xref: /openbmc/linux/kernel/sysctl.c (revision e1e38ea1)
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 #include <linux/pipe_fs_i.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 /*
149  * This is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs
150  * and hung_task_check_interval_secs
151  */
152 #ifdef CONFIG_DETECT_HUNG_TASK
153 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
154 #endif
155 
156 #ifdef CONFIG_INOTIFY_USER
157 #include <linux/inotify.h>
158 #endif
159 #ifdef CONFIG_SPARC
160 #endif
161 
162 #ifdef __hppa__
163 extern int pwrsw_enabled;
164 #endif
165 
166 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
167 extern int unaligned_enabled;
168 #endif
169 
170 #ifdef CONFIG_IA64
171 extern int unaligned_dump_stack;
172 #endif
173 
174 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
175 extern int no_unaligned_warning;
176 #endif
177 
178 #ifdef CONFIG_PROC_SYSCTL
179 
180 /**
181  * enum sysctl_writes_mode - supported sysctl write modes
182  *
183  * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
184  * 	to be written, and multiple writes on the same sysctl file descriptor
185  * 	will rewrite the sysctl value, regardless of file position. No warning
186  * 	is issued when the initial position is not 0.
187  * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
188  * 	not 0.
189  * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
190  * 	file position 0 and the value must be fully contained in the buffer
191  * 	sent to the write syscall. If dealing with strings respect the file
192  * 	position, but restrict this to the max length of the buffer, anything
193  * 	passed the max lenght will be ignored. Multiple writes will append
194  * 	to the buffer.
195  *
196  * These write modes control how current file position affects the behavior of
197  * updating sysctl values through the proc interface on each write.
198  */
199 enum sysctl_writes_mode {
200 	SYSCTL_WRITES_LEGACY		= -1,
201 	SYSCTL_WRITES_WARN		= 0,
202 	SYSCTL_WRITES_STRICT		= 1,
203 };
204 
205 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
206 
207 static int proc_do_cad_pid(struct ctl_table *table, int write,
208 		  void __user *buffer, size_t *lenp, loff_t *ppos);
209 static int proc_taint(struct ctl_table *table, int write,
210 			       void __user *buffer, size_t *lenp, loff_t *ppos);
211 #endif
212 
213 #ifdef CONFIG_PRINTK
214 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
215 				void __user *buffer, size_t *lenp, loff_t *ppos);
216 #endif
217 
218 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
219 		void __user *buffer, size_t *lenp, loff_t *ppos);
220 #ifdef CONFIG_COREDUMP
221 static int proc_dostring_coredump(struct ctl_table *table, int write,
222 		void __user *buffer, size_t *lenp, loff_t *ppos);
223 #endif
224 static int proc_dopipe_max_size(struct ctl_table *table, int write,
225 		void __user *buffer, size_t *lenp, loff_t *ppos);
226 
227 #ifdef CONFIG_MAGIC_SYSRQ
228 /* Note: sysrq code uses its own private copy */
229 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
230 
231 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
232 				void __user *buffer, size_t *lenp,
233 				loff_t *ppos)
234 {
235 	int error;
236 
237 	error = proc_dointvec(table, write, buffer, lenp, ppos);
238 	if (error)
239 		return error;
240 
241 	if (write)
242 		sysrq_toggle_support(__sysrq_enabled);
243 
244 	return 0;
245 }
246 
247 #endif
248 
249 static struct ctl_table kern_table[];
250 static struct ctl_table vm_table[];
251 static struct ctl_table fs_table[];
252 static struct ctl_table debug_table[];
253 static struct ctl_table dev_table[];
254 extern struct ctl_table random_table[];
255 #ifdef CONFIG_EPOLL
256 extern struct ctl_table epoll_table[];
257 #endif
258 
259 #ifdef CONFIG_FW_LOADER_USER_HELPER
260 extern struct ctl_table firmware_config_table[];
261 #endif
262 
263 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
264 int sysctl_legacy_va_layout;
265 #endif
266 
267 /* The default sysctl tables: */
268 
269 static struct ctl_table sysctl_base_table[] = {
270 	{
271 		.procname	= "kernel",
272 		.mode		= 0555,
273 		.child		= kern_table,
274 	},
275 	{
276 		.procname	= "vm",
277 		.mode		= 0555,
278 		.child		= vm_table,
279 	},
280 	{
281 		.procname	= "fs",
282 		.mode		= 0555,
283 		.child		= fs_table,
284 	},
285 	{
286 		.procname	= "debug",
287 		.mode		= 0555,
288 		.child		= debug_table,
289 	},
290 	{
291 		.procname	= "dev",
292 		.mode		= 0555,
293 		.child		= dev_table,
294 	},
295 	{ }
296 };
297 
298 #ifdef CONFIG_SCHED_DEBUG
299 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
300 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
301 static int min_wakeup_granularity_ns;			/* 0 usecs */
302 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
303 #ifdef CONFIG_SMP
304 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
305 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
306 #endif /* CONFIG_SMP */
307 #endif /* CONFIG_SCHED_DEBUG */
308 
309 #ifdef CONFIG_COMPACTION
310 static int min_extfrag_threshold;
311 static int max_extfrag_threshold = 1000;
312 #endif
313 
314 static struct ctl_table kern_table[] = {
315 	{
316 		.procname	= "sched_child_runs_first",
317 		.data		= &sysctl_sched_child_runs_first,
318 		.maxlen		= sizeof(unsigned int),
319 		.mode		= 0644,
320 		.proc_handler	= proc_dointvec,
321 	},
322 #ifdef CONFIG_SCHED_DEBUG
323 	{
324 		.procname	= "sched_min_granularity_ns",
325 		.data		= &sysctl_sched_min_granularity,
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_latency_ns",
334 		.data		= &sysctl_sched_latency,
335 		.maxlen		= sizeof(unsigned int),
336 		.mode		= 0644,
337 		.proc_handler	= sched_proc_update_handler,
338 		.extra1		= &min_sched_granularity_ns,
339 		.extra2		= &max_sched_granularity_ns,
340 	},
341 	{
342 		.procname	= "sched_wakeup_granularity_ns",
343 		.data		= &sysctl_sched_wakeup_granularity,
344 		.maxlen		= sizeof(unsigned int),
345 		.mode		= 0644,
346 		.proc_handler	= sched_proc_update_handler,
347 		.extra1		= &min_wakeup_granularity_ns,
348 		.extra2		= &max_wakeup_granularity_ns,
349 	},
350 #ifdef CONFIG_SMP
351 	{
352 		.procname	= "sched_tunable_scaling",
353 		.data		= &sysctl_sched_tunable_scaling,
354 		.maxlen		= sizeof(enum sched_tunable_scaling),
355 		.mode		= 0644,
356 		.proc_handler	= sched_proc_update_handler,
357 		.extra1		= &min_sched_tunable_scaling,
358 		.extra2		= &max_sched_tunable_scaling,
359 	},
360 	{
361 		.procname	= "sched_migration_cost_ns",
362 		.data		= &sysctl_sched_migration_cost,
363 		.maxlen		= sizeof(unsigned int),
364 		.mode		= 0644,
365 		.proc_handler	= proc_dointvec,
366 	},
367 	{
368 		.procname	= "sched_nr_migrate",
369 		.data		= &sysctl_sched_nr_migrate,
370 		.maxlen		= sizeof(unsigned int),
371 		.mode		= 0644,
372 		.proc_handler	= proc_dointvec,
373 	},
374 #ifdef CONFIG_SCHEDSTATS
375 	{
376 		.procname	= "sched_schedstats",
377 		.data		= NULL,
378 		.maxlen		= sizeof(unsigned int),
379 		.mode		= 0644,
380 		.proc_handler	= sysctl_schedstats,
381 		.extra1		= &zero,
382 		.extra2		= &one,
383 	},
384 #endif /* CONFIG_SCHEDSTATS */
385 #endif /* CONFIG_SMP */
386 #ifdef CONFIG_NUMA_BALANCING
387 	{
388 		.procname	= "numa_balancing_scan_delay_ms",
389 		.data		= &sysctl_numa_balancing_scan_delay,
390 		.maxlen		= sizeof(unsigned int),
391 		.mode		= 0644,
392 		.proc_handler	= proc_dointvec,
393 	},
394 	{
395 		.procname	= "numa_balancing_scan_period_min_ms",
396 		.data		= &sysctl_numa_balancing_scan_period_min,
397 		.maxlen		= sizeof(unsigned int),
398 		.mode		= 0644,
399 		.proc_handler	= proc_dointvec,
400 	},
401 	{
402 		.procname	= "numa_balancing_scan_period_max_ms",
403 		.data		= &sysctl_numa_balancing_scan_period_max,
404 		.maxlen		= sizeof(unsigned int),
405 		.mode		= 0644,
406 		.proc_handler	= proc_dointvec,
407 	},
408 	{
409 		.procname	= "numa_balancing_scan_size_mb",
410 		.data		= &sysctl_numa_balancing_scan_size,
411 		.maxlen		= sizeof(unsigned int),
412 		.mode		= 0644,
413 		.proc_handler	= proc_dointvec_minmax,
414 		.extra1		= &one,
415 	},
416 	{
417 		.procname	= "numa_balancing",
418 		.data		= NULL, /* filled in by handler */
419 		.maxlen		= sizeof(unsigned int),
420 		.mode		= 0644,
421 		.proc_handler	= sysctl_numa_balancing,
422 		.extra1		= &zero,
423 		.extra2		= &one,
424 	},
425 #endif /* CONFIG_NUMA_BALANCING */
426 #endif /* CONFIG_SCHED_DEBUG */
427 	{
428 		.procname	= "sched_rt_period_us",
429 		.data		= &sysctl_sched_rt_period,
430 		.maxlen		= sizeof(unsigned int),
431 		.mode		= 0644,
432 		.proc_handler	= sched_rt_handler,
433 	},
434 	{
435 		.procname	= "sched_rt_runtime_us",
436 		.data		= &sysctl_sched_rt_runtime,
437 		.maxlen		= sizeof(int),
438 		.mode		= 0644,
439 		.proc_handler	= sched_rt_handler,
440 	},
441 	{
442 		.procname	= "sched_rr_timeslice_ms",
443 		.data		= &sysctl_sched_rr_timeslice,
444 		.maxlen		= sizeof(int),
445 		.mode		= 0644,
446 		.proc_handler	= sched_rr_handler,
447 	},
448 #ifdef CONFIG_SCHED_AUTOGROUP
449 	{
450 		.procname	= "sched_autogroup_enabled",
451 		.data		= &sysctl_sched_autogroup_enabled,
452 		.maxlen		= sizeof(unsigned int),
453 		.mode		= 0644,
454 		.proc_handler	= proc_dointvec_minmax,
455 		.extra1		= &zero,
456 		.extra2		= &one,
457 	},
458 #endif
459 #ifdef CONFIG_CFS_BANDWIDTH
460 	{
461 		.procname	= "sched_cfs_bandwidth_slice_us",
462 		.data		= &sysctl_sched_cfs_bandwidth_slice,
463 		.maxlen		= sizeof(unsigned int),
464 		.mode		= 0644,
465 		.proc_handler	= proc_dointvec_minmax,
466 		.extra1		= &one,
467 	},
468 #endif
469 #ifdef CONFIG_PROVE_LOCKING
470 	{
471 		.procname	= "prove_locking",
472 		.data		= &prove_locking,
473 		.maxlen		= sizeof(int),
474 		.mode		= 0644,
475 		.proc_handler	= proc_dointvec,
476 	},
477 #endif
478 #ifdef CONFIG_LOCK_STAT
479 	{
480 		.procname	= "lock_stat",
481 		.data		= &lock_stat,
482 		.maxlen		= sizeof(int),
483 		.mode		= 0644,
484 		.proc_handler	= proc_dointvec,
485 	},
486 #endif
487 	{
488 		.procname	= "panic",
489 		.data		= &panic_timeout,
490 		.maxlen		= sizeof(int),
491 		.mode		= 0644,
492 		.proc_handler	= proc_dointvec,
493 	},
494 #ifdef CONFIG_COREDUMP
495 	{
