xref: /linux/kernel/sysctl.c (revision 2fe05e1139a555ae91f00a812cb9520e7d3022ab)
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,
371 	},
372 #ifdef CONFIG_SCHEDSTATS
373 	{
374 		.procname	= "sched_schedstats",
375 		.data		= NULL,
376 		.maxlen		= sizeof(unsigned int),
377 		.mode		= 0644,
378 		.proc_handler	= sysctl_schedstats,
379 		.extra1		= &zero,
380 		.extra2		= &one,
381 	},
382 #endif /* CONFIG_SCHEDSTATS */
383 #endif /* CONFIG_SMP */
384 #ifdef CONFIG_NUMA_BALANCING
385 	{
386 		.procname	= "numa_balancing_scan_delay_ms",
387 		.data		= &sysctl_numa_balancing_scan_delay,
388 		.maxlen		= sizeof(unsigned int),
389 		.mode		= 0644,
390 		.proc_handler	= proc_dointvec,
391 	},
392 	{
393 		.procname	= "numa_balancing_scan_period_min_ms",
394 		.data		= &sysctl_numa_balancing_scan_period_min,
395 		.maxlen		= sizeof(unsigned int),
396 		.mode		= 0644,
397 		.proc_handler	= proc_dointvec,
398 	},
399 	{
400 		.procname	= "numa_balancing_scan_period_max_ms",
401 		.data		= &sysctl_numa_balancing_scan_period_max,
402 		.maxlen		= sizeof(unsigned int),
403 		.mode		= 0644,
404 		.proc_handler	= proc_dointvec,
405 	},
406 	{
407 		.procname	= "numa_balancing_scan_size_mb",
408 		.data		= &sysctl_numa_balancing_scan_size,
409 		.maxlen		= sizeof(unsigned int),
410 		.mode		= 0644,
411 		.proc_handler	= proc_dointvec_minmax,
412 		.extra1		= &one,
413 	},
414 	{
415 		.procname	= "numa_balancing",
416 		.data		= NULL, /* filled in by handler */
417 		.maxlen		= sizeof(unsigned int),
418 		.mode		= 0644,
419 		.proc_handler	= sysctl_numa_balancing,
420 		.extra1		= &zero,
421 		.extra2		= &one,
422 	},
423 #endif /* CONFIG_NUMA_BALANCING */
424 #endif /* CONFIG_SCHED_DEBUG */
425 	{
426 		.procname	= "sched_rt_period_us",
427 		.data		= &sysctl_sched_rt_period,
428 		.maxlen		= sizeof(unsigned int),
429 		.mode		= 0644,
430 		.proc_handler	= sched_rt_handler,
431 	},
432 	{
433 		.procname	= "sched_rt_runtime_us",
434 		.data		= &sysctl_sched_rt_runtime,
435 		.maxlen		= sizeof(int),
436 		.mode		= 0644,
437 		.proc_handler	= sched_rt_handler,
438 	},
439 	{
440 		.procname	= "sched_rr_timeslice_ms",
441 		.data		= &sysctl_sched_rr_timeslice,
442 		.maxlen		= sizeof(int),
443 		.mode		= 0644,
444 		.proc_handler	= sched_rr_handler,
445 	},
446 #ifdef CONFIG_SCHED_AUTOGROUP
447 	{
448 		.procname	= "sched_autogroup_enabled",
449 		.data		= &sysctl_sched_autogroup_enabled,
450 		.maxlen		= sizeof(unsigned int),
451 		.mode		= 0644,
452 		.proc_handler	= proc_dointvec_minmax,
453 		.extra1		= &zero,
454 		.extra2		= &one,
455 	},
456 #endif
457 #ifdef CONFIG_CFS_BANDWIDTH
458 	{
459 		.procname	= "sched_cfs_bandwidth_slice_us",
460 		.data		= &sysctl_sched_cfs_bandwidth_slice,
461 		.maxlen		= sizeof(unsigned int),
462 		.mode		= 0644,
463 		.proc_handler	= proc_dointvec_minmax,
464 		.extra1		= &one,
465 	},
466 #endif
467 #ifdef CONFIG_PROVE_LOCKING
468 	{
469 		.procname	= "prove_locking",
470 		.data		= &prove_locking,
471 		.maxlen		= sizeof(int),
472 		.mode		= 0644,
473 		.proc_handler	= proc_dointvec,
474 	},
475 #endif
476 #ifdef CONFIG_LOCK_STAT
477 	{
478 		.procname	= "lock_stat",
479 		.data		= &lock_stat,
480 		.maxlen		= sizeof(int),
481 		.mode		= 0644,
482 		.proc_handler	= proc_dointvec,
483 	},
484 #endif
485 	{
486 		.procname	= "panic",
487 		.data		= &panic_timeout,
488 		.maxlen		= sizeof(int),
489 		.mode		= 0644,
490 		.proc_handler	= proc_dointvec,
491 	},
492 #ifdef CONFIG_COREDUMP
493 	{
494 		.procname	= "core_uses_pid",
495 		.data		= &core_uses_pid,
496 		.maxlen		= sizeof(int),
497 		.mode		= 0644,
498 		.proc_handler	= proc_dointvec,
499 	},
500 	{
501 		.procname	= "core_pattern",
502 		.data		= core_pattern,
503 		.maxlen		= CORENAME_MAX_SIZE,
504 		.mode		= 0644,
505 		.proc_handler	= proc_dostring_coredump,
506 	},
507 	{
508 		.procname	= "core_pipe_limit",
509 		.data		= &core_pipe_limit,
510 		.maxlen		= sizeof(unsigned int),
511 		.mode		= 0644,
512 		.proc_handler	= proc_dointvec,
513 	},
514 #endif
515 #ifdef CONFIG_PROC_SYSCTL
516 	{
517 		.procname	= "tainted",
518 		.maxlen 	= sizeof(long),
519 		.mode		= 0644,
520 		.proc_handler	= proc_taint,
521 	},
522 	{
523 		.procname	= "sysctl_writes_strict",
524 		.data		= &sysctl_writes_strict,
525 		.maxlen		= sizeof(int),
526 		.mode		= 0644,
527 		.proc_handler	= proc_dointvec_minmax,
528 		.extra1		= &neg_one,
529 		.extra2		= &one,
530 	},
531 #endif
532 #ifdef CONFIG_LATENCYTOP
533 	{
534 		.procname	= "latencytop",
535 		.data		= &latencytop_enabled,
536 		.maxlen		= sizeof(int),
537 		.mode		= 0644,
538 		.proc_handler	= sysctl_latencytop,
539 	},
540 #endif
541 #ifdef CONFIG_BLK_DEV_INITRD
542 	{
543 		.procname	= "real-root-dev",
544 		.data		= &real_root_dev,
545 		.maxlen		= sizeof(int),
546 		.mode		= 0644,
547 		.proc_handler	= proc_dointvec,
548 	},
549 #endif
550 	{
551 		.procname	= "print-fatal-signals",
552 		.data		= &print_fatal_signals,
553 		.maxlen		= sizeof(int),
554 		.mode		= 0644,
555 		.proc_handler	= proc_dointvec,
556 	},
557 #ifdef CONFIG_SPARC
558 	{
559 		.procname	= "reboot-cmd",
560 		.data		= reboot_command,
561 		.maxlen		= 256,
562 		.mode		= 0644,
563 		.proc_handler	= proc_dostring,
564 	},
565 	{
566 		.procname	= "stop-a",
567 		.data		= &stop_a_enabled,
568 		.maxlen		= sizeof (int),
569 		.mode		= 0644,
570 		.proc_handler	= proc_dointvec,
571 	},
572 	{
573 		.procname	= "scons-poweroff",
574 		.data		= &scons_pwroff,
575 		.maxlen		= sizeof (int),
576 		.mode		= 0644,
577 		.proc_handler	= proc_dointvec,
578 	},
579 #endif
580 #ifdef CONFIG_SPARC64
581 	{
582 		.procname	= "tsb-ratio",
583 		.data		= &sysctl_tsb_ratio,
584 		.maxlen		= sizeof (int),
585 		.mode		= 0644,
586 		.proc_handler	= proc_dointvec,
587 	},
588 #endif
589 #ifdef __hppa__
590 	{
591 		.procname	= "soft-power",
592 		.data		= &pwrsw_enabled,
593 		.maxlen		= sizeof (int),
594 	 	.mode		= 0644,
595 		.proc_handler	= proc_dointvec,
596 	},
597 #endif
598 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
599 	{
600 		.procname	= "unaligned-trap",
601 		.data		= &unaligned_enabled,
602 		.maxlen		= sizeof (int),
603 		.mode		= 0644,
604 		.proc_handler	= proc_dointvec,
605 	},
606 #endif
607 	{
608 		.procname	= "ctrl-alt-del",
609 		.data		= &C_A_D,
610 		.maxlen		= sizeof(int),
611 		.mode		= 0644,
612 		.proc_handler	= proc_dointvec,
613 	},
614 #ifdef CONFIG_FUNCTION_TRACER
615 	{
616 		.procname	= "ftrace_enabled",
617 		.data		= &ftrace_enabled,
618 		.maxlen		= sizeof(int),
619 		.mode		= 0644,
620 		.proc_handler	= ftrace_enable_sysctl,
621 	},
622 #endif
623 #ifdef CONFIG_STACK_TRACER
624 	{
625 		.procname	= "stack_tracer_enabled",
626 		.data		= &stack_tracer_enabled,
627 		.maxlen		= sizeof(int),
628 		.mode		= 0644,
629 		.proc_handler	= stack_trace_sysctl,
630 	},
631 #endif
632 #ifdef CONFIG_TRACING
633 	{
634 		.procname	= "ftrace_dump_on_oops",
635 		.data		= &ftrace_dump_on_oops,
636 		.maxlen		= sizeof(int),
637 		.mode		= 0644,
638 		.proc_handler	= proc_dointvec,
639 	},
640 	{
641 		.procname	= "traceoff_on_warning",
642 		.data		= &__disable_trace_on_warning,
643 		.maxlen		= sizeof(__disable_trace_on_warning),
644 		.mode		= 0644,
645 		.proc_handler	= proc_dointvec,
646 	},
647 	{
648 		.procname	= "tracepoint_printk",
649 		.data		= &tracepoint_printk,
650 		.maxlen		= sizeof(tracepoint_printk),
651 		.mode		= 0644,
652 		.proc_handler	= tracepoint_printk_sysctl,
653 	},
654 #endif
655 #ifdef CONFIG_KEXEC_CORE
656 	{
657 		.procname	= "kexec_load_disabled",
658 		.data		= &kexec_load_disabled,
659 		.maxlen		= sizeof(int),
660 		.mode		= 0644,
661 		/* only handle a transition from default "0" to "1" */
662 		.proc_handler	= proc_dointvec_minmax,
663 		.extra1		= &one,
664 		.extra2		= &one,
665 	},
666 #endif
667 #ifdef CONFIG_MODULES
668 	{
669 		.procname	= "modprobe",
670 		.data		= &modprobe_path,
671 		.maxlen		= KMOD_PATH_LEN,
672 		.mode		= 0644,
673 		.proc_handler	= proc_dostring,
674 	},
675 	{
676 		.procname	= "modules_disabled",
677 		.data		= &modules_disabled,
678 		.maxlen		= sizeof(int),
679 		.mode		= 0644,
680 		/* only handle a transition from default "0" to "1" */
681 		.proc_handler	= proc_dointvec_minmax,
682 		.extra1		= &one,
683 		.extra2		= &one,
684 	},
685 #endif
686 #ifdef CONFIG_UEVENT_HELPER
687 	{
688 		.procname	= "hotplug",
689 		.data		= &uevent_helper,
690 		.maxlen		= UEVENT_HELPER_PATH_LEN,
691 		.mode		= 0644,
692 		.proc_handler	= proc_dostring,
693 	},
694 #endif
695 #ifdef CONFIG_CHR_DEV_SG
696 	{
697 		.procname	= "sg-big-buff",
698 		.data		= &sg_big_buff,
699 		.maxlen		= sizeof (int),
700 		.mode		= 0444,
701 		.proc_handler	= proc_dointvec,
702 	},
703 #endif
704 #ifdef CONFIG_BSD_PROCESS_ACCT
705 	{
706 		.procname	= "acct",
707 		.data		= &acct_parm,
708 		.maxlen		= 3*sizeof(int),
709 		.mode		= 0644,
710 		.proc_handler	= proc_dointvec,
711 	},
712 #endif
713 #ifdef CONFIG_MAGIC_SYSRQ
714 	{
715 		.procname	= "sysrq",
716 		.data		= &__sysrq_enabled,
717 		.maxlen		= sizeof (int),
718 		.mode		= 0644,
719 		.proc_handler	= sysrq_sysctl_handler,
720 	},
721 #endif
722 #ifdef CONFIG_PROC_SYSCTL
723 	{
724 		.procname	= "cad_pid",
725 		.data		= NULL,
726 		.maxlen		= sizeof (int),
727 		.mode		= 0600,
728 		.proc_handler	= proc_do_cad_pid,
729 	},
730 #endif
731 	{
732 		.procname	= "threads-max",
733 		.data		= NULL,
734 		.maxlen		= sizeof(int),
735 		.mode		= 0644,
736 		.proc_handler	= sysctl_max_threads,
737 	},
738 	{
739 		.procname	= "random",
740 		.mode		= 0555,
741 		.child		= random_table,
742 	},
743 	{
744 		.procname	= "usermodehelper",
745 		.mode		= 0555,
746 		.child		= usermodehelper_table,
747 	},
748 	{
749 		.procname	= "overflowuid",
750 		.data		= &overflowuid,
751 		.maxlen		= sizeof(int),
752 		.mode		= 0644,
753 		.proc_handler	= proc_dointvec_minmax,
754 		.extra1		= &minolduid,
