xref: /linux/kernel/sysctl.c (revision e25883411e80f314f0f76c8eeacd1b47d7e4805b)
1 /*
2  * sysctl.c: General linux system control interface
3  *
4  * Begun 24 March 1995, Stephen Tweedie
5  * Added /proc support, Dec 1995
6  * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7  * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8  * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9  * Dynamic registration fixes, Stephen Tweedie.
10  * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11  * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12  *  Horn.
13  * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14  * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15  * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16  *  Wendling.
17  * The list_for_each() macro wasn't appropriate for the sysctl loop.
18  *  Removed it and replaced it with older style, 03/23/00, Bill Wendling
19  */
20 
21 #include <linux/module.h>
22 #include <linux/aio.h>
23 #include <linux/mm.h>
24 #include <linux/swap.h>
25 #include <linux/slab.h>
26 #include <linux/sysctl.h>
27 #include <linux/bitmap.h>
28 #include <linux/signal.h>
29 #include <linux/printk.h>
30 #include <linux/proc_fs.h>
31 #include <linux/security.h>
32 #include <linux/ctype.h>
33 #include <linux/kmemcheck.h>
34 #include <linux/kmemleak.h>
35 #include <linux/fs.h>
36 #include <linux/init.h>
37 #include <linux/kernel.h>
38 #include <linux/kobject.h>
39 #include <linux/net.h>
40 #include <linux/sysrq.h>
41 #include <linux/highuid.h>
42 #include <linux/writeback.h>
43 #include <linux/ratelimit.h>
44 #include <linux/compaction.h>
45 #include <linux/hugetlb.h>
46 #include <linux/initrd.h>
47 #include <linux/key.h>
48 #include <linux/times.h>
49 #include <linux/limits.h>
50 #include <linux/dcache.h>
51 #include <linux/dnotify.h>
52 #include <linux/syscalls.h>
53 #include <linux/vmstat.h>
54 #include <linux/nfs_fs.h>
55 #include <linux/acpi.h>
56 #include <linux/reboot.h>
57 #include <linux/ftrace.h>
58 #include <linux/perf_event.h>
59 #include <linux/kprobes.h>
60 #include <linux/pipe_fs_i.h>
61 #include <linux/oom.h>
62 #include <linux/kmod.h>
63 #include <linux/capability.h>
64 #include <linux/binfmts.h>
65 #include <linux/sched/sysctl.h>
66 #include <linux/sched/coredump.h>
67 #include <linux/kexec.h>
68 #include <linux/bpf.h>
69 #include <linux/mount.h>
70 
71 #include <linux/uaccess.h>
72 #include <asm/processor.h>
73 
74 #ifdef CONFIG_X86
75 #include <asm/nmi.h>
76 #include <asm/stacktrace.h>
77 #include <asm/io.h>
78 #endif
79 #ifdef CONFIG_SPARC
80 #include <asm/setup.h>
81 #endif
82 #ifdef CONFIG_BSD_PROCESS_ACCT
83 #include <linux/acct.h>
84 #endif
85 #ifdef CONFIG_RT_MUTEXES
86 #include <linux/rtmutex.h>
87 #endif
88 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
89 #include <linux/lockdep.h>
90 #endif
91 #ifdef CONFIG_CHR_DEV_SG
92 #include <scsi/sg.h>
93 #endif
94 
95 #ifdef CONFIG_LOCKUP_DETECTOR
96 #include <linux/nmi.h>
97 #endif
98 
99 #if defined(CONFIG_SYSCTL)
100 
101 /* External variables not in a header file. */
102 extern int suid_dumpable;
103 #ifdef CONFIG_COREDUMP
104 extern int core_uses_pid;
105 extern char core_pattern[];
106 extern unsigned int core_pipe_limit;
107 #endif
108 extern int pid_max;
109 extern int pid_max_min, pid_max_max;
110 extern int percpu_pagelist_fraction;
111 extern int latencytop_enabled;
112 extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
113 #ifndef CONFIG_MMU
114 extern int sysctl_nr_trim_pages;
115 #endif
116 
117 /* Constants used for minimum and  maximum */
118 #ifdef CONFIG_LOCKUP_DETECTOR
119 static int sixty = 60;
120 #endif
121 
122 static int __maybe_unused neg_one = -1;
123 
124 static int zero;
125 static int __maybe_unused one = 1;
126 static int __maybe_unused two = 2;
127 static int __maybe_unused four = 4;
128 static unsigned long one_ul = 1;
129 static int one_hundred = 100;
130 static int one_thousand = 1000;
131 #ifdef CONFIG_PRINTK
132 static int ten_thousand = 10000;
133 #endif
134 #ifdef CONFIG_PERF_EVENTS
135 static int six_hundred_forty_kb = 640 * 1024;
136 #endif
137 
138 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
139 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
140 
141 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
142 static int maxolduid = 65535;
143 static int minolduid;
144 
145 static int ngroups_max = NGROUPS_MAX;
146 static const int cap_last_cap = CAP_LAST_CAP;
147 
148 /*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */
149 #ifdef CONFIG_DETECT_HUNG_TASK
150 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
151 #endif
152 
153 #ifdef CONFIG_INOTIFY_USER
154 #include <linux/inotify.h>
155 #endif
156 #ifdef CONFIG_SPARC
157 #endif
158 
159 #ifdef __hppa__
160 extern int pwrsw_enabled;
161 #endif
162 
163 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
164 extern int unaligned_enabled;
165 #endif
166 
167 #ifdef CONFIG_IA64
168 extern int unaligned_dump_stack;
169 #endif
170 
171 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
172 extern int no_unaligned_warning;
173 #endif
174 
175 #ifdef CONFIG_PROC_SYSCTL
176 
177 /**
178  * enum sysctl_writes_mode - supported sysctl write modes
179  *
180  * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
181  * 	to be written, and multiple writes on the same sysctl file descriptor
182  * 	will rewrite the sysctl value, regardless of file position. No warning
183  * 	is issued when the initial position is not 0.
184  * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
185  * 	not 0.
186  * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
187  * 	file position 0 and the value must be fully contained in the buffer
188  * 	sent to the write syscall. If dealing with strings respect the file
189  * 	position, but restrict this to the max length of the buffer, anything
190  * 	passed the max lenght will be ignored. Multiple writes will append
191  * 	to the buffer.
192  *
193  * These write modes control how current file position affects the behavior of
194  * updating sysctl values through the proc interface on each write.
195  */
196 enum sysctl_writes_mode {
197 	SYSCTL_WRITES_LEGACY		= -1,
198 	SYSCTL_WRITES_WARN		= 0,
199 	SYSCTL_WRITES_STRICT		= 1,
200 };
201 
202 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
203 
204 static int proc_do_cad_pid(struct ctl_table *table, int write,
205 		  void __user *buffer, size_t *lenp, loff_t *ppos);
206 static int proc_taint(struct ctl_table *table, int write,
207 			       void __user *buffer, size_t *lenp, loff_t *ppos);
208 #endif
209 
210 #ifdef CONFIG_PRINTK
211 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
212 				void __user *buffer, size_t *lenp, loff_t *ppos);
213 #endif
214 
215 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
216 		void __user *buffer, size_t *lenp, loff_t *ppos);
217 #ifdef CONFIG_COREDUMP
218 static int proc_dostring_coredump(struct ctl_table *table, int write,
219 		void __user *buffer, size_t *lenp, loff_t *ppos);
220 #endif
221 
222 #ifdef CONFIG_MAGIC_SYSRQ
223 /* Note: sysrq code uses it's own private copy */
224 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
225 
226 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
227 				void __user *buffer, size_t *lenp,
228 				loff_t *ppos)
229 {
230 	int error;
231 
232 	error = proc_dointvec(table, write, buffer, lenp, ppos);
233 	if (error)
234 		return error;
235 
236 	if (write)
237 		sysrq_toggle_support(__sysrq_enabled);
238 
239 	return 0;
240 }
241 
242 #endif
243 
244 static struct ctl_table kern_table[];
245 static struct ctl_table vm_table[];
246 static struct ctl_table fs_table[];
247 static struct ctl_table debug_table[];
248 static struct ctl_table dev_table[];
249 extern struct ctl_table random_table[];
250 #ifdef CONFIG_EPOLL
251 extern struct ctl_table epoll_table[];
252 #endif
253 
254 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
255 int sysctl_legacy_va_layout;
256 #endif
257 
258 /* The default sysctl tables: */
259 
260 static struct ctl_table sysctl_base_table[] = {
261 	{
262 		.procname	= "kernel",
263 		.mode		= 0555,
264 		.child		= kern_table,
265 	},
266 	{
267 		.procname	= "vm",
268 		.mode		= 0555,
269 		.child		= vm_table,
270 	},
271 	{
272 		.procname	= "fs",
273 		.mode		= 0555,
274 		.child		= fs_table,
275 	},
276 	{
277 		.procname	= "debug",
278 		.mode		= 0555,
279 		.child		= debug_table,
280 	},
281 	{
282 		.procname	= "dev",
283 		.mode		= 0555,
284 		.child		= dev_table,
285 	},
286 	{ }
287 };
288 
289 #ifdef CONFIG_SCHED_DEBUG
290 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
291 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
292 static int min_wakeup_granularity_ns;			/* 0 usecs */
293 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
294 #ifdef CONFIG_SMP
295 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
296 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
297 #endif /* CONFIG_SMP */
298 #endif /* CONFIG_SCHED_DEBUG */
299 
300 #ifdef CONFIG_COMPACTION
301 static int min_extfrag_threshold;
302 static int max_extfrag_threshold = 1000;
303 #endif
304 
305 static struct ctl_table kern_table[] = {
306 	{
307 		.procname	= "sched_child_runs_first",
308 		.data		= &sysctl_sched_child_runs_first,
309 		.maxlen		= sizeof(unsigned int),
310 		.mode		= 0644,
311 		.proc_handler	= proc_dointvec,
312 	},
313 #ifdef CONFIG_SCHED_DEBUG
314 	{
315 		.procname	= "sched_min_granularity_ns",
316 		.data		= &sysctl_sched_min_granularity,
317 		.maxlen		= sizeof(unsigned int),
318 		.mode		= 0644,
319 		.proc_handler	= sched_proc_update_handler,
320 		.extra1		= &min_sched_granularity_ns,
321 		.extra2		= &max_sched_granularity_ns,
322 	},
323 	{
324 		.procname	= "sched_latency_ns",
325 		.data		= &sysctl_sched_latency,
326 		.maxlen		= sizeof(unsigned int),
327 		.mode		= 0644,
328 		.proc_handler	= sched_proc_update_handler,
329 		.extra1		= &min_sched_granularity_ns,
330 		.extra2		= &max_sched_granularity_ns,
331 	},
332 	{
333 		.procname	= "sched_wakeup_granularity_ns",
334 		.data		= &sysctl_sched_wakeup_granularity,
335 		.maxlen		= sizeof(unsigned int),
336 		.mode		= 0644,
337 		.proc_handler	= sched_proc_update_handler,
338 		.extra1		= &min_wakeup_granularity_ns,
339 		.extra2		= &max_wakeup_granularity_ns,
340 	},
341 #ifdef CONFIG_SMP
