xref: /openbmc/linux/kernel/sysctl.c (revision 7e035230)
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/mm.h>
23 #include <linux/swap.h>
24 #include <linux/slab.h>
25 #include <linux/sysctl.h>
26 #include <linux/bitmap.h>
27 #include <linux/signal.h>
28 #include <linux/printk.h>
29 #include <linux/proc_fs.h>
30 #include <linux/security.h>
31 #include <linux/ctype.h>
32 #include <linux/kmemcheck.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 
65 #include <asm/uaccess.h>
66 #include <asm/processor.h>
67 
68 #ifdef CONFIG_X86
69 #include <asm/nmi.h>
70 #include <asm/stacktrace.h>
71 #include <asm/io.h>
72 #endif
73 #ifdef CONFIG_SPARC
74 #include <asm/setup.h>
75 #endif
76 #ifdef CONFIG_BSD_PROCESS_ACCT
77 #include <linux/acct.h>
78 #endif
79 #ifdef CONFIG_RT_MUTEXES
80 #include <linux/rtmutex.h>
81 #endif
82 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
83 #include <linux/lockdep.h>
84 #endif
85 #ifdef CONFIG_CHR_DEV_SG
86 #include <scsi/sg.h>
87 #endif
88 
89 #ifdef CONFIG_LOCKUP_DETECTOR
90 #include <linux/nmi.h>
91 #endif
92 
93 
94 #if defined(CONFIG_SYSCTL)
95 
96 /* External variables not in a header file. */
97 extern int sysctl_overcommit_memory;
98 extern int sysctl_overcommit_ratio;
99 extern int max_threads;
100 extern int core_uses_pid;
101 extern int suid_dumpable;
102 extern char core_pattern[];
103 extern unsigned int core_pipe_limit;
104 extern int pid_max;
105 extern int min_free_kbytes;
106 extern int pid_max_min, pid_max_max;
107 extern int sysctl_drop_caches;
108 extern int percpu_pagelist_fraction;
109 extern int compat_log;
110 extern int latencytop_enabled;
111 extern int sysctl_nr_open_min, sysctl_nr_open_max;
112 #ifndef CONFIG_MMU
113 extern int sysctl_nr_trim_pages;
114 #endif
115 #ifdef CONFIG_BLOCK
116 extern int blk_iopoll_enabled;
117 #endif
118 
119 /* Constants used for minimum and  maximum */
120 #ifdef CONFIG_LOCKUP_DETECTOR
121 static int sixty = 60;
122 static int neg_one = -1;
123 #endif
124 
125 static int zero;
126 static int __maybe_unused one = 1;
127 static int __maybe_unused two = 2;
128 static int __maybe_unused three = 3;
129 static unsigned long one_ul = 1;
130 static int one_hundred = 100;
131 #ifdef CONFIG_PRINTK
132 static int ten_thousand = 10000;
133 #endif
134 
135 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
136 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
137 
138 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
139 static int maxolduid = 65535;
140 static int minolduid;
141 static int min_percpu_pagelist_fract = 8;
142 
143 static int ngroups_max = NGROUPS_MAX;
144 static const int cap_last_cap = CAP_LAST_CAP;
145 
146 #ifdef CONFIG_INOTIFY_USER
147 #include <linux/inotify.h>
148 #endif
149 #ifdef CONFIG_SPARC
150 #endif
151 
152 #ifdef CONFIG_SPARC64
153 extern int sysctl_tsb_ratio;
154 #endif
155 
156 #ifdef __hppa__
157 extern int pwrsw_enabled;
158 extern int unaligned_enabled;
159 #endif
160 
161 #ifdef CONFIG_IA64
162 extern int no_unaligned_warning;
163 extern int unaligned_dump_stack;
164 #endif
165 
166 #ifdef CONFIG_PROC_SYSCTL
167 static int proc_do_cad_pid(struct ctl_table *table, int write,
168 		  void __user *buffer, size_t *lenp, loff_t *ppos);
169 static int proc_taint(struct ctl_table *table, int write,