755 		.extra2		= &maxolduid,
756 	},
757 	{
758 		.procname	= "overflowgid",
759 		.data		= &overflowgid,
760 		.maxlen		= sizeof(int),
761 		.mode		= 0644,
762 		.proc_handler	= proc_dointvec_minmax,
763 		.extra1		= &minolduid,
764 		.extra2		= &maxolduid,
765 	},
766 #ifdef CONFIG_S390
767 #ifdef CONFIG_MATHEMU
768 	{
769 		.procname	= "ieee_emulation_warnings",
770 		.data		= &sysctl_ieee_emulation_warnings,
771 		.maxlen		= sizeof(int),
772 		.mode		= 0644,
773 		.proc_handler	= proc_dointvec,
774 	},
775 #endif
776 	{
777 		.procname	= "userprocess_debug",
778 		.data		= &show_unhandled_signals,
779 		.maxlen		= sizeof(int),
780 		.mode		= 0644,
781 		.proc_handler	= proc_dointvec,
782 	},
783 #endif
784 	{
785 		.procname	= "pid_max",
786 		.data		= &pid_max,
787 		.maxlen		= sizeof (int),
788 		.mode		= 0644,
789 		.proc_handler	= proc_dointvec_minmax,
790 		.extra1		= &pid_max_min,
791 		.extra2		= &pid_max_max,
792 	},
793 	{
794 		.procname	= "panic_on_oops",
795 		.data		= &panic_on_oops,
796 		.maxlen		= sizeof(int),
797 		.mode		= 0644,
798 		.proc_handler	= proc_dointvec,
799 	},
800 #if defined CONFIG_PRINTK
801 	{
802 		.procname	= "printk",
803 		.data		= &console_loglevel,
804 		.maxlen		= 4*sizeof(int),
805 		.mode		= 0644,
806 		.proc_handler	= proc_dointvec,
807 	},
808 	{
809 		.procname	= "printk_ratelimit",
810 		.data		= &printk_ratelimit_state.interval,
811 		.maxlen		= sizeof(int),
812 		.mode		= 0644,
813 		.proc_handler	= proc_dointvec_jiffies,
814 	},
815 	{
816 		.procname	= "printk_ratelimit_burst",
817 		.data		= &printk_ratelimit_state.burst,
818 		.maxlen		= sizeof(int),
819 		.mode		= 0644,
820 		.proc_handler	= proc_dointvec,
821 	},
822 	{
823 		.procname	= "printk_delay",
824 		.data		= &printk_delay_msec,
825 		.maxlen		= sizeof(int),
826 		.mode		= 0644,
827 		.proc_handler	= proc_dointvec_minmax,
828 		.extra1		= &zero,
829 		.extra2		= &ten_thousand,
830 	},
831 	{
832 		.procname	= "printk_devkmsg",
833 		.data		= devkmsg_log_str,
834 		.maxlen		= DEVKMSG_STR_MAX_SIZE,
835 		.mode		= 0644,
836 		.proc_handler	= devkmsg_sysctl_set_loglvl,
837 	},
838 	{
839 		.procname	= "dmesg_restrict",
840 		.data		= &dmesg_restrict,
841 		.maxlen		= sizeof(int),
842 		.mode		= 0644,
843 		.proc_handler	= proc_dointvec_minmax_sysadmin,
844 		.extra1		= &zero,
845 		.extra2		= &one,
846 	},
847 	{
848 		.procname	= "kptr_restrict",
849 		.data		= &kptr_restrict,
850 		.maxlen		= sizeof(int),
851 		.mode		= 0644,
852 		.proc_handler	= proc_dointvec_minmax_sysadmin,
853 		.extra1		= &zero,
854 		.extra2		= &two,
855 	},
856 #endif
857 	{
858 		.procname	= "ngroups_max",
859 		.data		= &ngroups_max,
860 		.maxlen		= sizeof (int),
861 		.mode		= 0444,
862 		.proc_handler	= proc_dointvec,
863 	},
864 	{
865 		.procname	= "cap_last_cap",
866 		.data		= (void *)&cap_last_cap,
867 		.maxlen		= sizeof(int),
868 		.mode		= 0444,
869 		.proc_handler	= proc_dointvec,
870 	},
871 #if defined(CONFIG_LOCKUP_DETECTOR)
872 	{
873 		.procname       = "watchdog",
874 		.data           = &watchdog_user_enabled,
875 		.maxlen         = sizeof (int),
876 		.mode           = 0644,
877 		.proc_handler   = proc_watchdog,
878 		.extra1		= &zero,
879 		.extra2		= &one,
880 	},
881 	{
882 		.procname	= "watchdog_thresh",
883 		.data		= &watchdog_thresh,
884 		.maxlen		= sizeof(int),
885 		.mode		= 0644,
886 		.proc_handler	= proc_watchdog_thresh,
887 		.extra1		= &zero,
888 		.extra2		= &sixty,
889 	},
890 	{
891 		.procname       = "nmi_watchdog",
892 		.data           = &nmi_watchdog_enabled,
893 		.maxlen         = sizeof (int),
894 		.mode           = 0644,
895 		.proc_handler   = proc_nmi_watchdog,
896 		.extra1		= &zero,
897 #if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR)
898 		.extra2		= &one,
899 #else
900 		.extra2		= &zero,
901 #endif
902 	},
903 	{
904 		.procname	= "watchdog_cpumask",
905 		.data		= &watchdog_cpumask_bits,
906 		.maxlen		= NR_CPUS,
907 		.mode		= 0644,
908 		.proc_handler	= proc_watchdog_cpumask,
909 	},
910 #ifdef CONFIG_SOFTLOCKUP_DETECTOR
911 	{
912 		.procname       = "soft_watchdog",
913 		.data           = &soft_watchdog_enabled,
914 		.maxlen         = sizeof (int),
915 		.mode           = 0644,
916 		.proc_handler   = proc_soft_watchdog,
917 		.extra1		= &zero,
918 		.extra2		= &one,
919 	},
920 	{
921 		.procname	= "softlockup_panic",
922 		.data		= &softlockup_panic,
923 		.maxlen		= sizeof(int),
924 		.mode		= 0644,
925 		.proc_handler	= proc_dointvec_minmax,
926 		.extra1		= &zero,
927 		.extra2		= &one,
928 	},
929 #ifdef CONFIG_SMP
930 	{
931 		.procname	= "softlockup_all_cpu_backtrace",
932 		.data		= &sysctl_softlockup_all_cpu_backtrace,
933 		.maxlen		= sizeof(int),
934 		.mode		= 0644,
935 		.proc_handler	= proc_dointvec_minmax,
936 		.extra1		= &zero,
937 		.extra2		= &one,
938 	},
939 #endif /* CONFIG_SMP */
940 #endif
941 #ifdef CONFIG_HARDLOCKUP_DETECTOR
942 	{
943 		.procname	= "hardlockup_panic",
944 		.data		= &hardlockup_panic,
945 		.maxlen		= sizeof(int),
946 		.mode		= 0644,
947 		.proc_handler	= proc_dointvec_minmax,
948 		.extra1		= &zero,
949 		.extra2		= &one,
950 	},
951 #ifdef CONFIG_SMP
952 	{
953 		.procname	= "hardlockup_all_cpu_backtrace",
954 		.data		= &sysctl_hardlockup_all_cpu_backtrace,
955 		.maxlen		= sizeof(int),
956 		.mode		= 0644,
957 		.proc_handler	= proc_dointvec_minmax,
958 		.extra1		= &zero,
959 		.extra2		= &one,
960 	},
961 #endif /* CONFIG_SMP */
962 #endif
963 #endif
964 
965 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
966 	{
967 		.procname       = "unknown_nmi_panic",
968 		.data           = &unknown_nmi_panic,
969 		.maxlen         = sizeof (int),
970 		.mode           = 0644,
971 		.proc_handler   = proc_dointvec,
972 	},
973 #endif
974 #if defined(CONFIG_X86)
975 	{
976 		.procname	= "panic_on_unrecovered_nmi",
977 		.data		= &panic_on_unrecovered_nmi,
978 		.maxlen		= sizeof(int),
979 		.mode		= 0644,
980 		.proc_handler	= proc_dointvec,
981 	},
982 	{
983 		.procname	= "panic_on_io_nmi",
984 		.data		= &panic_on_io_nmi,
985 		.maxlen		= sizeof(int),
986 		.mode		= 0644,
987 		.proc_handler	= proc_dointvec,
988 	},
989 #ifdef CONFIG_DEBUG_STACKOVERFLOW
990 	{
991 		.procname	= "panic_on_stackoverflow",
992 		.data		= &sysctl_panic_on_stackoverflow,
993 		.maxlen		= sizeof(int),
994 		.mode		= 0644,
995 		.proc_handler	= proc_dointvec,
996 	},
997 #endif
998 	{
999 		.procname	= "bootloader_type",
1000 		.data		= &bootloader_type,
1001 		.maxlen		= sizeof (int),
1002 		.mode		= 0444,
1003 		.proc_handler	= proc_dointvec,
1004 	},
1005 	{
1006 		.procname	= "bootloader_version",
1007 		.data		= &bootloader_version,
1008 		.maxlen		= sizeof (int),
1009 		.mode		= 0444,
1010 		.proc_handler	= proc_dointvec,
1011 	},
1012 	{
1013 		.procname	= "io_delay_type",
1014 		.data		= &io_delay_type,
1015 		.maxlen		= sizeof(int),
1016 		.mode		= 0644,
1017 		.proc_handler	= proc_dointvec,
1018 	},
1019 #endif
1020 #if defined(CONFIG_MMU)
1021 	{
1022 		.procname	= "randomize_va_space",
1023 		.data		= &randomize_va_space,
1024 		.maxlen		= sizeof(int),
1025 		.mode		= 0644,
1026 		.proc_handler	= proc_dointvec,
1027 	},
1028 #endif
1029 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1030 	{
1031 		.procname	= "spin_retry",
1032 		.data		= &spin_retry,
1033 		.maxlen		= sizeof (int),
1034 		.mode		= 0644,
1035 		.proc_handler	= proc_dointvec,
1036 	},
1037 #endif
1038 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1039 	{
1040 		.procname	= "acpi_video_flags",
1041 		.data		= &acpi_realmode_flags,
1042 		.maxlen		= sizeof (unsigned long),
1043 		.mode		= 0644,
1044 		.proc_handler	= proc_doulongvec_minmax,
1045 	},
1046 #endif
1047 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1048 	{
1049 		.procname	= "ignore-unaligned-usertrap",
1050 		.data		= &no_unaligned_warning,
1051 		.maxlen		= sizeof (int),
1052 	 	.mode		= 0644,
1053 		.proc_handler	= proc_dointvec,
1054 	},
1055 #endif
1056 #ifdef CONFIG_IA64
1057 	{
1058 		.procname	= "unaligned-dump-stack",
1059 		.data		= &unaligned_dump_stack,
1060 		.maxlen		= sizeof (int),
1061 		.mode		= 0644,
1062 		.proc_handler	= proc_dointvec,
1063 	},
1064 #endif
1065 #ifdef CONFIG_DETECT_HUNG_TASK
1066 	{
1067 		.procname	= "hung_task_panic",
1068 		.data		= &sysctl_hung_task_panic,
1069 		.maxlen		= sizeof(int),
1070 		.mode		= 0644,
1071 		.proc_handler	= proc_dointvec_minmax,
1072 		.extra1		= &zero,
1073 		.extra2		= &one,
1074 	},
1075 	{
1076 		.procname	= "hung_task_check_count",
1077 		.data		= &sysctl_hung_task_check_count,
1078 		.maxlen		= sizeof(int),
1079 		.mode		= 0644,
1080 		.proc_handler	= proc_dointvec_minmax,
1081 		.extra1		= &zero,
1082 	},
1083 	{
1084 		.procname	= "hung_task_timeout_secs",
1085 		.data		= &sysctl_hung_task_timeout_secs,
1086 		.maxlen		= sizeof(unsigned long),
1087 		.mode		= 0644,
1088 		.proc_handler	= proc_dohung_task_timeout_secs,
1089 		.extra2		= &hung_task_timeout_max,
1090 	},
1091 	{
1092 		.procname	= "hung_task_warnings",
1093 		.data		= &sysctl_hung_task_warnings,
1094 		.maxlen		= sizeof(int),
1095 		.mode		= 0644,
1096 		.proc_handler	= proc_dointvec_minmax,
1097 		.extra1		= &neg_one,
1098 	},
1099 #endif
1100 #ifdef CONFIG_RT_MUTEXES
1101 	{
1102 		.procname	= "max_lock_depth",
1103 		.data		= &max_lock_depth,
1104 		.maxlen		= sizeof(int),
1105 		.mode		= 0644,
1106 		.proc_handler	= proc_dointvec,
1107 	},
1108 #endif
1109 	{
1110 		.procname	= "poweroff_cmd",
1111 		.data		= &poweroff_cmd,
1112 		.maxlen		= POWEROFF_CMD_PATH_LEN,
1113 		.mode		= 0644,
1114 		.proc_handler	= proc_dostring,
1115 	},
1116 #ifdef CONFIG_KEYS
1117 	{
1118 		.procname	= "keys",
1119 		.mode		= 0555,
1120 		.child		= key_sysctls,
1121 	},
1122 #endif
1123 #ifdef CONFIG_PERF_EVENTS
1124 	/*
1125 	 * User-space scripts rely on the existence of this file
1126 	 * as a feature check for perf_events being enabled.