496 		.procname	= "core_uses_pid",
497 		.data		= &core_uses_pid,
498 		.maxlen		= sizeof(int),
499 		.mode		= 0644,
500 		.proc_handler	= proc_dointvec,
501 	},
502 	{
503 		.procname	= "core_pattern",
504 		.data		= core_pattern,
505 		.maxlen		= CORENAME_MAX_SIZE,
506 		.mode		= 0644,
507 		.proc_handler	= proc_dostring_coredump,
508 	},
509 	{
510 		.procname	= "core_pipe_limit",
511 		.data		= &core_pipe_limit,
512 		.maxlen		= sizeof(unsigned int),
513 		.mode		= 0644,
514 		.proc_handler	= proc_dointvec,
515 	},
516 #endif
517 #ifdef CONFIG_PROC_SYSCTL
518 	{
519 		.procname	= "tainted",
520 		.maxlen 	= sizeof(long),
521 		.mode		= 0644,
522 		.proc_handler	= proc_taint,
523 	},
524 	{
525 		.procname	= "sysctl_writes_strict",
526 		.data		= &sysctl_writes_strict,
527 		.maxlen		= sizeof(int),
528 		.mode		= 0644,
529 		.proc_handler	= proc_dointvec_minmax,
530 		.extra1		= &neg_one,
531 		.extra2		= &one,
532 	},
533 #endif
534 #ifdef CONFIG_LATENCYTOP
535 	{
536 		.procname	= "latencytop",
537 		.data		= &latencytop_enabled,
538 		.maxlen		= sizeof(int),
539 		.mode		= 0644,
540 		.proc_handler	= sysctl_latencytop,
541 	},
542 #endif
543 #ifdef CONFIG_BLK_DEV_INITRD
544 	{
545 		.procname	= "real-root-dev",
546 		.data		= &real_root_dev,
547 		.maxlen		= sizeof(int),
548 		.mode		= 0644,
549 		.proc_handler	= proc_dointvec,
550 	},
551 #endif
552 	{
553 		.procname	= "print-fatal-signals",
554 		.data		= &print_fatal_signals,
555 		.maxlen		= sizeof(int),
556 		.mode		= 0644,
557 		.proc_handler	= proc_dointvec,
558 	},
559 #ifdef CONFIG_SPARC
560 	{
561 		.procname	= "reboot-cmd",
562 		.data		= reboot_command,
563 		.maxlen		= 256,
564 		.mode		= 0644,
565 		.proc_handler	= proc_dostring,
566 	},
567 	{
568 		.procname	= "stop-a",
569 		.data		= &stop_a_enabled,
570 		.maxlen		= sizeof (int),
571 		.mode		= 0644,
572 		.proc_handler	= proc_dointvec,
573 	},
574 	{
575 		.procname	= "scons-poweroff",
576 		.data		= &scons_pwroff,
577 		.maxlen		= sizeof (int),
578 		.mode		= 0644,
579 		.proc_handler	= proc_dointvec,
580 	},
581 #endif
582 #ifdef CONFIG_SPARC64
583 	{
584 		.procname	= "tsb-ratio",
585 		.data		= &sysctl_tsb_ratio,
586 		.maxlen		= sizeof (int),
587 		.mode		= 0644,
588 		.proc_handler	= proc_dointvec,
589 	},
590 #endif
591 #ifdef __hppa__
592 	{
593 		.procname	= "soft-power",
594 		.data		= &pwrsw_enabled,
595 		.maxlen		= sizeof (int),
596 	 	.mode		= 0644,
597 		.proc_handler	= proc_dointvec,
598 	},
599 #endif
600 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
601 	{
602 		.procname	= "unaligned-trap",
603 		.data		= &unaligned_enabled,
604 		.maxlen		= sizeof (int),
605 		.mode		= 0644,
606 		.proc_handler	= proc_dointvec,
607 	},
608 #endif
609 	{
610 		.procname	= "ctrl-alt-del",
611 		.data		= &C_A_D,
612 		.maxlen		= sizeof(int),
613 		.mode		= 0644,
614 		.proc_handler	= proc_dointvec,
615 	},
616 #ifdef CONFIG_FUNCTION_TRACER
617 	{
618 		.procname	= "ftrace_enabled",
619 		.data		= &ftrace_enabled,
620 		.maxlen		= sizeof(int),
621 		.mode		= 0644,
622 		.proc_handler	= ftrace_enable_sysctl,
623 	},
624 #endif
625 #ifdef CONFIG_STACK_TRACER
626 	{
627 		.procname	= "stack_tracer_enabled",
628 		.data		= &stack_tracer_enabled,
629 		.maxlen		= sizeof(int),
630 		.mode		= 0644,
631 		.proc_handler	= stack_trace_sysctl,
632 	},
633 #endif
634 #ifdef CONFIG_TRACING
635 	{
636 		.procname	= "ftrace_dump_on_oops",
637 		.data		= &ftrace_dump_on_oops,
638 		.maxlen		= sizeof(int),
639 		.mode		= 0644,
640 		.proc_handler	= proc_dointvec,
641 	},
642 	{
643 		.procname	= "traceoff_on_warning",
644 		.data		= &__disable_trace_on_warning,
645 		.maxlen		= sizeof(__disable_trace_on_warning),
646 		.mode		= 0644,
647 		.proc_handler	= proc_dointvec,
648 	},
649 	{
650 		.procname	= "tracepoint_printk",
651 		.data		= &tracepoint_printk,
652 		.maxlen		= sizeof(tracepoint_printk),
653 		.mode		= 0644,
654 		.proc_handler	= tracepoint_printk_sysctl,
655 	},
656 #endif
657 #ifdef CONFIG_KEXEC_CORE
658 	{
659 		.procname	= "kexec_load_disabled",
660 		.data		= &kexec_load_disabled,
661 		.maxlen		= sizeof(int),
662 		.mode		= 0644,
663 		/* only handle a transition from default "0" to "1" */
664 		.proc_handler	= proc_dointvec_minmax,
665 		.extra1		= &one,
666 		.extra2		= &one,
667 	},
668 #endif
669 #ifdef CONFIG_MODULES
670 	{
671 		.procname	= "modprobe",
672 		.data		= &modprobe_path,
673 		.maxlen		= KMOD_PATH_LEN,
674 		.mode		= 0644,
675 		.proc_handler	= proc_dostring,
676 	},
677 	{
678 		.procname	= "modules_disabled",
679 		.data		= &modules_disabled,
680 		.maxlen		= sizeof(int),
681 		.mode		= 0644,
682 		/* only handle a transition from default "0" to "1" */
683 		.proc_handler	= proc_dointvec_minmax,
684 		.extra1		= &one,
685 		.extra2		= &one,
686 	},
687 #endif
688 #ifdef CONFIG_UEVENT_HELPER
689 	{
690 		.procname	= "hotplug",
691 		.data		= &uevent_helper,
692 		.maxlen		= UEVENT_HELPER_PATH_LEN,
693 		.mode		= 0644,
694 		.proc_handler	= proc_dostring,
695 	},
696 #endif
697 #ifdef CONFIG_CHR_DEV_SG
698 	{
699 		.procname	= "sg-big-buff",
700 		.data		= &sg_big_buff,
701 		.maxlen		= sizeof (int),
702 		.mode		= 0444,
703 		.proc_handler	= proc_dointvec,
704 	},
705 #endif
706 #ifdef CONFIG_BSD_PROCESS_ACCT
707 	{
708 		.procname	= "acct",
709 		.data		= &acct_parm,
710 		.maxlen		= 3*sizeof(int),
711 		.mode		= 0644,
712 		.proc_handler	= proc_dointvec,
713 	},
714 #endif
715 #ifdef CONFIG_MAGIC_SYSRQ
716 	{
717 		.procname	= "sysrq",
718 		.data		= &__sysrq_enabled,
719 		.maxlen		= sizeof (int),
720 		.mode		= 0644,
721 		.proc_handler	= sysrq_sysctl_handler,
722 	},
723 #endif
724 #ifdef CONFIG_PROC_SYSCTL
725 	{
726 		.procname	= "cad_pid",
727 		.data		= NULL,
728 		.maxlen		= sizeof (int),
729 		.mode		= 0600,
730 		.proc_handler	= proc_do_cad_pid,
731 	},
732 #endif
733 	{
734 		.procname	= "threads-max",
735 		.data		= NULL,
736 		.maxlen		= sizeof(int),
737 		.mode		= 0644,
738 		.proc_handler	= sysctl_max_threads,
739 	},
740 	{
741 		.procname	= "random",
742 		.mode		= 0555,
743 		.child		= random_table,
744 	},
745 	{
746 		.procname	= "usermodehelper",
747 		.mode		= 0555,
748 		.child		= usermodehelper_table,
749 	},
750 #ifdef CONFIG_FW_LOADER_USER_HELPER
751 	{
752 		.procname	= "firmware_config",
753 		.mode		= 0555,
754 		.child		= firmware_config_table,
755 	},
756 #endif
757 	{
758 		.procname	= "overflowuid",
759 		.data		= &overflowuid,
760 		.maxlen		= sizeof(int),
761 		.mode		= 0644,
762 		.proc_handler	= proc_dointvec_minmax,
763 		.extra1		= &minolduid,
764 		.extra2		= &maxolduid,
765 	},
766 	{
767 		.procname	= "overflowgid",
768 		.data		= &overflowgid,
769 		.maxlen		= sizeof(int),
770 		.mode		= 0644,
771 		.proc_handler	= proc_dointvec_minmax,
772 		.extra1		= &minolduid,
773 		.extra2		= &maxolduid,
774 	},
775 #ifdef CONFIG_S390
776 #ifdef CONFIG_MATHEMU
777 	{
778 		.procname	= "ieee_emulation_warnings",
779 		.data		= &sysctl_ieee_emulation_warnings,
780 		.maxlen		= sizeof(int),
781 		.mode		= 0644,
782 		.proc_handler	= proc_dointvec,
783 	},
784 #endif
785 	{
786 		.procname	= "userprocess_debug",
787 		.data		= &show_unhandled_signals,
788 		.maxlen		= sizeof(int),
789 		.mode		= 0644,
790 		.proc_handler	= proc_dointvec,
791 	},
792 #endif
793 	{
794 		.procname	= "pid_max",
795 		.data		= &pid_max,
796 		.maxlen		= sizeof (int),
797 		.mode		= 0644,
798 		.proc_handler	= proc_dointvec_minmax,
799 		.extra1		= &pid_max_min,
800 		.extra2		= &pid_max_max,
801 	},
802 	{
803 		.procname	= "panic_on_oops",
804 		.data		= &panic_on_oops,
805 		.maxlen		= sizeof(int),
806 		.mode		= 0644,
807 		.proc_handler	= proc_dointvec,
808 	},
809 #if defined CONFIG_PRINTK
810 	{
811 		.procname	= "printk",
812 		.data		= &console_loglevel,
813 		.maxlen		= 4*sizeof(int),
814 		.mode		= 0644,
815 		.proc_handler	= proc_dointvec,
816 	},
817 	{
818 		.procname	= "printk_ratelimit",
819 		.data		= &printk_ratelimit_state.interval,
820 		.maxlen		= sizeof(int),
821 		.mode		= 0644,
822 		.proc_handler	= proc_dointvec_jiffies,
823 	},
824 	{
825 		.procname	= "printk_ratelimit_burst",
826 		.data		= &printk_ratelimit_state.burst,
827 		.maxlen		= sizeof(int),
828 		.mode		= 0644,
829 		.proc_handler	= proc_dointvec,
830 	},
831 	{
832 		.procname	= "printk_delay",
833 		.data		= &printk_delay_msec,
834 		.maxlen		= sizeof(int),
835 		.mode		= 0644,
836 		.proc_handler	= proc_dointvec_minmax,
837 		.extra1		= &zero,
838 		.extra2		= &ten_thousand,
839 	},
840 	{
841 		.procname	= "printk_devkmsg",
842 		.data		= devkmsg_log_str,
843 		.maxlen		= DEVKMSG_STR_MAX_SIZE,
844 		.mode		= 0644,
845 		.proc_handler	= devkmsg_sysctl_set_loglvl,
846 	},
847 	{
848 		.procname	= "dmesg_restrict",
849 		.data		= &dmesg_restrict,
850 		.maxlen		= sizeof(int),
851 		.mode		= 0644,
852 		.proc_handler	= proc_dointvec_minmax_sysadmin,
853 		.extra1		= &zero,
854 		.extra2		= &one,
855 	},
856 	{
857 		.procname	= "kptr_restrict",
858 		.data		= &kptr_restrict,
859 		.maxlen		= sizeof(int),
860 		.mode		= 0644,
861 		.proc_handler	= proc_dointvec_minmax_sysadmin,
862 		.extra1		= &zero,
863 		.extra2		= &two,
864 	},
865 #endif
866 	{
867 		.procname	= "ngroups_max",
868 		.data		= &ngroups_max,
869 		.maxlen		= sizeof (int),
870 		.mode		= 0444,
871 		.proc_handler	= proc_dointvec,
872 	},
873 	{
874 		.procname	= "cap_last_cap",
875 		.data		= (void *)&cap_last_cap,
876 		.maxlen		= sizeof(int),
877 		.mode		= 0444,
878 		.proc_handler	= proc_dointvec,
879 	},
880 #if defined(CONFIG_LOCKUP_DETECTOR)
881 	{
882 		.procname       = "watchdog",
883 		.data		= &watchdog_user_enabled,
884 		.maxlen		= sizeof(int),
885 		.mode		= 0644,
886 		.proc_handler   = proc_watchdog,
887 		.extra1		= &zero,
888 		.extra2		= &one,
889 	},
890 	{
891 		.procname	= "watchdog_thresh",
892 		.data		= &watchdog_thresh,
893 		.maxlen		= sizeof(int),
894 		.mode		= 0644,
895 		.proc_handler	= proc_watchdog_thresh,
896 		.extra1		= &zero,
897 		.extra2		= &sixty,
898 	},
899 	{
900 		.procname       = "nmi_watchdog",
901 		.data		= &nmi_watchdog_user_enabled,
902 		.maxlen		= sizeof(int),
903 		.mode		= NMI_WATCHDOG_SYSCTL_PERM,
904 		.proc_handler   = proc_nmi_watchdog,
905 		.extra1		= &zero,
