xref: /linux/kernel/sysctl.c (revision e60e1ee60630cafef5e430c2ae364877e061d980)
1 /*
2  * sysctl.c: General linux system control interface
3  *
4  * Begun 24 March 1995, Stephen Tweedie
5  * Added /proc support, Dec 1995
6  * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7  * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8  * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9  * Dynamic registration fixes, Stephen Tweedie.
10  * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11  * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12  *  Horn.
13  * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14  * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15  * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16  *  Wendling.
17  * The list_for_each() macro wasn't appropriate for the sysctl loop.
18  *  Removed it and replaced it with older style, 03/23/00, Bill Wendling
19  */
20 
21 #include <linux/module.h>
22 #include <linux/aio.h>
23 #include <linux/mm.h>
24 #include <linux/swap.h>
25 #include <linux/slab.h>
26 #include <linux/sysctl.h>
27 #include <linux/bitmap.h>
28 #include <linux/signal.h>
29 #include <linux/printk.h>
30 #include <linux/proc_fs.h>
31 #include <linux/security.h>
32 #include <linux/ctype.h>
33 #include <linux/kmemleak.h>
34 #include <linux/fs.h>
35 #include <linux/init.h>
36 #include <linux/kernel.h>
37 #include <linux/kobject.h>
38 #include <linux/net.h>
39 #include <linux/sysrq.h>
40 #include <linux/highuid.h>
41 #include <linux/writeback.h>
42 #include <linux/ratelimit.h>
43 #include <linux/compaction.h>
44 #include <linux/hugetlb.h>
45 #include <linux/initrd.h>
46 #include <linux/key.h>
47 #include <linux/times.h>
48 #include <linux/limits.h>
49 #include <linux/dcache.h>
50 #include <linux/dnotify.h>
51 #include <linux/syscalls.h>
52 #include <linux/vmstat.h>
53 #include <linux/nfs_fs.h>
54 #include <linux/acpi.h>
55 #include <linux/reboot.h>
56 #include <linux/ftrace.h>
57 #include <linux/perf_event.h>
58 #include <linux/kprobes.h>
59 #include <linux/pipe_fs_i.h>
60 #include <linux/oom.h>
61 #include <linux/kmod.h>
62 #include <linux/capability.h>
63 #include <linux/binfmts.h>
64 #include <linux/sched/sysctl.h>
65 #include <linux/sched/coredump.h>
66 #include <linux/kexec.h>
67 #include <linux/bpf.h>
68 #include <linux/mount.h>
69 
70 #include <linux/uaccess.h>
71 #include <asm/processor.h>
72 
73 #ifdef CONFIG_X86
74 #include <asm/nmi.h>
75 #include <asm/stacktrace.h>
76 #include <asm/io.h>
77 #endif
78 #ifdef CONFIG_SPARC
79 #include <asm/setup.h>
80 #endif
81 #ifdef CONFIG_BSD_PROCESS_ACCT
82 #include <linux/acct.h>
83 #endif
84 #ifdef CONFIG_RT_MUTEXES
85 #include <linux/rtmutex.h>
86 #endif
87 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
88 #include <linux/lockdep.h>
89 #endif
90 #ifdef CONFIG_CHR_DEV_SG
91 #include <scsi/sg.h>
92 #endif
93 
94 #ifdef CONFIG_LOCKUP_DETECTOR
95 #include <linux/nmi.h>
96 #endif
97 
98 #if defined(CONFIG_SYSCTL)
99 
100 /* External variables not in a header file. */
101 extern int suid_dumpable;
102 #ifdef CONFIG_COREDUMP
103 extern int core_uses_pid;
104 extern char core_pattern[];
105 extern unsigned int core_pipe_limit;
106 #endif
107 extern int pid_max;
108 extern int pid_max_min, pid_max_max;
109 extern int percpu_pagelist_fraction;
110 extern int latencytop_enabled;
111 extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
112 #ifndef CONFIG_MMU
113 extern int sysctl_nr_trim_pages;
114 #endif
115 
116 /* Constants used for minimum and  maximum */
117 #ifdef CONFIG_LOCKUP_DETECTOR
118 static int sixty = 60;
119 #endif
120 
121 static int __maybe_unused neg_one = -1;
122 
123 static int zero;
124 static int __maybe_unused one = 1;
125 static int __maybe_unused two = 2;
126 static int __maybe_unused four = 4;
127 static unsigned long one_ul = 1;
128 static int one_hundred = 100;
129 static int one_thousand = 1000;
130 #ifdef CONFIG_PRINTK
131 static int ten_thousand = 10000;
132 #endif
133 #ifdef CONFIG_PERF_EVENTS
134 static int six_hundred_forty_kb = 640 * 1024;
135 #endif
136 
137 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
138 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
139 
140 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
141 static int maxolduid = 65535;
142 static int minolduid;
143 
144 static int ngroups_max = NGROUPS_MAX;
145 static const int cap_last_cap = CAP_LAST_CAP;
146 
147 /*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */
148 #ifdef CONFIG_DETECT_HUNG_TASK
149 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
150 #endif
151 
152 #ifdef CONFIG_INOTIFY_USER
153 #include <linux/inotify.h>
154 #endif
155 #ifdef CONFIG_SPARC
156 #endif
157 
158 #ifdef __hppa__
159 extern int pwrsw_enabled;
160 #endif
161 
162 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
163 extern int unaligned_enabled;
164 #endif
165 
166 #ifdef CONFIG_IA64
167 extern int unaligned_dump_stack;
168 #endif
169 
170 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
171 extern int no_unaligned_warning;
172 #endif
173 
174 #ifdef CONFIG_PROC_SYSCTL
175 
176 /**
177  * enum sysctl_writes_mode - supported sysctl write modes
178  *
179  * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
180  * 	to be written, and multiple writes on the same sysctl file descriptor
181  * 	will rewrite the sysctl value, regardless of file position. No warning
182  * 	is issued when the initial position is not 0.
183  * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
184  * 	not 0.
185  * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
186  * 	file position 0 and the value must be fully contained in the buffer
187  * 	sent to the write syscall. If dealing with strings respect the file
188  * 	position, but restrict this to the max length of the buffer, anything
189  * 	passed the max lenght will be ignored. Multiple writes will append
190  * 	to the buffer.
191  *
192  * These write modes control how current file position affects the behavior of
193  * updating sysctl values through the proc interface on each write.
194  */
195 enum sysctl_writes_mode {
196 	SYSCTL_WRITES_LEGACY		= -1,
197 	SYSCTL_WRITES_WARN		= 0,
198 	SYSCTL_WRITES_STRICT		= 1,
199 };
200 
201 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
202 
203 static int proc_do_cad_pid(struct ctl_table *table, int write,
204 		  void __user *buffer, size_t *lenp, loff_t *ppos);
205 static int proc_taint(struct ctl_table *table, int write,
206 			       void __user *buffer, size_t *lenp, loff_t *ppos);
207 #endif
208 
209 #ifdef CONFIG_PRINTK
210 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
211 				void __user *buffer, size_t *lenp, loff_t *ppos);
212 #endif
213 
214 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
215 		void __user *buffer, size_t *lenp, loff_t *ppos);
216 #ifdef CONFIG_COREDUMP
217 static int proc_dostring_coredump(struct ctl_table *table, int write,
218 		void __user *buffer, size_t *lenp, loff_t *ppos);
219 #endif
220 
221 #ifdef CONFIG_MAGIC_SYSRQ
222 /* Note: sysrq code uses it's own private copy */
223 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
224 
225 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
226 				void __user *buffer, size_t *lenp,
227 				loff_t *ppos)
228 {
229 	int error;
230 
231 	error = proc_dointvec(table, write, buffer, lenp, ppos);
232 	if (error)
233 		return error;
234 
235 	if (write)
236 		sysrq_toggle_support(__sysrq_enabled);
237 
238 	return 0;
239 }
240 
241 #endif
242 
243 static struct ctl_table kern_table[];
244 static struct ctl_table vm_table[];
245 static struct ctl_table fs_table[];
246 static struct ctl_table debug_table[];
247 static struct ctl_table dev_table[];
248 extern struct ctl_table random_table[];
249 #ifdef CONFIG_EPOLL
250 extern struct ctl_table epoll_table[];
251 #endif
252 
253 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
254 int sysctl_legacy_va_layout;
255 #endif
256 
257 /* The default sysctl tables: */
258 
259 static struct ctl_table sysctl_base_table[] = {
260 	{
261 		.procname	= "kernel",
262 		.mode		= 0555,
263 		.child		= kern_table,
264 	},
265 	{
266 		.procname	= "vm",
267 		.mode		= 0555,
268 		.child		= vm_table,
269 	},
270 	{
271 		.procname	= "fs",
272 		.mode		= 0555,
273 		.child		= fs_table,
274 	},
275 	{
276 		.procname	= "debug",
277 		.mode		= 0555,
278 		.child		= debug_table,
279 	},
280 	{
281 		.procname	= "dev",
282 		.mode		= 0555,
283 		.child		= dev_table,
284 	},
285 	{ }
286 };
287 
288 #ifdef CONFIG_SCHED_DEBUG
289 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
290 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
291 static int min_wakeup_granularity_ns;			/* 0 usecs */
292 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
293 #ifdef CONFIG_SMP