1127 	 *
1128 	 * So it's an ABI, do not remove!
1129 	 */
1130 	{
1131 		.procname	= "perf_event_paranoid",
1132 		.data		= &sysctl_perf_event_paranoid,
1133 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
1134 		.mode		= 0644,
1135 		.proc_handler	= proc_dointvec,
1136 	},
1137 	{
1138 		.procname	= "perf_event_mlock_kb",
1139 		.data		= &sysctl_perf_event_mlock,
1140 		.maxlen		= sizeof(sysctl_perf_event_mlock),
1141 		.mode		= 0644,
1142 		.proc_handler	= proc_dointvec,
1143 	},
1144 	{
1145 		.procname	= "perf_event_max_sample_rate",
1146 		.data		= &sysctl_perf_event_sample_rate,
1147 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
1148 		.mode		= 0644,
1149 		.proc_handler	= perf_proc_update_handler,
1150 		.extra1		= &one,
1151 	},
1152 	{
1153 		.procname	= "perf_cpu_time_max_percent",
1154 		.data		= &sysctl_perf_cpu_time_max_percent,
1155 		.maxlen		= sizeof(sysctl_perf_cpu_time_max_percent),
1156 		.mode		= 0644,
1157 		.proc_handler	= perf_cpu_time_max_percent_handler,
1158 		.extra1		= &zero,
1159 		.extra2		= &one_hundred,
1160 	},
1161 	{
1162 		.procname	= "perf_event_max_stack",
1163 		.data		= &sysctl_perf_event_max_stack,
1164 		.maxlen		= sizeof(sysctl_perf_event_max_stack),
1165 		.mode		= 0644,
1166 		.proc_handler	= perf_event_max_stack_handler,
1167 		.extra1		= &zero,
1168 		.extra2		= &six_hundred_forty_kb,
1169 	},
1170 	{
1171 		.procname	= "perf_event_max_contexts_per_stack",
1172 		.data		= &sysctl_perf_event_max_contexts_per_stack,
1173 		.maxlen		= sizeof(sysctl_perf_event_max_contexts_per_stack),
1174 		.mode		= 0644,
1175 		.proc_handler	= perf_event_max_stack_handler,
1176 		.extra1		= &zero,
1177 		.extra2		= &one_thousand,
1178 	},
1179 #endif
1180 #ifdef CONFIG_KMEMCHECK
1181 	{
1182 		.procname	= "kmemcheck",
1183 		.data		= &kmemcheck_enabled,
1184 		.maxlen		= sizeof(int),
1185 		.mode		= 0644,
1186 		.proc_handler	= proc_dointvec,
1187 	},
1188 #endif
1189 	{
1190 		.procname	= "panic_on_warn",
1191 		.data		= &panic_on_warn,
1192 		.maxlen		= sizeof(int),
1193 		.mode		= 0644,
1194 		.proc_handler	= proc_dointvec_minmax,
1195 		.extra1		= &zero,
1196 		.extra2		= &one,
1197 	},
1198 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1199 	{
1200 		.procname	= "timer_migration",
1201 		.data		= &sysctl_timer_migration,
1202 		.maxlen		= sizeof(unsigned int),
1203 		.mode		= 0644,
1204 		.proc_handler	= timer_migration_handler,
1205 		.extra1		= &zero,
1206 		.extra2		= &one,
1207 	},
1208 #endif
1209 #ifdef CONFIG_BPF_SYSCALL
1210 	{
1211 		.procname	= "unprivileged_bpf_disabled",
1212 		.data		= &sysctl_unprivileged_bpf_disabled,
1213 		.maxlen		= sizeof(sysctl_unprivileged_bpf_disabled),
1214 		.mode		= 0644,
1215 		/* only handle a transition from default "0" to "1" */
1216 		.proc_handler	= proc_dointvec_minmax,
1217 		.extra1		= &one,
1218 		.extra2		= &one,
1219 	},
1220 #endif
1221 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1222 	{
1223 		.procname	= "panic_on_rcu_stall",
1224 		.data		= &sysctl_panic_on_rcu_stall,
1225 		.maxlen		= sizeof(sysctl_panic_on_rcu_stall),
1226 		.mode		= 0644,
1227 		.proc_handler	= proc_dointvec_minmax,
1228 		.extra1		= &zero,
1229 		.extra2		= &one,
1230 	},
1231 #endif
1232 	{ }
1233 };
1234 
1235 static struct ctl_table vm_table[] = {
1236 	{
1237 		.procname	= "overcommit_memory",
1238 		.data		= &sysctl_overcommit_memory,
1239 		.maxlen		= sizeof(sysctl_overcommit_memory),
1240 		.mode		= 0644,
1241 		.proc_handler	= proc_dointvec_minmax,
1242 		.extra1		= &zero,
1243 		.extra2		= &two,
1244 	},
1245 	{
1246 		.procname	= "panic_on_oom",
1247 		.data		= &sysctl_panic_on_oom,
1248 		.maxlen		= sizeof(sysctl_panic_on_oom),
1249 		.mode		= 0644,
1250 		.proc_handler	= proc_dointvec_minmax,
1251 		.extra1		= &zero,
1252 		.extra2		= &two,
1253 	},
1254 	{
1255 		.procname	= "oom_kill_allocating_task",
1256 		.data		= &sysctl_oom_kill_allocating_task,
1257 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1258 		.mode		= 0644,
1259 		.proc_handler	= proc_dointvec,
1260 	},
1261 	{
1262 		.procname	= "oom_dump_tasks",
1263 		.data		= &sysctl_oom_dump_tasks,
1264 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1265 		.mode		= 0644,
1266 		.proc_handler	= proc_dointvec,
1267 	},
1268 	{
1269 		.procname	= "overcommit_ratio",
1270 		.data		= &sysctl_overcommit_ratio,
1271 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1272 		.mode		= 0644,
1273 		.proc_handler	= overcommit_ratio_handler,
1274 	},
1275 	{
1276 		.procname	= "overcommit_kbytes",
1277 		.data		= &sysctl_overcommit_kbytes,
1278 		.maxlen		= sizeof(sysctl_overcommit_kbytes),
1279 		.mode		= 0644,
1280 		.proc_handler	= overcommit_kbytes_handler,
1281 	},
1282 	{
1283 		.procname	= "page-cluster",
1284 		.data		= &page_cluster,
1285 		.maxlen		= sizeof(int),
1286 		.mode		= 0644,
1287 		.proc_handler	= proc_dointvec_minmax,
1288 		.extra1		= &zero,
1289 	},
1290 	{
1291 		.procname	= "dirty_background_ratio",
1292 		.data		= &dirty_background_ratio,
1293 		.maxlen		= sizeof(dirty_background_ratio),
1294 		.mode		= 0644,
1295 		.proc_handler	= dirty_background_ratio_handler,
1296 		.extra1		= &zero,
1297 		.extra2		= &one_hundred,
1298 	},
1299 	{
1300 		.procname	= "dirty_background_bytes",
1301 		.data		= &dirty_background_bytes,
1302 		.maxlen		= sizeof(dirty_background_bytes),
1303 		.mode		= 0644,
1304 		.proc_handler	= dirty_background_bytes_handler,
1305 		.extra1		= &one_ul,
1306 	},
1307 	{
1308 		.procname	= "dirty_ratio",
1309 		.data		= &vm_dirty_ratio,
1310 		.maxlen		= sizeof(vm_dirty_ratio),
1311 		.mode		= 0644,
1312 		.proc_handler	= dirty_ratio_handler,
1313 		.extra1		= &zero,
1314 		.extra2		= &one_hundred,
1315 	},
1316 	{
1317 		.procname	= "dirty_bytes",
1318 		.data		= &vm_dirty_bytes,
1319 		.maxlen		= sizeof(vm_dirty_bytes),
1320 		.mode		= 0644,
1321 		.proc_handler	= dirty_bytes_handler,
1322 		.extra1		= &dirty_bytes_min,
1323 	},
1324 	{
1325 		.procname	= "dirty_writeback_centisecs",
1326 		.data		= &dirty_writeback_interval,
1327 		.maxlen		= sizeof(dirty_writeback_interval),
1328 		.mode		= 0644,
1329 		.proc_handler	= dirty_writeback_centisecs_handler,
1330 	},
1331 	{
1332 		.procname	= "dirty_expire_centisecs",
1333 		.data		= &dirty_expire_interval,
1334 		.maxlen		= sizeof(dirty_expire_interval),
1335 		.mode		= 0644,
1336 		.proc_handler	= proc_dointvec_minmax,
1337 		.extra1		= &zero,
1338 	},
1339 	{
1340 		.procname	= "dirtytime_expire_seconds",
1341 		.data		= &dirtytime_expire_interval,
1342 		.maxlen		= sizeof(dirty_expire_interval),
1343 		.mode		= 0644,
1344 		.proc_handler	= dirtytime_interval_handler,
1345 		.extra1		= &zero,
1346 	},
1347 	{
1348 		.procname       = "nr_pdflush_threads",
1349 		.mode           = 0444 /* read-only */,
1350 		.proc_handler   = pdflush_proc_obsolete,
1351 	},
1352 	{
1353 		.procname	= "swappiness",
1354 		.data		= &vm_swappiness,
1355 		.maxlen		= sizeof(vm_swappiness),
1356 		.mode		= 0644,
1357 		.proc_handler	= proc_dointvec_minmax,
1358 		.extra1		= &zero,
1359 		.extra2		= &one_hundred,
1360 	},
1361 #ifdef CONFIG_HUGETLB_PAGE
1362 	{
1363 		.procname	= "nr_hugepages",
1364 		.data		= NULL,
1365 		.maxlen		= sizeof(unsigned long),
1366 		.mode		= 0644,
1367 		.proc_handler	= hugetlb_sysctl_handler,
1368 	},
1369 #ifdef CONFIG_NUMA
1370 	{
1371 		.procname       = "nr_hugepages_mempolicy",
1372 		.data           = NULL,
1373 		.maxlen         = sizeof(unsigned long),
1374 		.mode           = 0644,
1375 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1376 	},
1377 #endif
1378 	 {
1379 		.procname	= "hugetlb_shm_group",
1380 		.data		= &sysctl_hugetlb_shm_group,
1381 		.maxlen		= sizeof(gid_t),
1382 		.mode		= 0644,
1383 		.proc_handler	= proc_dointvec,
1384 	 },
1385 	 {
1386 		.procname	= "hugepages_treat_as_movable",
1387 		.data		= &hugepages_treat_as_movable,
1388 		.maxlen		= sizeof(int),
1389 		.mode		= 0644,
1390 		.proc_handler	= proc_dointvec,
1391 	},
1392 	{
1393 		.procname	= "nr_overcommit_hugepages",
1394 		.data		= NULL,
1395 		.maxlen		= sizeof(unsigned long),