906 		.extra2		= &one,
907 	},
908 	{
909 		.procname	= "watchdog_cpumask",
910 		.data		= &watchdog_cpumask_bits,
911 		.maxlen		= NR_CPUS,
912 		.mode		= 0644,
913 		.proc_handler	= proc_watchdog_cpumask,
914 	},
915 #ifdef CONFIG_SOFTLOCKUP_DETECTOR
916 	{
917 		.procname       = "soft_watchdog",
918 		.data		= &soft_watchdog_user_enabled,
919 		.maxlen		= sizeof(int),
920 		.mode		= 0644,
921 		.proc_handler   = proc_soft_watchdog,
922 		.extra1		= &zero,
923 		.extra2		= &one,
924 	},
925 	{
926 		.procname	= "softlockup_panic",
927 		.data		= &softlockup_panic,
928 		.maxlen		= sizeof(int),
929 		.mode		= 0644,
930 		.proc_handler	= proc_dointvec_minmax,
931 		.extra1		= &zero,
932 		.extra2		= &one,
933 	},
934 #ifdef CONFIG_SMP
935 	{
936 		.procname	= "softlockup_all_cpu_backtrace",
937 		.data		= &sysctl_softlockup_all_cpu_backtrace,
938 		.maxlen		= sizeof(int),
939 		.mode		= 0644,
940 		.proc_handler	= proc_dointvec_minmax,
941 		.extra1		= &zero,
942 		.extra2		= &one,
943 	},
944 #endif /* CONFIG_SMP */
945 #endif
946 #ifdef CONFIG_HARDLOCKUP_DETECTOR
947 	{
948 		.procname	= "hardlockup_panic",
949 		.data		= &hardlockup_panic,
950 		.maxlen		= sizeof(int),
951 		.mode		= 0644,
952 		.proc_handler	= proc_dointvec_minmax,
953 		.extra1		= &zero,
954 		.extra2		= &one,
955 	},
956 #ifdef CONFIG_SMP
957 	{
958 		.procname	= "hardlockup_all_cpu_backtrace",
959 		.data		= &sysctl_hardlockup_all_cpu_backtrace,
960 		.maxlen		= sizeof(int),
961 		.mode		= 0644,
962 		.proc_handler	= proc_dointvec_minmax,
963 		.extra1		= &zero,
964 		.extra2		= &one,
965 	},
966 #endif /* CONFIG_SMP */
967 #endif
968 #endif
969 
970 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
971 	{
972 		.procname       = "unknown_nmi_panic",
973 		.data           = &unknown_nmi_panic,
974 		.maxlen         = sizeof (int),
975 		.mode           = 0644,
976 		.proc_handler   = proc_dointvec,
977 	},
978 #endif
979 #if defined(CONFIG_X86)
980 	{
981 		.procname	= "panic_on_unrecovered_nmi",
982 		.data		= &panic_on_unrecovered_nmi,
983 		.maxlen		= sizeof(int),
984 		.mode		= 0644,
985 		.proc_handler	= proc_dointvec,
986 	},
987 	{
988 		.procname	= "panic_on_io_nmi",
989 		.data		= &panic_on_io_nmi,
990 		.maxlen		= sizeof(int),
991 		.mode		= 0644,
992 		.proc_handler	= proc_dointvec,
993 	},
994 #ifdef CONFIG_DEBUG_STACKOVERFLOW
995 	{
996 		.procname	= "panic_on_stackoverflow",
997 		.data		= &sysctl_panic_on_stackoverflow,
998 		.maxlen		= sizeof(int),
999 		.mode		= 0644,
1000 		.proc_handler	= proc_dointvec,
1001 	},
1002 #endif
1003 	{
1004 		.procname	= "bootloader_type",
1005 		.data		= &bootloader_type,
1006 		.maxlen		= sizeof (int),
1007 		.mode		= 0444,
1008 		.proc_handler	= proc_dointvec,
1009 	},
1010 	{
1011 		.procname	= "bootloader_version",
1012 		.data		= &bootloader_version,
1013 		.maxlen		= sizeof (int),
1014 		.mode		= 0444,
1015 		.proc_handler	= proc_dointvec,
1016 	},
1017 	{
1018 		.procname	= "io_delay_type",
1019 		.data		= &io_delay_type,
1020 		.maxlen		= sizeof(int),
1021 		.mode		= 0644,
1022 		.proc_handler	= proc_dointvec,
1023 	},
1024 #endif
1025 #if defined(CONFIG_MMU)
1026 	{
1027 		.procname	= "randomize_va_space",
1028 		.data		= &randomize_va_space,
1029 		.maxlen		= sizeof(int),
1030 		.mode		= 0644,
1031 		.proc_handler	= proc_dointvec,
1032 	},
1033 #endif
1034 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1035 	{
1036 		.procname	= "spin_retry",
1037 		.data		= &spin_retry,
1038 		.maxlen		= sizeof (int),
1039 		.mode		= 0644,
1040 		.proc_handler	= proc_dointvec,
1041 	},
1042 #endif
1043 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1044 	{
1045 		.procname	= "acpi_video_flags",
1046 		.data		= &acpi_realmode_flags,
1047 		.maxlen		= sizeof (unsigned long),
1048 		.mode		= 0644,
1049 		.proc_handler	= proc_doulongvec_minmax,
1050 	},
1051 #endif
1052 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1053 	{
1054 		.procname	= "ignore-unaligned-usertrap",
1055 		.data		= &no_unaligned_warning,
1056 		.maxlen		= sizeof (int),
1057 	 	.mode		= 0644,
1058 		.proc_handler	= proc_dointvec,
1059 	},
1060 #endif
1061 #ifdef CONFIG_IA64
1062 	{
1063 		.procname	= "unaligned-dump-stack",
1064 		.data		= &unaligned_dump_stack,
1065 		.maxlen		= sizeof (int),
1066 		.mode		= 0644,
1067 		.proc_handler	= proc_dointvec,
1068 	},
1069 #endif
1070 #ifdef CONFIG_DETECT_HUNG_TASK
1071 	{
1072 		.procname	= "hung_task_panic",
1073 		.data		= &sysctl_hung_task_panic,
1074 		.maxlen		= sizeof(int),
1075 		.mode		= 0644,
1076 		.proc_handler	= proc_dointvec_minmax,
1077 		.extra1		= &zero,
1078 		.extra2		= &one,
1079 	},
1080 	{
1081 		.procname	= "hung_task_check_count",
1082 		.data		= &sysctl_hung_task_check_count,
1083 		.maxlen		= sizeof(int),
1084 		.mode		= 0644,
1085 		.proc_handler	= proc_dointvec_minmax,
1086 		.extra1		= &zero,
1087 	},
1088 	{
1089 		.procname	= "hung_task_timeout_secs",
1090 		.data		= &sysctl_hung_task_timeout_secs,
1091 		.maxlen		= sizeof(unsigned long),
1092 		.mode		= 0644,
1093 		.proc_handler	= proc_dohung_task_timeout_secs,
1094 		.extra2		= &hung_task_timeout_max,
1095 	},
1096 	{
1097 		.procname	= "hung_task_check_interval_secs",
1098 		.data		= &sysctl_hung_task_check_interval_secs,
1099 		.maxlen		= sizeof(unsigned long),
1100 		.mode		= 0644,
1101 		.proc_handler	= proc_dohung_task_timeout_secs,
1102 		.extra2		= &hung_task_timeout_max,
1103 	},
1104 	{
1105 		.procname	= "hung_task_warnings",
1106 		.data		= &sysctl_hung_task_warnings,
1107 		.maxlen		= sizeof(int),
1108 		.mode		= 0644,
1109 		.proc_handler	= proc_dointvec_minmax,
1110 		.extra1		= &neg_one,
1111 	},
1112 #endif
1113 #ifdef CONFIG_RT_MUTEXES
1114 	{
1115 		.procname	= "max_lock_depth",
1116 		.data		= &max_lock_depth,
1117 		.maxlen		= sizeof(int),
1118 		.mode		= 0644,
1119 		.proc_handler	= proc_dointvec,
1120 	},
1121 #endif
1122 	{
1123 		.procname	= "poweroff_cmd",
1124 		.data		= &poweroff_cmd,
1125 		.maxlen		= POWEROFF_CMD_PATH_LEN,
1126 		.mode		= 0644,
1127 		.proc_handler	= proc_dostring,
1128 	},
1129 #ifdef CONFIG_KEYS
1130 	{
1131 		.procname	= "keys",
1132 		.mode		= 0555,
1133 		.child		= key_sysctls,
1134 	},
1135 #endif
1136 #ifdef CONFIG_PERF_EVENTS
1137 	/*
1138 	 * User-space scripts rely on the existence of this file
1139 	 * as a feature check for perf_events being enabled.
1140 	 *
1141 	 * So it's an ABI, do not remove!
1142 	 */
1143 	{
1144 		.procname	= "perf_event_paranoid",
1145 		.data		= &sysctl_perf_event_paranoid,
1146 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
1147 		.mode		= 0644,
1148 		.proc_handler	= proc_dointvec,
1149 	},
1150 	{
1151 		.procname	= "perf_event_mlock_kb",
1152 		.data		= &sysctl_perf_event_mlock,
1153 		.maxlen		= sizeof(sysctl_perf_event_mlock),
1154 		.mode		= 0644,
1155 		.proc_handler	= proc_dointvec,
1156 	},
1157 	{
1158 		.procname	= "perf_event_max_sample_rate",
1159 		.data		= &sysctl_perf_event_sample_rate,
1160 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
1161 		.mode		= 0644,
1162 		.proc_handler	= perf_proc_update_handler,
1163 		.extra1		= &one,
1164 	},
1165 	{
1166 		.procname	= "perf_cpu_time_max_percent",
1167 		.data		= &sysctl_perf_cpu_time_max_percent,
1168 		.maxlen		= sizeof(sysctl_perf_cpu_time_max_percent),
1169 		.mode		= 0644,
1170 		.proc_handler	= perf_cpu_time_max_percent_handler,
1171 		.extra1		= &zero,
1172 		.extra2		= &one_hundred,
1173 	},
1174 	{
1175 		.procname	= "perf_event_max_stack",
1176 		.data		= &sysctl_perf_event_max_stack,
1177 		.maxlen		= sizeof(sysctl_perf_event_max_stack),
1178 		.mode		= 0644,
1179 		.proc_handler	= perf_event_max_stack_handler,
1180 		.extra1		= &zero,
1181 		.extra2		= &six_hundred_forty_kb,
1182 	},
1183 	{
1184 		.procname	= "perf_event_max_contexts_per_stack",
1185 		.data		= &sysctl_perf_event_max_contexts_per_stack,
1186 		.maxlen		= sizeof(sysctl_perf_event_max_contexts_per_stack),
1187 		.mode		= 0644,
1188 		.proc_handler	= perf_event_max_stack_handler,
1189 		.extra1		= &zero,
1190 		.extra2		= &one_thousand,
1191 	},
1192 #endif
1193 	{
1194 		.procname	= "panic_on_warn",
1195 		.data		= &panic_on_warn,
1196 		.maxlen		= sizeof(int),
1197 		.mode		= 0644,
1198 		.proc_handler	= proc_dointvec_minmax,
1199 		.extra1		= &zero,
1200 		.extra2		= &one,
1201 	},
1202 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1203 	{
1204 		.procname	= "timer_migration",
1205 		.data		= &sysctl_timer_migration,
1206 		.maxlen		= sizeof(unsigned int),
1207 		.mode		= 0644,
1208 		.proc_handler	= timer_migration_handler,
1209 		.extra1		= &zero,
1210 		.extra2		= &one,
1211 	},
1212 #endif
1213 #ifdef CONFIG_BPF_SYSCALL
1214 	{
1215 		.procname	= "unprivileged_bpf_disabled",
1216 		.data		= &sysctl_unprivileged_bpf_disabled,
1217 		.maxlen		= sizeof(sysctl_unprivileged_bpf_disabled),
1218 		.mode		= 0644,
1219 		/* only handle a transition from default "0" to "1" */
1220 		.proc_handler	= proc_dointvec_minmax,
1221 		.extra1		= &one,
1222 		.extra2		= &one,
1223 	},
1224 #endif
1225 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1226 	{
1227 		.procname	= "panic_on_rcu_stall",
1228 		.data		= &sysctl_panic_on_rcu_stall,
1229 		.maxlen		= sizeof(sysctl_panic_on_rcu_stall),
1230 		.mode		= 0644,
1231 		.proc_handler	= proc_dointvec_minmax,
1232 		.extra1		= &zero,
1233 		.extra2		= &one,
1234 	},
1235 #endif
1236 	{ }
1237 };
1238 
1239 static struct ctl_table vm_table[] = {
1240 	{
1241 		.procname	= "overcommit_memory",
1242 		.data		= &sysctl_overcommit_memory,
1243 		.maxlen		= sizeof(sysctl_overcommit_memory),
1244 		.mode		= 0644,
1245 		.proc_handler	= proc_dointvec_minmax,
1246 		.extra1		= &zero,
1247 		.extra2		= &two,
1248 	},
1249 	{
1250 		.procname	= "panic_on_oom",
1251 		.data		= &sysctl_panic_on_oom,
1252 		.maxlen		= sizeof(sysctl_panic_on_oom),
1253 		.mode		= 0644,
1254 		.proc_handler	= proc_dointvec_minmax,
1255 		.extra1		= &zero,
1256 		.extra2		= &two,
1257 	},
1258 	{
1259 		.procname	= "oom_kill_allocating_task",
1260 		.data		= &sysctl_oom_kill_allocating_task,
1261 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1262 		.mode		= 0644,
1263 		.proc_handler	= proc_dointvec,
1264 	},
1265 	{
1266 		.procname	= "oom_dump_tasks",