342 	{
343 		.procname	= "sched_tunable_scaling",
344 		.data		= &sysctl_sched_tunable_scaling,
345 		.maxlen		= sizeof(enum sched_tunable_scaling),
346 		.mode		= 0644,
347 		.proc_handler	= sched_proc_update_handler,
348 		.extra1		= &min_sched_tunable_scaling,
349 		.extra2		= &max_sched_tunable_scaling,
350 	},
351 	{
352 		.procname	= "sched_migration_cost_ns",
353 		.data		= &sysctl_sched_migration_cost,
354 		.maxlen		= sizeof(unsigned int),
355 		.mode		= 0644,
356 		.proc_handler	= proc_dointvec,
357 	},
358 	{
359 		.procname	= "sched_nr_migrate",
360 		.data		= &sysctl_sched_nr_migrate,
361 		.maxlen		= sizeof(unsigned int),
362 		.mode		= 0644,
363 		.proc_handler	= proc_dointvec,
364 	},
365 	{
366 		.procname	= "sched_time_avg_ms",
367 		.data		= &sysctl_sched_time_avg,
368 		.maxlen		= sizeof(unsigned int),
369 		.mode		= 0644,
370 		.proc_handler	= proc_dointvec_minmax,
371 		.extra1		= &one,
372 	},
373 #ifdef CONFIG_SCHEDSTATS
374 	{
375 		.procname	= "sched_schedstats",
376 		.data		= NULL,
377 		.maxlen		= sizeof(unsigned int),
378 		.mode		= 0644,
379 		.proc_handler	= sysctl_schedstats,
380 		.extra1		= &zero,
381 		.extra2		= &one,
382 	},
383 #endif /* CONFIG_SCHEDSTATS */
384 #endif /* CONFIG_SMP */
385 #ifdef CONFIG_NUMA_BALANCING
386 	{
387 		.procname	= "numa_balancing_scan_delay_ms",
388 		.data		= &sysctl_numa_balancing_scan_delay,
389 		.maxlen		= sizeof(unsigned int),
390 		.mode		= 0644,
391 		.proc_handler	= proc_dointvec,
392 	},
393 	{
394 		.procname	= "numa_balancing_scan_period_min_ms",
395 		.data		= &sysctl_numa_balancing_scan_period_min,
396 		.maxlen		= sizeof(unsigned int),
397 		.mode		= 0644,
398 		.proc_handler	= proc_dointvec,
399 	},
400 	{
401 		.procname	= "numa_balancing_scan_period_max_ms",
402 		.data		= &sysctl_numa_balancing_scan_period_max,
403 		.maxlen		= sizeof(unsigned int),
404 		.mode		= 0644,
405 		.proc_handler	= proc_dointvec,
406 	},
407 	{
408 		.procname	= "numa_balancing_scan_size_mb",
409 		.data		= &sysctl_numa_balancing_scan_size,
410 		.maxlen		= sizeof(unsigned int),
411 		.mode		= 0644,
412 		.proc_handler	= proc_dointvec_minmax,
413 		.extra1		= &one,
414 	},
415 	{
416 		.procname	= "numa_balancing",
417 		.data		= NULL, /* filled in by handler */
418 		.maxlen		= sizeof(unsigned int),
419 		.mode		= 0644,
420 		.proc_handler	= sysctl_numa_balancing,
421 		.extra1		= &zero,
422 		.extra2		= &one,
423 	},
424 #endif /* CONFIG_NUMA_BALANCING */
425 #endif /* CONFIG_SCHED_DEBUG */
426 	{
427 		.procname	= "sched_rt_period_us",
428 		.data		= &sysctl_sched_rt_period,
429 		.maxlen		= sizeof(unsigned int),
430 		.mode		= 0644,
431 		.proc_handler	= sched_rt_handler,
432 	},
433 	{
434 		.procname	= "sched_rt_runtime_us",
435 		.data		= &sysctl_sched_rt_runtime,
436 		.maxlen		= sizeof(int),
437 		.mode		= 0644,
438 		.proc_handler	= sched_rt_handler,
439 	},
440 	{
441 		.procname	= "sched_rr_timeslice_ms",
442 		.data		= &sysctl_sched_rr_timeslice,
443 		.maxlen		= sizeof(int),
444 		.mode		= 0644,
445 		.proc_handler	= sched_rr_handler,
446 	},
447 #ifdef CONFIG_SCHED_AUTOGROUP
448 	{
449 		.procname	= "sched_autogroup_enabled",
450 		.data		= &sysctl_sched_autogroup_enabled,
451 		.maxlen		= sizeof(unsigned int),
452 		.mode		= 0644,
453 		.proc_handler	= proc_dointvec_minmax,
454 		.extra1		= &zero,
455 		.extra2		= &one,
456 	},
457 #endif
458 #ifdef CONFIG_CFS_BANDWIDTH
459 	{
460 		.procname	= "sched_cfs_bandwidth_slice_us",
461 		.data		= &sysctl_sched_cfs_bandwidth_slice,
462 		.maxlen		= sizeof(unsigned int),
463 		.mode		= 0644,
464 		.proc_handler	= proc_dointvec_minmax,
465 		.extra1		= &one,
466 	},
467 #endif
468 #ifdef CONFIG_PROVE_LOCKING
469 	{
470 		.procname	= "prove_locking",
471 		.data		= &prove_locking,
472 		.maxlen		= sizeof(int),
473 		.mode		= 0644,
474 		.proc_handler	= proc_dointvec,
475 	},
476 #endif
477 #ifdef CONFIG_LOCK_STAT
478 	{
479 		.procname	= "lock_stat",
480 		.data		= &lock_stat,
481 		.maxlen		= sizeof(int),
482 		.mode		= 0644,
483 		.proc_handler	= proc_dointvec,
484 	},
485 #endif
486 	{
487 		.procname	= "panic",
488 		.data		= &panic_timeout,
489 		.maxlen		= sizeof(int),
490 		.mode		= 0644,
491 		.proc_handler	= proc_dointvec,
492 	},
493 #ifdef CONFIG_COREDUMP
494 	{
495 		.procname	= "core_uses_pid",
496 		.data		= &core_uses_pid,
497 		.maxlen		= sizeof(int),
498 		.mode		= 0644,
499 		.proc_handler	= proc_dointvec,
500 	},
501 	{
502 		.procname	= "core_pattern",
503 		.data		= core_pattern,
504 		.maxlen		= CORENAME_MAX_SIZE,
505 		.mode		= 0644,
506 		.proc_handler	= proc_dostring_coredump,
507 	},
508 	{
509 		.procname	= "core_pipe_limit",
510 		.data		= &core_pipe_limit,
511 		.maxlen		= sizeof(unsigned int),
512 		.mode		= 0644,
513 		.proc_handler	= proc_dointvec,
514 	},
515 #endif
516 #ifdef CONFIG_PROC_SYSCTL
517 	{
518 		.procname	= "tainted",
519 		.maxlen 	= sizeof(long),
520 		.mode		= 0644,
521 		.proc_handler	= proc_taint,
522 	},
523 	{
524 		.procname	= "sysctl_writes_strict",
525 		.data		= &sysctl_writes_strict,
526 		.maxlen		= sizeof(int),
527 		.mode		= 0644,
528 		.proc_handler	= proc_dointvec_minmax,
529 		.extra1		= &neg_one,
530 		.extra2		= &one,
531 	},
532 #endif
533 #ifdef CONFIG_LATENCYTOP
534 	{
535 		.procname	= "latencytop",
536 		.data		= &latencytop_enabled,
537 		.maxlen		= sizeof(int),
538 		.mode		= 0644,
539 		.proc_handler	= sysctl_latencytop,
540 	},
541 #endif
542 #ifdef CONFIG_BLK_DEV_INITRD
543 	{
544 		.procname	= "real-root-dev",
545 		.data		= &real_root_dev,
546 		.maxlen		= sizeof(int),
547 		.mode		= 0644,
548 		.proc_handler	= proc_dointvec,
549 	},
550 #endif
551 	{
552 		.procname	= "print-fatal-signals",
553 		.data		= &print_fatal_signals,
554 		.maxlen		= sizeof(int),
555 		.mode		= 0644,
556 		.proc_handler	= proc_dointvec,
557 	},
558 #ifdef CONFIG_SPARC
559 	{
560 		.procname	= "reboot-cmd",
561 		.data		= reboot_command,
562 		.maxlen		= 256,
563 		.mode		= 0644,
564 		.proc_handler	= proc_dostring,
565 	},
566 	{
567 		.procname	= "stop-a",
568 		.data		= &stop_a_enabled,
569 		.maxlen		= sizeof (int),
570 		.mode		= 0644,
571 		.proc_handler	= proc_dointvec,
572 	},
573 	{
574 		.procname	= "scons-poweroff",
575 		.data		= &scons_pwroff,
576 		.maxlen		= sizeof (int),
577 		.mode		= 0644,
578 		.proc_handler	= proc_dointvec,
579 	},
580 #endif
581 #ifdef CONFIG_SPARC64
582 	{
583 		.procname	= "tsb-ratio",
584 		.data		= &sysctl_tsb_ratio,
585 		.maxlen		= sizeof (int),
586 		.mode		= 0644,
587 		.proc_handler	= proc_dointvec,
588 	},
589 #endif
590 #ifdef __hppa__
591 	{
592 		.procname	= "soft-power",
593 		.data		= &pwrsw_enabled,
594 		.maxlen		= sizeof (int),
595 	 	.mode		= 0644,
596 		.proc_handler	= proc_dointvec,
597 	},
598 #endif
599 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
600 	{
601 		.procname	= "unaligned-trap",
602 		.data		= &unaligned_enabled,
603 		.maxlen		= sizeof (int),
604 		.mode		= 0644,
605 		.proc_handler	= proc_dointvec,
606 	},
607 #endif
608 	{
609 		.procname	= "ctrl-alt-del",
610 		.data		= &C_A_D,
611 		.maxlen		= sizeof(int),
612 		.mode		= 0644,
613 		.proc_handler	= proc_dointvec,
614 	},
615 #ifdef CONFIG_FUNCTION_TRACER
616 	{
617 		.procname	= "ftrace_enabled",
618 		.data		= &ftrace_enabled,
619 		.maxlen		= sizeof(int),
620 		.mode		= 0644,
621 		.proc_handler	= ftrace_enable_sysctl,
622 	},
623 #endif
624 #ifdef CONFIG_STACK_TRACER
625 	{
626 		.procname	= "stack_tracer_enabled",
627 		.data		= &stack_tracer_enabled,
628 		.maxlen		= sizeof(int),
629 		.mode		= 0644,
630 		.proc_handler	= stack_trace_sysctl,
631 	},
632 #endif
633 #ifdef CONFIG_TRACING
634 	{
635 		.procname	= "ftrace_dump_on_oops",
636 		.data		= &ftrace_dump_on_oops,
637 		.maxlen		= sizeof(int),
638 		.mode		= 0644,
639 		.proc_handler	= proc_dointvec,
640 	},
641 	{
642 		.procname	= "traceoff_on_warning",
643 		.data		= &__disable_trace_on_warning,
644 		.maxlen		= sizeof(__disable_trace_on_warning),
645 		.mode		= 0644,
646 		.proc_handler	= proc_dointvec,
647 	},
648 	{
649 		.procname	= "tracepoint_printk",
650 		.data		= &tracepoint_printk,
651 		.maxlen		= sizeof(tracepoint_printk),
652 		.mode		= 0644,
653 		.proc_handler	= tracepoint_printk_sysctl,
654 	},
655 #endif
656 #ifdef CONFIG_KEXEC_CORE
657 	{
658 		.procname	= "kexec_load_disabled",
659 		.data		= &kexec_load_disabled,
660 		.maxlen		= sizeof(int),
661 		.mode		= 0644,
662 		/* only handle a transition from default "0" to "1" */
663 		.proc_handler	= proc_dointvec_minmax,
664 		.extra1		= &one,
665 		.extra2		= &one,
666 	},
667 #endif
668 #ifdef CONFIG_MODULES
669 	{
670 		.procname	= "modprobe",
671 		.data		= &modprobe_path,
672 		.maxlen		= KMOD_PATH_LEN,
673 		.mode		= 0644,
674 		.proc_handler	= proc_dostring,
675 	},
676 	{
677 		.procname	= "modules_disabled",
678 		.data		= &modules_disabled,
679 		.maxlen		= sizeof(int),
680 		.mode		= 0644,
681 		/* only handle a transition from default "0" to "1" */
682 		.proc_handler	= proc_dointvec_minmax,
683 		.extra1		= &one,
684 		.extra2		= &one,
685 	},
686 #endif
687 #ifdef CONFIG_UEVENT_HELPER
688 	{
689 		.procname	= "hotplug",
690 		.data		= &uevent_helper,
691 		.maxlen		= UEVENT_HELPER_PATH_LEN,
692 		.mode		= 0644,
693 		.proc_handler	= proc_dostring,
694 	},
695 #endif
696 #ifdef CONFIG_CHR_DEV_SG
697 	{
698 		.procname	= "sg-big-buff",
699 		.data		= &sg_big_buff,
700 		.maxlen		= sizeof (int),
701 		.mode		= 0444,
702 		.proc_handler	= proc_dointvec,
703 	},
704 #endif
705 #ifdef CONFIG_BSD_PROCESS_ACCT
706 	{
707 		.procname	= "acct",
708 		.data		= &acct_parm,
709 		.maxlen		= 3*sizeof(int),
710 		.mode		= 0644,
711 		.proc_handler	= proc_dointvec,
712 	},
713 #endif
714 #ifdef CONFIG_MAGIC_SYSRQ
715 	{
716 		.procname	= "sysrq",
717 		.data		= &__sysrq_enabled,
718 		.maxlen		= sizeof (int),
719 		.mode		= 0644,
720 		.proc_handler	= sysrq_sysctl_handler,
721 	},
722 #endif
723 #ifdef CONFIG_PROC_SYSCTL
724 	{
725 		.procname	= "cad_pid",
726 		.data		= NULL,
727 		.maxlen		= sizeof (int),
728 		.mode		= 0600,
729 		.proc_handler	= proc_do_cad_pid,
730 	},
731 #endif
732 	{
733 		.procname	= "threads-max",
734 		.data		= NULL,
735 		.maxlen		= sizeof(int),
736 		.mode		= 0644,
737 		.proc_handler	= sysctl_max_threads,
738 	},
739 	{
740 		.procname	= "random",