1396 		.mode		= 0644,
1397 		.proc_handler	= hugetlb_overcommit_handler,
1398 	},
1399 #endif
1400 	{
1401 		.procname	= "lowmem_reserve_ratio",
1402 		.data		= &sysctl_lowmem_reserve_ratio,
1403 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1404 		.mode		= 0644,
1405 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1406 	},
1407 	{
1408 		.procname	= "drop_caches",
1409 		.data		= &sysctl_drop_caches,
1410 		.maxlen		= sizeof(int),
1411 		.mode		= 0644,
1412 		.proc_handler	= drop_caches_sysctl_handler,
1413 		.extra1		= &one,
1414 		.extra2		= &four,
1415 	},
1416 #ifdef CONFIG_COMPACTION
1417 	{
1418 		.procname	= "compact_memory",
1419 		.data		= &sysctl_compact_memory,
1420 		.maxlen		= sizeof(int),
1421 		.mode		= 0200,
1422 		.proc_handler	= sysctl_compaction_handler,
1423 	},
1424 	{
1425 		.procname	= "extfrag_threshold",
1426 		.data		= &sysctl_extfrag_threshold,
1427 		.maxlen		= sizeof(int),
1428 		.mode		= 0644,
1429 		.proc_handler	= sysctl_extfrag_handler,
1430 		.extra1		= &min_extfrag_threshold,
1431 		.extra2		= &max_extfrag_threshold,
1432 	},
1433 	{
1434 		.procname	= "compact_unevictable_allowed",
1435 		.data		= &sysctl_compact_unevictable_allowed,
1436 		.maxlen		= sizeof(int),
1437 		.mode		= 0644,
1438 		.proc_handler	= proc_dointvec,
1439 		.extra1		= &zero,
1440 		.extra2		= &one,
1441 	},
1442 
1443 #endif /* CONFIG_COMPACTION */
1444 	{
1445 		.procname	= "min_free_kbytes",
1446 		.data		= &min_free_kbytes,
1447 		.maxlen		= sizeof(min_free_kbytes),
1448 		.mode		= 0644,
1449 		.proc_handler	= min_free_kbytes_sysctl_handler,
1450 		.extra1		= &zero,
1451 	},
1452 	{
1453 		.procname	= "watermark_scale_factor",
1454 		.data		= &watermark_scale_factor,
1455 		.maxlen		= sizeof(watermark_scale_factor),
1456 		.mode		= 0644,
1457 		.proc_handler	= watermark_scale_factor_sysctl_handler,
1458 		.extra1		= &one,
1459 		.extra2		= &one_thousand,
1460 	},
1461 	{
1462 		.procname	= "percpu_pagelist_fraction",
1463 		.data		= &percpu_pagelist_fraction,
1464 		.maxlen		= sizeof(percpu_pagelist_fraction),
1465 		.mode		= 0644,
1466 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1467 		.extra1		= &zero,
1468 	},
1469 #ifdef CONFIG_MMU
1470 	{
1471 		.procname	= "max_map_count",
1472 		.data		= &sysctl_max_map_count,
1473 		.maxlen		= sizeof(sysctl_max_map_count),
1474 		.mode		= 0644,
1475 		.proc_handler	= proc_dointvec_minmax,
1476 		.extra1		= &zero,
1477 	},
1478 #else
1479 	{
1480 		.procname	= "nr_trim_pages",
1481 		.data		= &sysctl_nr_trim_pages,
1482 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1483 		.mode		= 0644,
1484 		.proc_handler	= proc_dointvec_minmax,
1485 		.extra1		= &zero,
1486 	},
1487 #endif
1488 	{
1489 		.procname	= "laptop_mode",
1490 		.data		= &laptop_mode,
1491 		.maxlen		= sizeof(laptop_mode),
1492 		.mode		= 0644,
1493 		.proc_handler	= proc_dointvec_jiffies,
1494 	},
1495 	{
1496 		.procname	= "block_dump",
1497 		.data		= &block_dump,
1498 		.maxlen		= sizeof(block_dump),
1499 		.mode		= 0644,
1500 		.proc_handler	= proc_dointvec,
1501 		.extra1		= &zero,
1502 	},
1503 	{
1504 		.procname	= "vfs_cache_pressure",
1505 		.data		= &sysctl_vfs_cache_pressure,
1506 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1507 		.mode		= 0644,
1508 		.proc_handler	= proc_dointvec,
1509 		.extra1		= &zero,
1510 	},
1511 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1512 	{
1513 		.procname	= "legacy_va_layout",
1514 		.data		= &sysctl_legacy_va_layout,
1515 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1516 		.mode		= 0644,
1517 		.proc_handler	= proc_dointvec,
1518 		.extra1		= &zero,
1519 	},
1520 #endif
1521 #ifdef CONFIG_NUMA
1522 	{
1523 		.procname	= "zone_reclaim_mode",
1524 		.data		= &node_reclaim_mode,
1525 		.maxlen		= sizeof(node_reclaim_mode),
1526 		.mode		= 0644,
1527 		.proc_handler	= proc_dointvec,
1528 		.extra1		= &zero,
1529 	},
1530 	{
1531 		.procname	= "min_unmapped_ratio",
1532 		.data		= &sysctl_min_unmapped_ratio,
1533 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1534 		.mode		= 0644,
1535 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1536 		.extra1		= &zero,
1537 		.extra2		= &one_hundred,
1538 	},
1539 	{
1540 		.procname	= "min_slab_ratio",
1541 		.data		= &sysctl_min_slab_ratio,
1542 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1543 		.mode		= 0644,
1544 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1545 		.extra1		= &zero,
1546 		.extra2		= &one_hundred,
1547 	},
1548 #endif
1549 #ifdef CONFIG_SMP
1550 	{
1551 		.procname	= "stat_interval",
1552 		.data		= &sysctl_stat_interval,
1553 		.maxlen		= sizeof(sysctl_stat_interval),
1554 		.mode		= 0644,
1555 		.proc_handler	= proc_dointvec_jiffies,
1556 	},
1557 	{
1558 		.procname	= "stat_refresh",
1559 		.data		= NULL,
1560 		.maxlen		= 0,
1561 		.mode		= 0600,
1562 		.proc_handler	= vmstat_refresh,
1563 	},
1564 #endif
1565 #ifdef CONFIG_MMU
1566 	{
1567 		.procname	= "mmap_min_addr",
1568 		.data		= &dac_mmap_min_addr,
1569 		.maxlen		= sizeof(unsigned long),
1570 		.mode		= 0644,
1571 		.proc_handler	= mmap_min_addr_handler,
1572 	},
1573 #endif
1574 #ifdef CONFIG_NUMA
1575 	{
1576 		.procname	= "numa_zonelist_order",
1577 		.data		= &numa_zonelist_order,
1578 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1579 		.mode		= 0644,
1580 		.proc_handler	= numa_zonelist_order_handler,
1581 	},
1582 #endif
1583 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1584    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1585 	{
1586 		.procname	= "vdso_enabled",
1587 #ifdef CONFIG_X86_32
1588 		.data		= &vdso32_enabled,
1589 		.maxlen		= sizeof(vdso32_enabled),
1590 #else
1591 		.data		= &vdso_enabled,
1592 		.maxlen		= sizeof(vdso_enabled),
1593 #endif
1594 		.mode		= 0644,
1595 		.proc_handler	= proc_dointvec,
1596 		.extra1		= &zero,
1597 	},
1598 #endif
1599 #ifdef CONFIG_HIGHMEM
1600 	{
1601 		.procname	= "highmem_is_dirtyable",
1602 		.data		= &vm_highmem_is_dirtyable,
1603 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1604 		.mode		= 0644,
1605 		.proc_handler	= proc_dointvec_minmax,
1606 		.extra1		= &zero,
1607 		.extra2		= &one,
1608 	},
1609 #endif
1610 #ifdef CONFIG_MEMORY_FAILURE
1611 	{
1612 		.procname	= "memory_failure_early_kill",
1613 		.data		= &sysctl_memory_failure_early_kill,
1614 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1615 		.mode		= 0644,
1616 		.proc_handler	= proc_dointvec_minmax,
1617 		.extra1		= &zero,
1618 		.extra2		= &one,
1619 	},
1620 	{
1621 		.procname	= "memory_failure_recovery",
1622 		.data		= &sysctl_memory_failure_recovery,
1623 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1624 		.mode		= 0644,
1625 		.proc_handler	= proc_dointvec_minmax,
1626 		.extra1		= &zero,
1627 		.extra2		= &one,
1628 	},
1629 #endif
1630 	{
1631 		.procname	= "user_reserve_kbytes",
1632 		.data		= &sysctl_user_reserve_kbytes,
1633 		.maxlen		= sizeof(sysctl_user_reserve_kbytes),
1634 		.mode		= 0644,
1635 		.proc_handler	= proc_doulongvec_minmax,
1636 	},
1637 	{
1638 		.procname	= "admin_reserve_kbytes",
1639 		.data		= &sysctl_admin_reserve_kbytes,
1640 		.maxlen		= sizeof(sysctl_admin_reserve_kbytes),
1641 		.mode		= 0644,
1642 		.proc_handler	= proc_doulongvec_minmax,
1643 	},
1644 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1645 	{
1646 		.procname	= "mmap_rnd_bits",
1647 		.data		= &mmap_rnd_bits,
1648 		.maxlen		= sizeof(mmap_rnd_bits),
1649 		.mode		= 0600,
1650 		.proc_handler	= proc_dointvec_minmax,
1651 		.extra1		= (void *)&mmap_rnd_bits_min,
1652 		.extra2		= (void *)&mmap_rnd_bits_max,
1653 	},
1654 #endif
1655 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1656 	{
1657 		.procname	= "mmap_rnd_compat_bits",
1658 		.data		= &mmap_rnd_compat_bits,
1659 		.maxlen		= sizeof(mmap_rnd_compat_bits),
1660 		.mode		= 0600,
1661 		.proc_handler	= proc_dointvec_minmax,
1662 		.extra1		= (void *)&mmap_rnd_compat_bits_min,
1663 		.extra2		= (void *)&mmap_rnd_compat_bits_max,
1664 	},
1665 #endif
1666 	{ }
1667 };
1668 
1669 static struct ctl_table fs_table[] = {