294 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
295 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
296 #endif /* CONFIG_SMP */
297 #endif /* CONFIG_SCHED_DEBUG */
298 
299 #ifdef CONFIG_COMPACTION
300 static int min_extfrag_threshold;
301 static int max_extfrag_threshold = 1000;
302 #endif
303 
304 static struct ctl_table kern_table[] = {
305 	{
306 		.procname	= "sched_child_runs_first",
307 		.data		= &sysctl_sched_child_runs_first,
308 		.maxlen		= sizeof(unsigned int),
309 		.mode		= 0644,
310 		.proc_handler	= proc_dointvec,
311 	},
312 #ifdef CONFIG_SCHED_DEBUG
313 	{
314 		.procname	= "sched_min_granularity_ns",
315 		.data		= &sysctl_sched_min_granularity,
316 		.maxlen		= sizeof(unsigned int),
317 		.mode		= 0644,
318 		.proc_handler	= sched_proc_update_handler,
319 		.extra1		= &min_sched_granularity_ns,
320 		.extra2		= &max_sched_granularity_ns,
321 	},
322 	{
323 		.procname	= "sched_latency_ns",
324 		.data		= &sysctl_sched_latency,
325 		.maxlen		= sizeof(unsigned int),
326 		.mode		= 0644,
327 		.proc_handler	= sched_proc_update_handler,
328 		.extra1		= &min_sched_granularity_ns,
329 		.extra2		= &max_sched_granularity_ns,
330 	},
331 	{
332 		.procname	= "sched_wakeup_granularity_ns",
333 		.data		= &sysctl_sched_wakeup_granularity,
334 		.maxlen		= sizeof(unsigned int),
335 		.mode		= 0644,
336 		.proc_handler	= sched_proc_update_handler,
337 		.extra1		= &min_wakeup_granularity_ns,
338 		.extra2		= &max_wakeup_granularity_ns,
339 	},
340 #ifdef CONFIG_SMP
341 	{
342 		.procname	= "sched_tunable_scaling",
343 		.data		= &sysctl_sched_tunable_scaling,
344 		.maxlen		= sizeof(enum sched_tunable_scaling),
345 		.mode		= 0644,
346 		.proc_handler	= sched_proc_update_handler,
347 		.extra1		= &min_sched_tunable_scaling,
348 		.extra2		= &max_sched_tunable_scaling,
349 	},
350 	{
351 		.procname	= "sched_migration_cost_ns",
352 		.data		= &sysctl_sched_migration_cost,
353 		.maxlen		= sizeof(unsigned int),
354 		.mode		= 0644,
355 		.proc_handler	= proc_dointvec,
356 	},
357 	{
358 		.procname	= "sched_nr_migrate",
359 		.data		= &sysctl_sched_nr_migrate,
360 		.maxlen		= sizeof(unsigned int),
361 		.mode		= 0644,
362 		.proc_handler	= proc_dointvec,
363 	},
364 	{
365 		.procname	= "sched_time_avg_ms",
366 		.data		= &sysctl_sched_time_avg,
367 		.maxlen		= sizeof(unsigned int),
368 		.mode		= 0644,
369 		.proc_handler	= proc_dointvec_minmax,
370 		.extra1		= &one,
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_user_enabled,
893 		.maxlen		= sizeof(int),
894 		.mode		= NMI_WATCHDOG_SYSCTL_PERM,
895 		.proc_handler   = proc_nmi_watchdog,
896 		.extra1		= &zero,
897 		.extra2		= &one,
898 	},
899 	{
900 		.procname	= "watchdog_cpumask",
901 		.data		= &watchdog_cpumask_bits,
902 		.maxlen		= NR_CPUS,
903 		.mode		= 0644,
904 		.proc_handler	= proc_watchdog_cpumask,
905 	},
906 #ifdef CONFIG_SOFTLOCKUP_DETECTOR
907 	{
908 		.procname       = "soft_watchdog",
909 		.data		= &soft_watchdog_user_enabled,
910 		.maxlen		= sizeof(int),
911 		.mode		= 0644,
912 		.proc_handler   = proc_soft_watchdog,
913 		.extra1		= &zero,
914 		.extra2		= &one,
915 	},
916 	{
917 		.procname	= "softlockup_panic",
918 		.data		= &softlockup_panic,
919 		.maxlen		= sizeof(int),
920 		.mode		= 0644,
921 		.proc_handler	= proc_dointvec_minmax,
922 		.extra1		= &zero,
923 		.extra2		= &one,
924 	},
925 #ifdef CONFIG_SMP
926 	{
927 		.procname	= "softlockup_all_cpu_backtrace",
928 		.data		= &sysctl_softlockup_all_cpu_backtrace,
929 		.maxlen		= sizeof(int),
930 		.mode		= 0644,
931 		.proc_handler	= proc_dointvec_minmax,
932 		.extra1		= &zero,
933 		.extra2		= &one,
934 	},
935 #endif /* CONFIG_SMP */
936 #endif
937 #ifdef CONFIG_HARDLOCKUP_DETECTOR
938 	{
939 		.procname	= "hardlockup_panic",
940 		.data		= &hardlockup_panic,
941 		.maxlen		= sizeof(int),
942 		.mode		= 0644,
943 		.proc_handler	= proc_dointvec_minmax,
944 		.extra1		= &zero,
945 		.extra2		= &one,
946 	},
947 #ifdef CONFIG_SMP
948 	{
949 		.procname	= "hardlockup_all_cpu_backtrace",
950 		.data		= &sysctl_hardlockup_all_cpu_backtrace,
951 		.maxlen		= sizeof(int),
952 		.mode		= 0644,
953 		.proc_handler	= proc_dointvec_minmax,
954 		.extra1		= &zero,
955 		.extra2		= &one,
956 	},
957 #endif /* CONFIG_SMP */
958 #endif
959 #endif
960 
961 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
962 	{
963 		.procname       = "unknown_nmi_panic",
964 		.data           = &unknown_nmi_panic,
965 		.maxlen         = sizeof (int),
966 		.mode           = 0644,
967 		.proc_handler   = proc_dointvec,
968 	},
969 #endif
970 #if defined(CONFIG_X86)
971 	{
972 		.procname	= "panic_on_unrecovered_nmi",
973 		.data		= &panic_on_unrecovered_nmi,
974 		.maxlen		= sizeof(int),
975 		.mode		= 0644,
976 		.proc_handler	= proc_dointvec,
977 	},
978 	{
979 		.procname	= "panic_on_io_nmi",
980 		.data		= &panic_on_io_nmi,
981 		.maxlen		= sizeof(int),
982 		.mode		= 0644,
983 		.proc_handler	= proc_dointvec,
984 	},
985 #ifdef CONFIG_DEBUG_STACKOVERFLOW
986 	{
987 		.procname	= "panic_on_stackoverflow",
988 		.data		= &sysctl_panic_on_stackoverflow,
989 		.maxlen		= sizeof(int),
990 		.mode		= 0644,
991 		.proc_handler	= proc_dointvec,
992 	},
993 #endif
994 	{
995 		.procname	= "bootloader_type",
996 		.data		= &bootloader_type,
997 		.maxlen		= sizeof (int),
998 		.mode		= 0444,
999 		.proc_handler	= proc_dointvec,
1000 	},
1001 	{
1002 		.procname	= "bootloader_version",
1003 		.data		= &bootloader_version,
1004 		.maxlen		= sizeof (int),
1005 		.mode		= 0444,
1006 		.proc_handler	= proc_dointvec,
1007 	},
1008 	{
1009 		.procname	= "io_delay_type",
1010 		.data		= &io_delay_type,
1011 		.maxlen		= sizeof(int),
1012 		.mode		= 0644,
1013 		.proc_handler	= proc_dointvec,
1014 	},
1015 #endif
1016 #if defined(CONFIG_MMU)
1017 	{
1018 		.procname	= "randomize_va_space",
1019 		.data		= &randomize_va_space,
1020 		.maxlen		= sizeof(int),
1021 		.mode		= 0644,
1022 		.proc_handler	= proc_dointvec,
1023 	},
1024 #endif
1025 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1026 	{
1027 		.procname	= "spin_retry",
1028 		.data		= &spin_retry,
1029 		.maxlen		= sizeof (int),
1030 		.mode		= 0644,
1031 		.proc_handler	= proc_dointvec,
1032 	},
1033 #endif
1034 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1035 	{
1036 		.procname	= "acpi_video_flags",
1037 		.data		= &acpi_realmode_flags,
1038 		.maxlen		= sizeof (unsigned long),
1039 		.mode		= 0644,
1040 		.proc_handler	= proc_doulongvec_minmax,
1041 	},
1042 #endif
1043 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1044 	{
1045 		.procname	= "ignore-unaligned-usertrap",
1046 		.data		= &no_unaligned_warning,
1047 		.maxlen		= sizeof (int),
1048 	 	.mode		= 0644,
1049 		.proc_handler	= proc_dointvec,
1050 	},
1051 #endif
1052 #ifdef CONFIG_IA64
1053 	{
1054 		.procname	= "unaligned-dump-stack",
1055 		.data		= &unaligned_dump_stack,
1056 		.maxlen		= sizeof (int),
1057 		.mode		= 0644,
1058 		.proc_handler	= proc_dointvec,
1059 	},
1060 #endif
1061 #ifdef CONFIG_DETECT_HUNG_TASK
1062 	{
1063 		.procname	= "hung_task_panic",
1064 		.data		= &sysctl_hung_task_panic,
1065 		.maxlen		= sizeof(int),
1066 		.mode		= 0644,
1067 		.proc_handler	= proc_dointvec_minmax,
1068 		.extra1		= &zero,
1069 		.extra2		= &one,
1070 	},
1071 	{
1072 		.procname	= "hung_task_check_count",
1073 		.data		= &sysctl_hung_task_check_count,
1074 		.maxlen		= sizeof(int),
1075 		.mode		= 0644,
1076 		.proc_handler	= proc_dointvec_minmax,
1077 		.extra1		= &zero,
1078 	},
1079 	{
1080 		.procname	= "hung_task_timeout_secs",
1081 		.data		= &sysctl_hung_task_timeout_secs,
1082 		.maxlen		= sizeof(unsigned long),
1083 		.mode		= 0644,
1084 		.proc_handler	= proc_dohung_task_timeout_secs,
1085 		.extra2		= &hung_task_timeout_max,
1086 	},
1087 	{
1088 		.procname	= "hung_task_warnings",
1089 		.data		= &sysctl_hung_task_warnings,
1090 		.maxlen		= sizeof(int),
1091 		.mode		= 0644,
1092 		.proc_handler	= proc_dointvec_minmax,
1093 		.extra1		= &neg_one,
1094 	},
1095 #endif
1096 #ifdef CONFIG_RT_MUTEXES
1097 	{
1098 		.procname	= "max_lock_depth",
1099 		.data		= &max_lock_depth,
1100 		.maxlen		= sizeof(int),
1101 		.mode		= 0644,
1102 		.proc_handler	= proc_dointvec,
1103 	},
1104 #endif
1105 	{
1106 		.procname	= "poweroff_cmd",
1107 		.data		= &poweroff_cmd,
1108 		.maxlen		= POWEROFF_CMD_PATH_LEN,
1109 		.mode		= 0644,
1110 		.proc_handler	= proc_dostring,
1111 	},
1112 #ifdef CONFIG_KEYS
1113 	{
1114 		.procname	= "keys",
1115 		.mode		= 0555,
1116 		.child		= key_sysctls,
1117 	},
1118 #endif
1119 #ifdef CONFIG_PERF_EVENTS
1120 	/*
1121 	 * User-space scripts rely on the existence of this file
1122 	 * as a feature check for perf_events being enabled.
1123 	 *
1124 	 * So it's an ABI, do not remove!