741 		.mode		= 0555,
742 		.child		= random_table,
743 	},
744 	{
745 		.procname	= "usermodehelper",
746 		.mode		= 0555,
747 		.child		= usermodehelper_table,
748 	},
749 	{
750 		.procname	= "overflowuid",
751 		.data		= &overflowuid,
752 		.maxlen		= sizeof(int),
753 		.mode		= 0644,
754 		.proc_handler	= proc_dointvec_minmax,
755 		.extra1		= &minolduid,
756 		.extra2		= &maxolduid,
757 	},
758 	{
759 		.procname	= "overflowgid",
760 		.data		= &overflowgid,
761 		.maxlen		= sizeof(int),
762 		.mode		= 0644,
763 		.proc_handler	= proc_dointvec_minmax,
764 		.extra1		= &minolduid,
765 		.extra2		= &maxolduid,
766 	},
767 #ifdef CONFIG_S390
768 #ifdef CONFIG_MATHEMU
769 	{
770 		.procname	= "ieee_emulation_warnings",
771 		.data		= &sysctl_ieee_emulation_warnings,
772 		.maxlen		= sizeof(int),
773 		.mode		= 0644,
774 		.proc_handler	= proc_dointvec,
775 	},
776 #endif
777 	{
778 		.procname	= "userprocess_debug",
779 		.data		= &show_unhandled_signals,
780 		.maxlen		= sizeof(int),
781 		.mode		= 0644,
782 		.proc_handler	= proc_dointvec,
783 	},
784 #endif
785 	{
786 		.procname	= "pid_max",
787 		.data		= &pid_max,
788 		.maxlen		= sizeof (int),
789 		.mode		= 0644,
790 		.proc_handler	= proc_dointvec_minmax,
791 		.extra1		= &pid_max_min,
792 		.extra2		= &pid_max_max,
793 	},
794 	{
795 		.procname	= "panic_on_oops",
796 		.data		= &panic_on_oops,
797 		.maxlen		= sizeof(int),
798 		.mode		= 0644,
799 		.proc_handler	= proc_dointvec,
800 	},
801 #if defined CONFIG_PRINTK
802 	{
803 		.procname	= "printk",
804 		.data		= &console_loglevel,
805 		.maxlen		= 4*sizeof(int),
806 		.mode		= 0644,
807 		.proc_handler	= proc_dointvec,
808 	},
809 	{
810 		.procname	= "printk_ratelimit",
811 		.data		= &printk_ratelimit_state.interval,
812 		.maxlen		= sizeof(int),
813 		.mode		= 0644,
814 		.proc_handler	= proc_dointvec_jiffies,
815 	},
816 	{
817 		.procname	= "printk_ratelimit_burst",
818 		.data		= &printk_ratelimit_state.burst,
819 		.maxlen		= sizeof(int),
820 		.mode		= 0644,
821 		.proc_handler	= proc_dointvec,
822 	},
823 	{
824 		.procname	= "printk_delay",
825 		.data		= &printk_delay_msec,
826 		.maxlen		= sizeof(int),
827 		.mode		= 0644,
828 		.proc_handler	= proc_dointvec_minmax,
829 		.extra1		= &zero,
830 		.extra2		= &ten_thousand,
831 	},
832 	{
833 		.procname	= "printk_devkmsg",
834 		.data		= devkmsg_log_str,
835 		.maxlen		= DEVKMSG_STR_MAX_SIZE,
836 		.mode		= 0644,
837 		.proc_handler	= devkmsg_sysctl_set_loglvl,
838 	},
839 	{
840 		.procname	= "dmesg_restrict",
841 		.data		= &dmesg_restrict,
842 		.maxlen		= sizeof(int),
843 		.mode		= 0644,
844 		.proc_handler	= proc_dointvec_minmax_sysadmin,
845 		.extra1		= &zero,
846 		.extra2		= &one,
847 	},
848 	{
849 		.procname	= "kptr_restrict",
850 		.data		= &kptr_restrict,
851 		.maxlen		= sizeof(int),
852 		.mode		= 0644,
853 		.proc_handler	= proc_dointvec_minmax_sysadmin,
854 		.extra1		= &zero,
855 		.extra2		= &two,
856 	},
857 #endif
858 	{
859 		.procname	= "ngroups_max",
860 		.data		= &ngroups_max,
861 		.maxlen		= sizeof (int),
862 		.mode		= 0444,
863 		.proc_handler	= proc_dointvec,
864 	},
865 	{
866 		.procname	= "cap_last_cap",
867 		.data		= (void *)&cap_last_cap,
868 		.maxlen		= sizeof(int),
869 		.mode		= 0444,
870 		.proc_handler	= proc_dointvec,
871 	},
872 #if defined(CONFIG_LOCKUP_DETECTOR)
873 	{
874 		.procname       = "watchdog",
875 		.data           = &watchdog_user_enabled,
876 		.maxlen         = sizeof (int),
877 		.mode           = 0644,
878 		.proc_handler   = proc_watchdog,
879 		.extra1		= &zero,
880 		.extra2		= &one,
881 	},
882 	{
883 		.procname	= "watchdog_thresh",
884 		.data		= &watchdog_thresh,
885 		.maxlen		= sizeof(int),
886 		.mode		= 0644,
887 		.proc_handler	= proc_watchdog_thresh,
888 		.extra1		= &zero,
889 		.extra2		= &sixty,
890 	},
891 	{
892 		.procname       = "nmi_watchdog",
893 		.data           = &nmi_watchdog_enabled,
894 		.maxlen         = sizeof (int),
895 		.mode           = 0644,
896 		.proc_handler   = proc_nmi_watchdog,
897 		.extra1		= &zero,
898 #if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR)
899 		.extra2		= &one,
900 #else
901 		.extra2		= &zero,
902 #endif
903 	},
904 	{
905 		.procname	= "watchdog_cpumask",
906 		.data		= &watchdog_cpumask_bits,
907 		.maxlen		= NR_CPUS,
908 		.mode		= 0644,
909 		.proc_handler	= proc_watchdog_cpumask,
910 	},
911 #ifdef CONFIG_SOFTLOCKUP_DETECTOR
912 	{
913 		.procname       = "soft_watchdog",
914 		.data           = &soft_watchdog_enabled,
915 		.maxlen         = sizeof (int),
916 		.mode           = 0644,
917 		.proc_handler   = proc_soft_watchdog,
918 		.extra1		= &zero,
919 		.extra2		= &one,
920 	},
921 	{
922 		.procname	= "softlockup_panic",
923 		.data		= &softlockup_panic,
924 		.maxlen		= sizeof(int),
925 		.mode		= 0644,
926 		.proc_handler	= proc_dointvec_minmax,
927 		.extra1		= &zero,
928 		.extra2		= &one,
929 	},
930 #ifdef CONFIG_SMP
931 	{
932 		.procname	= "softlockup_all_cpu_backtrace",
933 		.data		= &sysctl_softlockup_all_cpu_backtrace,
934 		.maxlen		= sizeof(int),
935 		.mode		= 0644,
936 		.proc_handler	= proc_dointvec_minmax,
937 		.extra1		= &zero,
938 		.extra2		= &one,
939 	},
940 #endif /* CONFIG_SMP */
941 #endif
942 #ifdef CONFIG_HARDLOCKUP_DETECTOR
943 	{
944 		.procname	= "hardlockup_panic",
945 		.data		= &hardlockup_panic,
946 		.maxlen		= sizeof(int),
947 		.mode		= 0644,
948 		.proc_handler	= proc_dointvec_minmax,
949 		.extra1		= &zero,
950 		.extra2		= &one,
951 	},
952 #ifdef CONFIG_SMP
953 	{
954 		.procname	= "hardlockup_all_cpu_backtrace",
955 		.data		= &sysctl_hardlockup_all_cpu_backtrace,
956 		.maxlen		= sizeof(int),
957 		.mode		= 0644,
958 		.proc_handler	= proc_dointvec_minmax,
959 		.extra1		= &zero,
960 		.extra2		= &one,
961 	},
962 #endif /* CONFIG_SMP */
963 #endif
964 #endif
965 
966 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
967 	{
968 		.procname       = "unknown_nmi_panic",
969 		.data           = &unknown_nmi_panic,
970 		.maxlen         = sizeof (int),
971 		.mode           = 0644,
972 		.proc_handler   = proc_dointvec,
973 	},
974 #endif
975 #if defined(CONFIG_X86)
976 	{
977 		.procname	= "panic_on_unrecovered_nmi",
978 		.data		= &panic_on_unrecovered_nmi,
979 		.maxlen		= sizeof(int),
980 		.mode		= 0644,
981 		.proc_handler	= proc_dointvec,
982 	},
983 	{
984 		.procname	= "panic_on_io_nmi",
985 		.data		= &panic_on_io_nmi,
986 		.maxlen		= sizeof(int),
987 		.mode		= 0644,
988 		.proc_handler	= proc_dointvec,
989 	},
990 #ifdef CONFIG_DEBUG_STACKOVERFLOW
991 	{
992 		.procname	= "panic_on_stackoverflow",
993 		.data		= &sysctl_panic_on_stackoverflow,
994 		.maxlen		= sizeof(int),
995 		.mode		= 0644,
996 		.proc_handler	= proc_dointvec,
997 	},
998 #endif
999 	{
1000 		.procname	= "bootloader_type",
1001 		.data		= &bootloader_type,
1002 		.maxlen		= sizeof (int),
1003 		.mode		= 0444,
1004 		.proc_handler	= proc_dointvec,
1005 	},
1006 	{
1007 		.procname	= "bootloader_version",
1008 		.data		= &bootloader_version,
1009 		.maxlen		= sizeof (int),
1010 		.mode		= 0444,
1011 		.proc_handler	= proc_dointvec,
1012 	},
1013 	{
1014 		.procname	= "io_delay_type",
1015 		.data		= &io_delay_type,
1016 		.maxlen		= sizeof(int),
1017 		.mode		= 0644,
1018 		.proc_handler	= proc_dointvec,
1019 	},
1020 #endif
1021 #if defined(CONFIG_MMU)
1022 	{
1023 		.procname	= "randomize_va_space",
1024 		.data		= &randomize_va_space,
1025 		.maxlen		= sizeof(int),
1026 		.mode		= 0644,
1027 		.proc_handler	= proc_dointvec,
1028 	},
1029 #endif
1030 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1031 	{
1032 		.procname	= "spin_retry",
1033 		.data		= &spin_retry,
1034 		.maxlen		= sizeof (int),
1035 		.mode		= 0644,
1036 		.proc_handler	= proc_dointvec,
1037 	},
1038 #endif
1039 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1040 	{
1041 		.procname	= "acpi_video_flags",
1042 		.data		= &acpi_realmode_flags,
1043 		.maxlen		= sizeof (unsigned long),
1044 		.mode		= 0644,
1045 		.proc_handler	= proc_doulongvec_minmax,
1046 	},
1047 #endif
1048 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1049 	{
1050 		.procname	= "ignore-unaligned-usertrap",
1051 		.data		= &no_unaligned_warning,
1052 		.maxlen		= sizeof (int),
1053 	 	.mode		= 0644,
1054 		.proc_handler	= proc_dointvec,
1055 	},
1056 #endif
1057 #ifdef CONFIG_IA64
1058 	{
1059 		.procname	= "unaligned-dump-stack",
1060 		.data		= &unaligned_dump_stack,
1061 		.maxlen		= sizeof (int),
1062 		.mode		= 0644,
1063 		.proc_handler	= proc_dointvec,
1064 	},
1065 #endif
1066 #ifdef CONFIG_DETECT_HUNG_TASK
1067 	{
1068 		.procname	= "hung_task_panic",
1069 		.data		= &sysctl_hung_task_panic,
1070 		.maxlen		= sizeof(int),
1071 		.mode		= 0644,
1072 		.proc_handler	= proc_dointvec_minmax,
1073 		.extra1		= &zero,
1074 		.extra2		= &one,
1075 	},
1076 	{
1077 		.procname	= "hung_task_check_count",
1078 		.data		= &sysctl_hung_task_check_count,
1079 		.maxlen		= sizeof(int),
1080 		.mode		= 0644,
1081 		.proc_handler	= proc_dointvec_minmax,
1082 		.extra1		= &zero,
1083 	},
1084 	{
1085 		.procname	= "hung_task_timeout_secs",
1086 		.data		= &sysctl_hung_task_timeout_secs,
1087 		.maxlen		= sizeof(unsigned long),
1088 		.mode		= 0644,
1089 		.proc_handler	= proc_dohung_task_timeout_secs,
1090 		.extra2		= &hung_task_timeout_max,
1091 	},
1092 	{
1093 		.procname	= "hung_task_warnings",
1094 		.data		= &sysctl_hung_task_warnings,
1095 		.maxlen		= sizeof(int),
1096 		.mode		= 0644,
1097 		.proc_handler	= proc_dointvec_minmax,
1098 		.extra1		= &neg_one,
1099 	},
1100 #endif
1101 #ifdef CONFIG_RT_MUTEXES
1102 	{
1103 		.procname	= "max_lock_depth",
1104 		.data		= &max_lock_depth,
1105 		.maxlen		= sizeof(int),
1106 		.mode		= 0644,
1107 		.proc_handler	= proc_dointvec,
1108 	},
1109 #endif
1110 	{
1111 		.procname	= "poweroff_cmd",
1112 		.data		= &poweroff_cmd,
1113 		.maxlen		= POWEROFF_CMD_PATH_LEN,
1114 		.mode		= 0644,
1115 		.proc_handler	= proc_dostring,
1116 	},
1117 #ifdef CONFIG_KEYS
1118 	{
1119 		.procname	= "keys",
1120 		.mode		= 0555,
1121 		.child		= key_sysctls,
1122 	},
1123 #endif
1124 #ifdef CONFIG_PERF_EVENTS
1125 	/*
1126 	 * User-space scripts rely on the existence of this file
1127 	 * as a feature check for perf_events being enabled.
1128 	 *
1129 	 * So it's an ABI, do not remove!