1670 	{
1671 		.procname	= "inode-nr",
1672 		.data		= &inodes_stat,
1673 		.maxlen		= 2*sizeof(long),
1674 		.mode		= 0444,
1675 		.proc_handler	= proc_nr_inodes,
1676 	},
1677 	{
1678 		.procname	= "inode-state",
1679 		.data		= &inodes_stat,
1680 		.maxlen		= 7*sizeof(long),
1681 		.mode		= 0444,
1682 		.proc_handler	= proc_nr_inodes,
1683 	},
1684 	{
1685 		.procname	= "file-nr",
1686 		.data		= &files_stat,
1687 		.maxlen		= sizeof(files_stat),
1688 		.mode		= 0444,
1689 		.proc_handler	= proc_nr_files,
1690 	},
1691 	{
1692 		.procname	= "file-max",
1693 		.data		= &files_stat.max_files,
1694 		.maxlen		= sizeof(files_stat.max_files),
1695 		.mode		= 0644,
1696 		.proc_handler	= proc_doulongvec_minmax,
1697 	},
1698 	{
1699 		.procname	= "nr_open",
1700 		.data		= &sysctl_nr_open,
1701 		.maxlen		= sizeof(unsigned int),
1702 		.mode		= 0644,
1703 		.proc_handler	= proc_dointvec_minmax,
1704 		.extra1		= &sysctl_nr_open_min,
1705 		.extra2		= &sysctl_nr_open_max,
1706 	},
1707 	{
1708 		.procname	= "dentry-state",
1709 		.data		= &dentry_stat,
1710 		.maxlen		= 6*sizeof(long),
1711 		.mode		= 0444,
1712 		.proc_handler	= proc_nr_dentry,
1713 	},
1714 	{
1715 		.procname	= "overflowuid",
1716 		.data		= &fs_overflowuid,
1717 		.maxlen		= sizeof(int),
1718 		.mode		= 0644,
1719 		.proc_handler	= proc_dointvec_minmax,
1720 		.extra1		= &minolduid,
1721 		.extra2		= &maxolduid,
1722 	},
1723 	{
1724 		.procname	= "overflowgid",
1725 		.data		= &fs_overflowgid,
1726 		.maxlen		= sizeof(int),
1727 		.mode		= 0644,
1728 		.proc_handler	= proc_dointvec_minmax,
1729 		.extra1		= &minolduid,
1730 		.extra2		= &maxolduid,
1731 	},
1732 #ifdef CONFIG_FILE_LOCKING
1733 	{
1734 		.procname	= "leases-enable",
1735 		.data		= &leases_enable,
1736 		.maxlen		= sizeof(int),
1737 		.mode		= 0644,
1738 		.proc_handler	= proc_dointvec,
1739 	},
1740 #endif
1741 #ifdef CONFIG_DNOTIFY
1742 	{
1743 		.procname	= "dir-notify-enable",
1744 		.data		= &dir_notify_enable,
1745 		.maxlen		= sizeof(int),
1746 		.mode		= 0644,
1747 		.proc_handler	= proc_dointvec,
1748 	},
1749 #endif
1750 #ifdef CONFIG_MMU
1751 #ifdef CONFIG_FILE_LOCKING
1752 	{
1753 		.procname	= "lease-break-time",
1754 		.data		= &lease_break_time,
1755 		.maxlen		= sizeof(int),
1756 		.mode		= 0644,
1757 		.proc_handler	= proc_dointvec,
1758 	},
1759 #endif
1760 #ifdef CONFIG_AIO
1761 	{
1762 		.procname	= "aio-nr",
1763 		.data		= &aio_nr,
1764 		.maxlen		= sizeof(aio_nr),
1765 		.mode		= 0444,
1766 		.proc_handler	= proc_doulongvec_minmax,
1767 	},
1768 	{
1769 		.procname	= "aio-max-nr",
1770 		.data		= &aio_max_nr,
1771 		.maxlen		= sizeof(aio_max_nr),
1772 		.mode		= 0644,
1773 		.proc_handler	= proc_doulongvec_minmax,
1774 	},
1775 #endif /* CONFIG_AIO */
1776 #ifdef CONFIG_INOTIFY_USER
1777 	{
1778 		.procname	= "inotify",
1779 		.mode		= 0555,
1780 		.child		= inotify_table,
1781 	},
1782 #endif
1783 #ifdef CONFIG_EPOLL
1784 	{
1785 		.procname	= "epoll",
1786 		.mode		= 0555,
1787 		.child		= epoll_table,
1788 	},
1789 #endif
1790 #endif
1791 	{
1792 		.procname	= "protected_symlinks",
1793 		.data		= &sysctl_protected_symlinks,
1794 		.maxlen		= sizeof(int),
1795 		.mode		= 0600,
1796 		.proc_handler	= proc_dointvec_minmax,
1797 		.extra1		= &zero,
1798 		.extra2		= &one,
1799 	},
1800 	{
1801 		.procname	= "protected_hardlinks",
1802 		.data		= &sysctl_protected_hardlinks,
1803 		.maxlen		= sizeof(int),
1804 		.mode		= 0600,
1805 		.proc_handler	= proc_dointvec_minmax,
1806 		.extra1		= &zero,
1807 		.extra2		= &one,
1808 	},
1809 	{
1810 		.procname	= "suid_dumpable",
1811 		.data		= &suid_dumpable,
1812 		.maxlen		= sizeof(int),
1813 		.mode		= 0644,
1814 		.proc_handler	= proc_dointvec_minmax_coredump,
1815 		.extra1		= &zero,
1816 		.extra2		= &two,
1817 	},
1818 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1819 	{
1820 		.procname	= "binfmt_misc",
1821 		.mode		= 0555,
1822 		.child		= sysctl_mount_point,
1823 	},
1824 #endif
1825 	{
1826 		.procname	= "pipe-max-size",
1827 		.data		= &pipe_max_size,
1828 		.maxlen		= sizeof(int),
1829 		.mode		= 0644,
1830 		.proc_handler	= &pipe_proc_fn,
1831 		.extra1		= &pipe_min_size,
1832 	},
1833 	{
1834 		.procname	= "pipe-user-pages-hard",
1835 		.data		= &pipe_user_pages_hard,
1836 		.maxlen		= sizeof(pipe_user_pages_hard),
1837 		.mode		= 0644,
1838 		.proc_handler	= proc_doulongvec_minmax,
1839 	},
1840 	{
1841 		.procname	= "pipe-user-pages-soft",
1842 		.data		= &pipe_user_pages_soft,
1843 		.maxlen		= sizeof(pipe_user_pages_soft),
1844 		.mode		= 0644,
1845 		.proc_handler	= proc_doulongvec_minmax,
1846 	},
1847 	{
1848 		.procname	= "mount-max",
1849 		.data		= &sysctl_mount_max,
1850 		.maxlen		= sizeof(unsigned int),
1851 		.mode		= 0644,
1852 		.proc_handler	= proc_dointvec_minmax,
1853 		.extra1		= &one,
1854 	},
1855 	{ }
1856 };
1857 
1858 static struct ctl_table debug_table[] = {
1859 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1860 	{
1861 		.procname	= "exception-trace",
1862 		.data		= &show_unhandled_signals,
1863 		.maxlen		= sizeof(int),
1864 		.mode		= 0644,
1865 		.proc_handler	= proc_dointvec
1866 	},
1867 #endif
1868 #if defined(CONFIG_OPTPROBES)
1869 	{
1870 		.procname	= "kprobes-optimization",
1871 		.data		= &sysctl_kprobes_optimization,
1872 		.maxlen		= sizeof(int),
1873 		.mode		= 0644,
1874 		.proc_handler	= proc_kprobes_optimization_handler,
1875 		.extra1		= &zero,
1876 		.extra2		= &one,
1877 	},
1878 #endif
1879 	{ }
1880 };
1881 
1882 static struct ctl_table dev_table[] = {
1883 	{ }
1884 };
1885 
1886 int __init sysctl_init(void)
1887 {
1888 	struct ctl_table_header *hdr;
1889 
1890 	hdr = register_sysctl_table(sysctl_base_table);
1891 	kmemleak_not_leak(hdr);
1892 	return 0;
1893 }
1894 
1895 #endif /* CONFIG_SYSCTL */
1896 
1897 /*
1898  * /proc/sys support
1899  */
1900 
1901 #ifdef CONFIG_PROC_SYSCTL
1902 
1903 static int _proc_do_string(char *data, int maxlen, int write,
1904 			   char __user *buffer,
1905 			   size_t *lenp, loff_t *ppos)
1906 {
1907 	size_t len;
1908 	char __user *p;
1909 	char c;
1910 
1911 	if (!data || !maxlen || !*lenp) {
1912 		*lenp = 0;
1913 		return 0;
1914 	}
1915 
1916 	if (write) {
1917 		if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1918 			/* Only continue writes not past the end of buffer. */
1919 			len = strlen(data);
1920 			if (len > maxlen - 1)
1921 				len = maxlen - 1;
1922 
1923 			if (*ppos > len)
1924 				return 0;
1925 			len = *ppos;
1926 		} else {
1927 			/* Start writing from beginning of buffer. */
1928 			len = 0;
1929 		}
1930 
1931 		*ppos += *lenp;
1932 		p = buffer;
1933 		while ((p - buffer) < *lenp && len < maxlen - 1) {
1934 			if (get_user(c, p++))
1935 				return -EFAULT;
1936 			if (c == 0 || c == '\n')
1937 				break;
1938 			data[len++] = c;
1939 		}
1940 		data[len] = 0;
1941 	} else {
1942 		len = strlen(data);
1943 		if (len > maxlen)
1944 			len = maxlen;
1945 
1946 		if (*ppos > len) {
1947 			*lenp = 0;
1948 			return 0;
1949 		}
1950 
1951 		data += *ppos;
1952 		len  -= *ppos;
1953 
1954 		if (len > *lenp)
1955 			len = *lenp;
1956 		if (len)
1957 			if (copy_to_user(buffer, data, len))
1958 				return -EFAULT;
1959 		if (len < *lenp) {
1960 			if (put_user('\n', buffer + len))
1961 				return -EFAULT;
1962 			len++;
1963 		}
1964 		*lenp = len;
1965 		*ppos += len;
1966 	}
1967 	return 0;
1968 }
1969 
1970 static void warn_sysctl_write(struct ctl_table *table)
1971 {
1972 	pr_warn_once("%s wrote to %s when file position was not 0!\n"
1973 		"This will not be supported in the future. To silence this\n"
1974 		"warning, set kernel.sysctl_writes_strict = -1\n",
1975 		current->comm, table->procname);
1976 }
1977 
1978 /**
1979  * proc_first_pos_non_zero_ignore - check if firs position is allowed
1980  * @ppos: file position
1981  * @table: the sysctl table
1982  *
1983  * Returns true if the first position is non-zero and the sysctl_writes_strict
1984  * mode indicates this is not allowed for numeric input types. String proc
1985  * hadlers can ignore the return value.