1125 	 */
1126 	{
1127 		.procname	= "perf_event_paranoid",
1128 		.data		= &sysctl_perf_event_paranoid,
1129 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
1130 		.mode		= 0644,
1131 		.proc_handler	= proc_dointvec,
1132 	},
1133 	{
1134 		.procname	= "perf_event_mlock_kb",
1135 		.data		= &sysctl_perf_event_mlock,
1136 		.maxlen		= sizeof(sysctl_perf_event_mlock),
1137 		.mode		= 0644,
1138 		.proc_handler	= proc_dointvec,
1139 	},
1140 	{
1141 		.procname	= "perf_event_max_sample_rate",
1142 		.data		= &sysctl_perf_event_sample_rate,
1143 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
1144 		.mode		= 0644,
1145 		.proc_handler	= perf_proc_update_handler,
1146 		.extra1		= &one,
1147 	},
1148 	{
1149 		.procname	= "perf_cpu_time_max_percent",
1150 		.data		= &sysctl_perf_cpu_time_max_percent,
1151 		.maxlen		= sizeof(sysctl_perf_cpu_time_max_percent),
1152 		.mode		= 0644,
1153 		.proc_handler	= perf_cpu_time_max_percent_handler,
1154 		.extra1		= &zero,
1155 		.extra2		= &one_hundred,
1156 	},
1157 	{
1158 		.procname	= "perf_event_max_stack",
1159 		.data		= &sysctl_perf_event_max_stack,
1160 		.maxlen		= sizeof(sysctl_perf_event_max_stack),
1161 		.mode		= 0644,
1162 		.proc_handler	= perf_event_max_stack_handler,
1163 		.extra1		= &zero,
1164 		.extra2		= &six_hundred_forty_kb,
1165 	},
1166 	{
1167 		.procname	= "perf_event_max_contexts_per_stack",
1168 		.data		= &sysctl_perf_event_max_contexts_per_stack,
1169 		.maxlen		= sizeof(sysctl_perf_event_max_contexts_per_stack),
1170 		.mode		= 0644,
1171 		.proc_handler	= perf_event_max_stack_handler,
1172 		.extra1		= &zero,
1173 		.extra2		= &one_thousand,
1174 	},
1175 #endif
1176 	{
1177 		.procname	= "panic_on_warn",
1178 		.data		= &panic_on_warn,
1179 		.maxlen		= sizeof(int),
1180 		.mode		= 0644,
1181 		.proc_handler	= proc_dointvec_minmax,
1182 		.extra1		= &zero,
1183 		.extra2		= &one,
1184 	},
1185 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1186 	{
1187 		.procname	= "timer_migration",
1188 		.data		= &sysctl_timer_migration,
1189 		.maxlen		= sizeof(unsigned int),
1190 		.mode		= 0644,
1191 		.proc_handler	= timer_migration_handler,
1192 		.extra1		= &zero,
1193 		.extra2		= &one,
1194 	},
1195 #endif
1196 #ifdef CONFIG_BPF_SYSCALL
1197 	{
1198 		.procname	= "unprivileged_bpf_disabled",
1199 		.data		= &sysctl_unprivileged_bpf_disabled,
1200 		.maxlen		= sizeof(sysctl_unprivileged_bpf_disabled),
1201 		.mode		= 0644,
1202 		/* only handle a transition from default "0" to "1" */
1203 		.proc_handler	= proc_dointvec_minmax,
1204 		.extra1		= &one,
1205 		.extra2		= &one,
1206 	},
1207 #endif
1208 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1209 	{
1210 		.procname	= "panic_on_rcu_stall",
1211 		.data		= &sysctl_panic_on_rcu_stall,
1212 		.maxlen		= sizeof(sysctl_panic_on_rcu_stall),
1213 		.mode		= 0644,
1214 		.proc_handler	= proc_dointvec_minmax,
1215 		.extra1		= &zero,
1216 		.extra2		= &one,
1217 	},
1218 #endif
1219 	{ }
1220 };
1221 
1222 static struct ctl_table vm_table[] = {
1223 	{
1224 		.procname	= "overcommit_memory",
1225 		.data		= &sysctl_overcommit_memory,
1226 		.maxlen		= sizeof(sysctl_overcommit_memory),
1227 		.mode		= 0644,
1228 		.proc_handler	= proc_dointvec_minmax,
1229 		.extra1		= &zero,
1230 		.extra2		= &two,
1231 	},
1232 	{
1233 		.procname	= "panic_on_oom",
1234 		.data		= &sysctl_panic_on_oom,
1235 		.maxlen		= sizeof(sysctl_panic_on_oom),
1236 		.mode		= 0644,
1237 		.proc_handler	= proc_dointvec_minmax,
1238 		.extra1		= &zero,
1239 		.extra2		= &two,
1240 	},
1241 	{
1242 		.procname	= "oom_kill_allocating_task",
1243 		.data		= &sysctl_oom_kill_allocating_task,
1244 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1245 		.mode		= 0644,
1246 		.proc_handler	= proc_dointvec,
1247 	},
1248 	{
1249 		.procname	= "oom_dump_tasks",
1250 		.data		= &sysctl_oom_dump_tasks,
1251 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1252 		.mode		= 0644,
1253 		.proc_handler	= proc_dointvec,
1254 	},
1255 	{
1256 		.procname	= "overcommit_ratio",
1257 		.data		= &sysctl_overcommit_ratio,
1258 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1259 		.mode		= 0644,
1260 		.proc_handler	= overcommit_ratio_handler,
1261 	},
1262 	{
1263 		.procname	= "overcommit_kbytes",
1264 		.data		= &sysctl_overcommit_kbytes,
1265 		.maxlen		= sizeof(sysctl_overcommit_kbytes),
1266 		.mode		= 0644,
1267 		.proc_handler	= overcommit_kbytes_handler,
1268 	},
1269 	{
1270 		.procname	= "page-cluster",
1271 		.data		= &page_cluster,
1272 		.maxlen		= sizeof(int),
1273 		.mode		= 0644,
1274 		.proc_handler	= proc_dointvec_minmax,
1275 		.extra1		= &zero,
1276 	},
1277 	{
1278 		.procname	= "dirty_background_ratio",
1279 		.data		= &dirty_background_ratio,
1280 		.maxlen		= sizeof(dirty_background_ratio),
1281 		.mode		= 0644,
1282 		.proc_handler	= dirty_background_ratio_handler,
1283 		.extra1		= &zero,
1284 		.extra2		= &one_hundred,
1285 	},
1286 	{
1287 		.procname	= "dirty_background_bytes",
1288 		.data		= &dirty_background_bytes,
1289 		.maxlen		= sizeof(dirty_background_bytes),
1290 		.mode		= 0644,
1291 		.proc_handler	= dirty_background_bytes_handler,
1292 		.extra1		= &one_ul,
1293 	},
1294 	{
1295 		.procname	= "dirty_ratio",
1296 		.data		= &vm_dirty_ratio,
1297 		.maxlen		= sizeof(vm_dirty_ratio),
1298 		.mode		= 0644,
1299 		.proc_handler	= dirty_ratio_handler,
1300 		.extra1		= &zero,
1301 		.extra2		= &one_hundred,
1302 	},
1303 	{
1304 		.procname	= "dirty_bytes",
1305 		.data		= &vm_dirty_bytes,
1306 		.maxlen		= sizeof(vm_dirty_bytes),
1307 		.mode		= 0644,
1308 		.proc_handler	= dirty_bytes_handler,
1309 		.extra1		= &dirty_bytes_min,
1310 	},
1311 	{
1312 		.procname	= "dirty_writeback_centisecs",
1313 		.data		= &dirty_writeback_interval,
1314 		.maxlen		= sizeof(dirty_writeback_interval),
1315 		.mode		= 0644,
1316 		.proc_handler	= dirty_writeback_centisecs_handler,
1317 	},
1318 	{
1319 		.procname	= "dirty_expire_centisecs",
1320 		.data		= &dirty_expire_interval,
1321 		.maxlen		= sizeof(dirty_expire_interval),
1322 		.mode		= 0644,
1323 		.proc_handler	= proc_dointvec_minmax,
1324 		.extra1		= &zero,
1325 	},
1326 	{
1327 		.procname	= "dirtytime_expire_seconds",
1328 		.data		= &dirtytime_expire_interval,
1329 		.maxlen		= sizeof(dirty_expire_interval),
1330 		.mode		= 0644,
1331 		.proc_handler	= dirtytime_interval_handler,
1332 		.extra1		= &zero,
1333 	},
1334 	{
1335 		.procname	= "swappiness",
1336 		.data		= &vm_swappiness,
1337 		.maxlen		= sizeof(vm_swappiness),
1338 		.mode		= 0644,
1339 		.proc_handler	= proc_dointvec_minmax,
1340 		.extra1		= &zero,
1341 		.extra2		= &one_hundred,
1342 	},
1343 #ifdef CONFIG_HUGETLB_PAGE
1344 	{
1345 		.procname	= "nr_hugepages",
1346 		.data		= NULL,
1347 		.maxlen		= sizeof(unsigned long),
1348 		.mode		= 0644,
1349 		.proc_handler	= hugetlb_sysctl_handler,
1350 	},
1351 #ifdef CONFIG_NUMA
1352 	{
1353 		.procname       = "nr_hugepages_mempolicy",
1354 		.data           = NULL,
1355 		.maxlen         = sizeof(unsigned long),
1356 		.mode           = 0644,
1357 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1358 	},
1359 	{
1360 		.procname		= "numa_stat",
1361 		.data			= &sysctl_vm_numa_stat,
1362 		.maxlen			= sizeof(int),
1363 		.mode			= 0644,
1364 		.proc_handler	= sysctl_vm_numa_stat_handler,
1365 		.extra1			= &zero,
1366 		.extra2			= &one,
1367 	},
1368 #endif
1369 	 {
1370 		.procname	= "hugetlb_shm_group",
1371 		.data		= &sysctl_hugetlb_shm_group,
1372 		.maxlen		= sizeof(gid_t),
1373 		.mode		= 0644,
1374 		.proc_handler	= proc_dointvec,
1375 	 },
1376 	 {
1377 		.procname	= "hugepages_treat_as_movable",
1378 		.data		= &hugepages_treat_as_movable,
1379 		.maxlen		= sizeof(int),
1380 		.mode		= 0644,
1381 		.proc_handler	= proc_dointvec,
1382 	},
1383 	{
1384 		.procname	= "nr_overcommit_hugepages",
1385 		.data		= NULL,
1386 		.maxlen		= sizeof(unsigned long),
1387 		.mode		= 0644,
1388 		.proc_handler	= hugetlb_overcommit_handler,
1389 	},
1390 #endif
1391 	{
1392 		.procname	= "lowmem_reserve_ratio",
1393 		.data		= &sysctl_lowmem_reserve_ratio,
1394 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1395 		.mode		= 0644,
1396 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1397 	},
1398 	{
1399 		.procname	= "drop_caches",
1400 		.data		= &sysctl_drop_caches,
1401 		.maxlen		= sizeof(int),
1402 		.mode		= 0644,
1403 		.proc_handler	= drop_caches_sysctl_handler,
1404 		.extra1		= &one,
1405 		.extra2		= &four,
1406 	},
1407 #ifdef CONFIG_COMPACTION
1408 	{
1409 		.procname	= "compact_memory",
1410 		.data		= &sysctl_compact_memory,
1411 		.maxlen		= sizeof(int),
1412 		.mode		= 0200,
1413 		.proc_handler	= sysctl_compaction_handler,
1414 	},
1415 	{
1416 		.procname	= "extfrag_threshold",
1417 		.data		= &sysctl_extfrag_threshold,
1418 		.maxlen		= sizeof(int),
1419 		.mode		= 0644,