1130 	 */
1131 	{
1132 		.procname	= "perf_event_paranoid",
1133 		.data		= &sysctl_perf_event_paranoid,
1134 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
1135 		.mode		= 0644,
1136 		.proc_handler	= proc_dointvec,
1137 	},
1138 	{
1139 		.procname	= "perf_event_mlock_kb",
1140 		.data		= &sysctl_perf_event_mlock,
1141 		.maxlen		= sizeof(sysctl_perf_event_mlock),
1142 		.mode		= 0644,
1143 		.proc_handler	= proc_dointvec,
1144 	},
1145 	{
1146 		.procname	= "perf_event_max_sample_rate",
1147 		.data		= &sysctl_perf_event_sample_rate,
1148 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
1149 		.mode		= 0644,
1150 		.proc_handler	= perf_proc_update_handler,
1151 		.extra1		= &one,
1152 	},
1153 	{
1154 		.procname	= "perf_cpu_time_max_percent",
1155 		.data		= &sysctl_perf_cpu_time_max_percent,
1156 		.maxlen		= sizeof(sysctl_perf_cpu_time_max_percent),
1157 		.mode		= 0644,
1158 		.proc_handler	= perf_cpu_time_max_percent_handler,
1159 		.extra1		= &zero,
1160 		.extra2		= &one_hundred,
1161 	},
1162 	{
1163 		.procname	= "perf_event_max_stack",
1164 		.data		= &sysctl_perf_event_max_stack,
1165 		.maxlen		= sizeof(sysctl_perf_event_max_stack),
1166 		.mode		= 0644,
1167 		.proc_handler	= perf_event_max_stack_handler,
1168 		.extra1		= &zero,
1169 		.extra2		= &six_hundred_forty_kb,
1170 	},
1171 	{
1172 		.procname	= "perf_event_max_contexts_per_stack",
1173 		.data		= &sysctl_perf_event_max_contexts_per_stack,
1174 		.maxlen		= sizeof(sysctl_perf_event_max_contexts_per_stack),
1175 		.mode		= 0644,
1176 		.proc_handler	= perf_event_max_stack_handler,
1177 		.extra1		= &zero,
1178 		.extra2		= &one_thousand,
1179 	},
1180 #endif
1181 #ifdef CONFIG_KMEMCHECK
1182 	{
1183 		.procname	= "kmemcheck",
1184 		.data		= &kmemcheck_enabled,
1185 		.maxlen		= sizeof(int),
1186 		.mode		= 0644,
1187 		.proc_handler	= proc_dointvec,
1188 	},
1189 #endif
1190 	{
1191 		.procname	= "panic_on_warn",
1192 		.data		= &panic_on_warn,
1193 		.maxlen		= sizeof(int),
1194 		.mode		= 0644,
1195 		.proc_handler	= proc_dointvec_minmax,
1196 		.extra1		= &zero,
1197 		.extra2		= &one,
1198 	},
1199 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1200 	{
1201 		.procname	= "timer_migration",
1202 		.data		= &sysctl_timer_migration,
1203 		.maxlen		= sizeof(unsigned int),
1204 		.mode		= 0644,
1205 		.proc_handler	= timer_migration_handler,
1206 		.extra1		= &zero,
1207 		.extra2		= &one,
1208 	},
1209 #endif
1210 #ifdef CONFIG_BPF_SYSCALL
1211 	{
1212 		.procname	= "unprivileged_bpf_disabled",
1213 		.data		= &sysctl_unprivileged_bpf_disabled,
1214 		.maxlen		= sizeof(sysctl_unprivileged_bpf_disabled),
1215 		.mode		= 0644,
1216 		/* only handle a transition from default "0" to "1" */
1217 		.proc_handler	= proc_dointvec_minmax,
1218 		.extra1		= &one,
1219 		.extra2		= &one,
1220 	},
1221 #endif
1222 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1223 	{
1224 		.procname	= "panic_on_rcu_stall",
1225 		.data		= &sysctl_panic_on_rcu_stall,
1226 		.maxlen		= sizeof(sysctl_panic_on_rcu_stall),
1227 		.mode		= 0644,
1228 		.proc_handler	= proc_dointvec_minmax,
1229 		.extra1		= &zero,
1230 		.extra2		= &one,
1231 	},
1232 #endif
1233 	{ }
1234 };
1235 
1236 static struct ctl_table vm_table[] = {
1237 	{
1238 		.procname	= "overcommit_memory",
1239 		.data		= &sysctl_overcommit_memory,
1240 		.maxlen		= sizeof(sysctl_overcommit_memory),
1241 		.mode		= 0644,
1242 		.proc_handler	= proc_dointvec_minmax,
1243 		.extra1		= &zero,
1244 		.extra2		= &two,
1245 	},
1246 	{
1247 		.procname	= "panic_on_oom",
1248 		.data		= &sysctl_panic_on_oom,
1249 		.maxlen		= sizeof(sysctl_panic_on_oom),
1250 		.mode		= 0644,
1251 		.proc_handler	= proc_dointvec_minmax,
1252 		.extra1		= &zero,
1253 		.extra2		= &two,
1254 	},
1255 	{
1256 		.procname	= "oom_kill_allocating_task",
1257 		.data		= &sysctl_oom_kill_allocating_task,
1258 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1259 		.mode		= 0644,
1260 		.proc_handler	= proc_dointvec,
1261 	},
1262 	{
1263 		.procname	= "oom_dump_tasks",
1264 		.data		= &sysctl_oom_dump_tasks,
1265 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1266 		.mode		= 0644,
1267 		.proc_handler	= proc_dointvec,
1268 	},
1269 	{
1270 		.procname	= "overcommit_ratio",
1271 		.data		= &sysctl_overcommit_ratio,
1272 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1273 		.mode		= 0644,
1274 		.proc_handler	= overcommit_ratio_handler,
1275 	},
1276 	{
1277 		.procname	= "overcommit_kbytes",
1278 		.data		= &sysctl_overcommit_kbytes,
1279 		.maxlen		= sizeof(sysctl_overcommit_kbytes),
1280 		.mode		= 0644,
1281 		.proc_handler	= overcommit_kbytes_handler,
1282 	},
1283 	{
1284 		.procname	= "page-cluster",
1285 		.data		= &page_cluster,
1286 		.maxlen		= sizeof(int),
1287 		.mode		= 0644,
1288 		.proc_handler	= proc_dointvec_minmax,
1289 		.extra1		= &zero,
1290 	},
1291 	{
1292 		.procname	= "dirty_background_ratio",
1293 		.data		= &dirty_background_ratio,
1294 		.maxlen		= sizeof(dirty_background_ratio),
1295 		.mode		= 0644,
1296 		.proc_handler	= dirty_background_ratio_handler,
1297 		.extra1		= &zero,
1298 		.extra2		= &one_hundred,
1299 	},
1300 	{
1301 		.procname	= "dirty_background_bytes",
1302 		.data		= &dirty_background_bytes,
1303 		.maxlen		= sizeof(dirty_background_bytes),
1304 		.mode		= 0644,
1305 		.proc_handler	= dirty_background_bytes_handler,
1306 		.extra1		= &one_ul,
1307 	},
1308 	{
1309 		.procname	= "dirty_ratio",
1310 		.data		= &vm_dirty_ratio,
1311 		.maxlen		= sizeof(vm_dirty_ratio),
1312 		.mode		= 0644,
1313 		.proc_handler	= dirty_ratio_handler,
1314 		.extra1		= &zero,
1315 		.extra2		= &one_hundred,
1316 	},
1317 	{
1318 		.procname	= "dirty_bytes",
1319 		.data		= &vm_dirty_bytes,
1320 		.maxlen		= sizeof(vm_dirty_bytes),
1321 		.mode		= 0644,
1322 		.proc_handler	= dirty_bytes_handler,
1323 		.extra1		= &dirty_bytes_min,
1324 	},
1325 	{
1326 		.procname	= "dirty_writeback_centisecs",
1327 		.data		= &dirty_writeback_interval,
1328 		.maxlen		= sizeof(dirty_writeback_interval),
1329 		.mode		= 0644,
1330 		.proc_handler	= dirty_writeback_centisecs_handler,
1331 	},
1332 	{
1333 		.procname	= "dirty_expire_centisecs",
1334 		.data		= &dirty_expire_interval,
1335 		.maxlen		= sizeof(dirty_expire_interval),
1336 		.mode		= 0644,
1337 		.proc_handler	= proc_dointvec_minmax,
1338 		.extra1		= &zero,
1339 	},
1340 	{
1341 		.procname	= "dirtytime_expire_seconds",
1342 		.data		= &dirtytime_expire_interval,
1343 		.maxlen		= sizeof(dirty_expire_interval),
1344 		.mode		= 0644,
1345 		.proc_handler	= dirtytime_interval_handler,
1346 		.extra1		= &zero,
1347 	},
1348 	{
1349 		.procname       = "nr_pdflush_threads",
1350 		.mode           = 0444 /* read-only */,
1351 		.proc_handler   = pdflush_proc_obsolete,
1352 	},
1353 	{
1354 		.procname	= "swappiness",
1355 		.data		= &vm_swappiness,
1356 		.maxlen		= sizeof(vm_swappiness),
1357 		.mode		= 0644,
1358 		.proc_handler	= proc_dointvec_minmax,
1359 		.extra1		= &zero,
1360 		.extra2		= &one_hundred,
1361 	},
1362 #ifdef CONFIG_HUGETLB_PAGE
1363 	{
1364 		.procname	= "nr_hugepages",
1365 		.data		= NULL,
1366 		.maxlen		= sizeof(unsigned long),
1367 		.mode		= 0644,
1368 		.proc_handler	= hugetlb_sysctl_handler,
1369 	},
1370 #ifdef CONFIG_NUMA
1371 	{
1372 		.procname       = "nr_hugepages_mempolicy",
1373 		.data           = NULL,
1374 		.maxlen         = sizeof(unsigned long),
1375 		.mode           = 0644,
1376 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1377 	},
1378 #endif
1379 	 {
1380 		.procname	= "hugetlb_shm_group",
1381 		.data		= &sysctl_hugetlb_shm_group,
1382 		.maxlen		= sizeof(gid_t),
1383 		.mode		= 0644,
1384 		.proc_handler	= proc_dointvec,
1385 	 },
1386 	 {
1387 		.procname	= "hugepages_treat_as_movable",
1388 		.data		= &hugepages_treat_as_movable,
1389 		.maxlen		= sizeof(int),
1390 		.mode		= 0644,
1391 		.proc_handler	= proc_dointvec,
1392 	},
1393 	{
1394 		.procname	= "nr_overcommit_hugepages",
1395 		.data		= NULL,
1396 		.maxlen		= sizeof(unsigned long),
1397 		.mode		= 0644,
1398 		.proc_handler	= hugetlb_overcommit_handler,
1399 	},
1400 #endif
1401 	{
1402 		.procname	= "lowmem_reserve_ratio",
1403 		.data		= &sysctl_lowmem_reserve_ratio,
1404 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1405 		.mode		= 0644,
1406 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1407 	},
1408 	{
1409 		.procname	= "drop_caches",
1410 		.data		= &sysctl_drop_caches,
1411 		.maxlen		= sizeof(int),
1412 		.mode		= 0644,
1413 		.proc_handler	= drop_caches_sysctl_handler,
1414 		.extra1		= &one,
1415 		.extra2		= &four,
1416 	},
1417 #ifdef CONFIG_COMPACTION
1418 	{
1419 		.procname	= "compact_memory",
1420 		.data		= &sysctl_compact_memory,
1421 		.maxlen		= sizeof(int),
1422 		.mode		= 0200,
1423 		.proc_handler	= sysctl_compaction_handler,
1424 	},
1425 	{
1426 		.procname	= "extfrag_threshold",
1427 		.data		= &sysctl_extfrag_threshold,
1428 		.maxlen		= sizeof(int),
1429 		.mode		= 0644,
1430 		.proc_handler	= sysctl_extfrag_handler,
1431 		.extra1		= &min_extfrag_threshold,
1432 		.extra2		= &max_extfrag_threshold,
1433 	},
1434 	{
1435 		.procname	= "compact_unevictable_allowed",
1436 		.data		= &sysctl_compact_unevictable_allowed,
1437 		.maxlen		= sizeof(int),
1438 		.mode		= 0644,
1439 		.proc_handler	= proc_dointvec,
1440 		.extra1		= &zero,
1441 		.extra2		= &one,
1442 	},
1443 
1444 #endif /* CONFIG_COMPACTION */
1445 	{
1446 		.procname	= "min_free_kbytes",
1447 		.data		= &min_free_kbytes,
1448 		.maxlen		= sizeof(min_free_kbytes),
1449 		.mode		= 0644,
1450 		.proc_handler	= min_free_kbytes_sysctl_handler,
1451 		.extra1		= &zero,
1452 	},
1453 	{
1454 		.procname	= "watermark_scale_factor",
1455 		.data		= &watermark_scale_factor,
1456 		.maxlen		= sizeof(watermark_scale_factor),
1457 		.mode		= 0644,
1458 		.proc_handler	= watermark_scale_factor_sysctl_handler,
1459 		.extra1		= &one,
1460 		.extra2		= &one_thousand,
1461 	},
1462 	{
1463 		.procname	= "percpu_pagelist_fraction",
1464 		.data		= &percpu_pagelist_fraction,
1465 		.maxlen		= sizeof(percpu_pagelist_fraction),
1466 		.mode		= 0644,
1467 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1468 		.extra1		= &zero,
1469 	},
1470 #ifdef CONFIG_MMU
1471 	{
1472 		.procname	= "max_map_count",
1473 		.data		= &sysctl_max_map_count,
1474 		.maxlen		= sizeof(sysctl_max_map_count),
1475 		.mode		= 0644,
1476 		.proc_handler	= proc_dointvec_minmax,
1477 		.extra1		= &zero,
1478 	},
1479 #else
1480 	{
1481 		.procname	= "nr_trim_pages",
1482 		.data		= &sysctl_nr_trim_pages,
1483 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1484 		.mode		= 0644,
1485 		.proc_handler	= proc_dointvec_minmax,
1486 		.extra1		= &zero,
1487 	},
1488 #endif
1489 	{
1490 		.procname	= "laptop_mode",
1491 		.data		= &laptop_mode,
1492 		.maxlen		= sizeof(laptop_mode),
1493 		.mode		= 0644,
1494 		.proc_handler	= proc_dointvec_jiffies,
1495 	},
1496 	{
1497 		.procname	= "block_dump",
1498 		.data		= &block_dump,
1499 		.maxlen		= sizeof(block_dump),
1500 		.mode		= 0644,
1501 		.proc_handler	= proc_dointvec,
1502 		.extra1		= &zero,
1503 	},
1504 	{
1505 		.procname	= "vfs_cache_pressure",
1506 		.data		= &sysctl_vfs_cache_pressure,
1507 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1508 		.mode		= 0644,
1509 		.proc_handler	= proc_dointvec,
1510 		.extra1		= &zero,
1511 	},
1512 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1513 	{
1514 		.procname	= "legacy_va_layout",
1515 		.data		= &sysctl_legacy_va_layout,
1516 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1517 		.mode		= 0644,
1518 		.proc_handler	= proc_dointvec,
1519 		.extra1		= &zero,
1520 	},
1521 #endif
1522 #ifdef CONFIG_NUMA
1523 	{