1267 		.data		= &sysctl_oom_dump_tasks,
1268 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1269 		.mode		= 0644,
1270 		.proc_handler	= proc_dointvec,
1271 	},
1272 	{
1273 		.procname	= "overcommit_ratio",
1274 		.data		= &sysctl_overcommit_ratio,
1275 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1276 		.mode		= 0644,
1277 		.proc_handler	= overcommit_ratio_handler,
1278 	},
1279 	{
1280 		.procname	= "overcommit_kbytes",
1281 		.data		= &sysctl_overcommit_kbytes,
1282 		.maxlen		= sizeof(sysctl_overcommit_kbytes),
1283 		.mode		= 0644,
1284 		.proc_handler	= overcommit_kbytes_handler,
1285 	},
1286 	{
1287 		.procname	= "page-cluster",
1288 		.data		= &page_cluster,
1289 		.maxlen		= sizeof(int),
1290 		.mode		= 0644,
1291 		.proc_handler	= proc_dointvec_minmax,
1292 		.extra1		= &zero,
1293 	},
1294 	{
1295 		.procname	= "dirty_background_ratio",
1296 		.data		= &dirty_background_ratio,
1297 		.maxlen		= sizeof(dirty_background_ratio),
1298 		.mode		= 0644,
1299 		.proc_handler	= dirty_background_ratio_handler,
1300 		.extra1		= &zero,
1301 		.extra2		= &one_hundred,
1302 	},
1303 	{
1304 		.procname	= "dirty_background_bytes",
1305 		.data		= &dirty_background_bytes,
1306 		.maxlen		= sizeof(dirty_background_bytes),
1307 		.mode		= 0644,
1308 		.proc_handler	= dirty_background_bytes_handler,
1309 		.extra1		= &one_ul,
1310 	},
1311 	{
1312 		.procname	= "dirty_ratio",
1313 		.data		= &vm_dirty_ratio,
1314 		.maxlen		= sizeof(vm_dirty_ratio),
1315 		.mode		= 0644,
1316 		.proc_handler	= dirty_ratio_handler,
1317 		.extra1		= &zero,
1318 		.extra2		= &one_hundred,
1319 	},
1320 	{
1321 		.procname	= "dirty_bytes",
1322 		.data		= &vm_dirty_bytes,
1323 		.maxlen		= sizeof(vm_dirty_bytes),
1324 		.mode		= 0644,
1325 		.proc_handler	= dirty_bytes_handler,
1326 		.extra1		= &dirty_bytes_min,
1327 	},
1328 	{
1329 		.procname	= "dirty_writeback_centisecs",
1330 		.data		= &dirty_writeback_interval,
1331 		.maxlen		= sizeof(dirty_writeback_interval),
1332 		.mode		= 0644,
1333 		.proc_handler	= dirty_writeback_centisecs_handler,
1334 	},
1335 	{
1336 		.procname	= "dirty_expire_centisecs",
1337 		.data		= &dirty_expire_interval,
1338 		.maxlen		= sizeof(dirty_expire_interval),
1339 		.mode		= 0644,
1340 		.proc_handler	= proc_dointvec_minmax,
1341 		.extra1		= &zero,
1342 	},
1343 	{
1344 		.procname	= "dirtytime_expire_seconds",
1345 		.data		= &dirtytime_expire_interval,
1346 		.maxlen		= sizeof(dirtytime_expire_interval),
1347 		.mode		= 0644,
1348 		.proc_handler	= dirtytime_interval_handler,
1349 		.extra1		= &zero,
1350 	},
1351 	{
1352 		.procname	= "swappiness",
1353 		.data		= &vm_swappiness,
1354 		.maxlen		= sizeof(vm_swappiness),
1355 		.mode		= 0644,
1356 		.proc_handler	= proc_dointvec_minmax,
1357 		.extra1		= &zero,
1358 		.extra2		= &one_hundred,
1359 	},
1360 #ifdef CONFIG_HUGETLB_PAGE
1361 	{
1362 		.procname	= "nr_hugepages",
1363 		.data		= NULL,
1364 		.maxlen		= sizeof(unsigned long),
1365 		.mode		= 0644,
1366 		.proc_handler	= hugetlb_sysctl_handler,
1367 	},
1368 #ifdef CONFIG_NUMA
1369 	{
1370 		.procname       = "nr_hugepages_mempolicy",
1371 		.data           = NULL,
1372 		.maxlen         = sizeof(unsigned long),
1373 		.mode           = 0644,
1374 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1375 	},
1376 	{
1377 		.procname		= "numa_stat",
1378 		.data			= &sysctl_vm_numa_stat,
1379 		.maxlen			= sizeof(int),
1380 		.mode			= 0644,
1381 		.proc_handler	= sysctl_vm_numa_stat_handler,
1382 		.extra1			= &zero,
1383 		.extra2			= &one,
1384 	},
1385 #endif
1386 	 {
1387 		.procname	= "hugetlb_shm_group",
1388 		.data		= &sysctl_hugetlb_shm_group,
1389 		.maxlen		= sizeof(gid_t),
1390 		.mode		= 0644,
1391 		.proc_handler	= proc_dointvec,
1392 	 },
1393 	{
1394 		.procname	= "nr_overcommit_hugepages",
1395 		.data		= NULL,
1396 		.maxlen		= sizeof(unsigned long),
1397 		.mode		= 0644,
1398 		.proc_handler	= hugetlb_overcommit_handler,
1399 	},
1400 #endif
1401 	{
1402 		.procname	= "lowmem_reserve_ratio",
1403 		.data		= &sysctl_lowmem_reserve_ratio,
1404 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1405 		.mode		= 0644,
1406 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1407 	},
1408 	{
1409 		.procname	= "drop_caches",
1410 		.data		= &sysctl_drop_caches,
1411 		.maxlen		= sizeof(int),
1412 		.mode		= 0644,
1413 		.proc_handler	= drop_caches_sysctl_handler,
1414 		.extra1		= &one,
1415 		.extra2		= &four,
1416 	},
1417 #ifdef CONFIG_COMPACTION
1418 	{
1419 		.procname	= "compact_memory",
1420 		.data		= &sysctl_compact_memory,
1421 		.maxlen		= sizeof(int),
1422 		.mode		= 0200,
1423 		.proc_handler	= sysctl_compaction_handler,
1424 	},
1425 	{
1426 		.procname	= "extfrag_threshold",
1427 		.data		= &sysctl_extfrag_threshold,
1428 		.maxlen		= sizeof(int),
1429 		.mode		= 0644,
1430 		.proc_handler	= sysctl_extfrag_handler,
1431 		.extra1		= &min_extfrag_threshold,
1432 		.extra2		= &max_extfrag_threshold,
1433 	},
1434 	{
1435 		.procname	= "compact_unevictable_allowed",
1436 		.data		= &sysctl_compact_unevictable_allowed,
1437 		.maxlen		= sizeof(int),
1438 		.mode		= 0644,
1439 		.proc_handler	= proc_dointvec,
1440 		.extra1		= &zero,
1441 		.extra2		= &one,
1442 	},
1443 
1444 #endif /* CONFIG_COMPACTION */
1445 	{
1446 		.procname	= "min_free_kbytes",
1447 		.data		= &min_free_kbytes,
1448 		.maxlen		= sizeof(min_free_kbytes),
1449 		.mode		= 0644,
1450 		.proc_handler	= min_free_kbytes_sysctl_handler,
1451 		.extra1		= &zero,
1452 	},
1453 	{
1454 		.procname	= "watermark_scale_factor",
1455 		.data		= &watermark_scale_factor,
1456 		.maxlen		= sizeof(watermark_scale_factor),
1457 		.mode		= 0644,
1458 		.proc_handler	= watermark_scale_factor_sysctl_handler,
1459 		.extra1		= &one,
1460 		.extra2		= &one_thousand,
1461 	},
1462 	{
1463 		.procname	= "percpu_pagelist_fraction",
1464 		.data		= &percpu_pagelist_fraction,
1465 		.maxlen		= sizeof(percpu_pagelist_fraction),
1466 		.mode		= 0644,
1467 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1468 		.extra1		= &zero,
1469 	},
1470 #ifdef CONFIG_MMU
1471 	{
1472 		.procname	= "max_map_count",
1473 		.data		= &sysctl_max_map_count,
1474 		.maxlen		= sizeof(sysctl_max_map_count),
1475 		.mode		= 0644,
1476 		.proc_handler	= proc_dointvec_minmax,
1477 		.extra1		= &zero,
1478 	},
1479 #else
1480 	{
1481 		.procname	= "nr_trim_pages",
1482 		.data		= &sysctl_nr_trim_pages,
1483 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1484 		.mode		= 0644,
1485 		.proc_handler	= proc_dointvec_minmax,
1486 		.extra1		= &zero,
1487 	},
1488 #endif
1489 	{
1490 		.procname	= "laptop_mode",
1491 		.data		= &laptop_mode,
1492 		.maxlen		= sizeof(laptop_mode),
1493 		.mode		= 0644,
1494 		.proc_handler	= proc_dointvec_jiffies,
1495 	},
1496 	{
1497 		.procname	= "block_dump",
1498 		.data		= &block_dump,
1499 		.maxlen		= sizeof(block_dump),
1500 		.mode		= 0644,
1501 		.proc_handler	= proc_dointvec,
1502 		.extra1		= &zero,
1503 	},
1504 	{
1505 		.procname	= "vfs_cache_pressure",
1506 		.data		= &sysctl_vfs_cache_pressure,
1507 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1508 		.mode		= 0644,
1509 		.proc_handler	= proc_dointvec,
1510 		.extra1		= &zero,
1511 	},
1512 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1513 	{
1514 		.procname	= "legacy_va_layout",
1515 		.data		= &sysctl_legacy_va_layout,
1516 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1517 		.mode		= 0644,
1518 		.proc_handler	= proc_dointvec,
1519 		.extra1		= &zero,
1520 	},
1521 #endif
1522 #ifdef CONFIG_NUMA
1523 	{
1524 		.procname	= "zone_reclaim_mode",
1525 		.data		= &node_reclaim_mode,
1526 		.maxlen		= sizeof(node_reclaim_mode),
1527 		.mode		= 0644,
1528 		.proc_handler	= proc_dointvec,
1529 		.extra1		= &zero,
1530 	},
1531 	{
1532 		.procname	= "min_unmapped_ratio",
1533 		.data		= &sysctl_min_unmapped_ratio,
1534 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1535 		.mode		= 0644,
1536 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1537 		.extra1		= &zero,
1538 		.extra2		= &one_hundred,
1539 	},
1540 	{
1541 		.procname	= "min_slab_ratio",
1542 		.data		= &sysctl_min_slab_ratio,
1543 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1544 		.mode		= 0644,
1545 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1546 		.extra1		= &zero,
1547 		.extra2		= &one_hundred,
1548 	},
1549 #endif
1550 #ifdef CONFIG_SMP
1551 	{
1552 		.procname	= "stat_interval",
1553 		.data		= &sysctl_stat_interval,
1554 		.maxlen		= sizeof(sysctl_stat_interval),
1555 		.mode		= 0644,
1556 		.proc_handler	= proc_dointvec_jiffies,
1557 	},
1558 	{
1559 		.procname	= "stat_refresh",
1560 		.data		= NULL,
1561 		.maxlen		= 0,
1562 		.mode		= 0600,
1563 		.proc_handler	= vmstat_refresh,
1564 	},
1565 #endif
1566 #ifdef CONFIG_MMU
1567 	{
1568 		.procname	= "mmap_min_addr",
1569 		.data		= &dac_mmap_min_addr,
1570 		.maxlen		= sizeof(unsigned long),
1571 		.mode		= 0644,
1572 		.proc_handler	= mmap_min_addr_handler,
1573 	},
1574 #endif
1575 #ifdef CONFIG_NUMA
1576 	{
1577 		.procname	= "numa_zonelist_order",
1578 		.data		= &numa_zonelist_order,
1579 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1580 		.mode		= 0644,
1581 		.proc_handler	= numa_zonelist_order_handler,
1582 	},
1583 #endif
1584 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1585    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1586 	{
1587 		.procname	= "vdso_enabled",
1588 #ifdef CONFIG_X86_32
1589 		.data		= &vdso32_enabled,
1590 		.maxlen		= sizeof(vdso32_enabled),
1591 #else
1592 		.data		= &vdso_enabled,
1593 		.maxlen		= sizeof(vdso_enabled),
1594 #endif
1595 		.mode		= 0644,
1596 		.proc_handler	= proc_dointvec,
1597 		.extra1		= &zero,
1598 	},
1599 #endif
1600 #ifdef CONFIG_HIGHMEM
1601 	{
1602 		.procname	= "highmem_is_dirtyable",
1603 		.data		= &vm_highmem_is_dirtyable,
1604 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1605 		.mode		= 0644,
1606 		.proc_handler	= proc_dointvec_minmax,
1607 		.extra1		= &zero,
1608 		.extra2		= &one,
1609 	},
1610 #endif
1611 #ifdef CONFIG_MEMORY_FAILURE
1612 	{
1613 		.procname	= "memory_failure_early_kill",
1614 		.data		= &sysctl_memory_failure_early_kill,
1615 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1616 		.mode		= 0644,