1420 		.proc_handler	= sysctl_extfrag_handler,
1421 		.extra1		= &min_extfrag_threshold,
1422 		.extra2		= &max_extfrag_threshold,
1423 	},
1424 	{
1425 		.procname	= "compact_unevictable_allowed",
1426 		.data		= &sysctl_compact_unevictable_allowed,
1427 		.maxlen		= sizeof(int),
1428 		.mode		= 0644,
1429 		.proc_handler	= proc_dointvec,
1430 		.extra1		= &zero,
1431 		.extra2		= &one,
1432 	},
1433 
1434 #endif /* CONFIG_COMPACTION */
1435 	{
1436 		.procname	= "min_free_kbytes",
1437 		.data		= &min_free_kbytes,
1438 		.maxlen		= sizeof(min_free_kbytes),
1439 		.mode		= 0644,
1440 		.proc_handler	= min_free_kbytes_sysctl_handler,
1441 		.extra1		= &zero,
1442 	},
1443 	{
1444 		.procname	= "watermark_scale_factor",
1445 		.data		= &watermark_scale_factor,
1446 		.maxlen		= sizeof(watermark_scale_factor),
1447 		.mode		= 0644,
1448 		.proc_handler	= watermark_scale_factor_sysctl_handler,
1449 		.extra1		= &one,
1450 		.extra2		= &one_thousand,
1451 	},
1452 	{
1453 		.procname	= "percpu_pagelist_fraction",
1454 		.data		= &percpu_pagelist_fraction,
1455 		.maxlen		= sizeof(percpu_pagelist_fraction),
1456 		.mode		= 0644,
1457 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1458 		.extra1		= &zero,
1459 	},
1460 #ifdef CONFIG_MMU
1461 	{
1462 		.procname	= "max_map_count",
1463 		.data		= &sysctl_max_map_count,
1464 		.maxlen		= sizeof(sysctl_max_map_count),
1465 		.mode		= 0644,
1466 		.proc_handler	= proc_dointvec_minmax,
1467 		.extra1		= &zero,
1468 	},
1469 #else
1470 	{
1471 		.procname	= "nr_trim_pages",
1472 		.data		= &sysctl_nr_trim_pages,
1473 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1474 		.mode		= 0644,
1475 		.proc_handler	= proc_dointvec_minmax,
1476 		.extra1		= &zero,
1477 	},
1478 #endif
1479 	{
1480 		.procname	= "laptop_mode",
1481 		.data		= &laptop_mode,
1482 		.maxlen		= sizeof(laptop_mode),
1483 		.mode		= 0644,
1484 		.proc_handler	= proc_dointvec_jiffies,
1485 	},
1486 	{
1487 		.procname	= "block_dump",
1488 		.data		= &block_dump,
1489 		.maxlen		= sizeof(block_dump),
1490 		.mode		= 0644,
1491 		.proc_handler	= proc_dointvec,
1492 		.extra1		= &zero,
1493 	},
1494 	{
1495 		.procname	= "vfs_cache_pressure",
1496 		.data		= &sysctl_vfs_cache_pressure,
1497 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1498 		.mode		= 0644,
1499 		.proc_handler	= proc_dointvec,
1500 		.extra1		= &zero,
1501 	},
1502 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1503 	{
1504 		.procname	= "legacy_va_layout",
1505 		.data		= &sysctl_legacy_va_layout,
1506 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1507 		.mode		= 0644,
1508 		.proc_handler	= proc_dointvec,
1509 		.extra1		= &zero,
1510 	},
1511 #endif
1512 #ifdef CONFIG_NUMA
1513 	{
1514 		.procname	= "zone_reclaim_mode",
1515 		.data		= &node_reclaim_mode,
1516 		.maxlen		= sizeof(node_reclaim_mode),
1517 		.mode		= 0644,
1518 		.proc_handler	= proc_dointvec,
1519 		.extra1		= &zero,
1520 	},
1521 	{
1522 		.procname	= "min_unmapped_ratio",
1523 		.data		= &sysctl_min_unmapped_ratio,
1524 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1525 		.mode		= 0644,
1526 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1527 		.extra1		= &zero,
1528 		.extra2		= &one_hundred,
1529 	},
1530 	{
1531 		.procname	= "min_slab_ratio",
1532 		.data		= &sysctl_min_slab_ratio,
1533 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1534 		.mode		= 0644,
1535 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1536 		.extra1		= &zero,
1537 		.extra2		= &one_hundred,
1538 	},
1539 #endif
1540 #ifdef CONFIG_SMP
1541 	{
1542 		.procname	= "stat_interval",
1543 		.data		= &sysctl_stat_interval,
1544 		.maxlen		= sizeof(sysctl_stat_interval),
1545 		.mode		= 0644,
1546 		.proc_handler	= proc_dointvec_jiffies,
1547 	},
1548 	{
1549 		.procname	= "stat_refresh",
1550 		.data		= NULL,
1551 		.maxlen		= 0,
1552 		.mode		= 0600,
1553 		.proc_handler	= vmstat_refresh,
1554 	},
1555 #endif
1556 #ifdef CONFIG_MMU
1557 	{
1558 		.procname	= "mmap_min_addr",
1559 		.data		= &dac_mmap_min_addr,
1560 		.maxlen		= sizeof(unsigned long),
1561 		.mode		= 0644,
1562 		.proc_handler	= mmap_min_addr_handler,
1563 	},
1564 #endif
1565 #ifdef CONFIG_NUMA
1566 	{
1567 		.procname	= "numa_zonelist_order",
1568 		.data		= &numa_zonelist_order,
1569 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1570 		.mode		= 0644,
1571 		.proc_handler	= numa_zonelist_order_handler,
1572 	},
1573 #endif
1574 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1575    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1576 	{
1577 		.procname	= "vdso_enabled",
1578 #ifdef CONFIG_X86_32
1579 		.data		= &vdso32_enabled,
1580 		.maxlen		= sizeof(vdso32_enabled),
1581 #else
1582 		.data		= &vdso_enabled,
1583 		.maxlen		= sizeof(vdso_enabled),
1584 #endif
1585 		.mode		= 0644,
1586 		.proc_handler	= proc_dointvec,
1587 		.extra1		= &zero,
1588 	},
1589 #endif
1590 #ifdef CONFIG_HIGHMEM
1591 	{
1592 		.procname	= "highmem_is_dirtyable",
1593 		.data		= &vm_highmem_is_dirtyable,
1594 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1595 		.mode		= 0644,
1596 		.proc_handler	= proc_dointvec_minmax,
1597 		.extra1		= &zero,
1598 		.extra2		= &one,
1599 	},
1600 #endif
1601 #ifdef CONFIG_MEMORY_FAILURE
1602 	{
1603 		.procname	= "memory_failure_early_kill",
1604 		.data		= &sysctl_memory_failure_early_kill,
1605 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1606 		.mode		= 0644,
1607 		.proc_handler	= proc_dointvec_minmax,
1608 		.extra1		= &zero,
1609 		.extra2		= &one,
1610 	},
1611 	{
1612 		.procname	= "memory_failure_recovery",
1613 		.data		= &sysctl_memory_failure_recovery,
1614 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1615 		.mode		= 0644,
1616 		.proc_handler	= proc_dointvec_minmax,
1617 		.extra1		= &zero,
1618 		.extra2		= &one,
1619 	},
1620 #endif
1621 	{
1622 		.procname	= "user_reserve_kbytes",
1623 		.data		= &sysctl_user_reserve_kbytes,
1624 		.maxlen		= sizeof(sysctl_user_reserve_kbytes),
1625 		.mode		= 0644,
1626 		.proc_handler	= proc_doulongvec_minmax,
1627 	},
1628 	{
1629 		.procname	= "admin_reserve_kbytes",
1630 		.data		= &sysctl_admin_reserve_kbytes,
1631 		.maxlen		= sizeof(sysctl_admin_reserve_kbytes),
1632 		.mode		= 0644,
1633 		.proc_handler	= proc_doulongvec_minmax,
1634 	},
1635 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1636 	{
1637 		.procname	= "mmap_rnd_bits",
1638 		.data		= &mmap_rnd_bits,
1639 		.maxlen		= sizeof(mmap_rnd_bits),
1640 		.mode		= 0600,
1641 		.proc_handler	= proc_dointvec_minmax,
1642 		.extra1		= (void *)&mmap_rnd_bits_min,
1643 		.extra2		= (void *)&mmap_rnd_bits_max,
1644 	},
1645 #endif
1646 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1647 	{
1648 		.procname	= "mmap_rnd_compat_bits",
1649 		.data		= &mmap_rnd_compat_bits,
1650 		.maxlen		= sizeof(mmap_rnd_compat_bits),
1651 		.mode		= 0600,
1652 		.proc_handler	= proc_dointvec_minmax,
1653 		.extra1		= (void *)&mmap_rnd_compat_bits_min,
1654 		.extra2		= (void *)&mmap_rnd_compat_bits_max,
1655 	},
1656 #endif
1657 	{ }
1658 };
1659 
1660 static struct ctl_table fs_table[] = {
1661 	{
1662 		.procname	= "inode-nr",
1663 		.data		= &inodes_stat,
1664 		.maxlen		= 2*sizeof(long),
1665 		.mode		= 0444,
1666 		.proc_handler	= proc_nr_inodes,
1667 	},
1668 	{
1669 		.procname	= "inode-state",
1670 		.data		= &inodes_stat,
1671 		.maxlen		= 7*sizeof(long),
1672 		.mode		= 0444,
1673 		.proc_handler	= proc_nr_inodes,
1674 	},
1675 	{
1676 		.procname	= "file-nr",
1677 		.data		= &files_stat,
1678 		.maxlen		= sizeof(files_stat),
1679 		.mode		= 0444,
1680 		.proc_handler	= proc_nr_files,
1681 	},
1682 	{
1683 		.procname	= "file-max",
1684 		.data		= &files_stat.max_files,
1685 		.maxlen		= sizeof(files_stat.max_files),
1686 		.mode		= 0644,
1687 		.proc_handler	= proc_doulongvec_minmax,
1688 	},
1689 	{
1690 		.procname	= "nr_open",
1691 		.data		= &sysctl_nr_open,
1692 		.maxlen		= sizeof(unsigned int),
1693 		.mode		= 0644,
1694 		.proc_handler	= proc_dointvec_minmax,
1695 		.extra1		= &sysctl_nr_open_min,
1696 		.extra2		= &sysctl_nr_open_max,
1697 	},
1698 	{
1699 		.procname	= "dentry-state",
1700 		.data		= &dentry_stat,
1701 		.maxlen		= 6*sizeof(long),
1702 		.mode		= 0444,
1703 		.proc_handler	= proc_nr_dentry,
1704 	},
1705 	{
1706 		.procname	= "overflowuid",
1707 		.data		= &fs_overflowuid,
1708 		.maxlen		= sizeof(int),
1709 		.mode		= 0644,
1710 		.proc_handler	= proc_dointvec_minmax,
1711 		.extra1		= &minolduid,
1712 		.extra2		= &maxolduid,
1713 	},
1714 	{
1715 		.procname	= "overflowgid",
1716 		.data		= &fs_overflowgid,
1717 		.maxlen		= sizeof(int),
1718 		.mode		= 0644,
1719 		.proc_handler	= proc_dointvec_minmax,
1720 		.extra1		= &minolduid,
1721 		.extra2		= &maxolduid,
1722 	},
1723 #ifdef CONFIG_FILE_LOCKING
1724 	{
1725 		.procname	= "leases-enable",