1524 		.procname	= "zone_reclaim_mode",
1525 		.data		= &node_reclaim_mode,
1526 		.maxlen		= sizeof(node_reclaim_mode),
1527 		.mode		= 0644,
1528 		.proc_handler	= proc_dointvec,
1529 		.extra1		= &zero,
1530 	},
1531 	{
1532 		.procname	= "min_unmapped_ratio",
1533 		.data		= &sysctl_min_unmapped_ratio,
1534 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1535 		.mode		= 0644,
1536 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1537 		.extra1		= &zero,
1538 		.extra2		= &one_hundred,
1539 	},
1540 	{
1541 		.procname	= "min_slab_ratio",
1542 		.data		= &sysctl_min_slab_ratio,
1543 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1544 		.mode		= 0644,
1545 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1546 		.extra1		= &zero,
1547 		.extra2		= &one_hundred,
1548 	},
1549 #endif
1550 #ifdef CONFIG_SMP
1551 	{
1552 		.procname	= "stat_interval",
1553 		.data		= &sysctl_stat_interval,
1554 		.maxlen		= sizeof(sysctl_stat_interval),
1555 		.mode		= 0644,
1556 		.proc_handler	= proc_dointvec_jiffies,
1557 	},
1558 	{
1559 		.procname	= "stat_refresh",
1560 		.data		= NULL,
1561 		.maxlen		= 0,
1562 		.mode		= 0600,
1563 		.proc_handler	= vmstat_refresh,
1564 	},
1565 #endif
1566 #ifdef CONFIG_MMU
1567 	{
1568 		.procname	= "mmap_min_addr",
1569 		.data		= &dac_mmap_min_addr,
1570 		.maxlen		= sizeof(unsigned long),
1571 		.mode		= 0644,
1572 		.proc_handler	= mmap_min_addr_handler,
1573 	},
1574 #endif
1575 #ifdef CONFIG_NUMA
1576 	{
1577 		.procname	= "numa_zonelist_order",
1578 		.data		= &numa_zonelist_order,
1579 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1580 		.mode		= 0644,
1581 		.proc_handler	= numa_zonelist_order_handler,
1582 	},
1583 #endif
1584 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1585    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1586 	{
1587 		.procname	= "vdso_enabled",
1588 #ifdef CONFIG_X86_32
1589 		.data		= &vdso32_enabled,
1590 		.maxlen		= sizeof(vdso32_enabled),
1591 #else
1592 		.data		= &vdso_enabled,
1593 		.maxlen		= sizeof(vdso_enabled),
1594 #endif
1595 		.mode		= 0644,
1596 		.proc_handler	= proc_dointvec,
1597 		.extra1		= &zero,
1598 	},
1599 #endif
1600 #ifdef CONFIG_HIGHMEM
1601 	{
1602 		.procname	= "highmem_is_dirtyable",
1603 		.data		= &vm_highmem_is_dirtyable,
1604 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1605 		.mode		= 0644,
1606 		.proc_handler	= proc_dointvec_minmax,
1607 		.extra1		= &zero,
1608 		.extra2		= &one,
1609 	},
1610 #endif
1611 #ifdef CONFIG_MEMORY_FAILURE
1612 	{
1613 		.procname	= "memory_failure_early_kill",
1614 		.data		= &sysctl_memory_failure_early_kill,
1615 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1616 		.mode		= 0644,
1617 		.proc_handler	= proc_dointvec_minmax,
1618 		.extra1		= &zero,
1619 		.extra2		= &one,
1620 	},
1621 	{
1622 		.procname	= "memory_failure_recovery",
1623 		.data		= &sysctl_memory_failure_recovery,
1624 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1625 		.mode		= 0644,
1626 		.proc_handler	= proc_dointvec_minmax,
1627 		.extra1		= &zero,
1628 		.extra2		= &one,
1629 	},
1630 #endif
1631 	{
1632 		.procname	= "user_reserve_kbytes",
1633 		.data		= &sysctl_user_reserve_kbytes,
1634 		.maxlen		= sizeof(sysctl_user_reserve_kbytes),
1635 		.mode		= 0644,
1636 		.proc_handler	= proc_doulongvec_minmax,
1637 	},
1638 	{
1639 		.procname	= "admin_reserve_kbytes",
1640 		.data		= &sysctl_admin_reserve_kbytes,
1641 		.maxlen		= sizeof(sysctl_admin_reserve_kbytes),
1642 		.mode		= 0644,
1643 		.proc_handler	= proc_doulongvec_minmax,
1644 	},
1645 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1646 	{
1647 		.procname	= "mmap_rnd_bits",
1648 		.data		= &mmap_rnd_bits,
1649 		.maxlen		= sizeof(mmap_rnd_bits),
1650 		.mode		= 0600,
1651 		.proc_handler	= proc_dointvec_minmax,
1652 		.extra1		= (void *)&mmap_rnd_bits_min,
1653 		.extra2		= (void *)&mmap_rnd_bits_max,
1654 	},
1655 #endif
1656 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1657 	{
1658 		.procname	= "mmap_rnd_compat_bits",
1659 		.data		= &mmap_rnd_compat_bits,
1660 		.maxlen		= sizeof(mmap_rnd_compat_bits),
1661 		.mode		= 0600,
1662 		.proc_handler	= proc_dointvec_minmax,
1663 		.extra1		= (void *)&mmap_rnd_compat_bits_min,
1664 		.extra2		= (void *)&mmap_rnd_compat_bits_max,
1665 	},
1666 #endif
1667 	{ }
1668 };
1669 
1670 static struct ctl_table fs_table[] = {
1671 	{
1672 		.procname	= "inode-nr",
1673 		.data		= &inodes_stat,
1674 		.maxlen		= 2*sizeof(long),
1675 		.mode		= 0444,
1676 		.proc_handler	= proc_nr_inodes,
1677 	},
1678 	{
1679 		.procname	= "inode-state",
1680 		.data		= &inodes_stat,
1681 		.maxlen		= 7*sizeof(long),
1682 		.mode		= 0444,
1683 		.proc_handler	= proc_nr_inodes,
1684 	},
1685 	{
1686 		.procname	= "file-nr",
1687 		.data		= &files_stat,
1688 		.maxlen		= sizeof(files_stat),
1689 		.mode		= 0444,
1690 		.proc_handler	= proc_nr_files,
1691 	},
1692 	{
1693 		.procname	= "file-max",
1694 		.data		= &files_stat.max_files,
1695 		.maxlen		= sizeof(files_stat.max_files),
1696 		.mode		= 0644,
1697 		.proc_handler	= proc_doulongvec_minmax,
1698 	},
1699 	{
1700 		.procname	= "nr_open",
1701 		.data		= &sysctl_nr_open,
1702 		.maxlen		= sizeof(unsigned int),
1703 		.mode		= 0644,
1704 		.proc_handler	= proc_dointvec_minmax,
1705 		.extra1		= &sysctl_nr_open_min,
1706 		.extra2		= &sysctl_nr_open_max,
1707 	},
1708 	{
1709 		.procname	= "dentry-state",
1710 		.data		= &dentry_stat,
1711 		.maxlen		= 6*sizeof(long),
1712 		.mode		= 0444,
1713 		.proc_handler	= proc_nr_dentry,
1714 	},
1715 	{
1716 		.procname	= "overflowuid",
1717 		.data		= &fs_overflowuid,
1718 		.maxlen		= sizeof(int),
1719 		.mode		= 0644,
1720 		.proc_handler	= proc_dointvec_minmax,
1721 		.extra1		= &minolduid,
1722 		.extra2		= &maxolduid,
1723 	},
1724 	{
1725 		.procname	= "overflowgid",
1726 		.data		= &fs_overflowgid,
1727 		.maxlen		= sizeof(int),
1728 		.mode		= 0644,
1729 		.proc_handler	= proc_dointvec_minmax,
1730 		.extra1		= &minolduid,
1731 		.extra2		= &maxolduid,
1732 	},
1733 #ifdef CONFIG_FILE_LOCKING
1734 	{
1735 		.procname	= "leases-enable",
1736 		.data		= &leases_enable,
1737 		.maxlen		= sizeof(int),
1738 		.mode		= 0644,
1739 		.proc_handler	= proc_dointvec,
1740 	},
1741 #endif
1742 #ifdef CONFIG_DNOTIFY
1743 	{
1744 		.procname	= "dir-notify-enable",
1745 		.data		= &dir_notify_enable,
1746 		.maxlen		= sizeof(int),
1747 		.mode		= 0644,
1748 		.proc_handler	= proc_dointvec,
1749 	},
1750 #endif
1751 #ifdef CONFIG_MMU
1752 #ifdef CONFIG_FILE_LOCKING
1753 	{
1754 		.procname	= "lease-break-time",
1755 		.data		= &lease_break_time,
1756 		.maxlen		= sizeof(int),
1757 		.mode		= 0644,
1758 		.proc_handler	= proc_dointvec,
1759 	},
1760 #endif
1761 #ifdef CONFIG_AIO
1762 	{
1763 		.procname	= "aio-nr",
1764 		.data		= &aio_nr,
1765 		.maxlen		= sizeof(aio_nr),
1766 		.mode		= 0444,
1767 		.proc_handler	= proc_doulongvec_minmax,
1768 	},
1769 	{
1770 		.procname	= "aio-max-nr",
1771 		.data		= &aio_max_nr,
1772 		.maxlen		= sizeof(aio_max_nr),
1773 		.mode		= 0644,
1774 		.proc_handler	= proc_doulongvec_minmax,
1775 	},
1776 #endif /* CONFIG_AIO */
1777 #ifdef CONFIG_INOTIFY_USER
1778 	{
1779 		.procname	= "inotify",
1780 		.mode		= 0555,
1781 		.child		= inotify_table,
1782 	},
1783 #endif
1784 #ifdef CONFIG_EPOLL
1785 	{
1786 		.procname	= "epoll",
1787 		.mode		= 0555,
1788 		.child		= epoll_table,
1789 	},
1790 #endif
1791 #endif
1792 	{
1793 		.procname	= "protected_symlinks",
1794 		.data		= &sysctl_protected_symlinks,
1795 		.maxlen		= sizeof(int),
1796 		.mode		= 0600,
1797 		.proc_handler	= proc_dointvec_minmax,
1798 		.extra1		= &zero,
1799 		.extra2		= &one,
1800 	},
1801 	{
1802 		.procname	= "protected_hardlinks",
1803 		.data		= &sysctl_protected_hardlinks,
1804 		.maxlen		= sizeof(int),
1805 		.mode		= 0600,
1806 		.proc_handler	= proc_dointvec_minmax,
1807 		.extra1		= &zero,
1808 		.extra2		= &one,
1809 	},
1810 	{
1811 		.procname	= "suid_dumpable",
1812 		.data		= &suid_dumpable,
1813 		.maxlen		= sizeof(int),
1814 		.mode		= 0644,
1815 		.proc_handler	= proc_dointvec_minmax_coredump,
1816 		.extra1		= &zero,
1817 		.extra2		= &two,
1818 	},
1819 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1820 	{
1821 		.procname	= "binfmt_misc",
1822 		.mode		= 0555,
1823 		.child		= sysctl_mount_point,
1824 	},
1825 #endif
1826 	{
1827 		.procname	= "pipe-max-size",
1828 		.data		= &pipe_max_size,
1829 		.maxlen		= sizeof(int),
1830 		.mode		= 0644,
1831 		.proc_handler	= &pipe_proc_fn,
1832 		.extra1		= &pipe_min_size,
1833 	},
1834 	{
1835 		.procname	= "pipe-user-pages-hard",
1836 		.data		= &pipe_user_pages_hard,
1837 		.maxlen		= sizeof(pipe_user_pages_hard),
1838 		.mode		= 0644,
1839 		.proc_handler	= proc_doulongvec_minmax,
1840 	},
1841 	{
1842 		.procname	= "pipe-user-pages-soft",
1843 		.data		= &pipe_user_pages_soft,
1844 		.maxlen		= sizeof(pipe_user_pages_soft),
1845 		.mode		= 0644,
1846 		.proc_handler	= proc_doulongvec_minmax,
1847 	},
1848 	{
1849 		.procname	= "mount-max",
1850 		.data		= &sysctl_mount_max,
1851 		.maxlen		= sizeof(unsigned int),
1852 		.mode		= 0644,
1853 		.proc_handler	= proc_dointvec_minmax,
1854 		.extra1		= &one,
1855 	},
1856 	{ }
1857 };
1858 
1859 static struct ctl_table debug_table[] = {
1860 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1861 	{
1862 		.procname	= "exception-trace",
1863 		.data		= &show_unhandled_signals,
1864 		.maxlen		= sizeof(int),
1865 		.mode		= 0644,
1866 		.proc_handler	= proc_dointvec
1867 	},
1868 #endif
1869 #if defined(CONFIG_OPTPROBES)
1870 	{
1871 		.procname	= "kprobes-optimization",
1872 		.data		= &sysctl_kprobes_optimization,
1873 		.maxlen		= sizeof(int),
1874 		.mode		= 0644,
1875 		.proc_handler	= proc_kprobes_optimization_handler,
1876 		.extra1		= &zero,
1877 		.extra2		= &one,
1878 	},
1879 #endif
1880 	{ }
1881 };
1882 
1883 static struct ctl_table dev_table[] = {
1884 	{ }
1885 };
1886 
1887 int __init sysctl_init(void)
1888 {
1889 	struct ctl_table_header *hdr;
1890 
1891 	hdr = register_sysctl_table(sysctl_base_table);
1892 	kmemleak_not_leak(hdr);
1893 	return 0;
1894 }
1895 
1896 #endif /* CONFIG_SYSCTL */
1897 
1898 /*
1899  * /proc/sys support
1900  */
1901 
1902 #ifdef CONFIG_PROC_SYSCTL
1903 
1904 static int _proc_do_string(char *data, int maxlen, int write,
1905 			   char __user *buffer,
1906 			   size_t *lenp, loff_t *ppos)
1907 {
1908 	size_t len;
1909 	char __user *p;
1910 	char c;
1911 
1912 	if (!data || !maxlen || !*lenp) {
1913 		*lenp = 0;
1914 		return 0;
1915 	}
1916 
1917 	if (write) {
1918 		if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1919 			/* Only continue writes not past the end of buffer. */
1920 			len = strlen(data);
1921 			if (len > maxlen - 1)
1922 				len = maxlen - 1;
1923 
1924 			if (*ppos > len)
1925 				return 0;
1926 			len = *ppos;
1927 		} else {
1928 			/* Start writing from beginning of buffer. */
1929 			len = 0;
1930 		}
1931 
1932 		*ppos += *lenp;
1933 		p = buffer;
1934 		while ((p - buffer) < *lenp && len < maxlen - 1) {
1935 			if (get_user(c, p++))
1936 				return -EFAULT;
1937 			if (c == 0 || c == '\n')
1938 				break;
1939 			data[len++] = c;
1940 		}
1941 		data[len] = 0;
1942 	} else {
1943 		len = strlen(data);
1944 		if (len > maxlen)
1945 			len = maxlen;
1946 
1947 		if (*ppos > len) {
1948 			*lenp = 0;
1949 			return 0;
1950 		}
1951 
1952 		data += *ppos;
1953 		len  -= *ppos;
1954 
1955 		if (len > *lenp)
1956 			len = *lenp;
1957 		if (len)
1958 			if (copy_to_user(buffer, data, len))
1959 				return -EFAULT;
1960 		if (len < *lenp) {
1961 			if (put_user('\n', buffer + len))
1962 				return -EFAULT;
1963 			len++;
1964 		}
1965 		*lenp = len;
1966 		*ppos += len;
1967 	}