1617 		.proc_handler	= proc_dointvec_minmax,
1618 		.extra1		= &zero,
1619 		.extra2		= &one,
1620 	},
1621 	{
1622 		.procname	= "memory_failure_recovery",
1623 		.data		= &sysctl_memory_failure_recovery,
1624 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1625 		.mode		= 0644,
1626 		.proc_handler	= proc_dointvec_minmax,
1627 		.extra1		= &zero,
1628 		.extra2		= &one,
1629 	},
1630 #endif
1631 	{
1632 		.procname	= "user_reserve_kbytes",
1633 		.data		= &sysctl_user_reserve_kbytes,
1634 		.maxlen		= sizeof(sysctl_user_reserve_kbytes),
1635 		.mode		= 0644,
1636 		.proc_handler	= proc_doulongvec_minmax,
1637 	},
1638 	{
1639 		.procname	= "admin_reserve_kbytes",
1640 		.data		= &sysctl_admin_reserve_kbytes,
1641 		.maxlen		= sizeof(sysctl_admin_reserve_kbytes),
1642 		.mode		= 0644,
1643 		.proc_handler	= proc_doulongvec_minmax,
1644 	},
1645 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1646 	{
1647 		.procname	= "mmap_rnd_bits",
1648 		.data		= &mmap_rnd_bits,
1649 		.maxlen		= sizeof(mmap_rnd_bits),
1650 		.mode		= 0600,
1651 		.proc_handler	= proc_dointvec_minmax,
1652 		.extra1		= (void *)&mmap_rnd_bits_min,
1653 		.extra2		= (void *)&mmap_rnd_bits_max,
1654 	},
1655 #endif
1656 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1657 	{
1658 		.procname	= "mmap_rnd_compat_bits",
1659 		.data		= &mmap_rnd_compat_bits,
1660 		.maxlen		= sizeof(mmap_rnd_compat_bits),
1661 		.mode		= 0600,
1662 		.proc_handler	= proc_dointvec_minmax,
1663 		.extra1		= (void *)&mmap_rnd_compat_bits_min,
1664 		.extra2		= (void *)&mmap_rnd_compat_bits_max,
1665 	},
1666 #endif
1667 	{ }
1668 };
1669 
1670 static struct ctl_table fs_table[] = {
1671 	{
1672 		.procname	= "inode-nr",
1673 		.data		= &inodes_stat,
1674 		.maxlen		= 2*sizeof(long),
1675 		.mode		= 0444,
1676 		.proc_handler	= proc_nr_inodes,
1677 	},
1678 	{
1679 		.procname	= "inode-state",
1680 		.data		= &inodes_stat,
1681 		.maxlen		= 7*sizeof(long),
1682 		.mode		= 0444,
1683 		.proc_handler	= proc_nr_inodes,
1684 	},
1685 	{
1686 		.procname	= "file-nr",
1687 		.data		= &files_stat,
1688 		.maxlen		= sizeof(files_stat),
1689 		.mode		= 0444,
1690 		.proc_handler	= proc_nr_files,
1691 	},
1692 	{
1693 		.procname	= "file-max",
1694 		.data		= &files_stat.max_files,
1695 		.maxlen		= sizeof(files_stat.max_files),
1696 		.mode		= 0644,
1697 		.proc_handler	= proc_doulongvec_minmax,
1698 	},
1699 	{
1700 		.procname	= "nr_open",
1701 		.data		= &sysctl_nr_open,
1702 		.maxlen		= sizeof(unsigned int),
1703 		.mode		= 0644,
1704 		.proc_handler	= proc_dointvec_minmax,
1705 		.extra1		= &sysctl_nr_open_min,
1706 		.extra2		= &sysctl_nr_open_max,
1707 	},
1708 	{
1709 		.procname	= "dentry-state",
1710 		.data		= &dentry_stat,
1711 		.maxlen		= 6*sizeof(long),
1712 		.mode		= 0444,
1713 		.proc_handler	= proc_nr_dentry,
1714 	},
1715 	{
1716 		.procname	= "overflowuid",
1717 		.data		= &fs_overflowuid,
1718 		.maxlen		= sizeof(int),
1719 		.mode		= 0644,
1720 		.proc_handler	= proc_dointvec_minmax,
1721 		.extra1		= &minolduid,
1722 		.extra2		= &maxolduid,
1723 	},
1724 	{
1725 		.procname	= "overflowgid",
1726 		.data		= &fs_overflowgid,
1727 		.maxlen		= sizeof(int),
1728 		.mode		= 0644,
1729 		.proc_handler	= proc_dointvec_minmax,
1730 		.extra1		= &minolduid,
1731 		.extra2		= &maxolduid,
1732 	},
1733 #ifdef CONFIG_FILE_LOCKING
1734 	{
1735 		.procname	= "leases-enable",
1736 		.data		= &leases_enable,
1737 		.maxlen		= sizeof(int),
1738 		.mode		= 0644,
1739 		.proc_handler	= proc_dointvec,
1740 	},
1741 #endif
1742 #ifdef CONFIG_DNOTIFY
1743 	{
1744 		.procname	= "dir-notify-enable",
1745 		.data		= &dir_notify_enable,
1746 		.maxlen		= sizeof(int),
1747 		.mode		= 0644,
1748 		.proc_handler	= proc_dointvec,
1749 	},
1750 #endif
1751 #ifdef CONFIG_MMU
1752 #ifdef CONFIG_FILE_LOCKING
1753 	{
1754 		.procname	= "lease-break-time",
1755 		.data		= &lease_break_time,
1756 		.maxlen		= sizeof(int),
1757 		.mode		= 0644,
1758 		.proc_handler	= proc_dointvec,
1759 	},
1760 #endif
1761 #ifdef CONFIG_AIO
1762 	{
1763 		.procname	= "aio-nr",
1764 		.data		= &aio_nr,
1765 		.maxlen		= sizeof(aio_nr),
1766 		.mode		= 0444,
1767 		.proc_handler	= proc_doulongvec_minmax,
1768 	},
1769 	{
1770 		.procname	= "aio-max-nr",
1771 		.data		= &aio_max_nr,
1772 		.maxlen		= sizeof(aio_max_nr),
1773 		.mode		= 0644,
1774 		.proc_handler	= proc_doulongvec_minmax,
1775 	},
1776 #endif /* CONFIG_AIO */
1777 #ifdef CONFIG_INOTIFY_USER
1778 	{
1779 		.procname	= "inotify",
1780 		.mode		= 0555,
1781 		.child		= inotify_table,
1782 	},
1783 #endif
1784 #ifdef CONFIG_EPOLL
1785 	{
1786 		.procname	= "epoll",
1787 		.mode		= 0555,
1788 		.child		= epoll_table,
1789 	},
1790 #endif
1791 #endif
1792 	{
1793 		.procname	= "protected_symlinks",
1794 		.data		= &sysctl_protected_symlinks,
1795 		.maxlen		= sizeof(int),
1796 		.mode		= 0600,
1797 		.proc_handler	= proc_dointvec_minmax,
1798 		.extra1		= &zero,
1799 		.extra2		= &one,
1800 	},
1801 	{
1802 		.procname	= "protected_hardlinks",
1803 		.data		= &sysctl_protected_hardlinks,
1804 		.maxlen		= sizeof(int),
1805 		.mode		= 0600,
1806 		.proc_handler	= proc_dointvec_minmax,
1807 		.extra1		= &zero,
1808 		.extra2		= &one,
1809 	},
1810 	{
1811 		.procname	= "protected_fifos",
1812 		.data		= &sysctl_protected_fifos,
1813 		.maxlen		= sizeof(int),
1814 		.mode		= 0600,
1815 		.proc_handler	= proc_dointvec_minmax,
1816 		.extra1		= &zero,
1817 		.extra2		= &two,
1818 	},
1819 	{
1820 		.procname	= "protected_regular",
1821 		.data		= &sysctl_protected_regular,
1822 		.maxlen		= sizeof(int),
1823 		.mode		= 0600,
1824 		.proc_handler	= proc_dointvec_minmax,
1825 		.extra1		= &zero,
1826 		.extra2		= &two,
1827 	},
1828 	{
1829 		.procname	= "suid_dumpable",
1830 		.data		= &suid_dumpable,
1831 		.maxlen		= sizeof(int),
1832 		.mode		= 0644,
1833 		.proc_handler	= proc_dointvec_minmax_coredump,
1834 		.extra1		= &zero,
1835 		.extra2		= &two,
1836 	},
1837 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1838 	{
1839 		.procname	= "binfmt_misc",
1840 		.mode		= 0555,
1841 		.child		= sysctl_mount_point,
1842 	},
1843 #endif
1844 	{
1845 		.procname	= "pipe-max-size",
1846 		.data		= &pipe_max_size,
1847 		.maxlen		= sizeof(pipe_max_size),
1848 		.mode		= 0644,
1849 		.proc_handler	= proc_dopipe_max_size,
1850 	},
1851 	{
1852 		.procname	= "pipe-user-pages-hard",
1853 		.data		= &pipe_user_pages_hard,
1854 		.maxlen		= sizeof(pipe_user_pages_hard),
1855 		.mode		= 0644,
1856 		.proc_handler	= proc_doulongvec_minmax,
1857 	},
1858 	{
1859 		.procname	= "pipe-user-pages-soft",
1860 		.data		= &pipe_user_pages_soft,
1861 		.maxlen		= sizeof(pipe_user_pages_soft),
1862 		.mode		= 0644,
1863 		.proc_handler	= proc_doulongvec_minmax,
1864 	},
1865 	{
1866 		.procname	= "mount-max",
1867 		.data		= &sysctl_mount_max,
1868 		.maxlen		= sizeof(unsigned int),
1869 		.mode		= 0644,
1870 		.proc_handler	= proc_dointvec_minmax,
1871 		.extra1		= &one,
1872 	},
1873 	{ }
1874 };
1875 
1876 static struct ctl_table debug_table[] = {
1877 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1878 	{
1879 		.procname	= "exception-trace",
1880 		.data		= &show_unhandled_signals,
1881 		.maxlen		= sizeof(int),
1882 		.mode		= 0644,
1883 		.proc_handler	= proc_dointvec
1884 	},
1885 #endif
1886 #if defined(CONFIG_OPTPROBES)
1887 	{
1888 		.procname	= "kprobes-optimization",
1889 		.data		= &sysctl_kprobes_optimization,
1890 		.maxlen		= sizeof(int),
1891 		.mode		= 0644,
1892 		.proc_handler	= proc_kprobes_optimization_handler,
1893 		.extra1		= &zero,
1894 		.extra2		= &one,
1895 	},
1896 #endif
1897 	{ }
1898 };
1899 
1900 static struct ctl_table dev_table[] = {
1901 	{ }
1902 };
1903 
1904 int __init sysctl_init(void)
1905 {
1906 	struct ctl_table_header *hdr;
1907 
1908 	hdr = register_sysctl_table(sysctl_base_table);
1909 	kmemleak_not_leak(hdr);
1910 	return 0;
1911 }
1912 
1913 #endif /* CONFIG_SYSCTL */
1914 
1915 /*
1916  * /proc/sys support
1917  */
1918 
1919 #ifdef CONFIG_PROC_SYSCTL
1920 
1921 static int _proc_do_string(char *data, int maxlen, int write,
1922 			   char __user *buffer,
1923 			   size_t *lenp, loff_t *ppos)
1924 {
1925 	size_t len;
1926 	char __user *p;
1927 	char c;
1928 
1929 	if (!data || !maxlen || !*lenp) {
1930 		*lenp = 0;
1931 		return 0;
1932 	}
1933 
1934 	if (write) {
1935 		if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1936 			/* Only continue writes not past the end of buffer. */
1937 			len = strlen(data);
1938 			if (len > maxlen - 1)
1939 				len = maxlen - 1;
1940 
1941 			if (*ppos > len)
1942 				return 0;
1943 			len = *ppos;
1944 		} else {
1945 			/* Start writing from beginning of buffer. */
1946 			len = 0;
1947 		}
1948 
1949 		*ppos += *lenp;
1950 		p = buffer;
1951 		while ((p - buffer) < *lenp && len < maxlen - 1) {
1952 			if (get_user(c, p++))
1953 				return -EFAULT;
1954 			if (c == 0 || c == '\n')
1955 				break;
1956 			data[len++] = c;
1957 		}
1958 		data[len] = 0;
1959 	} else {
1960 		len = strlen(data);
1961 		if (len > maxlen)
1962 			len = maxlen;
1963 
1964 		if (*ppos > len) {
1965 			*lenp = 0;
1966 			return 0;
1967 		}
1968 
1969 		data += *ppos;
1970 		len  -= *ppos;
1971 
1972 		if (len > *lenp)
1973 			len = *lenp;
1974 		if (len)
1975 			if (copy_to_user(buffer, data, len))
1976 				return -EFAULT;
1977 		if (len < *lenp) {
1978 			if (put_user('\n', buffer + len))
1979 				return -EFAULT;
1980 			len++;
1981 		}
1982 		*lenp = len;
1983 		*ppos += len;
1984 	}
1985 	return 0;
1986 }
1987 
1988 static void warn_sysctl_write(struct ctl_table *table)
1989 {
1990 	pr_warn_once("%s wrote to %s when file position was not 0!\n"
1991 		"This will not be supported in the future. To silence this\n"
1992 		"warning, set kernel.sysctl_writes_strict = -1\n",
1993 		current->comm, table->procname);
1994 }
1995 
1996 /**
1997  * proc_first_pos_non_zero_ignore - check if first position is allowed
1998  * @ppos: file position
1999  * @table: the sysctl table
2000  *
2001  * Returns true if the first position is non-zero and the sysctl_writes_strict
2002  * mode indicates this is not allowed for numeric input types. String proc
2003  * handlers can ignore the return value.