170 			       void __user *buffer, size_t *lenp, loff_t *ppos);
171 #endif
172 
173 #ifdef CONFIG_PRINTK
174 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
175 				void __user *buffer, size_t *lenp, loff_t *ppos);
176 #endif
177 
178 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
179 		void __user *buffer, size_t *lenp, loff_t *ppos);
180 static int proc_dostring_coredump(struct ctl_table *table, int write,
181 		void __user *buffer, size_t *lenp, loff_t *ppos);
182 
183 #ifdef CONFIG_MAGIC_SYSRQ
184 /* Note: sysrq code uses it's own private copy */
185 static int __sysrq_enabled = SYSRQ_DEFAULT_ENABLE;
186 
187 static int sysrq_sysctl_handler(ctl_table *table, int write,
188 				void __user *buffer, size_t *lenp,
189 				loff_t *ppos)
190 {
191 	int error;
192 
193 	error = proc_dointvec(table, write, buffer, lenp, ppos);
194 	if (error)
195 		return error;
196 
197 	if (write)
198 		sysrq_toggle_support(__sysrq_enabled);
199 
200 	return 0;
201 }
202 
203 #endif
204 
205 static struct ctl_table kern_table[];
206 static struct ctl_table vm_table[];
207 static struct ctl_table fs_table[];
208 static struct ctl_table debug_table[];
209 static struct ctl_table dev_table[];
210 extern struct ctl_table random_table[];
211 #ifdef CONFIG_EPOLL
212 extern struct ctl_table epoll_table[];
213 #endif
214 
215 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
216 int sysctl_legacy_va_layout;
217 #endif
218 
219 /* The default sysctl tables: */
220 
221 static struct ctl_table sysctl_base_table[] = {
222 	{
223 		.procname	= "kernel",
224 		.mode		= 0555,
225 		.child		= kern_table,
226 	},
227 	{
228 		.procname	= "vm",
229 		.mode		= 0555,
230 		.child		= vm_table,
231 	},
232 	{
233 		.procname	= "fs",
234 		.mode		= 0555,
235 		.child		= fs_table,
236 	},
237 	{
238 		.procname	= "debug",
239 		.mode		= 0555,
240 		.child		= debug_table,
241 	},
242 	{
243 		.procname	= "dev",
244 		.mode		= 0555,
245 		.child		= dev_table,
246 	},
247 	{ }
248 };
249 
250 #ifdef CONFIG_SCHED_DEBUG
251 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
252 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
253 static int min_wakeup_granularity_ns;			/* 0 usecs */
254 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
255 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
256 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
257 #endif
258 
259 #ifdef CONFIG_COMPACTION
260 static int min_extfrag_threshold;
261 static int max_extfrag_threshold = 1000;
262 #endif
263 
264 static struct ctl_table kern_table[] = {
265 	{
266 		.procname	= "sched_child_runs_first",
267 		.data		= &sysctl_sched_child_runs_first,
268 		.maxlen		= sizeof(unsigned int),
269 		.mode		= 0644,
270 		.proc_handler	= proc_dointvec,
271 	},
272 #ifdef CONFIG_SCHED_DEBUG
273 	{
274 		.procname	= "sched_min_granularity_ns",
275 		.data		= &sysctl_sched_min_granularity,
276 		.maxlen		= sizeof(unsigned int),
277 		.mode		= 0644,
278 		.proc_handler	= sched_proc_update_handler,
279 		.extra1		= &min_sched_granularity_ns,
280 		.extra2		= &max_sched_granularity_ns,
281 	},
282 	{
283 		.procname	= "sched_latency_ns",
284 		.data		= &sysctl_sched_latency,
285 		.maxlen		= sizeof(unsigned int),