1968 	return 0;
1969 }
1970 
1971 static void warn_sysctl_write(struct ctl_table *table)
1972 {
1973 	pr_warn_once("%s wrote to %s when file position was not 0!\n"
1974 		"This will not be supported in the future. To silence this\n"
1975 		"warning, set kernel.sysctl_writes_strict = -1\n",
1976 		current->comm, table->procname);
1977 }
1978 
1979 /**
1980  * proc_first_pos_non_zero_ignore - check if firs position is allowed
1981  * @ppos: file position
1982  * @table: the sysctl table
1983  *
1984  * Returns true if the first position is non-zero and the sysctl_writes_strict
1985  * mode indicates this is not allowed for numeric input types. String proc
1986  * hadlers can ignore the return value.
1987  */
1988 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
1989 					   struct ctl_table *table)
1990 {
1991 	if (!*ppos)
1992 		return false;
1993 
1994 	switch (sysctl_writes_strict) {
1995 	case SYSCTL_WRITES_STRICT:
1996 		return true;
1997 	case SYSCTL_WRITES_WARN:
1998 		warn_sysctl_write(table);
1999 		return false;
2000 	default:
2001 		return false;
2002 	}
2003 }
2004 
2005 /**
2006  * proc_dostring - read a string sysctl
2007  * @table: the sysctl table
2008  * @write: %TRUE if this is a write to the sysctl file
2009  * @buffer: the user buffer
2010  * @lenp: the size of the user buffer
2011  * @ppos: file position
2012  *
2013  * Reads/writes a string from/to the user buffer. If the kernel
2014  * buffer provided is not large enough to hold the string, the
2015  * string is truncated. The copied string is %NULL-terminated.
2016  * If the string is being read by the user process, it is copied
2017  * and a newline '\n' is added. It is truncated if the buffer is
2018  * not large enough.
2019  *
2020  * Returns 0 on success.
2021  */
2022 int proc_dostring(struct ctl_table *table, int write,
2023 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2024 {
2025 	if (write)
2026 		proc_first_pos_non_zero_ignore(ppos, table);
2027 
2028 	return _proc_do_string((char *)(table->data), table->maxlen, write,
2029 			       (char __user *)buffer, lenp, ppos);
2030 }
2031 
2032 static size_t proc_skip_spaces(char **buf)
2033 {
2034 	size_t ret;
2035 	char *tmp = skip_spaces(*buf);
2036 	ret = tmp - *buf;
2037 	*buf = tmp;
2038 	return ret;
2039 }
2040 
2041 static void proc_skip_char(char **buf, size_t *size, const char v)
2042 {
2043 	while (*size) {
2044 		if (**buf != v)
2045 			break;
2046 		(*size)--;
2047 		(*buf)++;
2048 	}
2049 }
2050 
2051 #define TMPBUFLEN 22
2052 /**
2053  * proc_get_long - reads an ASCII formatted integer from a user buffer
2054  *
2055  * @buf: a kernel buffer
2056  * @size: size of the kernel buffer
2057  * @val: this is where the number will be stored
2058  * @neg: set to %TRUE if number is negative
2059  * @perm_tr: a vector which contains the allowed trailers
2060  * @perm_tr_len: size of the perm_tr vector
2061  * @tr: pointer to store the trailer character
2062  *
2063  * In case of success %0 is returned and @buf and @size are updated with
2064  * the amount of bytes read. If @tr is non-NULL and a trailing
2065  * character exists (size is non-zero after returning from this
2066  * function), @tr is updated with the trailing character.
2067  */
2068 static int proc_get_long(char **buf, size_t *size,
2069 			  unsigned long *val, bool *neg,
2070 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
2071 {
2072 	int len;
2073 	char *p, tmp[TMPBUFLEN];
2074 
2075 	if (!*size)
2076 		return -EINVAL;
2077 
2078 	len = *size;
2079 	if (len > TMPBUFLEN - 1)
2080 		len = TMPBUFLEN - 1;
2081 
2082 	memcpy(tmp, *buf, len);
2083 
2084 	tmp[len] = 0;
2085 	p = tmp;
2086 	if (*p == '-' && *size > 1) {
2087 		*neg = true;
2088 		p++;
2089 	} else
2090 		*neg = false;
2091 	if (!isdigit(*p))
2092 		return -EINVAL;
2093 
2094 	*val = simple_strtoul(p, &p, 0);
2095 
2096 	len = p - tmp;
2097 
2098 	/* We don't know if the next char is whitespace thus we may accept
2099 	 * invalid integers (e.g. 1234...a) or two integers instead of one
2100 	 * (e.g. 123...1). So lets not allow such large numbers. */
2101 	if (len == TMPBUFLEN - 1)
2102 		return -EINVAL;
2103 
2104 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2105 		return -EINVAL;
2106 
2107 	if (tr && (len < *size))
2108 		*tr = *p;
2109 
2110 	*buf += len;
2111 	*size -= len;
2112 
2113 	return 0;
2114 }
2115 
2116 /**
2117  * proc_put_long - converts an integer to a decimal ASCII formatted string
2118  *
2119  * @buf: the user buffer
2120  * @size: the size of the user buffer
2121  * @val: the integer to be converted
2122  * @neg: sign of the number, %TRUE for negative
2123  *
2124  * In case of success %0 is returned and @buf and @size are updated with
2125  * the amount of bytes written.
2126  */
2127 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2128 			  bool neg)
2129 {
2130 	int len;
2131 	char tmp[TMPBUFLEN], *p = tmp;
2132 
2133 	sprintf(p, "%s%lu", neg ? "-" : "", val);
2134 	len = strlen(tmp);
2135 	if (len > *size)
2136 		len = *size;
2137 	if (copy_to_user(*buf, tmp, len))
2138 		return -EFAULT;
2139 	*size -= len;
2140 	*buf += len;
2141 	return 0;
2142 }
2143 #undef TMPBUFLEN
2144 
2145 static int proc_put_char(void __user **buf, size_t *size, char c)
2146 {
2147 	if (*size) {
2148 		char __user **buffer = (char __user **)buf;
2149 		if (put_user(c, *buffer))
2150 			return -EFAULT;
2151 		(*size)--, (*buffer)++;
2152 		*buf = *buffer;
2153 	}
2154 	return 0;
2155 }
2156 
2157 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2158 				 int *valp,
2159 				 int write, void *data)
2160 {
2161 	if (write) {
2162 		if (*negp) {
2163 			if (*lvalp > (unsigned long) INT_MAX + 1)
2164 				return -EINVAL;
2165 			*valp = -*lvalp;
2166 		} else {
2167 			if (*lvalp > (unsigned long) INT_MAX)
2168 				return -EINVAL;
2169 			*valp = *lvalp;
2170 		}
2171 	} else {
2172 		int val = *valp;
2173 		if (val < 0) {
2174 			*negp = true;
2175 			*lvalp = -(unsigned long)val;
2176 		} else {
2177 			*negp = false;
2178 			*lvalp = (unsigned long)val;
2179 		}
2180 	}
2181 	return 0;
2182 }
2183 
2184 static int do_proc_douintvec_conv(unsigned long *lvalp,
2185 				  unsigned int *valp,
2186 				  int write, void *data)
2187 {
2188 	if (write) {
2189 		if (*lvalp > UINT_MAX)
2190 			return -EINVAL;
2191 		if (*lvalp > UINT_MAX)
2192 			return -EINVAL;
2193 		*valp = *lvalp;
2194 	} else {
2195 		unsigned int val = *valp;
2196 		*lvalp = (unsigned long)val;
2197 	}
2198 	return 0;
2199 }
2200 
2201 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2202 
2203 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2204 		  int write, void __user *buffer,
2205 		  size_t *lenp, loff_t *ppos,
2206 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2207 			      int write, void *data),
2208 		  void *data)
2209 {
2210 	int *i, vleft, first = 1, err = 0;
2211 	size_t left;
2212 	char *kbuf = NULL, *p;
2213 
2214 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2215 		*lenp = 0;
2216 		return 0;
2217 	}
2218 
2219 	i = (int *) tbl_data;
2220 	vleft = table->maxlen / sizeof(*i);
2221 	left = *lenp;
2222 
2223 	if (!conv)
2224 		conv = do_proc_dointvec_conv;
2225 
2226 	if (write) {
2227 		if (proc_first_pos_non_zero_ignore(ppos, table))
2228 			goto out;
2229 
2230 		if (left > PAGE_SIZE - 1)
2231 			left = PAGE_SIZE - 1;
2232 		p = kbuf = memdup_user_nul(buffer, left);
2233 		if (IS_ERR(kbuf))
2234 			return PTR_ERR(kbuf);
2235 	}
2236 
2237 	for (; left && vleft--; i++, first=0) {
2238 		unsigned long lval;
2239 		bool neg;
2240 
2241 		if (write) {
2242 			left -= proc_skip_spaces(&p);
2243 
2244 			if (!left)
2245 				break;
2246 			err = proc_get_long(&p, &left, &lval, &neg,
2247 					     proc_wspace_sep,
2248 					     sizeof(proc_wspace_sep), NULL);
2249 			if (err)
2250 				break;
2251 			if (conv(&neg, &lval, i, 1, data)) {
2252 				err = -EINVAL;
2253 				break;
2254 			}
2255 		} else {
2256 			if (conv(&neg, &lval, i, 0, data)) {
2257 				err = -EINVAL;
2258 				break;
2259 			}
2260 			if (!first)
2261 				err = proc_put_char(&buffer, &left, '\t');
2262 			if (err)
2263 				break;
2264 			err = proc_put_long(&buffer, &left, lval, neg);
2265 			if (err)
2266 				break;
2267 		}
2268 	}
2269 
2270 	if (!write && !first && left && !err)
2271 		err = proc_put_char(&buffer, &left, '\n');
2272 	if (write && !err && left)
2273 		left -= proc_skip_spaces(&p);
2274 	if (write) {
2275 		kfree(kbuf);
2276 		if (first)
2277 			return err ? : -EINVAL;
2278 	}
2279 	*lenp -= left;
2280 out:
2281 	*ppos += *lenp;
2282 	return err;
2283 }
2284 
2285 static int do_proc_dointvec(struct ctl_table *table, int write,
2286 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2287 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2288 			      int write, void *data),
2289 		  void *data)
2290 {
2291 	return __do_proc_dointvec(table->data, table, write,
2292 			buffer, lenp, ppos, conv, data);
2293 }
2294 
2295 static int do_proc_douintvec_w(unsigned int *tbl_data,
2296 			       struct ctl_table *table,
2297 			       void __user *buffer,
2298 			       size_t *lenp, loff_t *ppos,
2299 			       int (*conv)(unsigned long *lvalp,
2300 					   unsigned int *valp,
2301 					   int write, void *data),
2302 			       void *data)
2303 {
2304 	unsigned long lval;
2305 	int err = 0;
2306 	size_t left;
2307 	bool neg;
2308 	char *kbuf = NULL, *p;
2309 
2310 	left = *lenp;
2311 
2312 	if (proc_first_pos_non_zero_ignore(ppos, table))
2313 		goto bail_early;
2314 
2315 	if (left > PAGE_SIZE - 1)
2316 		left = PAGE_SIZE - 1;
2317 
2318 	p = kbuf = memdup_user_nul(buffer, left);
2319 	if (IS_ERR(kbuf))
2320 		return -EINVAL;
2321 
2322 	left -= proc_skip_spaces(&p);
2323 	if (!left) {
2324 		err = -EINVAL;
2325 		goto out_free;
2326 	}
2327 
2328 	err = proc_get_long(&p, &left, &lval, &neg,
2329 			     proc_wspace_sep,
2330 			     sizeof(proc_wspace_sep), NULL);
2331 	if (err || neg) {
2332 		err = -EINVAL;
2333 		goto out_free;
2334 	}
2335 
2336 	if (conv(&lval, tbl_data, 1, data)) {
2337 		err = -EINVAL;
2338 		goto out_free;
2339 	}
2340 
2341 	if (!err && left)
2342 		left -= proc_skip_spaces(&p);
2343 
2344 out_free:
2345 	kfree(kbuf);
2346 	if (err)
2347 		return -EINVAL;
2348 
2349 	return 0;
2350 
2351 	/* This is in keeping with old __do_proc_dointvec() */
2352 bail_early:
2353 	*ppos += *lenp;
2354 	return err;
2355 }
2356 
2357 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2358 			       size_t *lenp, loff_t *ppos,
2359 			       int (*conv)(unsigned long *lvalp,
2360 					   unsigned int *valp,
2361 					   int write, void *data),
2362 			       void *data)
2363 {
2364 	unsigned long lval;
2365 	int err = 0;
2366 	size_t left;
2367 
2368 	left = *lenp;
2369 
2370 	if (conv(&lval, tbl_data, 0, data)) {
2371 		err = -EINVAL;
2372 		goto out;
2373 	}
2374 
2375 	err = proc_put_long(&buffer, &left, lval, false);
2376 	if (err || !left)
2377 		goto out;
2378 
2379 	err = proc_put_char(&buffer, &left, '\n');
2380 
2381 out:
2382 	*lenp -= left;
2383 	*ppos += *lenp;
2384 
2385 	return err;
2386 }
2387 
2388 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2389 			       int write, void __user *buffer,
2390 			       size_t *lenp, loff_t *ppos,
2391 			       int (*conv)(unsigned long *lvalp,
2392 					   unsigned int *valp,
2393 					   int write, void *data),
2394 			       void *data)
2395 {
2396 	unsigned int *i, vleft;
2397 
2398 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2399 		*lenp = 0;
2400 		return 0;
2401 	}
2402 
2403 	i = (unsigned int *) tbl_data;
2404 	vleft = table->maxlen / sizeof(*i);
2405 
2406 	/*
2407 	 * Arrays are not supported, keep this simple. *Do not* add
2408 	 * support for them.