1726 		.data		= &leases_enable,
1727 		.maxlen		= sizeof(int),
1728 		.mode		= 0644,
1729 		.proc_handler	= proc_dointvec,
1730 	},
1731 #endif
1732 #ifdef CONFIG_DNOTIFY
1733 	{
1734 		.procname	= "dir-notify-enable",
1735 		.data		= &dir_notify_enable,
1736 		.maxlen		= sizeof(int),
1737 		.mode		= 0644,
1738 		.proc_handler	= proc_dointvec,
1739 	},
1740 #endif
1741 #ifdef CONFIG_MMU
1742 #ifdef CONFIG_FILE_LOCKING
1743 	{
1744 		.procname	= "lease-break-time",
1745 		.data		= &lease_break_time,
1746 		.maxlen		= sizeof(int),
1747 		.mode		= 0644,
1748 		.proc_handler	= proc_dointvec,
1749 	},
1750 #endif
1751 #ifdef CONFIG_AIO
1752 	{
1753 		.procname	= "aio-nr",
1754 		.data		= &aio_nr,
1755 		.maxlen		= sizeof(aio_nr),
1756 		.mode		= 0444,
1757 		.proc_handler	= proc_doulongvec_minmax,
1758 	},
1759 	{
1760 		.procname	= "aio-max-nr",
1761 		.data		= &aio_max_nr,
1762 		.maxlen		= sizeof(aio_max_nr),
1763 		.mode		= 0644,
1764 		.proc_handler	= proc_doulongvec_minmax,
1765 	},
1766 #endif /* CONFIG_AIO */
1767 #ifdef CONFIG_INOTIFY_USER
1768 	{
1769 		.procname	= "inotify",
1770 		.mode		= 0555,
1771 		.child		= inotify_table,
1772 	},
1773 #endif
1774 #ifdef CONFIG_EPOLL
1775 	{
1776 		.procname	= "epoll",
1777 		.mode		= 0555,
1778 		.child		= epoll_table,
1779 	},
1780 #endif
1781 #endif
1782 	{
1783 		.procname	= "protected_symlinks",
1784 		.data		= &sysctl_protected_symlinks,
1785 		.maxlen		= sizeof(int),
1786 		.mode		= 0600,
1787 		.proc_handler	= proc_dointvec_minmax,
1788 		.extra1		= &zero,
1789 		.extra2		= &one,
1790 	},
1791 	{
1792 		.procname	= "protected_hardlinks",
1793 		.data		= &sysctl_protected_hardlinks,
1794 		.maxlen		= sizeof(int),
1795 		.mode		= 0600,
1796 		.proc_handler	= proc_dointvec_minmax,
1797 		.extra1		= &zero,
1798 		.extra2		= &one,
1799 	},
1800 	{
1801 		.procname	= "suid_dumpable",
1802 		.data		= &suid_dumpable,
1803 		.maxlen		= sizeof(int),
1804 		.mode		= 0644,
1805 		.proc_handler	= proc_dointvec_minmax_coredump,
1806 		.extra1		= &zero,
1807 		.extra2		= &two,
1808 	},
1809 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1810 	{
1811 		.procname	= "binfmt_misc",
1812 		.mode		= 0555,
1813 		.child		= sysctl_mount_point,
1814 	},
1815 #endif
1816 	{
1817 		.procname	= "pipe-max-size",
1818 		.data		= &pipe_max_size,
1819 		.maxlen		= sizeof(int),
1820 		.mode		= 0644,
1821 		.proc_handler	= &pipe_proc_fn,
1822 		.extra1		= &pipe_min_size,
1823 	},
1824 	{
1825 		.procname	= "pipe-user-pages-hard",
1826 		.data		= &pipe_user_pages_hard,
1827 		.maxlen		= sizeof(pipe_user_pages_hard),
1828 		.mode		= 0644,
1829 		.proc_handler	= proc_doulongvec_minmax,
1830 	},
1831 	{
1832 		.procname	= "pipe-user-pages-soft",
1833 		.data		= &pipe_user_pages_soft,
1834 		.maxlen		= sizeof(pipe_user_pages_soft),
1835 		.mode		= 0644,
1836 		.proc_handler	= proc_doulongvec_minmax,
1837 	},
1838 	{
1839 		.procname	= "mount-max",
1840 		.data		= &sysctl_mount_max,
1841 		.maxlen		= sizeof(unsigned int),
1842 		.mode		= 0644,
1843 		.proc_handler	= proc_dointvec_minmax,
1844 		.extra1		= &one,
1845 	},
1846 	{ }
1847 };
1848 
1849 static struct ctl_table debug_table[] = {
1850 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1851 	{
1852 		.procname	= "exception-trace",
1853 		.data		= &show_unhandled_signals,
1854 		.maxlen		= sizeof(int),
1855 		.mode		= 0644,
1856 		.proc_handler	= proc_dointvec
1857 	},
1858 #endif
1859 #if defined(CONFIG_OPTPROBES)
1860 	{
1861 		.procname	= "kprobes-optimization",
1862 		.data		= &sysctl_kprobes_optimization,
1863 		.maxlen		= sizeof(int),
1864 		.mode		= 0644,
1865 		.proc_handler	= proc_kprobes_optimization_handler,
1866 		.extra1		= &zero,
1867 		.extra2		= &one,
1868 	},
1869 #endif
1870 	{ }
1871 };
1872 
1873 static struct ctl_table dev_table[] = {
1874 	{ }
1875 };
1876 
1877 int __init sysctl_init(void)
1878 {
1879 	struct ctl_table_header *hdr;
1880 
1881 	hdr = register_sysctl_table(sysctl_base_table);
1882 	kmemleak_not_leak(hdr);
1883 	return 0;
1884 }
1885 
1886 #endif /* CONFIG_SYSCTL */
1887 
1888 /*
1889  * /proc/sys support
1890  */
1891 
1892 #ifdef CONFIG_PROC_SYSCTL
1893 
1894 static int _proc_do_string(char *data, int maxlen, int write,
1895 			   char __user *buffer,
1896 			   size_t *lenp, loff_t *ppos)
1897 {
1898 	size_t len;
1899 	char __user *p;
1900 	char c;
1901 
1902 	if (!data || !maxlen || !*lenp) {
1903 		*lenp = 0;
1904 		return 0;
1905 	}
1906 
1907 	if (write) {
1908 		if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1909 			/* Only continue writes not past the end of buffer. */
1910 			len = strlen(data);
1911 			if (len > maxlen - 1)
1912 				len = maxlen - 1;
1913 
1914 			if (*ppos > len)
1915 				return 0;
1916 			len = *ppos;
1917 		} else {
1918 			/* Start writing from beginning of buffer. */
1919 			len = 0;
1920 		}
1921 
1922 		*ppos += *lenp;
1923 		p = buffer;
1924 		while ((p - buffer) < *lenp && len < maxlen - 1) {
1925 			if (get_user(c, p++))
1926 				return -EFAULT;
1927 			if (c == 0 || c == '\n')
1928 				break;
1929 			data[len++] = c;
1930 		}
1931 		data[len] = 0;
1932 	} else {
1933 		len = strlen(data);
1934 		if (len > maxlen)
1935 			len = maxlen;
1936 
1937 		if (*ppos > len) {
1938 			*lenp = 0;
1939 			return 0;
1940 		}
1941 
1942 		data += *ppos;
1943 		len  -= *ppos;
1944 
1945 		if (len > *lenp)
1946 			len = *lenp;
1947 		if (len)
1948 			if (copy_to_user(buffer, data, len))
1949 				return -EFAULT;
1950 		if (len < *lenp) {
1951 			if (put_user('\n', buffer + len))
1952 				return -EFAULT;
1953 			len++;
1954 		}
1955 		*lenp = len;
1956 		*ppos += len;
1957 	}
1958 	return 0;
1959 }
1960 
1961 static void warn_sysctl_write(struct ctl_table *table)
1962 {
1963 	pr_warn_once("%s wrote to %s when file position was not 0!\n"
1964 		"This will not be supported in the future. To silence this\n"
1965 		"warning, set kernel.sysctl_writes_strict = -1\n",
1966 		current->comm, table->procname);
1967 }
1968 
1969 /**
1970  * proc_first_pos_non_zero_ignore - check if firs position is allowed
1971  * @ppos: file position
1972  * @table: the sysctl table
1973  *
1974  * Returns true if the first position is non-zero and the sysctl_writes_strict
1975  * mode indicates this is not allowed for numeric input types. String proc
1976  * hadlers can ignore the return value.
1977  */
1978 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
1979 					   struct ctl_table *table)
1980 {
1981 	if (!*ppos)
1982 		return false;
1983 
1984 	switch (sysctl_writes_strict) {
1985 	case SYSCTL_WRITES_STRICT:
1986 		return true;
1987 	case SYSCTL_WRITES_WARN:
1988 		warn_sysctl_write(table);
1989 		return false;
1990 	default:
1991 		return false;
1992 	}
1993 }
1994 
1995 /**
1996  * proc_dostring - read a string sysctl
1997  * @table: the sysctl table
1998  * @write: %TRUE if this is a write to the sysctl file
1999  * @buffer: the user buffer
2000  * @lenp: the size of the user buffer
2001  * @ppos: file position
2002  *
2003  * Reads/writes a string from/to the user buffer. If the kernel
2004  * buffer provided is not large enough to hold the string, the
2005  * string is truncated. The copied string is %NULL-terminated.
2006  * If the string is being read by the user process, it is copied
2007  * and a newline '\n' is added. It is truncated if the buffer is
2008  * not large enough.
2009  *
2010  * Returns 0 on success.
2011  */
2012 int proc_dostring(struct ctl_table *table, int write,
2013 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2014 {
2015 	if (write)
2016 		proc_first_pos_non_zero_ignore(ppos, table);
2017 
2018 	return _proc_do_string((char *)(table->data), table->maxlen, write,
2019 			       (char __user *)buffer, lenp, ppos);
2020 }
2021 
2022 static size_t proc_skip_spaces(char **buf)
2023 {
2024 	size_t ret;
2025 	char *tmp = skip_spaces(*buf);
2026 	ret = tmp - *buf;
2027 	*buf = tmp;
2028 	return ret;
2029 }
2030 
2031 static void proc_skip_char(char **buf, size_t *size, const char v)
2032 {
2033 	while (*size) {
2034 		if (**buf != v)
2035 			break;
2036 		(*size)--;
2037 		(*buf)++;
2038 	}
2039 }
2040 
2041 #define TMPBUFLEN 22
2042 /**
2043  * proc_get_long - reads an ASCII formatted integer from a user buffer
2044  *
2045  * @buf: a kernel buffer
2046  * @size: size of the kernel buffer
2047  * @val: this is where the number will be stored
2048  * @neg: set to %TRUE if number is negative
2049  * @perm_tr: a vector which contains the allowed trailers
2050  * @perm_tr_len: size of the perm_tr vector
2051  * @tr: pointer to store the trailer character
2052  *
2053  * In case of success %0 is returned and @buf and @size are updated with
2054  * the amount of bytes read. If @tr is non-NULL and a trailing
2055  * character exists (size is non-zero after returning from this
2056  * function), @tr is updated with the trailing character.