1986  */
1987 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
1988 					   struct ctl_table *table)
1989 {
1990 	if (!*ppos)
1991 		return false;
1992 
1993 	switch (sysctl_writes_strict) {
1994 	case SYSCTL_WRITES_STRICT:
1995 		return true;
1996 	case SYSCTL_WRITES_WARN:
1997 		warn_sysctl_write(table);
1998 		return false;
1999 	default:
2000 		return false;
2001 	}
2002 }
2003 
2004 /**
2005  * proc_dostring - read a string sysctl
2006  * @table: the sysctl table
2007  * @write: %TRUE if this is a write to the sysctl file
2008  * @buffer: the user buffer
2009  * @lenp: the size of the user buffer
2010  * @ppos: file position
2011  *
2012  * Reads/writes a string from/to the user buffer. If the kernel
2013  * buffer provided is not large enough to hold the string, the
2014  * string is truncated. The copied string is %NULL-terminated.
2015  * If the string is being read by the user process, it is copied
2016  * and a newline '\n' is added. It is truncated if the buffer is
2017  * not large enough.
2018  *
2019  * Returns 0 on success.
2020  */
2021 int proc_dostring(struct ctl_table *table, int write,
2022 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2023 {
2024 	if (write)
2025 		proc_first_pos_non_zero_ignore(ppos, table);
2026 
2027 	return _proc_do_string((char *)(table->data), table->maxlen, write,
2028 			       (char __user *)buffer, lenp, ppos);
2029 }
2030 
2031 static size_t proc_skip_spaces(char **buf)
2032 {
2033 	size_t ret;
2034 	char *tmp = skip_spaces(*buf);
2035 	ret = tmp - *buf;
2036 	*buf = tmp;
2037 	return ret;
2038 }
2039 
2040 static void proc_skip_char(char **buf, size_t *size, const char v)
2041 {
2042 	while (*size) {
2043 		if (**buf != v)
2044 			break;
2045 		(*size)--;
2046 		(*buf)++;
2047 	}
2048 }
2049 
2050 #define TMPBUFLEN 22
2051 /**
2052  * proc_get_long - reads an ASCII formatted integer from a user buffer
2053  *
2054  * @buf: a kernel buffer
2055  * @size: size of the kernel buffer
2056  * @val: this is where the number will be stored
2057  * @neg: set to %TRUE if number is negative
2058  * @perm_tr: a vector which contains the allowed trailers
2059  * @perm_tr_len: size of the perm_tr vector
2060  * @tr: pointer to store the trailer character
2061  *
2062  * In case of success %0 is returned and @buf and @size are updated with
2063  * the amount of bytes read. If @tr is non-NULL and a trailing
2064  * character exists (size is non-zero after returning from this
2065  * function), @tr is updated with the trailing character.
2066  */
2067 static int proc_get_long(char **buf, size_t *size,
2068 			  unsigned long *val, bool *neg,
2069 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
2070 {
2071 	int len;
2072 	char *p, tmp[TMPBUFLEN];
2073 
2074 	if (!*size)
2075 		return -EINVAL;
2076 
2077 	len = *size;
2078 	if (len > TMPBUFLEN - 1)
2079 		len = TMPBUFLEN - 1;
2080 
2081 	memcpy(tmp, *buf, len);
2082 
2083 	tmp[len] = 0;
2084 	p = tmp;
2085 	if (*p == '-' && *size > 1) {
2086 		*neg = true;
2087 		p++;
2088 	} else
2089 		*neg = false;
2090 	if (!isdigit(*p))
2091 		return -EINVAL;
2092 
2093 	*val = simple_strtoul(p, &p, 0);
2094 
2095 	len = p - tmp;
2096 
2097 	/* We don't know if the next char is whitespace thus we may accept
2098 	 * invalid integers (e.g. 1234...a) or two integers instead of one
2099 	 * (e.g. 123...1). So lets not allow such large numbers. */
2100 	if (len == TMPBUFLEN - 1)
2101 		return -EINVAL;
2102 
2103 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2104 		return -EINVAL;
2105 
2106 	if (tr && (len < *size))
2107 		*tr = *p;
2108 
2109 	*buf += len;
2110 	*size -= len;
2111 
2112 	return 0;
2113 }
2114 
2115 /**
2116  * proc_put_long - converts an integer to a decimal ASCII formatted string
2117  *
2118  * @buf: the user buffer
2119  * @size: the size of the user buffer
2120  * @val: the integer to be converted
2121  * @neg: sign of the number, %TRUE for negative
2122  *
2123  * In case of success %0 is returned and @buf and @size are updated with
2124  * the amount of bytes written.
2125  */
2126 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2127 			  bool neg)
2128 {
2129 	int len;
2130 	char tmp[TMPBUFLEN], *p = tmp;
2131 
2132 	sprintf(p, "%s%lu", neg ? "-" : "", val);
2133 	len = strlen(tmp);
2134 	if (len > *size)
2135 		len = *size;
2136 	if (copy_to_user(*buf, tmp, len))
2137 		return -EFAULT;
2138 	*size -= len;
2139 	*buf += len;
2140 	return 0;
2141 }
2142 #undef TMPBUFLEN
2143 
2144 static int proc_put_char(void __user **buf, size_t *size, char c)
2145 {
2146 	if (*size) {
2147 		char __user **buffer = (char __user **)buf;
2148 		if (put_user(c, *buffer))
2149 			return -EFAULT;
2150 		(*size)--, (*buffer)++;
2151 		*buf = *buffer;
2152 	}
2153 	return 0;
2154 }
2155 
2156 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2157 				 int *valp,
2158 				 int write, void *data)
2159 {
2160 	if (write) {
2161 		if (*negp) {
2162 			if (*lvalp > (unsigned long) INT_MAX + 1)
2163 				return -EINVAL;
2164 			*valp = -*lvalp;
2165 		} else {
2166 			if (*lvalp > (unsigned long) INT_MAX)
2167 				return -EINVAL;
2168 			*valp = *lvalp;
2169 		}
2170 	} else {
2171 		int val = *valp;
2172 		if (val < 0) {
2173 			*negp = true;
2174 			*lvalp = -(unsigned long)val;
2175 		} else {
2176 			*negp = false;
2177 			*lvalp = (unsigned long)val;
2178 		}
2179 	}
2180 	return 0;
2181 }
2182 
2183 static int do_proc_douintvec_conv(unsigned long *lvalp,
2184 				  unsigned int *valp,
2185 				  int write, void *data)
2186 {
2187 	if (write) {
2188 		if (*lvalp > UINT_MAX)
2189 			return -EINVAL;
2190 		if (*lvalp > UINT_MAX)
2191 			return -EINVAL;
2192 		*valp = *lvalp;
2193 	} else {
2194 		unsigned int val = *valp;
2195 		*lvalp = (unsigned long)val;
2196 	}
2197 	return 0;
2198 }
2199 
2200 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2201 
2202 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2203 		  int write, void __user *buffer,
2204 		  size_t *lenp, loff_t *ppos,
2205 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2206 			      int write, void *data),
2207 		  void *data)
2208 {
2209 	int *i, vleft, first = 1, err = 0;
2210 	size_t left;
2211 	char *kbuf = NULL, *p;
2212 
2213 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2214 		*lenp = 0;
2215 		return 0;
2216 	}
2217 
2218 	i = (int *) tbl_data;
2219 	vleft = table->maxlen / sizeof(*i);
2220 	left = *lenp;
2221 
2222 	if (!conv)
2223 		conv = do_proc_dointvec_conv;
2224 
2225 	if (write) {
2226 		if (proc_first_pos_non_zero_ignore(ppos, table))
2227 			goto out;
2228 
2229 		if (left > PAGE_SIZE - 1)
2230 			left = PAGE_SIZE - 1;
2231 		p = kbuf = memdup_user_nul(buffer, left);
2232 		if (IS_ERR(kbuf))
2233 			return PTR_ERR(kbuf);
2234 	}
2235 
2236 	for (; left && vleft--; i++, first=0) {
2237 		unsigned long lval;
2238 		bool neg;
2239 
2240 		if (write) {
2241 			left -= proc_skip_spaces(&p);
2242 
2243 			if (!left)
2244 				break;
2245 			err = proc_get_long(&p, &left, &lval, &neg,
2246 					     proc_wspace_sep,
2247 					     sizeof(proc_wspace_sep), NULL);
2248 			if (err)
2249 				break;
2250 			if (conv(&neg, &lval, i, 1, data)) {
2251 				err = -EINVAL;
2252 				break;
2253 			}
2254 		} else {
2255 			if (conv(&neg, &lval, i, 0, data)) {
2256 				err = -EINVAL;
2257 				break;
2258 			}
2259 			if (!first)
2260 				err = proc_put_char(&buffer, &left, '\t');
2261 			if (err)
2262 				break;
2263 			err = proc_put_long(&buffer, &left, lval, neg);
2264 			if (err)
2265 				break;
2266 		}
2267 	}
2268 
2269 	if (!write && !first && left && !err)
2270 		err = proc_put_char(&buffer, &left, '\n');
2271 	if (write && !err && left)
2272 		left -= proc_skip_spaces(&p);
2273 	if (write) {
2274 		kfree(kbuf);
2275 		if (first)
2276 			return err ? : -EINVAL;
2277 	}
2278 	*lenp -= left;
2279 out:
2280 	*ppos += *lenp;
2281 	return err;
2282 }
2283 
2284 static int do_proc_dointvec(struct ctl_table *table, int write,
2285 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2286 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2287 			      int write, void *data),
2288 		  void *data)
2289 {
2290 	return __do_proc_dointvec(table->data, table, write,
2291 			buffer, lenp, ppos, conv, data);
2292 }
2293 
2294 static int do_proc_douintvec_w(unsigned int *tbl_data,
2295 			       struct ctl_table *table,
2296 			       void __user *buffer,
2297 			       size_t *lenp, loff_t *ppos,
2298 			       int (*conv)(unsigned long *lvalp,
2299 					   unsigned int *valp,
2300 					   int write, void *data),
2301 			       void *data)
2302 {
2303 	unsigned long lval;
2304 	int err = 0;
2305 	size_t left;
2306 	bool neg;
2307 	char *kbuf = NULL, *p;
2308 
2309 	left = *lenp;
2310 
2311 	if (proc_first_pos_non_zero_ignore(ppos, table))
2312 		goto bail_early;
2313 
2314 	if (left > PAGE_SIZE - 1)
2315 		left = PAGE_SIZE - 1;
2316 
2317 	p = kbuf = memdup_user_nul(buffer, left);
2318 	if (IS_ERR(kbuf))
2319 		return -EINVAL;
2320 
2321 	left -= proc_skip_spaces(&p);
2322 	if (!left) {
2323 		err = -EINVAL;
2324 		goto out_free;
2325 	}
2326 
2327 	err = proc_get_long(&p, &left, &lval, &neg,
2328 			     proc_wspace_sep,
2329 			     sizeof(proc_wspace_sep), NULL);
2330 	if (err || neg) {
2331 		err = -EINVAL;
2332 		goto out_free;
2333 	}
2334 
2335 	if (conv(&lval, tbl_data, 1, data)) {
2336 		err = -EINVAL;
2337 		goto out_free;
2338 	}
2339 
2340 	if (!err && left)
2341 		left -= proc_skip_spaces(&p);
2342 
2343 out_free:
2344 	kfree(kbuf);
2345 	if (err)
2346 		return -EINVAL;
2347 
2348 	return 0;
2349 
2350 	/* This is in keeping with old __do_proc_dointvec() */
2351 bail_early:
2352 	*ppos += *lenp;
2353 	return err;
2354 }
2355 
2356 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2357 			       size_t *lenp, loff_t *ppos,
2358 			       int (*conv)(unsigned long *lvalp,
2359 					   unsigned int *valp,
2360 					   int write, void *data),
2361 			       void *data)
2362 {
2363 	unsigned long lval;
2364 	int err = 0;
2365 	size_t left;
2366 
2367 	left = *lenp;
2368 
2369 	if (conv(&lval, tbl_data, 0, data)) {
2370 		err = -EINVAL;
2371 		goto out;
2372 	}
2373 
2374 	err = proc_put_long(&buffer, &left, lval, false);
2375 	if (err || !left)
2376 		goto out;
2377 
2378 	err = proc_put_char(&buffer, &left, '\n');
2379 
2380 out:
2381 	*lenp -= left;
2382 	*ppos += *lenp;
2383 
2384 	return err;
2385 }
2386 
2387 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2388 			       int write, void __user *buffer,
2389 			       size_t *lenp, loff_t *ppos,
2390 			       int (*conv)(unsigned long *lvalp,
2391 					   unsigned int *valp,
2392 					   int write, void *data),
2393 			       void *data)
2394 {
2395 	unsigned int *i, vleft;
2396 
2397 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2398 		*lenp = 0;
2399 		return 0;
2400 	}
2401 
2402 	i = (unsigned int *) tbl_data;
2403 	vleft = table->maxlen / sizeof(*i);
2404 
2405 	/*
2406 	 * Arrays are not supported, keep this simple. *Do not* add
2407 	 * support for them.