286 		.mode		= 0644,
287 		.proc_handler	= sched_proc_update_handler,
288 		.extra1		= &min_sched_granularity_ns,
289 		.extra2		= &max_sched_granularity_ns,
290 	},
291 	{
292 		.procname	= "sched_wakeup_granularity_ns",
293 		.data		= &sysctl_sched_wakeup_granularity,
294 		.maxlen		= sizeof(unsigned int),
295 		.mode		= 0644,
296 		.proc_handler	= sched_proc_update_handler,
297 		.extra1		= &min_wakeup_granularity_ns,
298 		.extra2		= &max_wakeup_granularity_ns,
299 	},
300 	{
301 		.procname	= "sched_tunable_scaling",
302 		.data		= &sysctl_sched_tunable_scaling,
303 		.maxlen		= sizeof(enum sched_tunable_scaling),
304 		.mode		= 0644,
305 		.proc_handler	= sched_proc_update_handler,
306 		.extra1		= &min_sched_tunable_scaling,
307 		.extra2		= &max_sched_tunable_scaling,
308 	},
309 	{
310 		.procname	= "sched_migration_cost",
311 		.data		= &sysctl_sched_migration_cost,
312 		.maxlen		= sizeof(unsigned int),
313 		.mode		= 0644,
314 		.proc_handler	= proc_dointvec,
315 	},
316 	{
317 		.procname	= "sched_nr_migrate",
318 		.data		= &sysctl_sched_nr_migrate,
319 		.maxlen		= sizeof(unsigned int),
320 		.mode		= 0644,
321 		.proc_handler	= proc_dointvec,
322 	},
323 	{
324 		.procname	= "sched_time_avg",
325 		.data		= &sysctl_sched_time_avg,
326 		.maxlen		= sizeof(unsigned int),
327 		.mode		= 0644,
328 		.proc_handler	= proc_dointvec,
329 	},
330 	{
331 		.procname	= "sched_shares_window",
332 		.data		= &sysctl_sched_shares_window,
333 		.maxlen		= sizeof(unsigned int),
334 		.mode		= 0644,
335 		.proc_handler	= proc_dointvec,
336 	},
337 	{
338 		.procname	= "timer_migration",
339 		.data		= &sysctl_timer_migration,
340 		.maxlen		= sizeof(unsigned int),
341 		.mode		= 0644,
342 		.proc_handler	= proc_dointvec_minmax,
343 		.extra1		= &zero,
344 		.extra2		= &one,
345 	},
346 #endif
347 	{
348 		.procname	= "sched_rt_period_us",
349 		.data		= &sysctl_sched_rt_period,
350 		.maxlen		= sizeof(unsigned int),
351 		.mode		= 0644,
352 		.proc_handler	= sched_rt_handler,
353 	},
354 	{
355 		.procname	= "sched_rt_runtime_us",
356 		.data		= &sysctl_sched_rt_runtime,
357 		.maxlen		= sizeof(int),
358 		.mode		= 0644,
359 		.proc_handler	= sched_rt_handler,
360 	},
361 #ifdef CONFIG_SCHED_AUTOGROUP
362 	{
363 		.procname	= "sched_autogroup_enabled",
364 		.data		= &sysctl_sched_autogroup_enabled,
365 		.maxlen		= sizeof(unsigned int),
366 		.mode		= 0644,
367 		.proc_handler	= proc_dointvec_minmax,
368 		.extra1		= &zero,
369 		.extra2		= &one,
370 	},
371 #endif
372 #ifdef CONFIG_CFS_BANDWIDTH
373 	{
374 		.procname	= "sched_cfs_bandwidth_slice_us",
375 		.data		= &sysctl_sched_cfs_bandwidth_slice,
376 		.maxlen		= sizeof(unsigned int),
377 		.mode		= 0644,
378 		.proc_handler	= proc_dointvec_minmax,
379 		.extra1		= &one,
380 	},
381 #endif
382 #ifdef CONFIG_PROVE_LOCKING
383 	{
384 		.procname	= "prove_locking",
385 		.data		= &prove_locking,
386 		.maxlen		= sizeof(int),
387 		.mode		= 0644,
388 		.proc_handler	= proc_dointvec,
389 	},
390 #endif
391 #ifdef CONFIG_LOCK_STAT
392 	{
393 		.procname	= "lock_stat",
394 		.data		= &lock_stat,
395 		.maxlen		= sizeof(int),
396 		.mode		= 0644,
397 		.proc_handler	= proc_dointvec,
398 	},
399 #endif
400 	{
401 		.procname	= "panic",