2409 	 */
2410 	if (vleft != 1) {
2411 		*lenp = 0;
2412 		return -EINVAL;
2413 	}
2414 
2415 	if (!conv)
2416 		conv = do_proc_douintvec_conv;
2417 
2418 	if (write)
2419 		return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2420 					   conv, data);
2421 	return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2422 }
2423 
2424 static int do_proc_douintvec(struct ctl_table *table, int write,
2425 			     void __user *buffer, size_t *lenp, loff_t *ppos,
2426 			     int (*conv)(unsigned long *lvalp,
2427 					 unsigned int *valp,
2428 					 int write, void *data),
2429 			     void *data)
2430 {
2431 	return __do_proc_douintvec(table->data, table, write,
2432 				   buffer, lenp, ppos, conv, data);
2433 }
2434 
2435 /**
2436  * proc_dointvec - read a vector of integers
2437  * @table: the sysctl table
2438  * @write: %TRUE if this is a write to the sysctl file
2439  * @buffer: the user buffer
2440  * @lenp: the size of the user buffer
2441  * @ppos: file position
2442  *
2443  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2444  * values from/to the user buffer, treated as an ASCII string.
2445  *
2446  * Returns 0 on success.
2447  */
2448 int proc_dointvec(struct ctl_table *table, int write,
2449 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2450 {
2451 	return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2452 }
2453 
2454 /**
2455  * proc_douintvec - read a vector of unsigned integers
2456  * @table: the sysctl table
2457  * @write: %TRUE if this is a write to the sysctl file
2458  * @buffer: the user buffer
2459  * @lenp: the size of the user buffer
2460  * @ppos: file position
2461  *
2462  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2463  * values from/to the user buffer, treated as an ASCII string.
2464  *
2465  * Returns 0 on success.
2466  */
2467 int proc_douintvec(struct ctl_table *table, int write,
2468 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2469 {
2470 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
2471 				 do_proc_douintvec_conv, NULL);
2472 }
2473 
2474 /*
2475  * Taint values can only be increased
2476  * This means we can safely use a temporary.
2477  */
2478 static int proc_taint(struct ctl_table *table, int write,
2479 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2480 {
2481 	struct ctl_table t;
2482 	unsigned long tmptaint = get_taint();
2483 	int err;
2484 
2485 	if (write && !capable(CAP_SYS_ADMIN))
2486 		return -EPERM;
2487 
2488 	t = *table;
2489 	t.data = &tmptaint;
2490 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2491 	if (err < 0)
2492 		return err;
2493 
2494 	if (write) {
2495 		/*
2496 		 * Poor man's atomic or. Not worth adding a primitive
2497 		 * to everyone's atomic.h for this
2498 		 */
2499 		int i;
2500 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2501 			if ((tmptaint >> i) & 1)
2502 				add_taint(i, LOCKDEP_STILL_OK);
2503 		}
2504 	}
2505 
2506 	return err;
2507 }
2508 
2509 #ifdef CONFIG_PRINTK
2510 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2511 				void __user *buffer, size_t *lenp, loff_t *ppos)
2512 {
2513 	if (write && !capable(CAP_SYS_ADMIN))
2514 		return -EPERM;
2515 
2516 	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2517 }
2518 #endif
2519 
2520 struct do_proc_dointvec_minmax_conv_param {
2521 	int *min;
2522 	int *max;
2523 };
2524 
2525 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2526 					int *valp,
2527 					int write, void *data)
2528 {
2529 	struct do_proc_dointvec_minmax_conv_param *param = data;
2530 	if (write) {
2531 		int val = *negp ? -*lvalp : *lvalp;
2532 		if ((param->min && *param->min > val) ||
2533 		    (param->max && *param->max < val))
2534 			return -EINVAL;
2535 		*valp = val;
2536 	} else {
2537 		int val = *valp;
2538 		if (val < 0) {
2539 			*negp = true;
2540 			*lvalp = -(unsigned long)val;
2541 		} else {
2542 			*negp = false;
2543 			*lvalp = (unsigned long)val;
2544 		}
2545 	}
2546 	return 0;
2547 }
2548 
2549 /**
2550  * proc_dointvec_minmax - read a vector of integers with min/max values
2551  * @table: the sysctl table
2552  * @write: %TRUE if this is a write to the sysctl file
2553  * @buffer: the user buffer
2554  * @lenp: the size of the user buffer
2555  * @ppos: file position
2556  *
2557  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2558  * values from/to the user buffer, treated as an ASCII string.
2559  *
2560  * This routine will ensure the values are within the range specified by
2561  * table->extra1 (min) and table->extra2 (max).
2562  *
2563  * Returns 0 on success.
2564  */
2565 int proc_dointvec_minmax(struct ctl_table *table, int write,
2566 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2567 {
2568 	struct do_proc_dointvec_minmax_conv_param param = {
2569 		.min = (int *) table->extra1,
2570 		.max = (int *) table->extra2,
2571 	};
2572 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2573 				do_proc_dointvec_minmax_conv, &param);
2574 }
2575 
2576 struct do_proc_douintvec_minmax_conv_param {
2577 	unsigned int *min;
2578 	unsigned int *max;
2579 };
2580 
2581 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2582 					 unsigned int *valp,
2583 					 int write, void *data)
2584 {
2585 	struct do_proc_douintvec_minmax_conv_param *param = data;
2586 
2587 	if (write) {
2588 		unsigned int val = *lvalp;
2589 
2590 		if ((param->min && *param->min > val) ||
2591 		    (param->max && *param->max < val))
2592 			return -ERANGE;
2593 
2594 		if (*lvalp > UINT_MAX)
2595 			return -EINVAL;
2596 		*valp = val;
2597 	} else {
2598 		unsigned int val = *valp;
2599 		*lvalp = (unsigned long) val;
2600 	}
2601 
2602 	return 0;
2603 }
2604 
2605 /**
2606  * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2607  * @table: the sysctl table
2608  * @write: %TRUE if this is a write to the sysctl file
2609  * @buffer: the user buffer
2610  * @lenp: the size of the user buffer
2611  * @ppos: file position
2612  *
2613  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2614  * values from/to the user buffer, treated as an ASCII string. Negative
2615  * strings are not allowed.
2616  *
2617  * This routine will ensure the values are within the range specified by
2618  * table->extra1 (min) and table->extra2 (max). There is a final sanity
2619  * check for UINT_MAX to avoid having to support wrap around uses from
2620  * userspace.
2621  *
2622  * Returns 0 on success.
2623  */
2624 int proc_douintvec_minmax(struct ctl_table *table, int write,
2625 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2626 {
2627 	struct do_proc_douintvec_minmax_conv_param param = {
2628 		.min = (unsigned int *) table->extra1,
2629 		.max = (unsigned int *) table->extra2,
2630 	};
2631 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
2632 				 do_proc_douintvec_minmax_conv, &param);
2633 }
2634 
2635 static void validate_coredump_safety(void)
2636 {
2637 #ifdef CONFIG_COREDUMP
2638 	if (suid_dumpable == SUID_DUMP_ROOT &&
2639 	    core_pattern[0] != '/' && core_pattern[0] != '|') {
2640 		printk(KERN_WARNING
2641 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2642 "Pipe handler or fully qualified core dump path required.\n"
2643 "Set kernel.core_pattern before fs.suid_dumpable.\n"
2644 		);
2645 	}
2646 #endif
2647 }
2648 
2649 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2650 		void __user *buffer, size_t *lenp, loff_t *ppos)
2651 {
2652 	int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2653 	if (!error)
2654 		validate_coredump_safety();
2655 	return error;
2656 }
2657 
2658 #ifdef CONFIG_COREDUMP
2659 static int proc_dostring_coredump(struct ctl_table *table, int write,
2660 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2661 {
2662 	int error = proc_dostring(table, write, buffer, lenp, ppos);
2663 	if (!error)
2664 		validate_coredump_safety();
2665 	return error;
2666 }
2667 #endif
2668 
2669 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2670 				     void __user *buffer,
2671 				     size_t *lenp, loff_t *ppos,
2672 				     unsigned long convmul,
2673 				     unsigned long convdiv)
2674 {
2675 	unsigned long *i, *min, *max;
2676 	int vleft, first = 1, err = 0;
2677 	size_t left;
2678 	char *kbuf = NULL, *p;
2679 
2680 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2681 		*lenp = 0;
2682 		return 0;
2683 	}
2684 
2685 	i = (unsigned long *) data;
2686 	min = (unsigned long *) table->extra1;
2687 	max = (unsigned long *) table->extra2;
2688 	vleft = table->maxlen / sizeof(unsigned long);
2689 	left = *lenp;
2690 
2691 	if (write) {
2692 		if (proc_first_pos_non_zero_ignore(ppos, table))
2693 			goto out;
2694 
2695 		if (left > PAGE_SIZE - 1)
2696 			left = PAGE_SIZE - 1;
2697 		p = kbuf = memdup_user_nul(buffer, left);
2698 		if (IS_ERR(kbuf))
2699 			return PTR_ERR(kbuf);
2700 	}
2701 
2702 	for (; left && vleft--; i++, first = 0) {
2703 		unsigned long val;
2704 
2705 		if (write) {
2706 			bool neg;
2707 
2708 			left -= proc_skip_spaces(&p);
2709 
2710 			err = proc_get_long(&p, &left, &val, &neg,
2711 					     proc_wspace_sep,
2712 					     sizeof(proc_wspace_sep), NULL);
2713 			if (err)
2714 				break;
2715 			if (neg)
2716 				continue;
2717 			val = convmul * val / convdiv;
2718 			if ((min && val < *min) || (max && val > *max))
2719 				continue;
2720 			*i = val;
2721 		} else {
2722 			val = convdiv * (*i) / convmul;
2723 			if (!first) {
2724 				err = proc_put_char(&buffer, &left, '\t');
2725 				if (err)
2726 					break;
2727 			}
2728 			err = proc_put_long(&buffer, &left, val, false);
2729 			if (err)
2730 				break;
2731 		}
2732 	}
2733 
2734 	if (!write && !first && left && !err)
2735 		err = proc_put_char(&buffer, &left, '\n');
2736 	if (write && !err)
2737 		left -= proc_skip_spaces(&p);
2738 	if (write) {
2739 		kfree(kbuf);
2740 		if (first)
2741 			return err ? : -EINVAL;
2742 	}
2743 	*lenp -= left;
2744 out:
2745 	*ppos += *lenp;
2746 	return err;
2747 }
2748 
2749 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2750 				     void __user *buffer,
2751 				     size_t *lenp, loff_t *ppos,
2752 				     unsigned long convmul,
2753 				     unsigned long convdiv)
2754 {
2755 	return __do_proc_doulongvec_minmax(table->data, table, write,
2756 			buffer, lenp, ppos, convmul, convdiv);
2757 }
2758 
2759 /**
2760  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2761  * @table: the sysctl table
2762  * @write: %TRUE if this is a write to the sysctl file
2763  * @buffer: the user buffer
2764  * @lenp: the size of the user buffer
2765  * @ppos: file position
2766  *
2767  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2768  * values from/to the user buffer, treated as an ASCII string.