2408 	 */
2409 	if (vleft != 1) {
2410 		*lenp = 0;
2411 		return -EINVAL;
2412 	}
2413 
2414 	if (!conv)
2415 		conv = do_proc_douintvec_conv;
2416 
2417 	if (write)
2418 		return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2419 					   conv, data);
2420 	return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2421 }
2422 
2423 static int do_proc_douintvec(struct ctl_table *table, int write,
2424 			     void __user *buffer, size_t *lenp, loff_t *ppos,
2425 			     int (*conv)(unsigned long *lvalp,
2426 					 unsigned int *valp,
2427 					 int write, void *data),
2428 			     void *data)
2429 {
2430 	return __do_proc_douintvec(table->data, table, write,
2431 				   buffer, lenp, ppos, conv, data);
2432 }
2433 
2434 /**
2435  * proc_dointvec - read a vector of integers
2436  * @table: the sysctl table
2437  * @write: %TRUE if this is a write to the sysctl file
2438  * @buffer: the user buffer
2439  * @lenp: the size of the user buffer
2440  * @ppos: file position
2441  *
2442  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2443  * values from/to the user buffer, treated as an ASCII string.
2444  *
2445  * Returns 0 on success.
2446  */
2447 int proc_dointvec(struct ctl_table *table, int write,
2448 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2449 {
2450 	return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2451 }
2452 
2453 /**
2454  * proc_douintvec - read a vector of unsigned integers
2455  * @table: the sysctl table
2456  * @write: %TRUE if this is a write to the sysctl file
2457  * @buffer: the user buffer
2458  * @lenp: the size of the user buffer
2459  * @ppos: file position
2460  *
2461  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2462  * values from/to the user buffer, treated as an ASCII string.
2463  *
2464  * Returns 0 on success.
2465  */
2466 int proc_douintvec(struct ctl_table *table, int write,
2467 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2468 {
2469 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
2470 				 do_proc_douintvec_conv, NULL);
2471 }
2472 
2473 /*
2474  * Taint values can only be increased
2475  * This means we can safely use a temporary.
2476  */
2477 static int proc_taint(struct ctl_table *table, int write,
2478 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2479 {
2480 	struct ctl_table t;
2481 	unsigned long tmptaint = get_taint();
2482 	int err;
2483 
2484 	if (write && !capable(CAP_SYS_ADMIN))
2485 		return -EPERM;
2486 
2487 	t = *table;
2488 	t.data = &tmptaint;
2489 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2490 	if (err < 0)
2491 		return err;
2492 
2493 	if (write) {
2494 		/*
2495 		 * Poor man's atomic or. Not worth adding a primitive
2496 		 * to everyone's atomic.h for this
2497 		 */
2498 		int i;
2499 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2500 			if ((tmptaint >> i) & 1)
2501 				add_taint(i, LOCKDEP_STILL_OK);
2502 		}
2503 	}
2504 
2505 	return err;
2506 }
2507 
2508 #ifdef CONFIG_PRINTK
2509 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2510 				void __user *buffer, size_t *lenp, loff_t *ppos)
2511 {
2512 	if (write && !capable(CAP_SYS_ADMIN))
2513 		return -EPERM;
2514 
2515 	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2516 }
2517 #endif
2518 
2519 struct do_proc_dointvec_minmax_conv_param {
2520 	int *min;
2521 	int *max;
2522 };
2523 
2524 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2525 					int *valp,
2526 					int write, void *data)
2527 {
2528 	struct do_proc_dointvec_minmax_conv_param *param = data;
2529 	if (write) {
2530 		int val = *negp ? -*lvalp : *lvalp;
2531 		if ((param->min && *param->min > val) ||
2532 		    (param->max && *param->max < val))
2533 			return -EINVAL;
2534 		*valp = val;
2535 	} else {
2536 		int val = *valp;
2537 		if (val < 0) {
2538 			*negp = true;
2539 			*lvalp = -(unsigned long)val;
2540 		} else {
2541 			*negp = false;
2542 			*lvalp = (unsigned long)val;
2543 		}
2544 	}
2545 	return 0;
2546 }
2547 
2548 /**
2549  * proc_dointvec_minmax - read a vector of integers with min/max values
2550  * @table: the sysctl table
2551  * @write: %TRUE if this is a write to the sysctl file
2552  * @buffer: the user buffer
2553  * @lenp: the size of the user buffer
2554  * @ppos: file position
2555  *
2556  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2557  * values from/to the user buffer, treated as an ASCII string.
2558  *
2559  * This routine will ensure the values are within the range specified by
2560  * table->extra1 (min) and table->extra2 (max).
2561  *
2562  * Returns 0 on success.
2563  */
2564 int proc_dointvec_minmax(struct ctl_table *table, int write,
2565 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2566 {
2567 	struct do_proc_dointvec_minmax_conv_param param = {
2568 		.min = (int *) table->extra1,
2569 		.max = (int *) table->extra2,
2570 	};
2571 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2572 				do_proc_dointvec_minmax_conv, &param);
2573 }
2574 
2575 struct do_proc_douintvec_minmax_conv_param {
2576 	unsigned int *min;
2577 	unsigned int *max;
2578 };
2579 
2580 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2581 					 unsigned int *valp,
2582 					 int write, void *data)
2583 {
2584 	struct do_proc_douintvec_minmax_conv_param *param = data;
2585 
2586 	if (write) {
2587 		unsigned int val = *lvalp;
2588 
2589 		if ((param->min && *param->min > val) ||
2590 		    (param->max && *param->max < val))
2591 			return -ERANGE;
2592 
2593 		if (*lvalp > UINT_MAX)
2594 			return -EINVAL;
2595 		*valp = val;
2596 	} else {
2597 		unsigned int val = *valp;
2598 		*lvalp = (unsigned long) val;
2599 	}
2600 
2601 	return 0;
2602 }
2603 
2604 /**
2605  * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2606  * @table: the sysctl table
2607  * @write: %TRUE if this is a write to the sysctl file
2608  * @buffer: the user buffer
2609  * @lenp: the size of the user buffer
2610  * @ppos: file position
2611  *
2612  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2613  * values from/to the user buffer, treated as an ASCII string. Negative
2614  * strings are not allowed.
2615  *
2616  * This routine will ensure the values are within the range specified by
2617  * table->extra1 (min) and table->extra2 (max). There is a final sanity
2618  * check for UINT_MAX to avoid having to support wrap around uses from
2619  * userspace.
2620  *
2621  * Returns 0 on success.
2622  */
2623 int proc_douintvec_minmax(struct ctl_table *table, int write,
2624 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2625 {
2626 	struct do_proc_douintvec_minmax_conv_param param = {
2627 		.min = (unsigned int *) table->extra1,
2628 		.max = (unsigned int *) table->extra2,
2629 	};
2630 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
2631 				 do_proc_douintvec_minmax_conv, &param);
2632 }
2633 
2634 static void validate_coredump_safety(void)
2635 {
2636 #ifdef CONFIG_COREDUMP
2637 	if (suid_dumpable == SUID_DUMP_ROOT &&
2638 	    core_pattern[0] != '/' && core_pattern[0] != '|') {
2639 		printk(KERN_WARNING
2640 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2641 "Pipe handler or fully qualified core dump path required.\n"
2642 "Set kernel.core_pattern before fs.suid_dumpable.\n"
2643 		);
2644 	}
2645 #endif
2646 }
2647 
2648 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2649 		void __user *buffer, size_t *lenp, loff_t *ppos)
2650 {
2651 	int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2652 	if (!error)
2653 		validate_coredump_safety();
2654 	return error;
2655 }
2656 
2657 #ifdef CONFIG_COREDUMP
2658 static int proc_dostring_coredump(struct ctl_table *table, int write,
2659 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2660 {
2661 	int error = proc_dostring(table, write, buffer, lenp, ppos);
2662 	if (!error)
2663 		validate_coredump_safety();
2664 	return error;
2665 }
2666 #endif
2667 
2668 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2669 				     void __user *buffer,
2670 				     size_t *lenp, loff_t *ppos,
2671 				     unsigned long convmul,
2672 				     unsigned long convdiv)
2673 {
2674 	unsigned long *i, *min, *max;
2675 	int vleft, first = 1, err = 0;
2676 	size_t left;
2677 	char *kbuf = NULL, *p;
2678 
2679 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2680 		*lenp = 0;
2681 		return 0;
2682 	}
2683 
2684 	i = (unsigned long *) data;
2685 	min = (unsigned long *) table->extra1;
2686 	max = (unsigned long *) table->extra2;
2687 	vleft = table->maxlen / sizeof(unsigned long);
2688 	left = *lenp;
2689 
2690 	if (write) {
2691 		if (proc_first_pos_non_zero_ignore(ppos, table))
2692 			goto out;
2693 
2694 		if (left > PAGE_SIZE - 1)
2695 			left = PAGE_SIZE - 1;
2696 		p = kbuf = memdup_user_nul(buffer, left);
2697 		if (IS_ERR(kbuf))
2698 			return PTR_ERR(kbuf);
2699 	}
2700 
2701 	for (; left && vleft--; i++, first = 0) {
2702 		unsigned long val;
2703 
2704 		if (write) {
2705 			bool neg;
2706 
2707 			left -= proc_skip_spaces(&p);
2708 
2709 			err = proc_get_long(&p, &left, &val, &neg,
2710 					     proc_wspace_sep,
2711 					     sizeof(proc_wspace_sep), NULL);
2712 			if (err)
2713 				break;
2714 			if (neg)
2715 				continue;
2716 			val = convmul * val / convdiv;
2717 			if ((min && val < *min) || (max && val > *max))
2718 				continue;
2719 			*i = val;
2720 		} else {
2721 			val = convdiv * (*i) / convmul;
2722 			if (!first) {
2723 				err = proc_put_char(&buffer, &left, '\t');
2724 				if (err)
2725 					break;
2726 			}
2727 			err = proc_put_long(&buffer, &left, val, false);
2728 			if (err)
2729 				break;
2730 		}
2731 	}
2732 
2733 	if (!write && !first && left && !err)
2734 		err = proc_put_char(&buffer, &left, '\n');
2735 	if (write && !err)
2736 		left -= proc_skip_spaces(&p);
2737 	if (write) {
2738 		kfree(kbuf);
2739 		if (first)
2740 			return err ? : -EINVAL;
2741 	}
2742 	*lenp -= left;
2743 out:
2744 	*ppos += *lenp;
2745 	return err;
2746 }
2747 
2748 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2749 				     void __user *buffer,
2750 				     size_t *lenp, loff_t *ppos,
2751 				     unsigned long convmul,
2752 				     unsigned long convdiv)
2753 {
2754 	return __do_proc_doulongvec_minmax(table->data, table, write,
2755 			buffer, lenp, ppos, convmul, convdiv);
2756 }
2757 
2758 /**
2759  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2760  * @table: the sysctl table
2761  * @write: %TRUE if this is a write to the sysctl file
2762  * @buffer: the user buffer
2763  * @lenp: the size of the user buffer
2764  * @ppos: file position
2765  *
2766  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2767  * values from/to the user buffer, treated as an ASCII string.