402 		.data		= &panic_timeout,
403 		.maxlen		= sizeof(int),
404 		.mode		= 0644,
405 		.proc_handler	= proc_dointvec,
406 	},
407 	{
408 		.procname	= "core_uses_pid",
409 		.data		= &core_uses_pid,
410 		.maxlen		= sizeof(int),
411 		.mode		= 0644,
412 		.proc_handler	= proc_dointvec,
413 	},
414 	{
415 		.procname	= "core_pattern",
416 		.data		= core_pattern,
417 		.maxlen		= CORENAME_MAX_SIZE,
418 		.mode		= 0644,
419 		.proc_handler	= proc_dostring_coredump,
420 	},
421 	{
422 		.procname	= "core_pipe_limit",
423 		.data		= &core_pipe_limit,
424 		.maxlen		= sizeof(unsigned int),
425 		.mode		= 0644,
426 		.proc_handler	= proc_dointvec,
427 	},
428 #ifdef CONFIG_PROC_SYSCTL
429 	{
430 		.procname	= "tainted",
431 		.maxlen 	= sizeof(long),
432 		.mode		= 0644,
433 		.proc_handler	= proc_taint,
434 	},
435 #endif
436 #ifdef CONFIG_LATENCYTOP
437 	{
438 		.procname	= "latencytop",
439 		.data		= &latencytop_enabled,
440 		.maxlen		= sizeof(int),
441 		.mode		= 0644,
442 		.proc_handler	= proc_dointvec,
443 	},
444 #endif
445 #ifdef CONFIG_BLK_DEV_INITRD
446 	{
447 		.procname	= "real-root-dev",
448 		.data		= &real_root_dev,
449 		.maxlen		= sizeof(int),
450 		.mode		= 0644,
451 		.proc_handler	= proc_dointvec,
452 	},
453 #endif
454 	{
455 		.procname	= "print-fatal-signals",
456 		.data		= &print_fatal_signals,
457 		.maxlen		= sizeof(int),
458 		.mode		= 0644,
459 		.proc_handler	= proc_dointvec,
460 	},
461 #ifdef CONFIG_SPARC
462 	{
463 		.procname	= "reboot-cmd",
464 		.data		= reboot_command,
465 		.maxlen		= 256,
466 		.mode		= 0644,
467 		.proc_handler	= proc_dostring,
468 	},
469 	{
470 		.procname	= "stop-a",
471 		.data		= &stop_a_enabled,
472 		.maxlen		= sizeof (int),
473 		.mode		= 0644,
474 		.proc_handler	= proc_dointvec,
475 	},
476 	{
477 		.procname	= "scons-poweroff",
478 		.data		= &scons_pwroff,
479 		.maxlen		= sizeof (int),
480 		.mode		= 0644,
481 		.proc_handler	= proc_dointvec,
482 	},
483 #endif
484 #ifdef CONFIG_SPARC64
485 	{
486 		.procname	= "tsb-ratio",
487 		.data		= &sysctl_tsb_ratio,
488 		.maxlen		= sizeof (int),
489 		.mode		= 0644,
490 		.proc_handler	= proc_dointvec,
491 	},
492 #endif
493 #ifdef __hppa__
494 	{
495 		.procname	= "soft-power",
496 		.data		= &pwrsw_enabled,
497 		.maxlen		= sizeof (int),
498 	 	.mode		= 0644,
499 		.proc_handler	= proc_dointvec,
500 	},
501 	{
502 		.procname	= "unaligned-trap",
503 		.data		= &unaligned_enabled,
504 		.maxlen		= sizeof (int),
505 		.mode		= 0644,
506 		.proc_handler	= proc_dointvec,
507 	},
508 #endif
509 	{
510 		.procname	= "ctrl-alt-del",
511 		.data		= &C_A_D,
512 		.maxlen		= sizeof(int),
513 		.mode		= 0644,
514 		.proc_handler	= proc_dointvec,
515 	},
516 #ifdef CONFIG_FUNCTION_TRACER
517 	{
518 		.procname	= "ftrace_enabled",
519 		.data		= &ftrace_enabled,
520 		.maxlen		= sizeof(int),
521 		.mode		= 0644,
522 		.proc_handler	= ftrace_enable_sysctl,
523 	},
524 #endif
525 #ifdef CONFIG_STACK_TRACER
526 	{
527 		.procname	= "stack_tracer_enabled",
528 		.data		= &stack_tracer_enabled,
529 		.maxlen		= sizeof(int),
530 		.mode		= 0644,
531 		.proc_handler	= stack_trace_sysctl,
532 	},
533 #endif
534 #ifdef CONFIG_TRACING
535 	{
536 		.procname	= "ftrace_dump_on_oops",
537 		.data		= &ftrace_dump_on_oops,