2769  *
2770  * This routine will ensure the values are within the range specified by
2771  * table->extra1 (min) and table->extra2 (max).
2772  *
2773  * Returns 0 on success.
2774  */
2775 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2776 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2777 {
2778     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2779 }
2780 
2781 /**
2782  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2783  * @table: the sysctl table
2784  * @write: %TRUE if this is a write to the sysctl file
2785  * @buffer: the user buffer
2786  * @lenp: the size of the user buffer
2787  * @ppos: file position
2788  *
2789  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2790  * values from/to the user buffer, treated as an ASCII string. The values
2791  * are treated as milliseconds, and converted to jiffies when they are stored.
2792  *
2793  * This routine will ensure the values are within the range specified by
2794  * table->extra1 (min) and table->extra2 (max).
2795  *
2796  * Returns 0 on success.
2797  */
2798 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2799 				      void __user *buffer,
2800 				      size_t *lenp, loff_t *ppos)
2801 {
2802     return do_proc_doulongvec_minmax(table, write, buffer,
2803 				     lenp, ppos, HZ, 1000l);
2804 }
2805 
2806 
2807 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2808 					 int *valp,
2809 					 int write, void *data)
2810 {
2811 	if (write) {
2812 		if (*lvalp > INT_MAX / HZ)
2813 			return 1;
2814 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2815 	} else {
2816 		int val = *valp;
2817 		unsigned long lval;
2818 		if (val < 0) {
2819 			*negp = true;
2820 			lval = -(unsigned long)val;
2821 		} else {
2822 			*negp = false;
2823 			lval = (unsigned long)val;
2824 		}
2825 		*lvalp = lval / HZ;
2826 	}
2827 	return 0;
2828 }
2829 
2830 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2831 						int *valp,
2832 						int write, void *data)
2833 {
2834 	if (write) {
2835 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2836 			return 1;
2837 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2838 	} else {
2839 		int val = *valp;
2840 		unsigned long lval;
2841 		if (val < 0) {
2842 			*negp = true;
2843 			lval = -(unsigned long)val;
2844 		} else {
2845 			*negp = false;
2846 			lval = (unsigned long)val;
2847 		}
2848 		*lvalp = jiffies_to_clock_t(lval);
2849 	}
2850 	return 0;
2851 }
2852 
2853 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2854 					    int *valp,
2855 					    int write, void *data)
2856 {
2857 	if (write) {
2858 		unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2859 
2860 		if (jif > INT_MAX)
2861 			return 1;
2862 		*valp = (int)jif;
2863 	} else {
2864 		int val = *valp;
2865 		unsigned long lval;
2866 		if (val < 0) {
2867 			*negp = true;
2868 			lval = -(unsigned long)val;
2869 		} else {
2870 			*negp = false;
2871 			lval = (unsigned long)val;
2872 		}
2873 		*lvalp = jiffies_to_msecs(lval);
2874 	}
2875 	return 0;
2876 }
2877 
2878 /**
2879  * proc_dointvec_jiffies - read a vector of integers as seconds
2880  * @table: the sysctl table
2881  * @write: %TRUE if this is a write to the sysctl file
2882  * @buffer: the user buffer
2883  * @lenp: the size of the user buffer
2884  * @ppos: file position
2885  *
2886  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2887  * values from/to the user buffer, treated as an ASCII string.
2888  * The values read are assumed to be in seconds, and are converted into
2889  * jiffies.
2890  *
2891  * Returns 0 on success.
2892  */
2893 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2894 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2895 {
2896     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2897 		    	    do_proc_dointvec_jiffies_conv,NULL);
2898 }
2899 
2900 /**
2901  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2902  * @table: the sysctl table
2903  * @write: %TRUE if this is a write to the sysctl file
2904  * @buffer: the user buffer
2905  * @lenp: the size of the user buffer
2906  * @ppos: pointer to the file position
2907  *
2908  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2909  * values from/to the user buffer, treated as an ASCII string.
2910  * The values read are assumed to be in 1/USER_HZ seconds, and
2911  * are converted into jiffies.
2912  *
2913  * Returns 0 on success.
2914  */
2915 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2916 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2917 {
2918     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2919 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2920 }
2921 
2922 /**
2923  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2924  * @table: the sysctl table
2925  * @write: %TRUE if this is a write to the sysctl file
2926  * @buffer: the user buffer
2927  * @lenp: the size of the user buffer
2928  * @ppos: file position
2929  * @ppos: the current position in the file
2930  *
2931  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2932  * values from/to the user buffer, treated as an ASCII string.
2933  * The values read are assumed to be in 1/1000 seconds, and
2934  * are converted into jiffies.
2935  *
2936  * Returns 0 on success.
2937  */
2938 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2939 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2940 {
2941 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2942 				do_proc_dointvec_ms_jiffies_conv, NULL);
2943 }
2944 
2945 static int proc_do_cad_pid(struct ctl_table *table, int write,
2946 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2947 {
2948 	struct pid *new_pid;
2949 	pid_t tmp;
2950 	int r;
2951 
2952 	tmp = pid_vnr(cad_pid);
2953 
2954 	r = __do_proc_dointvec(&tmp, table, write, buffer,
2955 			       lenp, ppos, NULL, NULL);
2956 	if (r || !write)
2957 		return r;
2958 
2959 	new_pid = find_get_pid(tmp);
2960 	if (!new_pid)
2961 		return -ESRCH;
2962 
2963 	put_pid(xchg(&cad_pid, new_pid));
2964 	return 0;
2965 }
2966 
2967 /**
2968  * proc_do_large_bitmap - read/write from/to a large bitmap
2969  * @table: the sysctl table
2970  * @write: %TRUE if this is a write to the sysctl file
2971  * @buffer: the user buffer
2972  * @lenp: the size of the user buffer
2973  * @ppos: file position
2974  *
2975  * The bitmap is stored at table->data and the bitmap length (in bits)
2976  * in table->maxlen.
2977  *
2978  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2979  * large bitmaps may be represented in a compact manner. Writing into
2980  * the file will clear the bitmap then update it with the given input.
2981  *
2982  * Returns 0 on success.
2983  */
2984 int proc_do_large_bitmap(struct ctl_table *table, int write,
2985 			 void __user *buffer, size_t *lenp, loff_t *ppos)
2986 {
2987 	int err = 0;
2988 	bool first = 1;
2989 	size_t left = *lenp;
2990 	unsigned long bitmap_len = table->maxlen;
2991 	unsigned long *bitmap = *(unsigned long **) table->data;
2992 	unsigned long *tmp_bitmap = NULL;
2993 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2994 
2995 	if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2996 		*lenp = 0;
2997 		return 0;
2998 	}
2999 
3000 	if (write) {
3001 		char *kbuf, *p;
3002 
3003 		if (left > PAGE_SIZE - 1)
3004 			left = PAGE_SIZE - 1;
3005 
3006 		p = kbuf = memdup_user_nul(buffer, left);
3007 		if (IS_ERR(kbuf))
3008 			return PTR_ERR(kbuf);
3009 
3010 		tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
3011 				     GFP_KERNEL);
3012 		if (!tmp_bitmap) {
3013 			kfree(kbuf);
3014 			return -ENOMEM;
3015 		}
3016 		proc_skip_char(&p, &left, '\n');
3017 		while (!err && left) {
3018 			unsigned long val_a, val_b;
3019 			bool neg;
3020 
3021 			err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3022 					     sizeof(tr_a), &c);
3023 			if (err)
3024 				break;
3025 			if (val_a >= bitmap_len || neg) {
3026 				err = -EINVAL;
3027 				break;
3028 			}
3029 
3030 			val_b = val_a;
3031 			if (left) {
3032 				p++;
3033 				left--;
3034 			}
3035 
3036 			if (c == '-') {
3037 				err = proc_get_long(&p, &left, &val_b,
3038 						     &neg, tr_b, sizeof(tr_b),
3039 						     &c);
3040 				if (err)
3041 					break;
3042 				if (val_b >= bitmap_len || neg ||
3043 				    val_a > val_b) {
3044 					err = -EINVAL;
3045 					break;
3046 				}
3047 				if (left) {
3048 					p++;
3049 					left--;
3050 				}
3051 			}
3052 
3053 			bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3054 			first = 0;
3055 			proc_skip_char(&p, &left, '\n');
3056 		}
3057 		kfree(kbuf);
3058 	} else {
3059 		unsigned long bit_a, bit_b = 0;
3060 
3061 		while (left) {
3062 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3063 			if (bit_a >= bitmap_len)
3064 				break;
3065 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
3066 						   bit_a + 1) - 1;
3067 
3068 			if (!first) {
3069 				err = proc_put_char(&buffer, &left, ',');
3070 				if (err)
3071 					break;
3072 			}
3073 			err = proc_put_long(&buffer, &left, bit_a, false);
3074 			if (err)
3075 				break;
3076 			if (bit_a != bit_b) {
3077 				err = proc_put_char(&buffer, &left, '-');
3078 				if (err)
3079 					break;
3080 				err = proc_put_long(&buffer, &left, bit_b, false);
3081 				if (err)
3082 					break;
3083 			}
3084 
3085 			first = 0; bit_b++;
3086 		}
3087 		if (!err)
3088 			err = proc_put_char(&buffer, &left, '\n');
3089 	}
3090 
3091 	if (!err) {
3092 		if (write) {
3093 			if (*ppos)
3094 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3095 			else
3096 				bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3097 		}
3098 		kfree(tmp_bitmap);
3099 		*lenp -= left;
3100 		*ppos += *lenp;
3101 		return 0;
3102 	} else {
3103 		kfree(tmp_bitmap);
3104 		return err;
3105 	}
3106 }
3107 
3108 #else /* CONFIG_PROC_SYSCTL */
3109 
3110 int proc_dostring(struct ctl_table *table, int write,
3111 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3112 {
3113 	return -ENOSYS;
3114 }
3115 
3116 int proc_dointvec(struct ctl_table *table, int write,
3117 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3118 {
3119 	return -ENOSYS;
3120 }
3121 
3122 int proc_douintvec(struct ctl_table *table, int write,
3123 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3124 {
3125 	return -ENOSYS;
3126 }
3127 
3128 int proc_dointvec_minmax(struct ctl_table *table, int write,
3129 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3130 {
3131 	return -ENOSYS;
3132 }
3133 
3134 int proc_douintvec_minmax(struct ctl_table *table, int write,
3135 			  void __user *buffer, size_t *lenp, loff_t *ppos)
3136 {
3137 	return -ENOSYS;
3138 }
3139 
3140 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3141 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3142 {
3143 	return -ENOSYS;
3144 }
3145 
3146 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3147 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3148 {
3149 	return -ENOSYS;
3150 }
3151 
3152 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3153 			     void __user *buffer, size_t *lenp, loff_t *ppos)
3154 {
3155 	return -ENOSYS;
3156 }
3157 
3158 int proc_doulongvec_minmax(struct ctl_table *table, int write,
3159 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3160 {
3161 	return -ENOSYS;
3162 }
3163 
3164 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3165 				      void __user *buffer,
3166 				      size_t *lenp, loff_t *ppos)
3167 {
3168     return -ENOSYS;
3169 }
3170 
3171 
3172 #endif /* CONFIG_PROC_SYSCTL */
3173 
3174 /*
3175  * No sense putting this after each symbol definition, twice,
3176  * exception granted :-)
3177  */
3178 EXPORT_SYMBOL(proc_dointvec);
3179 EXPORT_SYMBOL(proc_douintvec);
3180 EXPORT_SYMBOL(proc_dointvec_jiffies);
3181 EXPORT_SYMBOL(proc_dointvec_minmax);
3182 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3183 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3184 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3185 EXPORT_SYMBOL(proc_dostring);
3186 EXPORT_SYMBOL(proc_doulongvec_minmax);
3187 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3188