2057  */
2058 static int proc_get_long(char **buf, size_t *size,
2059 			  unsigned long *val, bool *neg,
2060 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
2061 {
2062 	int len;
2063 	char *p, tmp[TMPBUFLEN];
2064 
2065 	if (!*size)
2066 		return -EINVAL;
2067 
2068 	len = *size;
2069 	if (len > TMPBUFLEN - 1)
2070 		len = TMPBUFLEN - 1;
2071 
2072 	memcpy(tmp, *buf, len);
2073 
2074 	tmp[len] = 0;
2075 	p = tmp;
2076 	if (*p == '-' && *size > 1) {
2077 		*neg = true;
2078 		p++;
2079 	} else
2080 		*neg = false;
2081 	if (!isdigit(*p))
2082 		return -EINVAL;
2083 
2084 	*val = simple_strtoul(p, &p, 0);
2085 
2086 	len = p - tmp;
2087 
2088 	/* We don't know if the next char is whitespace thus we may accept
2089 	 * invalid integers (e.g. 1234...a) or two integers instead of one
2090 	 * (e.g. 123...1). So lets not allow such large numbers. */
2091 	if (len == TMPBUFLEN - 1)
2092 		return -EINVAL;
2093 
2094 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2095 		return -EINVAL;
2096 
2097 	if (tr && (len < *size))
2098 		*tr = *p;
2099 
2100 	*buf += len;
2101 	*size -= len;
2102 
2103 	return 0;
2104 }
2105 
2106 /**
2107  * proc_put_long - converts an integer to a decimal ASCII formatted string
2108  *
2109  * @buf: the user buffer
2110  * @size: the size of the user buffer
2111  * @val: the integer to be converted
2112  * @neg: sign of the number, %TRUE for negative
2113  *
2114  * In case of success %0 is returned and @buf and @size are updated with
2115  * the amount of bytes written.
2116  */
2117 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2118 			  bool neg)
2119 {
2120 	int len;
2121 	char tmp[TMPBUFLEN], *p = tmp;
2122 
2123 	sprintf(p, "%s%lu", neg ? "-" : "", val);
2124 	len = strlen(tmp);
2125 	if (len > *size)
2126 		len = *size;
2127 	if (copy_to_user(*buf, tmp, len))
2128 		return -EFAULT;
2129 	*size -= len;
2130 	*buf += len;
2131 	return 0;
2132 }
2133 #undef TMPBUFLEN
2134 
2135 static int proc_put_char(void __user **buf, size_t *size, char c)
2136 {
2137 	if (*size) {
2138 		char __user **buffer = (char __user **)buf;
2139 		if (put_user(c, *buffer))
2140 			return -EFAULT;
2141 		(*size)--, (*buffer)++;
2142 		*buf = *buffer;
2143 	}
2144 	return 0;
2145 }
2146 
2147 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2148 				 int *valp,
2149 				 int write, void *data)
2150 {
2151 	if (write) {
2152 		if (*negp) {
2153 			if (*lvalp > (unsigned long) INT_MAX + 1)
2154 				return -EINVAL;
2155 			*valp = -*lvalp;
2156 		} else {
2157 			if (*lvalp > (unsigned long) INT_MAX)
2158 				return -EINVAL;
2159 			*valp = *lvalp;
2160 		}
2161 	} else {
2162 		int val = *valp;
2163 		if (val < 0) {
2164 			*negp = true;
2165 			*lvalp = -(unsigned long)val;
2166 		} else {
2167 			*negp = false;
2168 			*lvalp = (unsigned long)val;
2169 		}
2170 	}
2171 	return 0;
2172 }
2173 
2174 static int do_proc_douintvec_conv(unsigned long *lvalp,
2175 				  unsigned int *valp,
2176 				  int write, void *data)
2177 {
2178 	if (write) {
2179 		if (*lvalp > UINT_MAX)
2180 			return -EINVAL;
2181 		*valp = *lvalp;
2182 	} else {
2183 		unsigned int val = *valp;
2184 		*lvalp = (unsigned long)val;
2185 	}
2186 	return 0;
2187 }
2188 
2189 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2190 
2191 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2192 		  int write, void __user *buffer,
2193 		  size_t *lenp, loff_t *ppos,
2194 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2195 			      int write, void *data),
2196 		  void *data)
2197 {
2198 	int *i, vleft, first = 1, err = 0;
2199 	size_t left;
2200 	char *kbuf = NULL, *p;
2201 
2202 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2203 		*lenp = 0;
2204 		return 0;
2205 	}
2206 
2207 	i = (int *) tbl_data;
2208 	vleft = table->maxlen / sizeof(*i);
2209 	left = *lenp;
2210 
2211 	if (!conv)
2212 		conv = do_proc_dointvec_conv;
2213 
2214 	if (write) {
2215 		if (proc_first_pos_non_zero_ignore(ppos, table))
2216 			goto out;
2217 
2218 		if (left > PAGE_SIZE - 1)
2219 			left = PAGE_SIZE - 1;
2220 		p = kbuf = memdup_user_nul(buffer, left);
2221 		if (IS_ERR(kbuf))
2222 			return PTR_ERR(kbuf);
2223 	}
2224 
2225 	for (; left && vleft--; i++, first=0) {
2226 		unsigned long lval;
2227 		bool neg;
2228 
2229 		if (write) {
2230 			left -= proc_skip_spaces(&p);
2231 
2232 			if (!left)
2233 				break;
2234 			err = proc_get_long(&p, &left, &lval, &neg,
2235 					     proc_wspace_sep,
2236 					     sizeof(proc_wspace_sep), NULL);
2237 			if (err)
2238 				break;
2239 			if (conv(&neg, &lval, i, 1, data)) {
2240 				err = -EINVAL;
2241 				break;
2242 			}
2243 		} else {
2244 			if (conv(&neg, &lval, i, 0, data)) {
2245 				err = -EINVAL;
2246 				break;
2247 			}
2248 			if (!first)
2249 				err = proc_put_char(&buffer, &left, '\t');
2250 			if (err)
2251 				break;
2252 			err = proc_put_long(&buffer, &left, lval, neg);
2253 			if (err)
2254 				break;
2255 		}
2256 	}
2257 
2258 	if (!write && !first && left && !err)
2259 		err = proc_put_char(&buffer, &left, '\n');
2260 	if (write && !err && left)
2261 		left -= proc_skip_spaces(&p);
2262 	if (write) {
2263 		kfree(kbuf);
2264 		if (first)
2265 			return err ? : -EINVAL;
2266 	}
2267 	*lenp -= left;
2268 out:
2269 	*ppos += *lenp;
2270 	return err;
2271 }
2272 
2273 static int do_proc_dointvec(struct ctl_table *table, int write,
2274 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2275 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2276 			      int write, void *data),
2277 		  void *data)
2278 {
2279 	return __do_proc_dointvec(table->data, table, write,
2280 			buffer, lenp, ppos, conv, data);
2281 }
2282 
2283 static int do_proc_douintvec_w(unsigned int *tbl_data,
2284 			       struct ctl_table *table,
2285 			       void __user *buffer,
2286 			       size_t *lenp, loff_t *ppos,
2287 			       int (*conv)(unsigned long *lvalp,
2288 					   unsigned int *valp,
2289 					   int write, void *data),
2290 			       void *data)
2291 {
2292 	unsigned long lval;
2293 	int err = 0;
2294 	size_t left;
2295 	bool neg;
2296 	char *kbuf = NULL, *p;
2297 
2298 	left = *lenp;
2299 
2300 	if (proc_first_pos_non_zero_ignore(ppos, table))
2301 		goto bail_early;
2302 
2303 	if (left > PAGE_SIZE - 1)
2304 		left = PAGE_SIZE - 1;
2305 
2306 	p = kbuf = memdup_user_nul(buffer, left);
2307 	if (IS_ERR(kbuf))
2308 		return -EINVAL;
2309 
2310 	left -= proc_skip_spaces(&p);
2311 	if (!left) {
2312 		err = -EINVAL;
2313 		goto out_free;
2314 	}
2315 
2316 	err = proc_get_long(&p, &left, &lval, &neg,
2317 			     proc_wspace_sep,
2318 			     sizeof(proc_wspace_sep), NULL);
2319 	if (err || neg) {
2320 		err = -EINVAL;
2321 		goto out_free;
2322 	}
2323 
2324 	if (conv(&lval, tbl_data, 1, data)) {
2325 		err = -EINVAL;
2326 		goto out_free;
2327 	}
2328 
2329 	if (!err && left)
2330 		left -= proc_skip_spaces(&p);
2331 
2332 out_free:
2333 	kfree(kbuf);
2334 	if (err)
2335 		return -EINVAL;
2336 
2337 	return 0;
2338 
2339 	/* This is in keeping with old __do_proc_dointvec() */
2340 bail_early:
2341 	*ppos += *lenp;
2342 	return err;
2343 }
2344 
2345 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2346 			       size_t *lenp, loff_t *ppos,
2347 			       int (*conv)(unsigned long *lvalp,
2348 					   unsigned int *valp,
2349 					   int write, void *data),
2350 			       void *data)
2351 {
2352 	unsigned long lval;
2353 	int err = 0;
2354 	size_t left;
2355 
2356 	left = *lenp;
2357 
2358 	if (conv(&lval, tbl_data, 0, data)) {
2359 		err = -EINVAL;
2360 		goto out;
2361 	}
2362 
2363 	err = proc_put_long(&buffer, &left, lval, false);
2364 	if (err || !left)
2365 		goto out;
2366 
2367 	err = proc_put_char(&buffer, &left, '\n');
2368 
2369 out:
2370 	*lenp -= left;
2371 	*ppos += *lenp;
2372 
2373 	return err;
2374 }
2375 
2376 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2377 			       int write, void __user *buffer,
2378 			       size_t *lenp, loff_t *ppos,
2379 			       int (*conv)(unsigned long *lvalp,
2380 					   unsigned int *valp,
2381 					   int write, void *data),
2382 			       void *data)
2383 {
2384 	unsigned int *i, vleft;
2385 
2386 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2387 		*lenp = 0;
2388 		return 0;
2389 	}
2390 
2391 	i = (unsigned int *) tbl_data;
2392 	vleft = table->maxlen / sizeof(*i);
2393 
2394 	/*
2395 	 * Arrays are not supported, keep this simple. *Do not* add
2396 	 * support for them.