2004  */
2005 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
2006 					   struct ctl_table *table)
2007 {
2008 	if (!*ppos)
2009 		return false;
2010 
2011 	switch (sysctl_writes_strict) {
2012 	case SYSCTL_WRITES_STRICT:
2013 		return true;
2014 	case SYSCTL_WRITES_WARN:
2015 		warn_sysctl_write(table);
2016 		return false;
2017 	default:
2018 		return false;
2019 	}
2020 }
2021 
2022 /**
2023  * proc_dostring - read a string sysctl
2024  * @table: the sysctl table
2025  * @write: %TRUE if this is a write to the sysctl file
2026  * @buffer: the user buffer
2027  * @lenp: the size of the user buffer
2028  * @ppos: file position
2029  *
2030  * Reads/writes a string from/to the user buffer. If the kernel
2031  * buffer provided is not large enough to hold the string, the
2032  * string is truncated. The copied string is %NULL-terminated.
2033  * If the string is being read by the user process, it is copied
2034  * and a newline '\n' is added. It is truncated if the buffer is
2035  * not large enough.
2036  *
2037  * Returns 0 on success.
2038  */
2039 int proc_dostring(struct ctl_table *table, int write,
2040 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2041 {
2042 	if (write)
2043 		proc_first_pos_non_zero_ignore(ppos, table);
2044 
2045 	return _proc_do_string((char *)(table->data), table->maxlen, write,
2046 			       (char __user *)buffer, lenp, ppos);
2047 }
2048 
2049 static size_t proc_skip_spaces(char **buf)
2050 {
2051 	size_t ret;
2052 	char *tmp = skip_spaces(*buf);
2053 	ret = tmp - *buf;
2054 	*buf = tmp;
2055 	return ret;
2056 }
2057 
2058 static void proc_skip_char(char **buf, size_t *size, const char v)
2059 {
2060 	while (*size) {
2061 		if (**buf != v)
2062 			break;
2063 		(*size)--;
2064 		(*buf)++;
2065 	}
2066 }
2067 
2068 #define TMPBUFLEN 22
2069 /**
2070  * proc_get_long - reads an ASCII formatted integer from a user buffer
2071  *
2072  * @buf: a kernel buffer
2073  * @size: size of the kernel buffer
2074  * @val: this is where the number will be stored
2075  * @neg: set to %TRUE if number is negative
2076  * @perm_tr: a vector which contains the allowed trailers
2077  * @perm_tr_len: size of the perm_tr vector
2078  * @tr: pointer to store the trailer character
2079  *
2080  * In case of success %0 is returned and @buf and @size are updated with
2081  * the amount of bytes read. If @tr is non-NULL and a trailing
2082  * character exists (size is non-zero after returning from this
2083  * function), @tr is updated with the trailing character.
2084  */
2085 static int proc_get_long(char **buf, size_t *size,
2086 			  unsigned long *val, bool *neg,
2087 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
2088 {
2089 	int len;
2090 	char *p, tmp[TMPBUFLEN];
2091 
2092 	if (!*size)
2093 		return -EINVAL;
2094 
2095 	len = *size;
2096 	if (len > TMPBUFLEN - 1)
2097 		len = TMPBUFLEN - 1;
2098 
2099 	memcpy(tmp, *buf, len);
2100 
2101 	tmp[len] = 0;
2102 	p = tmp;
2103 	if (*p == '-' && *size > 1) {
2104 		*neg = true;
2105 		p++;
2106 	} else
2107 		*neg = false;
2108 	if (!isdigit(*p))
2109 		return -EINVAL;
2110 
2111 	*val = simple_strtoul(p, &p, 0);
2112 
2113 	len = p - tmp;
2114 
2115 	/* We don't know if the next char is whitespace thus we may accept
2116 	 * invalid integers (e.g. 1234...a) or two integers instead of one
2117 	 * (e.g. 123...1). So lets not allow such large numbers. */
2118 	if (len == TMPBUFLEN - 1)
2119 		return -EINVAL;
2120 
2121 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2122 		return -EINVAL;
2123 
2124 	if (tr && (len < *size))
2125 		*tr = *p;
2126 
2127 	*buf += len;
2128 	*size -= len;
2129 
2130 	return 0;
2131 }
2132 
2133 /**
2134  * proc_put_long - converts an integer to a decimal ASCII formatted string
2135  *
2136  * @buf: the user buffer
2137  * @size: the size of the user buffer
2138  * @val: the integer to be converted
2139  * @neg: sign of the number, %TRUE for negative
2140  *
2141  * In case of success %0 is returned and @buf and @size are updated with
2142  * the amount of bytes written.
2143  */
2144 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2145 			  bool neg)
2146 {
2147 	int len;
2148 	char tmp[TMPBUFLEN], *p = tmp;
2149 
2150 	sprintf(p, "%s%lu", neg ? "-" : "", val);
2151 	len = strlen(tmp);
2152 	if (len > *size)
2153 		len = *size;
2154 	if (copy_to_user(*buf, tmp, len))
2155 		return -EFAULT;
2156 	*size -= len;
2157 	*buf += len;
2158 	return 0;
2159 }
2160 #undef TMPBUFLEN
2161 
2162 static int proc_put_char(void __user **buf, size_t *size, char c)
2163 {
2164 	if (*size) {
2165 		char __user **buffer = (char __user **)buf;
2166 		if (put_user(c, *buffer))
2167 			return -EFAULT;
2168 		(*size)--, (*buffer)++;
2169 		*buf = *buffer;
2170 	}
2171 	return 0;
2172 }
2173 
2174 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2175 				 int *valp,
2176 				 int write, void *data)
2177 {
2178 	if (write) {
2179 		if (*negp) {
2180 			if (*lvalp > (unsigned long) INT_MAX + 1)
2181 				return -EINVAL;
2182 			*valp = -*lvalp;
2183 		} else {
2184 			if (*lvalp > (unsigned long) INT_MAX)
2185 				return -EINVAL;
2186 			*valp = *lvalp;
2187 		}
2188 	} else {
2189 		int val = *valp;
2190 		if (val < 0) {
2191 			*negp = true;
2192 			*lvalp = -(unsigned long)val;
2193 		} else {
2194 			*negp = false;
2195 			*lvalp = (unsigned long)val;
2196 		}
2197 	}
2198 	return 0;
2199 }
2200 
2201 static int do_proc_douintvec_conv(unsigned long *lvalp,
2202 				  unsigned int *valp,
2203 				  int write, void *data)
2204 {
2205 	if (write) {
2206 		if (*lvalp > UINT_MAX)
2207 			return -EINVAL;
2208 		*valp = *lvalp;
2209 	} else {
2210 		unsigned int val = *valp;
2211 		*lvalp = (unsigned long)val;
2212 	}
2213 	return 0;
2214 }
2215 
2216 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2217 
2218 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2219 		  int write, void __user *buffer,
2220 		  size_t *lenp, loff_t *ppos,
2221 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2222 			      int write, void *data),
2223 		  void *data)
2224 {
2225 	int *i, vleft, first = 1, err = 0;
2226 	size_t left;
2227 	char *kbuf = NULL, *p;
2228 
2229 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2230 		*lenp = 0;
2231 		return 0;
2232 	}
2233 
2234 	i = (int *) tbl_data;
2235 	vleft = table->maxlen / sizeof(*i);
2236 	left = *lenp;
2237 
2238 	if (!conv)
2239 		conv = do_proc_dointvec_conv;
2240 
2241 	if (write) {
2242 		if (proc_first_pos_non_zero_ignore(ppos, table))
2243 			goto out;
2244 
2245 		if (left > PAGE_SIZE - 1)
2246 			left = PAGE_SIZE - 1;
2247 		p = kbuf = memdup_user_nul(buffer, left);
2248 		if (IS_ERR(kbuf))
2249 			return PTR_ERR(kbuf);
2250 	}
2251 
2252 	for (; left && vleft--; i++, first=0) {
2253 		unsigned long lval;
2254 		bool neg;
2255 
2256 		if (write) {
2257 			left -= proc_skip_spaces(&p);
2258 
2259 			if (!left)
2260 				break;
2261 			err = proc_get_long(&p, &left, &lval, &neg,
2262 					     proc_wspace_sep,
2263 					     sizeof(proc_wspace_sep), NULL);
2264 			if (err)
2265 				break;
2266 			if (conv(&neg, &lval, i, 1, data)) {
2267 				err = -EINVAL;
2268 				break;
2269 			}
2270 		} else {
2271 			if (conv(&neg, &lval, i, 0, data)) {
2272 				err = -EINVAL;
2273 				break;
2274 			}
2275 			if (!first)
2276 				err = proc_put_char(&buffer, &left, '\t');
2277 			if (err)
2278 				break;
2279 			err = proc_put_long(&buffer, &left, lval, neg);
2280 			if (err)
2281 				break;
2282 		}
2283 	}
2284 
2285 	if (!write && !first && left && !err)
2286 		err = proc_put_char(&buffer, &left, '\n');
2287 	if (write && !err && left)
2288 		left -= proc_skip_spaces(&p);
2289 	if (write) {
2290 		kfree(kbuf);
2291 		if (first)
2292 			return err ? : -EINVAL;
2293 	}
2294 	*lenp -= left;
2295 out:
2296 	*ppos += *lenp;
2297 	return err;
2298 }
2299 
2300 static int do_proc_dointvec(struct ctl_table *table, int write,
2301 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2302 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2303 			      int write, void *data),
2304 		  void *data)
2305 {
2306 	return __do_proc_dointvec(table->data, table, write,
2307 			buffer, lenp, ppos, conv, data);
2308 }
2309 
2310 static int do_proc_douintvec_w(unsigned int *tbl_data,
2311 			       struct ctl_table *table,
2312 			       void __user *buffer,
2313 			       size_t *lenp, loff_t *ppos,
2314 			       int (*conv)(unsigned long *lvalp,
2315 					   unsigned int *valp,
2316 					   int write, void *data),
2317 			       void *data)
2318 {
2319 	unsigned long lval;
2320 	int err = 0;
2321 	size_t left;
2322 	bool neg;
2323 	char *kbuf = NULL, *p;
2324 
2325 	left = *lenp;
2326 
2327 	if (proc_first_pos_non_zero_ignore(ppos, table))
2328 		goto bail_early;
2329 
2330 	if (left > PAGE_SIZE - 1)
2331 		left = PAGE_SIZE - 1;
2332 
2333 	p = kbuf = memdup_user_nul(buffer, left);
2334 	if (IS_ERR(kbuf))
2335 		return -EINVAL;
2336 
2337 	left -= proc_skip_spaces(&p);
2338 	if (!left) {
2339 		err = -EINVAL;
2340 		goto out_free;
2341 	}
2342 
2343 	err = proc_get_long(&p, &left, &lval, &neg,
2344 			     proc_wspace_sep,
2345 			     sizeof(proc_wspace_sep), NULL);
2346 	if (err || neg) {
2347 		err = -EINVAL;
2348 		goto out_free;
2349 	}
2350 
2351 	if (conv(&lval, tbl_data, 1, data)) {
2352 		err = -EINVAL;
2353 		goto out_free;
2354 	}
2355 
2356 	if (!err && left)
2357 		left -= proc_skip_spaces(&p);
2358 
2359 out_free:
2360 	kfree(kbuf);
2361 	if (err)
2362 		return -EINVAL;
2363 
2364 	return 0;
2365 
2366 	/* This is in keeping with old __do_proc_dointvec() */
2367 bail_early:
2368 	*ppos += *lenp;
2369 	return err;
2370 }
2371 
2372 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2373 			       size_t *lenp, loff_t *ppos,
2374 			       int (*conv)(unsigned long *lvalp,
2375 					   unsigned int *valp,
2376 					   int write, void *data),
2377 			       void *data)
2378 {
2379 	unsigned long lval;
2380 	int err = 0;
2381 	size_t left;
2382 
2383 	left = *lenp;
2384 
2385 	if (conv(&lval, tbl_data, 0, data)) {
2386 		err = -EINVAL;
2387 		goto out;
2388 	}
2389 
2390 	err = proc_put_long(&buffer, &left, lval, false);
2391 	if (err || !left)
2392 		goto out;
2393 
2394 	err = proc_put_char(&buffer, &left, '\n');
2395 
2396 out:
2397 	*lenp -= left;
2398 	*ppos += *lenp;
2399 
2400 	return err;
2401 }
2402 
2403 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2404 			       int write, void __user *buffer,
2405 			       size_t *lenp, loff_t *ppos,
2406 			       int (*conv)(unsigned long *lvalp,
2407 					   unsigned int *valp,
2408 					   int write, void *data),
2409 			       void *data)
2410 {
2411 	unsigned int *i, vleft;
2412 
2413 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2414 		*lenp = 0;
2415 		return 0;
2416 	}
2417 
2418 	i = (unsigned int *) tbl_data;
2419 	vleft = table->maxlen / sizeof(*i);
2420 
2421 	/*
2422 	 * Arrays are not supported, keep this simple. *Do not* add
2423 	 * support for them.