2768  *
2769  * This routine will ensure the values are within the range specified by
2770  * table->extra1 (min) and table->extra2 (max).
2771  *
2772  * Returns 0 on success.
2773  */
2774 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2775 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2776 {
2777     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2778 }
2779 
2780 /**
2781  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2782  * @table: the sysctl table
2783  * @write: %TRUE if this is a write to the sysctl file
2784  * @buffer: the user buffer
2785  * @lenp: the size of the user buffer
2786  * @ppos: file position
2787  *
2788  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2789  * values from/to the user buffer, treated as an ASCII string. The values
2790  * are treated as milliseconds, and converted to jiffies when they are stored.
2791  *
2792  * This routine will ensure the values are within the range specified by
2793  * table->extra1 (min) and table->extra2 (max).
2794  *
2795  * Returns 0 on success.
2796  */
2797 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2798 				      void __user *buffer,
2799 				      size_t *lenp, loff_t *ppos)
2800 {
2801     return do_proc_doulongvec_minmax(table, write, buffer,
2802 				     lenp, ppos, HZ, 1000l);
2803 }
2804 
2805 
2806 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2807 					 int *valp,
2808 					 int write, void *data)
2809 {
2810 	if (write) {
2811 		if (*lvalp > INT_MAX / HZ)
2812 			return 1;
2813 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2814 	} else {
2815 		int val = *valp;
2816 		unsigned long lval;
2817 		if (val < 0) {
2818 			*negp = true;
2819 			lval = -(unsigned long)val;
2820 		} else {
2821 			*negp = false;
2822 			lval = (unsigned long)val;
2823 		}
2824 		*lvalp = lval / HZ;
2825 	}
2826 	return 0;
2827 }
2828 
2829 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2830 						int *valp,
2831 						int write, void *data)
2832 {
2833 	if (write) {
2834 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2835 			return 1;
2836 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2837 	} else {
2838 		int val = *valp;
2839 		unsigned long lval;
2840 		if (val < 0) {
2841 			*negp = true;
2842 			lval = -(unsigned long)val;
2843 		} else {
2844 			*negp = false;
2845 			lval = (unsigned long)val;
2846 		}
2847 		*lvalp = jiffies_to_clock_t(lval);
2848 	}
2849 	return 0;
2850 }
2851 
2852 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2853 					    int *valp,
2854 					    int write, void *data)
2855 {
2856 	if (write) {
2857 		unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2858 
2859 		if (jif > INT_MAX)
2860 			return 1;
2861 		*valp = (int)jif;
2862 	} else {
2863 		int val = *valp;
2864 		unsigned long lval;
2865 		if (val < 0) {
2866 			*negp = true;
2867 			lval = -(unsigned long)val;
2868 		} else {
2869 			*negp = false;
2870 			lval = (unsigned long)val;
2871 		}
2872 		*lvalp = jiffies_to_msecs(lval);
2873 	}
2874 	return 0;
2875 }
2876 
2877 /**
2878  * proc_dointvec_jiffies - read a vector of integers as seconds
2879  * @table: the sysctl table
2880  * @write: %TRUE if this is a write to the sysctl file
2881  * @buffer: the user buffer
2882  * @lenp: the size of the user buffer
2883  * @ppos: file position
2884  *
2885  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2886  * values from/to the user buffer, treated as an ASCII string.
2887  * The values read are assumed to be in seconds, and are converted into
2888  * jiffies.
2889  *
2890  * Returns 0 on success.
2891  */
2892 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2893 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2894 {
2895     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2896 		    	    do_proc_dointvec_jiffies_conv,NULL);
2897 }
2898 
2899 /**
2900  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2901  * @table: the sysctl table
2902  * @write: %TRUE if this is a write to the sysctl file
2903  * @buffer: the user buffer
2904  * @lenp: the size of the user buffer
2905  * @ppos: pointer to the file position
2906  *
2907  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2908  * values from/to the user buffer, treated as an ASCII string.
2909  * The values read are assumed to be in 1/USER_HZ seconds, and
2910  * are converted into jiffies.
2911  *
2912  * Returns 0 on success.
2913  */
2914 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2915 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2916 {
2917     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2918 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2919 }
2920 
2921 /**
2922  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2923  * @table: the sysctl table
2924  * @write: %TRUE if this is a write to the sysctl file
2925  * @buffer: the user buffer
2926  * @lenp: the size of the user buffer
2927  * @ppos: file position
2928  * @ppos: the current position in the file
2929  *
2930  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2931  * values from/to the user buffer, treated as an ASCII string.
2932  * The values read are assumed to be in 1/1000 seconds, and
2933  * are converted into jiffies.
2934  *
2935  * Returns 0 on success.
2936  */
2937 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2938 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2939 {
2940 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2941 				do_proc_dointvec_ms_jiffies_conv, NULL);
2942 }
2943 
2944 static int proc_do_cad_pid(struct ctl_table *table, int write,
2945 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2946 {
2947 	struct pid *new_pid;
2948 	pid_t tmp;
2949 	int r;
2950 
2951 	tmp = pid_vnr(cad_pid);
2952 
2953 	r = __do_proc_dointvec(&tmp, table, write, buffer,
2954 			       lenp, ppos, NULL, NULL);
2955 	if (r || !write)
2956 		return r;
2957 
2958 	new_pid = find_get_pid(tmp);
2959 	if (!new_pid)
2960 		return -ESRCH;
2961 
2962 	put_pid(xchg(&cad_pid, new_pid));
2963 	return 0;
2964 }
2965 
2966 /**
2967  * proc_do_large_bitmap - read/write from/to a large bitmap
2968  * @table: the sysctl table
2969  * @write: %TRUE if this is a write to the sysctl file
2970  * @buffer: the user buffer
2971  * @lenp: the size of the user buffer
2972  * @ppos: file position
2973  *
2974  * The bitmap is stored at table->data and the bitmap length (in bits)
2975  * in table->maxlen.
2976  *
2977  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2978  * large bitmaps may be represented in a compact manner. Writing into
2979  * the file will clear the bitmap then update it with the given input.
2980  *
2981  * Returns 0 on success.
2982  */
2983 int proc_do_large_bitmap(struct ctl_table *table, int write,
2984 			 void __user *buffer, size_t *lenp, loff_t *ppos)
2985 {
2986 	int err = 0;
2987 	bool first = 1;
2988 	size_t left = *lenp;
2989 	unsigned long bitmap_len = table->maxlen;
2990 	unsigned long *bitmap = *(unsigned long **) table->data;
2991 	unsigned long *tmp_bitmap = NULL;
2992 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2993 
2994 	if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2995 		*lenp = 0;
2996 		return 0;
2997 	}
2998 
2999 	if (write) {
3000 		char *kbuf, *p;
3001 
3002 		if (left > PAGE_SIZE - 1)
3003 			left = PAGE_SIZE - 1;
3004 
3005 		p = kbuf = memdup_user_nul(buffer, left);
3006 		if (IS_ERR(kbuf))
3007 			return PTR_ERR(kbuf);
3008 
3009 		tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
3010 				     GFP_KERNEL);
3011 		if (!tmp_bitmap) {
3012 			kfree(kbuf);
3013 			return -ENOMEM;
3014 		}
3015 		proc_skip_char(&p, &left, '\n');
3016 		while (!err && left) {
3017 			unsigned long val_a, val_b;
3018 			bool neg;
3019 
3020 			err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3021 					     sizeof(tr_a), &c);
3022 			if (err)
3023 				break;
3024 			if (val_a >= bitmap_len || neg) {
3025 				err = -EINVAL;
3026 				break;
3027 			}
3028 
3029 			val_b = val_a;
3030 			if (left) {
3031 				p++;
3032 				left--;
3033 			}
3034 
3035 			if (c == '-') {
3036 				err = proc_get_long(&p, &left, &val_b,
3037 						     &neg, tr_b, sizeof(tr_b),
3038 						     &c);
3039 				if (err)
3040 					break;
3041 				if (val_b >= bitmap_len || neg ||
3042 				    val_a > val_b) {
3043 					err = -EINVAL;
3044 					break;
3045 				}
3046 				if (left) {
3047 					p++;
3048 					left--;
3049 				}
3050 			}
3051 
3052 			bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3053 			first = 0;
3054 			proc_skip_char(&p, &left, '\n');
3055 		}
3056 		kfree(kbuf);
3057 	} else {
3058 		unsigned long bit_a, bit_b = 0;
3059 
3060 		while (left) {
3061 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3062 			if (bit_a >= bitmap_len)
3063 				break;
3064 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
3065 						   bit_a + 1) - 1;
3066 
3067 			if (!first) {
3068 				err = proc_put_char(&buffer, &left, ',');
3069 				if (err)
3070 					break;
3071 			}
3072 			err = proc_put_long(&buffer, &left, bit_a, false);
3073 			if (err)
3074 				break;
3075 			if (bit_a != bit_b) {
3076 				err = proc_put_char(&buffer, &left, '-');
3077 				if (err)
3078 					break;
3079 				err = proc_put_long(&buffer, &left, bit_b, false);
3080 				if (err)
3081 					break;
3082 			}
3083 
3084 			first = 0; bit_b++;
3085 		}
3086 		if (!err)
3087 			err = proc_put_char(&buffer, &left, '\n');
3088 	}
3089 
3090 	if (!err) {
3091 		if (write) {
3092 			if (*ppos)
3093 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3094 			else
3095 				bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3096 		}
3097 		kfree(tmp_bitmap);
3098 		*lenp -= left;
3099 		*ppos += *lenp;
3100 		return 0;
3101 	} else {
3102 		kfree(tmp_bitmap);
3103 		return err;
3104 	}
3105 }
3106 
3107 #else /* CONFIG_PROC_SYSCTL */
3108 
3109 int proc_dostring(struct ctl_table *table, int write,
3110 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3111 {
3112 	return -ENOSYS;
3113 }
3114 
3115 int proc_dointvec(struct ctl_table *table, int write,
3116 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3117 {
3118 	return -ENOSYS;
3119 }
3120 
3121 int proc_douintvec(struct ctl_table *table, int write,
3122 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3123 {
3124 	return -ENOSYS;
3125 }
3126 
3127 int proc_dointvec_minmax(struct ctl_table *table, int write,
3128 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3129 {
3130 	return -ENOSYS;
3131 }
3132 
3133 int proc_douintvec_minmax(struct ctl_table *table, int write,
3134 			  void __user *buffer, size_t *lenp, loff_t *ppos)
3135 {
3136 	return -ENOSYS;
3137 }
3138 
3139 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3140 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3141 {
3142 	return -ENOSYS;
3143 }
3144 
3145 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3146 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3147 {
3148 	return -ENOSYS;
3149 }
3150 
3151 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3152 			     void __user *buffer, size_t *lenp, loff_t *ppos)
3153 {
3154 	return -ENOSYS;
3155 }
3156 
3157 int proc_doulongvec_minmax(struct ctl_table *table, int write,
3158 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3159 {
3160 	return -ENOSYS;
3161 }
3162 
3163 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3164 				      void __user *buffer,
3165 				      size_t *lenp, loff_t *ppos)
3166 {
3167     return -ENOSYS;
3168 }
3169 
3170 
3171 #endif /* CONFIG_PROC_SYSCTL */
3172 
3173 /*
3174  * No sense putting this after each symbol definition, twice,
3175  * exception granted :-)
3176  */
3177 EXPORT_SYMBOL(proc_dointvec);
3178 EXPORT_SYMBOL(proc_douintvec);
3179 EXPORT_SYMBOL(proc_dointvec_jiffies);
3180 EXPORT_SYMBOL(proc_dointvec_minmax);
3181 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3182 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3183 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3184 EXPORT_SYMBOL(proc_dostring);
3185 EXPORT_SYMBOL(proc_doulongvec_minmax);
3186 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3187