538 		.maxlen		= sizeof(int),
539 		.mode		= 0644,
540 		.proc_handler	= proc_dointvec,
541 	},
542 #endif
543 #ifdef CONFIG_MODULES
544 	{
545 		.procname	= "modprobe",
546 		.data		= &modprobe_path,
547 		.maxlen		= KMOD_PATH_LEN,
548 		.mode		= 0644,
549 		.proc_handler	= proc_dostring,
550 	},
551 	{
552 		.procname	= "modules_disabled",
553 		.data		= &modules_disabled,
554 		.maxlen		= sizeof(int),
555 		.mode		= 0644,
556 		/* only handle a transition from default "0" to "1" */
557 		.proc_handler	= proc_dointvec_minmax,
558 		.extra1		= &one,
559 		.extra2		= &one,
560 	},
561 #endif
562 #ifdef CONFIG_HOTPLUG
563 	{
564 		.procname	= "hotplug",
565 		.data		= &uevent_helper,
566 		.maxlen		= UEVENT_HELPER_PATH_LEN,
567 		.mode		= 0644,
568 		.proc_handler	= proc_dostring,
569 	},
570 #endif
571 #ifdef CONFIG_CHR_DEV_SG
572 	{
573 		.procname	= "sg-big-buff",
574 		.data		= &sg_big_buff,
575 		.maxlen		= sizeof (int),
576 		.mode		= 0444,
577 		.proc_handler	= proc_dointvec,
578 	},
579 #endif
580 #ifdef CONFIG_BSD_PROCESS_ACCT
581 	{
582 		.procname	= "acct",
583 		.data		= &acct_parm,
584 		.maxlen		= 3*sizeof(int),
585 		.mode		= 0644,
586 		.proc_handler	= proc_dointvec,
587 	},
588 #endif
589 #ifdef CONFIG_MAGIC_SYSRQ
590 	{
591 		.procname	= "sysrq",
592 		.data		= &__sysrq_enabled,
593 		.maxlen		= sizeof (int),
594 		.mode		= 0644,
595 		.proc_handler	= sysrq_sysctl_handler,
596 	},
597 #endif
598 #ifdef CONFIG_PROC_SYSCTL
599 	{
600 		.procname	= "cad_pid",
601 		.data		= NULL,
602 		.maxlen		= sizeof (int),
603 		.mode		= 0600,
604 		.proc_handler	= proc_do_cad_pid,
605 	},
606 #endif
607 	{
608 		.procname	= "threads-max",
609 		.data		= &max_threads,
610 		.maxlen		= sizeof(int),
611 		.mode		= 0644,
612 		.proc_handler	= proc_dointvec,
613 	},
614 	{
615 		.procname	= "random",
616 		.mode		= 0555,
617 		.child		= random_table,
618 	},
619 	{
620 		.procname	= "usermodehelper",
621 		.mode		= 0555,
622 		.child		= usermodehelper_table,
623 	},
624 	{
625 		.procname	= "overflowuid",
626 		.data		= &overflowuid,
627 		.maxlen		= sizeof(int),
628 		.mode		= 0644,
629 		.proc_handler	= proc_dointvec_minmax,
630 		.extra1		= &minolduid,
631 		.extra2		= &maxolduid,
632 	},
633 	{
634 		.procname	= "overflowgid",
635 		.data		= &overflowgid,
636 		.maxlen		= sizeof(int),
637 		.mode		= 0644,
638 		.proc_handler	= proc_dointvec_minmax,
639 		.extra1		= &minolduid,
640 		.extra2		= &maxolduid,
641 	},
642 #ifdef CONFIG_S390
643 #ifdef CONFIG_MATHEMU
644 	{
645 		.procname	= "ieee_emulation_warnings",
646 		.data		= &sysctl_ieee_emulation_warnings,
647 		.maxlen		= sizeof(int),
648 		.mode		= 0644,
649 		.proc_handler	= proc_dointvec,
650 	},
651 #endif
652 	{
653 		.procname	= "userprocess_debug",
654 		.data		= &show_unhandled_signals,
655 		.maxlen		= sizeof(int),
656 		.mode		= 0644,
657 		.proc_handler	= proc_dointvec,
658 	},
659 #endif
660 	{
661 		.procname	= "pid_max",
662 		.data		= &pid_max,
663 		.maxlen		= sizeof (int),
664 		.mode		= 0644,
665 		.proc_handler	= proc_dointvec_minmax,
666 		.extra1		= &pid_max_min,
667 		.extra2		= &pid_max_max,
668 	},
669 	{
670 		.procname	= "panic_on_oops",
671 		.data		= &panic_on_oops,
672 		.maxlen		= sizeof(int),