2424 	 */
2425 	if (vleft != 1) {
2426 		*lenp = 0;
2427 		return -EINVAL;
2428 	}
2429 
2430 	if (!conv)
2431 		conv = do_proc_douintvec_conv;
2432 
2433 	if (write)
2434 		return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2435 					   conv, data);
2436 	return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2437 }
2438 
2439 static int do_proc_douintvec(struct ctl_table *table, int write,
2440 			     void __user *buffer, size_t *lenp, loff_t *ppos,
2441 			     int (*conv)(unsigned long *lvalp,
2442 					 unsigned int *valp,
2443 					 int write, void *data),
2444 			     void *data)
2445 {
2446 	return __do_proc_douintvec(table->data, table, write,
2447 				   buffer, lenp, ppos, conv, data);
2448 }
2449 
2450 /**
2451  * proc_dointvec - read a vector of integers
2452  * @table: the sysctl table
2453  * @write: %TRUE if this is a write to the sysctl file
2454  * @buffer: the user buffer
2455  * @lenp: the size of the user buffer
2456  * @ppos: file position
2457  *
2458  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2459  * values from/to the user buffer, treated as an ASCII string.
2460  *
2461  * Returns 0 on success.
2462  */
2463 int proc_dointvec(struct ctl_table *table, int write,
2464 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2465 {
2466 	return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2467 }
2468 
2469 /**
2470  * proc_douintvec - read a vector of unsigned integers
2471  * @table: the sysctl table
2472  * @write: %TRUE if this is a write to the sysctl file
2473  * @buffer: the user buffer
2474  * @lenp: the size of the user buffer
2475  * @ppos: file position
2476  *
2477  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2478  * values from/to the user buffer, treated as an ASCII string.
2479  *
2480  * Returns 0 on success.
2481  */
2482 int proc_douintvec(struct ctl_table *table, int write,
2483 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2484 {
2485 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
2486 				 do_proc_douintvec_conv, NULL);
2487 }
2488 
2489 /*
2490  * Taint values can only be increased
2491  * This means we can safely use a temporary.
2492  */
2493 static int proc_taint(struct ctl_table *table, int write,
2494 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2495 {
2496 	struct ctl_table t;
2497 	unsigned long tmptaint = get_taint();
2498 	int err;
2499 
2500 	if (write && !capable(CAP_SYS_ADMIN))
2501 		return -EPERM;
2502 
2503 	t = *table;
2504 	t.data = &tmptaint;
2505 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2506 	if (err < 0)
2507 		return err;
2508 
2509 	if (write) {
2510 		/*
2511 		 * Poor man's atomic or. Not worth adding a primitive
2512 		 * to everyone's atomic.h for this
2513 		 */
2514 		int i;
2515 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2516 			if ((tmptaint >> i) & 1)
2517 				add_taint(i, LOCKDEP_STILL_OK);
2518 		}
2519 	}
2520 
2521 	return err;
2522 }
2523 
2524 #ifdef CONFIG_PRINTK
2525 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2526 				void __user *buffer, size_t *lenp, loff_t *ppos)
2527 {
2528 	if (write && !capable(CAP_SYS_ADMIN))
2529 		return -EPERM;
2530 
2531 	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2532 }
2533 #endif
2534 
2535 /**
2536  * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
2537  * @min: pointer to minimum allowable value
2538  * @max: pointer to maximum allowable value
2539  *
2540  * The do_proc_dointvec_minmax_conv_param structure provides the
2541  * minimum and maximum values for doing range checking for those sysctl
2542  * parameters that use the proc_dointvec_minmax() handler.
2543  */
2544 struct do_proc_dointvec_minmax_conv_param {
2545 	int *min;
2546 	int *max;
2547 };
2548 
2549 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2550 					int *valp,
2551 					int write, void *data)
2552 {
2553 	struct do_proc_dointvec_minmax_conv_param *param = data;
2554 	if (write) {
2555 		int val = *negp ? -*lvalp : *lvalp;
2556 		if ((param->min && *param->min > val) ||
2557 		    (param->max && *param->max < val))
2558 			return -EINVAL;
2559 		*valp = val;
2560 	} else {
2561 		int val = *valp;
2562 		if (val < 0) {
2563 			*negp = true;
2564 			*lvalp = -(unsigned long)val;
2565 		} else {
2566 			*negp = false;
2567 			*lvalp = (unsigned long)val;
2568 		}
2569 	}
2570 	return 0;
2571 }
2572 
2573 /**
2574  * proc_dointvec_minmax - read a vector of integers with min/max values
2575  * @table: the sysctl table
2576  * @write: %TRUE if this is a write to the sysctl file
2577  * @buffer: the user buffer
2578  * @lenp: the size of the user buffer
2579  * @ppos: file position
2580  *
2581  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2582  * values from/to the user buffer, treated as an ASCII string.
2583  *
2584  * This routine will ensure the values are within the range specified by
2585  * table->extra1 (min) and table->extra2 (max).
2586  *
2587  * Returns 0 on success or -EINVAL on write when the range check fails.
2588  */
2589 int proc_dointvec_minmax(struct ctl_table *table, int write,
2590 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2591 {
2592 	struct do_proc_dointvec_minmax_conv_param param = {
2593 		.min = (int *) table->extra1,
2594 		.max = (int *) table->extra2,
2595 	};
2596 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2597 				do_proc_dointvec_minmax_conv, &param);
2598 }
2599 
2600 /**
2601  * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
2602  * @min: pointer to minimum allowable value
2603  * @max: pointer to maximum allowable value
2604  *
2605  * The do_proc_douintvec_minmax_conv_param structure provides the
2606  * minimum and maximum values for doing range checking for those sysctl
2607  * parameters that use the proc_douintvec_minmax() handler.
2608  */
2609 struct do_proc_douintvec_minmax_conv_param {
2610 	unsigned int *min;
2611 	unsigned int *max;
2612 };
2613 
2614 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2615 					 unsigned int *valp,
2616 					 int write, void *data)
2617 {
2618 	struct do_proc_douintvec_minmax_conv_param *param = data;
2619 
2620 	if (write) {
2621 		unsigned int val = *lvalp;
2622 
2623 		if (*lvalp > UINT_MAX)
2624 			return -EINVAL;
2625 
2626 		if ((param->min && *param->min > val) ||
2627 		    (param->max && *param->max < val))
2628 			return -ERANGE;
2629 
2630 		*valp = val;
2631 	} else {
2632 		unsigned int val = *valp;
2633 		*lvalp = (unsigned long) val;
2634 	}
2635 
2636 	return 0;
2637 }
2638 
2639 /**
2640  * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2641  * @table: the sysctl table
2642  * @write: %TRUE if this is a write to the sysctl file
2643  * @buffer: the user buffer
2644  * @lenp: the size of the user buffer
2645  * @ppos: file position
2646  *
2647  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2648  * values from/to the user buffer, treated as an ASCII string. Negative
2649  * strings are not allowed.
2650  *
2651  * This routine will ensure the values are within the range specified by
2652  * table->extra1 (min) and table->extra2 (max). There is a final sanity
2653  * check for UINT_MAX to avoid having to support wrap around uses from
2654  * userspace.
2655  *
2656  * Returns 0 on success or -ERANGE on write when the range check fails.
2657  */
2658 int proc_douintvec_minmax(struct ctl_table *table, int write,
2659 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2660 {
2661 	struct do_proc_douintvec_minmax_conv_param param = {
2662 		.min = (unsigned int *) table->extra1,
2663 		.max = (unsigned int *) table->extra2,
2664 	};
2665 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
2666 				 do_proc_douintvec_minmax_conv, &param);
2667 }
2668 
2669 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
2670 					unsigned int *valp,
2671 					int write, void *data)
2672 {
2673 	if (write) {
2674 		unsigned int val;
2675 
2676 		val = round_pipe_size(*lvalp);
2677 		if (val == 0)
2678 			return -EINVAL;
2679 
2680 		*valp = val;
2681 	} else {
2682 		unsigned int val = *valp;
2683 		*lvalp = (unsigned long) val;
2684 	}
2685 
2686 	return 0;
2687 }
2688 
2689 static int proc_dopipe_max_size(struct ctl_table *table, int write,
2690 				void __user *buffer, size_t *lenp, loff_t *ppos)
2691 {
2692 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
2693 				 do_proc_dopipe_max_size_conv, NULL);
2694 }
2695 
2696 static void validate_coredump_safety(void)
2697 {
2698 #ifdef CONFIG_COREDUMP
2699 	if (suid_dumpable == SUID_DUMP_ROOT &&
2700 	    core_pattern[0] != '/' && core_pattern[0] != '|') {
2701 		printk(KERN_WARNING
2702 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2703 "Pipe handler or fully qualified core dump path required.\n"
2704 "Set kernel.core_pattern before fs.suid_dumpable.\n"
2705 		);
2706 	}
2707 #endif
2708 }
2709 
2710 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2711 		void __user *buffer, size_t *lenp, loff_t *ppos)
2712 {
2713 	int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2714 	if (!error)
2715 		validate_coredump_safety();
2716 	return error;
2717 }
2718 
2719 #ifdef CONFIG_COREDUMP
2720 static int proc_dostring_coredump(struct ctl_table *table, int write,
2721 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2722 {
2723 	int error = proc_dostring(table, write, buffer, lenp, ppos);
2724 	if (!error)
2725 		validate_coredump_safety();
2726 	return error;
2727 }
2728 #endif
2729 
2730 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2731 				     void __user *buffer,
2732 				     size_t *lenp, loff_t *ppos,
2733 				     unsigned long convmul,
2734 				     unsigned long convdiv)
2735 {
2736 	unsigned long *i, *min, *max;
2737 	int vleft, first = 1, err = 0;
2738 	size_t left;
2739 	char *kbuf = NULL, *p;
2740 
2741 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2742 		*lenp = 0;
2743 		return 0;
2744 	}
2745 
2746 	i = (unsigned long *) data;
2747 	min = (unsigned long *) table->extra1;
2748 	max = (unsigned long *) table->extra2;
2749 	vleft = table->maxlen / sizeof(unsigned long);
2750 	left = *lenp;
2751 
2752 	if (write) {
2753 		if (proc_first_pos_non_zero_ignore(ppos, table))
2754 			goto out;
2755 
2756 		if (left > PAGE_SIZE - 1)
2757 			left = PAGE_SIZE - 1;
2758 		p = kbuf = memdup_user_nul(buffer, left);
2759 		if (IS_ERR(kbuf))
2760 			return PTR_ERR(kbuf);
2761 	}
2762 
2763 	for (; left && vleft--; i++, first = 0) {
2764 		unsigned long val;
2765 
2766 		if (write) {
2767 			bool neg;
2768 
2769 			left -= proc_skip_spaces(&p);
2770 
2771 			err = proc_get_long(&p, &left, &val, &neg,
2772 					     proc_wspace_sep,
2773 					     sizeof(proc_wspace_sep), NULL);
2774 			if (err)
2775 				break;
2776 			if (neg)
2777 				continue;
2778 			val = convmul * val / convdiv;
2779 			if ((min && val < *min) || (max && val > *max))
2780 				continue;
2781 			*i = val;
2782 		} else {
2783 			val = convdiv * (*i) / convmul;
2784 			if (!first) {
2785 				err = proc_put_char(&buffer, &left, '\t');
2786 				if (err)
2787 					break;
2788 			}
2789 			err = proc_put_long(&buffer, &left, val, false);
2790 			if (err)
2791 				break;
2792 		}
2793 	}
2794 
2795 	if (!write && !first && left && !err)
2796 		err = proc_put_char(&buffer, &left, '\n');
2797 	if (write && !err)
2798 		left -= proc_skip_spaces(&p);
2799 	if (write) {
2800 		kfree(kbuf);
2801 		if (first)
2802 			return err ? : -EINVAL;
2803 	}
2804 	*lenp -= left;
2805 out:
2806 	*ppos += *lenp;
2807 	return err;
2808 }
2809 
2810 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2811 				     void __user *buffer,
2812 				     size_t *lenp, loff_t *ppos,
2813 				     unsigned long convmul,
2814 				     unsigned long convdiv)
2815 {
2816 	return __do_proc_doulongvec_minmax(table->data, table, write,
2817 			buffer, lenp, ppos, convmul, convdiv);
2818 }
2819 
2820 /**
2821  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2822  * @table: the sysctl table
2823  * @write: %TRUE if this is a write to the sysctl file
2824  * @buffer: the user buffer
2825  * @lenp: the size of the user buffer
2826  * @ppos: file position
2827  *
2828  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2829  * values from/to the user buffer, treated as an ASCII string.
