buffer, size_t *lenp, loff_t *ppos)
3243 {
3244 	return -ENOSYS;
3245 }
3246 
3247 int proc_doulongvec_minmax(struct ctl_table *table, int write,
3248 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3249 {
3250 	return -ENOSYS;
3251 }
3252 
3253 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3254 				      void __user *buffer,
3255 				      size_t *lenp, loff_t *ppos)
3256 {
3257     return -ENOSYS;
3258 }
3259 
3260 
3261 #endif /* CONFIG_PROC_SYSCTL */
3262 
3263 /*
3264  * No sense putting this after each symbol definition, twice,
3265  * exception granted :-)
3266  */
3267 EXPORT_SYMBOL(proc_dointvec);
3268 EXPORT_SYMBOL(proc_douintvec);
3269 EXPORT_SYMBOL(proc_dointvec_jiffies);
3270 EXPORT_SYMBOL(proc_dointvec_minmax);
3271 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3272 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3273 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3274 EXPORT_SYMBOL(proc_dostring);
3275 EXPORT_SYMBOL(proc_doulongvec_minmax);
3276 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3277