2397 	 */
2398 	if (vleft != 1) {
2399 		*lenp = 0;
2400 		return -EINVAL;
2401 	}
2402 
2403 	if (!conv)
2404 		conv = do_proc_douintvec_conv;
2405 
2406 	if (write)
2407 		return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2408 					   conv, data);
2409 	return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2410 }
2411 
2412 static int do_proc_douintvec(struct ctl_table *table, int write,
2413 			     void __user *buffer, size_t *lenp, loff_t *ppos,
2414 			     int (*conv)(unsigned long *lvalp,
2415 					 unsigned int *valp,
2416 					 int write, void *data),
2417 			     void *data)
2418 {
2419 	return __do_proc_douintvec(table->data, table, write,
2420 				   buffer, lenp, ppos, conv, data);
2421 }
2422 
2423 /**
2424  * proc_dointvec - read a vector of integers
2425  * @table: the sysctl table
2426  * @write: %TRUE if this is a write to the sysctl file
2427  * @buffer: the user buffer
2428  * @lenp: the size of the user buffer
2429  * @ppos: file position
2430  *
2431  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2432  * values from/to the user buffer, treated as an ASCII string.
2433  *
2434  * Returns 0 on success.
2435  */
2436 int proc_dointvec(struct ctl_table *table, int write,
2437 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2438 {
2439 	return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2440 }
2441 
2442 /**
2443  * proc_douintvec - read a vector of unsigned integers
2444  * @table: the sysctl table
2445  * @write: %TRUE if this is a write to the sysctl file
2446  * @buffer: the user buffer
2447  * @lenp: the size of the user buffer
2448  * @ppos: file position
2449  *
2450  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2451  * values from/to the user buffer, treated as an ASCII string.
2452  *
2453  * Returns 0 on success.
2454  */
2455 int proc_douintvec(struct ctl_table *table, int write,
2456 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2457 {
2458 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
2459 				 do_proc_douintvec_conv, NULL);
2460 }
2461 
2462 /*
2463  * Taint values can only be increased
2464  * This means we can safely use a temporary.
2465  */
2466 static int proc_taint(struct ctl_table *table, int write,
2467 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2468 {
2469 	struct ctl_table t;
2470 	unsigned long tmptaint = get_taint();
2471 	int err;
2472 
2473 	if (write && !capable(CAP_SYS_ADMIN))
2474 		return -EPERM;
2475 
2476 	t = *table;
2477 	t.data = &tmptaint;
2478 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2479 	if (err < 0)
2480 		return err;
2481 
2482 	if (write) {
2483 		/*
2484 		 * Poor man's atomic or. Not worth adding a primitive
2485 		 * to everyone's atomic.h for this
2486 		 */
2487 		int i;
2488 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2489 			if ((tmptaint >> i) & 1)
2490 				add_taint(i, LOCKDEP_STILL_OK);
2491 		}
2492 	}
2493 
2494 	return err;
2495 }
2496 
2497 #ifdef CONFIG_PRINTK
2498 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2499 				void __user *buffer, size_t *lenp, loff_t *ppos)
2500 {
2501 	if (write && !capable(CAP_SYS_ADMIN))
2502 		return -EPERM;
2503 
2504 	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2505 }
2506 #endif
2507 
2508 struct do_proc_dointvec_minmax_conv_param {
2509 	int *min;
2510 	int *max;
2511 };
2512 
2513 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2514 					int *valp,
2515 					int write, void *data)
2516 {
2517 	struct do_proc_dointvec_minmax_conv_param *param = data;
2518 	if (write) {
2519 		int val = *negp ? -*lvalp : *lvalp;
2520 		if ((param->min && *param->min > val) ||
2521 		    (param->max && *param->max < val))
2522 			return -EINVAL;
2523 		*valp = val;
2524 	} else {
2525 		int val = *valp;
2526 		if (val < 0) {
2527 			*negp = true;
2528 			*lvalp = -(unsigned long)val;
2529 		} else {
2530 			*negp = false;
2531 			*lvalp = (unsigned long)val;
2532 		}
2533 	}
2534 	return 0;
2535 }
2536 
2537 /**
2538  * proc_dointvec_minmax - read a vector of integers with min/max values
2539  * @table: the sysctl table
2540  * @write: %TRUE if this is a write to the sysctl file
2541  * @buffer: the user buffer
2542  * @lenp: the size of the user buffer
2543  * @ppos: file position
2544  *
2545  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2546  * values from/to the user buffer, treated as an ASCII string.
2547  *
2548  * This routine will ensure the values are within the range specified by
2549  * table->extra1 (min) and table->extra2 (max).
2550  *
2551  * Returns 0 on success.
2552  */
2553 int proc_dointvec_minmax(struct ctl_table *table, int write,
2554 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2555 {
2556 	struct do_proc_dointvec_minmax_conv_param param = {
2557 		.min = (int *) table->extra1,
2558 		.max = (int *) table->extra2,
2559 	};
2560 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2561 				do_proc_dointvec_minmax_conv, &param);
2562 }
2563 
2564 struct do_proc_douintvec_minmax_conv_param {
2565 	unsigned int *min;
2566 	unsigned int *max;
2567 };
2568 
2569 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2570 					 unsigned int *valp,
2571 					 int write, void *data)
2572 {
2573 	struct do_proc_douintvec_minmax_conv_param *param = data;
2574 
2575 	if (write) {
2576 		unsigned int val = *lvalp;
2577 
2578 		if ((param->min && *param->min > val) ||
2579 		    (param->max && *param->max < val))
2580 			return -ERANGE;
2581 
2582 		if (*lvalp > UINT_MAX)
2583 			return -EINVAL;
2584 		*valp = val;
2585 	} else {
2586 		unsigned int val = *valp;
2587 		*lvalp = (unsigned long) val;
2588 	}
2589 
2590 	return 0;
2591 }
2592 
2593 /**
2594  * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2595  * @table: the sysctl table
2596  * @write: %TRUE if this is a write to the sysctl file
2597  * @buffer: the user buffer
2598  * @lenp: the size of the user buffer
2599  * @ppos: file position
2600  *
2601  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2602  * values from/to the user buffer, treated as an ASCII string. Negative
2603  * strings are not allowed.
2604  *
2605  * This routine will ensure the values are within the range specified by
2606  * table->extra1 (min) and table->extra2 (max). There is a final sanity
2607  * check for UINT_MAX to avoid having to support wrap around uses from
2608  * userspace.
2609  *
2610  * Returns 0 on success.
2611  */
2612 int proc_douintvec_minmax(struct ctl_table *table, int write,
2613 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2614 {
2615 	struct do_proc_douintvec_minmax_conv_param param = {
2616 		.min = (unsigned int *) table->extra1,
2617 		.max = (unsigned int *) table->extra2,
2618 	};
2619 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
2620 				 do_proc_douintvec_minmax_conv, &param);
2621 }
2622 
2623 static void validate_coredump_safety(void)
2624 {
2625 #ifdef CONFIG_COREDUMP
2626 	if (suid_dumpable == SUID_DUMP_ROOT &&
2627 	    core_pattern[0] != '/' && core_pattern[0] != '|') {
2628 		printk(KERN_WARNING
2629 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2630 "Pipe handler or fully qualified core dump path required.\n"
2631 "Set kernel.core_pattern before fs.suid_dumpable.\n"
2632 		);
2633 	}
2634 #endif
2635 }
2636 
2637 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2638 		void __user *buffer, size_t *lenp, loff_t *ppos)
2639 {
2640 	int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2641 	if (!error)
2642 		validate_coredump_safety();
2643 	return error;
2644 }
2645 
2646 #ifdef CONFIG_COREDUMP
2647 static int proc_dostring_coredump(struct ctl_table *table, int write,
2648 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2649 {
2650 	int error = proc_dostring(table, write, buffer, lenp, ppos);
2651 	if (!error)
2652 		validate_coredump_safety();
2653 	return error;
2654 }
2655 #endif
2656 
2657 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2658 				     void __user *buffer,
2659 				     size_t *lenp, loff_t *ppos,
2660 				     unsigned long convmul,
2661 				     unsigned long convdiv)
2662 {
2663 	unsigned long *i, *min, *max;
2664 	int vleft, first = 1, err = 0;
2665 	size_t left;
2666 	char *kbuf = NULL, *p;
2667 
2668 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2669 		*lenp = 0;
2670 		return 0;
2671 	}
2672 
2673 	i = (unsigned long *) data;
2674 	min = (unsigned long *) table->extra1;
2675 	max = (unsigned long *) table->extra2;
2676 	vleft = table->maxlen / sizeof(unsigned long);
2677 	left = *lenp;
2678 
2679 	if (write) {
2680 		if (proc_first_pos_non_zero_ignore(ppos, table))
2681 			goto out;
2682 
2683 		if (left > PAGE_SIZE - 1)
2684 			left = PAGE_SIZE - 1;
2685 		p = kbuf = memdup_user_nul(buffer, left);
2686 		if (IS_ERR(kbuf))
2687 			return PTR_ERR(kbuf);
2688 	}
2689 
2690 	for (; left && vleft--; i++, first = 0) {
2691 		unsigned long val;
2692 
2693 		if (write) {
2694 			bool neg;
2695 
2696 			left -= proc_skip_spaces(&p);
2697 
2698 			err = proc_get_long(&p, &left, &val, &neg,
2699 					     proc_wspace_sep,
2700 					     sizeof(proc_wspace_sep), NULL);
2701 			if (err)
2702 				break;
2703 			if (neg)
2704 				continue;
2705 			val = convmul * val / convdiv;
2706 			if ((min && val < *min) || (max && val > *max))
2707 				continue;
2708 			*i = val;
2709 		} else {
2710 			val = convdiv * (*i) / convmul;
2711 			if (!first) {
2712 				err = proc_put_char(&buffer, &left, '\t');
2713 				if (err)
2714 					break;
2715 			}
2716 			err = proc_put_long(&buffer, &left, val, false);
2717 			if (err)
2718 				break;
2719 		}
2720 	}
2721 
2722 	if (!write && !first && left && !err)
2723 		err = proc_put_char(&buffer, &left, '\n');
2724 	if (write && !err)
2725 		left -= proc_skip_spaces(&p);
2726 	if (write) {
2727 		kfree(kbuf);
2728 		if (first)
2729 			return err ? : -EINVAL;
2730 	}
2731 	*lenp -= left;
2732 out:
2733 	*ppos += *lenp;
2734 	return err;
2735 }
2736 
2737 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2738 				     void __user *buffer,
2739 				     size_t *lenp, loff_t *ppos,
2740 				     unsigned long convmul,
2741 				     unsigned long convdiv)
2742 {
2743 	return __do_proc_doulongvec_minmax(table->data, table, write,
2744 			buffer, lenp, ppos, convmul, convdiv);
2745 }
2746 
2747 /**
2748  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2749  * @table: the sysctl table
2750  * @write: %TRUE if this is a write to the sysctl file
2751  * @buffer: the user buffer
2752  * @lenp: the size of the user buffer
2753  * @ppos: file position
2754  *
2755  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2756  * values from/to the user buffer, treated as an ASCII string.