2830  *
2831  * This routine will ensure the values are within the range specified by
2832  * table->extra1 (min) and table->extra2 (max).
2833  *
2834  * Returns 0 on success.
2835  */
2836 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2837 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2838 {
2839     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2840 }
2841 
2842 /**
2843  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2844  * @table: the sysctl table
2845  * @write: %TRUE if this is a write to the sysctl file
2846  * @buffer: the user buffer
2847  * @lenp: the size of the user buffer
2848  * @ppos: file position
2849  *
2850  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2851  * values from/to the user buffer, treated as an ASCII string. The values
2852  * are treated as milliseconds, and converted to jiffies when they are stored.
2853  *
2854  * This routine will ensure the values are within the range specified by
2855  * table->extra1 (min) and table->extra2 (max).
2856  *
2857  * Returns 0 on success.
2858  */
2859 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2860 				      void __user *buffer,
2861 				      size_t *lenp, loff_t *ppos)
2862 {
2863     return do_proc_doulongvec_minmax(table, write, buffer,
2864 				     lenp, ppos, HZ, 1000l);
2865 }
2866 
2867 
2868 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2869 					 int *valp,
2870 					 int write, void *data)
2871 {
2872 	if (write) {
2873 		if (*lvalp > INT_MAX / HZ)
2874 			return 1;
2875 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2876 	} else {
2877 		int val = *valp;
2878 		unsigned long lval;
2879 		if (val < 0) {
2880 			*negp = true;
2881 			lval = -(unsigned long)val;
2882 		} else {
2883 			*negp = false;
2884 			lval = (unsigned long)val;
2885 		}
2886 		*lvalp = lval / HZ;
2887 	}
2888 	return 0;
2889 }
2890 
2891 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2892 						int *valp,
2893 						int write, void *data)
2894 {
2895 	if (write) {
2896 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2897 			return 1;
2898 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2899 	} else {
2900 		int val = *valp;
2901 		unsigned long lval;
2902 		if (val < 0) {
2903 			*negp = true;
2904 			lval = -(unsigned long)val;
2905 		} else {
2906 			*negp = false;
2907 			lval = (unsigned long)val;
2908 		}
2909 		*lvalp = jiffies_to_clock_t(lval);
2910 	}
2911 	return 0;
2912 }
2913 
2914 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2915 					    int *valp,
2916 					    int write, void *data)
2917 {
2918 	if (write) {
2919 		unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2920 
2921 		if (jif > INT_MAX)
2922 			return 1;
2923 		*valp = (int)jif;
2924 	} else {
2925 		int val = *valp;
2926 		unsigned long lval;
2927 		if (val < 0) {
2928 			*negp = true;
2929 			lval = -(unsigned long)val;
2930 		} else {
2931 			*negp = false;
2932 			lval = (unsigned long)val;
2933 		}
2934 		*lvalp = jiffies_to_msecs(lval);
2935 	}
2936 	return 0;
2937 }
2938 
2939 /**
2940  * proc_dointvec_jiffies - read a vector of integers as seconds
2941  * @table: the sysctl table
2942  * @write: %TRUE if this is a write to the sysctl file
2943  * @buffer: the user buffer
2944  * @lenp: the size of the user buffer
2945  * @ppos: file position
2946  *
2947  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2948  * values from/to the user buffer, treated as an ASCII string.
2949  * The values read are assumed to be in seconds, and are converted into
2950  * jiffies.
2951  *
2952  * Returns 0 on success.
2953  */
2954 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2955 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2956 {
2957     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2958 		    	    do_proc_dointvec_jiffies_conv,NULL);
2959 }
2960 
2961 /**
2962  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2963  * @table: the sysctl table
2964  * @write: %TRUE if this is a write to the sysctl file
2965  * @buffer: the user buffer
2966  * @lenp: the size of the user buffer
2967  * @ppos: pointer to the file position
2968  *
2969  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2970  * values from/to the user buffer, treated as an ASCII string.
2971  * The values read are assumed to be in 1/USER_HZ seconds, and
2972  * are converted into jiffies.
2973  *
2974  * Returns 0 on success.
2975  */
2976 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2977 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2978 {
2979     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2980 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2981 }
2982 
2983 /**
2984  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2985  * @table: the sysctl table
2986  * @write: %TRUE if this is a write to the sysctl file
2987  * @buffer: the user buffer
2988  * @lenp: the size of the user buffer
2989  * @ppos: file position
2990  * @ppos: the current position in the file
2991  *
2992  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2993  * values from/to the user buffer, treated as an ASCII string.
2994  * The values read are assumed to be in 1/1000 seconds, and
2995  * are converted into jiffies.
2996  *
2997  * Returns 0 on success.
2998  */
2999 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3000 			     void __user *buffer, size_t *lenp, loff_t *ppos)
3001 {
3002 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
3003 				do_proc_dointvec_ms_jiffies_conv, NULL);
3004 }
3005 
3006 static int proc_do_cad_pid(struct ctl_table *table, int write,
3007 			   void __user *buffer, size_t *lenp, loff_t *ppos)
3008 {
3009 	struct pid *new_pid;
3010 	pid_t tmp;
3011 	int r;
3012 
3013 	tmp = pid_vnr(cad_pid);
3014 
3015 	r = __do_proc_dointvec(&tmp, table, write, buffer,
3016 			       lenp, ppos, NULL, NULL);
3017 	if (r || !write)
3018 		return r;
3019 
3020 	new_pid = find_get_pid(tmp);
3021 	if (!new_pid)
3022 		return -ESRCH;
3023 
3024 	put_pid(xchg(&cad_pid, new_pid));
3025 	return 0;
3026 }
3027 
3028 /**
3029  * proc_do_large_bitmap - read/write from/to a large bitmap
3030  * @table: the sysctl table
3031  * @write: %TRUE if this is a write to the sysctl file
3032  * @buffer: the user buffer
3033  * @lenp: the size of the user buffer
3034  * @ppos: file position
3035  *
3036  * The bitmap is stored at table->data and the bitmap length (in bits)
3037  * in table->maxlen.
3038  *
3039  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
3040  * large bitmaps may be represented in a compact manner. Writing into
3041  * the file will clear the bitmap then update it with the given input.
3042  *
3043  * Returns 0 on success.
3044  */
3045 int proc_do_large_bitmap(struct ctl_table *table, int write,
3046 			 void __user *buffer, size_t *lenp, loff_t *ppos)
3047 {
3048 	int err = 0;
3049 	bool first = 1;
3050 	size_t left = *lenp;
3051 	unsigned long bitmap_len = table->maxlen;
3052 	unsigned long *bitmap = *(unsigned long **) table->data;
3053 	unsigned long *tmp_bitmap = NULL;
3054 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
3055 
3056 	if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
3057 		*lenp = 0;
3058 		return 0;
3059 	}
3060 
3061 	if (write) {
3062 		char *kbuf, *p;
3063 
3064 		if (left > PAGE_SIZE - 1)
3065 			left = PAGE_SIZE - 1;
3066 
3067 		p = kbuf = memdup_user_nul(buffer, left);
3068 		if (IS_ERR(kbuf))
3069 			return PTR_ERR(kbuf);
3070 
3071 		tmp_bitmap = kcalloc(BITS_TO_LONGS(bitmap_len),
3072 				     sizeof(unsigned long),
3073 				     GFP_KERNEL);
3074 		if (!tmp_bitmap) {
3075 			kfree(kbuf);
3076 			return -ENOMEM;
3077 		}
3078 		proc_skip_char(&p, &left, '\n');
3079 		while (!err && left) {
3080 			unsigned long val_a, val_b;
3081 			bool neg;
3082 
3083 			err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3084 					     sizeof(tr_a), &c);
3085 			if (err)
3086 				break;
3087 			if (val_a >= bitmap_len || neg) {
3088 				err = -EINVAL;
3089 				break;
3090 			}
3091 
3092 			val_b = val_a;
3093 			if (left) {
3094 				p++;
3095 				left--;
3096 			}
3097 
3098 			if (c == '-') {
3099 				err = proc_get_long(&p, &left, &val_b,
3100 						     &neg, tr_b, sizeof(tr_b),
3101 						     &c);
3102 				if (err)
3103 					break;
3104 				if (val_b >= bitmap_len || neg ||
3105 				    val_a > val_b) {
3106 					err = -EINVAL;
3107 					break;
3108 				}
3109 				if (left) {
3110 					p++;
3111 					left--;
3112 				}
3113 			}
3114 
3115 			bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3116 			first = 0;
3117 			proc_skip_char(&p, &left, '\n');
3118 		}
3119 		kfree(kbuf);
3120 	} else {
3121 		unsigned long bit_a, bit_b = 0;
3122 
3123 		while (left) {
3124 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3125 			if (bit_a >= bitmap_len)
3126 				break;
3127 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
3128 						   bit_a + 1) - 1;
3129 
3130 			if (!first) {
3131 				err = proc_put_char(&buffer, &left, ',');
3132 				if (err)
3133 					break;
3134 			}
3135 			err = proc_put_long(&buffer, &left, bit_a, false);
3136 			if (err)
3137 				break;
3138 			if (bit_a != bit_b) {
3139 				err = proc_put_char(&buffer, &left, '-');
3140 				if (err)
3141 					break;
3142 				err = proc_put_long(&buffer, &left, bit_b, false);
3143 				if (err)
3144 					break;
3145 			}
3146 
3147 			first = 0; bit_b++;
3148 		}
3149 		if (!err)
3150 			err = proc_put_char(&buffer, &left, '\n');
3151 	}
3152 
3153 	if (!err) {
3154 		if (write) {
3155 			if (*ppos)
3156 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3157 			else
3158 				bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3159 		}
3160 		*lenp -= left;
3161 		*ppos += *lenp;
3162 	}
3163 
3164 	kfree(tmp_bitmap);
3165 	return err;
3166 }
3167 
3168 #else /* CONFIG_PROC_SYSCTL */
3169 
3170 int proc_dostring(struct ctl_table *table, int write,
3171 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3172 {
3173 	return -ENOSYS;
3174 }
3175 
3176 int proc_dointvec(struct ctl_table *table, int write,
3177 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3178 {
3179 	return -ENOSYS;
3180 }
3181 
3182 int proc_douintvec(struct ctl_table *table, int write,
3183 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3184 {
3185 	return -ENOSYS;
3186 }
3187 
3188 int proc_dointvec_minmax(struct ctl_table *table, int write,
3189 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3190 {
3191 	return -ENOSYS;
3192 }
3193 
3194 int proc_douintvec_minmax(struct ctl_table *table, int write,
3195 			  void __user *buffer, size_t *lenp, loff_t *ppos)
3196 {
3197 	return -ENOSYS;
3198 }
3199 
3200 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3201 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3202 {
3203 	return -ENOSYS;
3204 }
3205 
3206 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3207 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3208 {
3209 	return -ENOSYS;
3210 }
3211 
3212 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3213 			     void __user *buffer, size_t *lenp, loff_t *ppos)
3214 {
3215 	return -ENOSYS;
3216 }
3217 
3218 int proc_doulongvec_minmax(struct ctl_table *table, int write,
3219 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3220 {
3221 	return -ENOSYS;
3222 }
3223 
3224 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3225 				      void __user *buffer,
3226 				      size_t *lenp, loff_t *ppos)
3227 {
3228     return -ENOSYS;
3229 }
3230 
3231 
3232 #endif /* CONFIG_PROC_SYSCTL */
3233 
3234 /*
3235  * No sense putting this after each symbol definition, twice,
3236  * exception granted :-)
3237  */
3238 EXPORT_SYMBOL(proc_dointvec);
3239 EXPORT_SYMBOL(proc_douintvec);
3240 EXPORT_SYMBOL(proc_dointvec_jiffies);
3241 EXPORT_SYMBOL(proc_dointvec_minmax);
3242 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3243 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3244 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3245 EXPORT_SYMBOL(proc_dostring);
3246 EXPORT_SYMBOL(proc_doulongvec_minmax);
3247 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3248