2757  *
2758  * This routine will ensure the values are within the range specified by
2759  * table->extra1 (min) and table->extra2 (max).
2760  *
2761  * Returns 0 on success.
2762  */
2763 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2764 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2765 {
2766     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2767 }
2768 
2769 /**
2770  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2771  * @table: the sysctl table
2772  * @write: %TRUE if this is a write to the sysctl file
2773  * @buffer: the user buffer
2774  * @lenp: the size of the user buffer
2775  * @ppos: file position
2776  *
2777  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2778  * values from/to the user buffer, treated as an ASCII string. The values
2779  * are treated as milliseconds, and converted to jiffies when they are stored.
2780  *
2781  * This routine will ensure the values are within the range specified by
2782  * table->extra1 (min) and table->extra2 (max).
2783  *
2784  * Returns 0 on success.
2785  */
2786 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2787 				      void __user *buffer,
2788 				      size_t *lenp, loff_t *ppos)
2789 {
2790     return do_proc_doulongvec_minmax(table, write, buffer,
2791 				     lenp, ppos, HZ, 1000l);
2792 }
2793 
2794 
2795 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2796 					 int *valp,
2797 					 int write, void *data)
2798 {
2799 	if (write) {
2800 		if (*lvalp > INT_MAX / HZ)
2801 			return 1;
2802 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2803 	} else {
2804 		int val = *valp;
2805 		unsigned long lval;
2806 		if (val < 0) {
2807 			*negp = true;
2808 			lval = -(unsigned long)val;
2809 		} else {
2810 			*negp = false;
2811 			lval = (unsigned long)val;
2812 		}
2813 		*lvalp = lval / HZ;
2814 	}
2815 	return 0;
2816 }
2817 
2818 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2819 						int *valp,
2820 						int write, void *data)
2821 {
2822 	if (write) {
2823 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2824 			return 1;
2825 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2826 	} else {
2827 		int val = *valp;
2828 		unsigned long lval;
2829 		if (val < 0) {
2830 			*negp = true;
2831 			lval = -(unsigned long)val;
2832 		} else {
2833 			*negp = false;
2834 			lval = (unsigned long)val;
2835 		}
2836 		*lvalp = jiffies_to_clock_t(lval);
2837 	}
2838 	return 0;
2839 }
2840 
2841 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2842 					    int *valp,
2843 					    int write, void *data)
2844 {
2845 	if (write) {
2846 		unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2847 
2848 		if (jif > INT_MAX)
2849 			return 1;
2850 		*valp = (int)jif;
2851 	} else {
2852 		int val = *valp;
2853 		unsigned long lval;
2854 		if (val < 0) {
2855 			*negp = true;
2856 			lval = -(unsigned long)val;
2857 		} else {
2858 			*negp = false;
2859 			lval = (unsigned long)val;
2860 		}
2861 		*lvalp = jiffies_to_msecs(lval);
2862 	}
2863 	return 0;
2864 }
2865 
2866 /**
2867  * proc_dointvec_jiffies - read a vector of integers as seconds
2868  * @table: the sysctl table
2869  * @write: %TRUE if this is a write to the sysctl file
2870  * @buffer: the user buffer
2871  * @lenp: the size of the user buffer
2872  * @ppos: file position
2873  *
2874  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2875  * values from/to the user buffer, treated as an ASCII string.
2876  * The values read are assumed to be in seconds, and are converted into
2877  * jiffies.
2878  *
2879  * Returns 0 on success.
2880  */
2881 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2882 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2883 {
2884     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2885 		    	    do_proc_dointvec_jiffies_conv,NULL);
2886 }
2887 
2888 /**
2889  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2890  * @table: the sysctl table
2891  * @write: %TRUE if this is a write to the sysctl file
2892  * @buffer: the user buffer
2893  * @lenp: the size of the user buffer
2894  * @ppos: pointer to the file position
2895  *
2896  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2897  * values from/to the user buffer, treated as an ASCII string.
2898  * The values read are assumed to be in 1/USER_HZ seconds, and
2899  * are converted into jiffies.
2900  *
2901  * Returns 0 on success.
2902  */
2903 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2904 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2905 {
2906     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2907 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2908 }
2909 
2910 /**
2911  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2912  * @table: the sysctl table
2913  * @write: %TRUE if this is a write to the sysctl file
2914  * @buffer: the user buffer
2915  * @lenp: the size of the user buffer
2916  * @ppos: file position
2917  * @ppos: the current position in the file
2918  *
2919  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2920  * values from/to the user buffer, treated as an ASCII string.
2921  * The values read are assumed to be in 1/1000 seconds, and
2922  * are converted into jiffies.
2923  *
2924  * Returns 0 on success.
2925  */
2926 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2927 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2928 {
2929 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2930 				do_proc_dointvec_ms_jiffies_conv, NULL);
2931 }
2932 
2933 static int proc_do_cad_pid(struct ctl_table *table, int write,
2934 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2935 {
2936 	struct pid *new_pid;
2937 	pid_t tmp;
2938 	int r;
2939 
2940 	tmp = pid_vnr(cad_pid);
2941 
2942 	r = __do_proc_dointvec(&tmp, table, write, buffer,
2943 			       lenp, ppos, NULL, NULL);
2944 	if (r || !write)
2945 		return r;
2946 
2947 	new_pid = find_get_pid(tmp);
2948 	if (!new_pid)
2949 		return -ESRCH;
2950 
2951 	put_pid(xchg(&cad_pid, new_pid));
2952 	return 0;
2953 }
2954 
2955 /**
2956  * proc_do_large_bitmap - read/write from/to a large bitmap
2957  * @table: the sysctl table
2958  * @write: %TRUE if this is a write to the sysctl file
2959  * @buffer: the user buffer
2960  * @lenp: the size of the user buffer
2961  * @ppos: file position
2962  *
2963  * The bitmap is stored at table->data and the bitmap length (in bits)
2964  * in table->maxlen.
2965  *
2966  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2967  * large bitmaps may be represented in a compact manner. Writing into
2968  * the file will clear the bitmap then update it with the given input.
2969  *
2970  * Returns 0 on success.
2971  */
2972 int proc_do_large_bitmap(struct ctl_table *table, int write,
2973 			 void __user *buffer, size_t *lenp, loff_t *ppos)
2974 {
2975 	int err = 0;
2976 	bool first = 1;
2977 	size_t left = *lenp;
2978 	unsigned long bitmap_len = table->maxlen;
2979 	unsigned long *bitmap = *(unsigned long **) table->data;
2980 	unsigned long *tmp_bitmap = NULL;
2981 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2982 
2983 	if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2984 		*lenp = 0;
2985 		return 0;
2986 	}
2987 
2988 	if (write) {
2989 		char *kbuf, *p;
2990 
2991 		if (left > PAGE_SIZE - 1)
2992 			left = PAGE_SIZE - 1;
2993 
2994 		p = kbuf = memdup_user_nul(buffer, left);
2995 		if (IS_ERR(kbuf))
2996 			return PTR_ERR(kbuf);
2997 
2998 		tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2999 				     GFP_KERNEL);
3000 		if (!tmp_bitmap) {
3001 			kfree(kbuf);
3002 			return -ENOMEM;
3003 		}
3004 		proc_skip_char(&p, &left, '\n');
3005 		while (!err && left) {
3006 			unsigned long val_a, val_b;
3007 			bool neg;
3008 
3009 			err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3010 					     sizeof(tr_a), &c);
3011 			if (err)
3012 				break;
3013 			if (val_a >= bitmap_len || neg) {
3014 				err = -EINVAL;
3015 				break;
3016 			}
3017 
3018 			val_b = val_a;
3019 			if (left) {
3020 				p++;
3021 				left--;
3022 			}
3023 
3024 			if (c == '-') {
3025 				err = proc_get_long(&p, &left, &val_b,
3026 						     &neg, tr_b, sizeof(tr_b),
3027 						     &c);
3028 				if (err)
3029 					break;
3030 				if (val_b >= bitmap_len || neg ||
3031 				    val_a > val_b) {
3032 					err = -EINVAL;
3033 					break;
3034 				}
3035 				if (left) {
3036 					p++;
3037 					left--;
3038 				}
3039 			}
3040 
3041 			bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3042 			first = 0;
3043 			proc_skip_char(&p, &left, '\n');
3044 		}
3045 		kfree(kbuf);
3046 	} else {
3047 		unsigned long bit_a, bit_b = 0;
3048 
3049 		while (left) {
3050 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3051 			if (bit_a >= bitmap_len)
3052 				break;
3053 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
3054 						   bit_a + 1) - 1;
3055 
3056 			if (!first) {
3057 				err = proc_put_char(&buffer, &left, ',');
3058 				if (err)
3059 					break;
3060 			}
3061 			err = proc_put_long(&buffer, &left, bit_a, false);
3062 			if (err)
3063 				break;
3064 			if (bit_a != bit_b) {
3065 				err = proc_put_char(&buffer, &left, '-');
3066 				if (err)
3067 					break;
3068 				err = proc_put_long(&buffer, &left, bit_b, false);
3069 				if (err)
3070 					break;
3071 			}
3072 
3073 			first = 0; bit_b++;
3074 		}
3075 		if (!err)
3076 			err = proc_put_char(&buffer, &left, '\n');
3077 	}
3078 
3079 	if (!err) {
3080 		if (write) {
3081 			if (*ppos)
3082 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3083 			else
3084 				bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3085 		}
3086 		kfree(tmp_bitmap);
3087 		*lenp -= left;
3088 		*ppos += *lenp;
3089 		return 0;
3090 	} else {
3091 		kfree(tmp_bitmap);
3092 		return err;
3093 	}
3094 }
3095 
3096 #else /* CONFIG_PROC_SYSCTL */
3097 
3098 int proc_dostring(struct ctl_table *table, int write,
3099 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3100 {
3101 	return -ENOSYS;
3102 }
3103 
3104 int proc_dointvec(struct ctl_table *table, int write,
3105 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3106 {
3107 	return -ENOSYS;
3108 }
3109 
3110 int proc_douintvec(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_minmax(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_minmax(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_jiffies(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_dointvec_userhz_jiffies(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_ms_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_doulongvec_minmax(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_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3153 				      void __user *buffer,
3154 				      size_t *lenp, loff_t *ppos)
3155 {
3156     return -ENOSYS;
3157 }
3158 
3159 
3160 #endif /* CONFIG_PROC_SYSCTL */
3161 
3162 /*
3163  * No sense putting this after each symbol definition, twice,
3164  * exception granted :-)
3165  */
3166 EXPORT_SYMBOL(proc_dointvec);
3167 EXPORT_SYMBOL(proc_douintvec);
3168 EXPORT_SYMBOL(proc_dointvec_jiffies);
3169 EXPORT_SYMBOL(proc_dointvec_minmax);
3170 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3171 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3172 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3173 EXPORT_SYMBOL(proc_dostring);
3174 EXPORT_SYMBOL(proc_doulongvec_minmax);
3175 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3176