673 		.mode		= 0644,
674 		.proc_handler	= proc_dointvec,
675 	},
676 #if defined CONFIG_PRINTK
677 	{
678 		.procname	= "printk",
679 		.data		= &console_loglevel,
680 		.maxlen		= 4*sizeof(int),
681 		.mode		= 0644,
682 		.proc_handler	= proc_dointvec,
683 	},
684 	{
685 		.procname	= "printk_ratelimit",
686 		.data		= &printk_ratelimit_state.interval,
687 		.maxlen		= sizeof(int),
688 		.mode		= 0644,
689 		.proc_handler	= proc_dointvec_jiffies,
690 	},
691 	{
692 		.procname	= "printk_ratelimit_burst",
693 		.data		= &printk_ratelimit_state.burst,
694 		.maxlen		= sizeof(int),
695 		.mode		= 0644,
696 		.proc_handler	= proc_dointvec,
697 	},
698 	{
699 		.procname	= "printk_delay",
700 		.data		= &printk_delay_msec,
701 		.maxlen		= sizeof(int),
702 		.mode		= 0644,
703 		.proc_handler	= proc_dointvec_minmax,
704 		.extra1		= &zero,
705 		.extra2		= &ten_thousand,
706 	},
707 	{
708 		.procname	= "dmesg_restrict",
709 		.data		= &dmesg_restrict,
710 		.maxlen		= sizeof(int),
711 		.mode		= 0644,
712 		.proc_handler	= proc_dointvec_minmax_sysadmin,
713 		.extra1		= &zero,
714 		.extra2		= &one,
715 	},
716 	{
717 		.procname	= "kptr_restrict",
718 		.data		= &kptr_restrict,
719 		.maxlen		= sizeof(int),
720 		.mode		= 0644,
721 		.proc_handler	= proc_dointvec_minmax_sysadmin,
722 		.extra1		= &zero,
723 		.extra2		= &two,
724 	},
725 #endif
726 	{
727 		.procname	= "ngroups_max",
728 		.data		= &ngroups_max,
729 		.maxlen		= sizeof (int),
730 		.mode		= 0444,
731 		.proc_handler	= proc_dointvec,
732 	},
733 	{
734 		.procname	= "cap_last_cap",
735 		.data		= (void *)&cap_last_cap,
736 		.maxlen		= sizeof(int),
737 		.mode		= 0444,
738 		.proc_handler	= proc_dointvec,
739 	},
740 #if defined(CONFIG_LOCKUP_DETECTOR)
741 	{
742 		.procname       = "watchdog",
743 		.data           = &watchdog_enabled,
744 		.maxlen         = sizeof (int),
745 		.mode           = 0644,
746 		.proc_handler   = proc_dowatchdog,
747 		.extra1		= &zero,
748 		.extra2		= &one,
749 	},
750 	{
751 		.procname	= "watchdog_thresh",
752 		.data		= &watchdog_thresh,
753 		.maxlen		= sizeof(int),
754 		.mode		= 0644,
755 		.proc_handler	= proc_dowatchdog,
756 		.extra1		= &neg_one,
757 		.extra2		= &sixty,
758 	},
759 	{
760 		.procname	= "softlockup_panic",
761 		.data		= &softlockup_panic,
762 		.maxlen		= sizeof(int),
763 		.mode		= 0644,
764 		.proc_handler	= proc_dointvec_minmax,
765 		.extra1		= &zero,
766 		.extra2		= &one,
767 	},
768 	{
769 		.procname       = "nmi_watchdog",
770 		.data           = &watchdog_enabled,
771 		.maxlen         = sizeof (int),
772 		.mode           = 0644,
773 		.proc_handler   = proc_dowatchdog,
774 		.extra1		= &zero,
775 		.extra2		= &one,
776 	},
777 #endif
778 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
779 	{
780 		.procname       = "unknown_nmi_panic",
781 		.data           = &unknown_nmi_panic,
782 		.maxlen         = sizeof (int),
783 		.mode           = 0644,
784 		.proc_handler   = proc_dointvec,
785 	},
786 #endif
787 #if defined(CONFIG_X86)
788 	{
789 		.procname	= "panic_on_unrecovered_nmi",
790 		.data		= &panic_on_unrecovered_nmi,
791 		.maxlen		= sizeof(int),
792 		.mode		= 0644,
793 		.proc_handler	= proc_dointvec,
794 	},
795 	{
796 		.procname	= "panic_on_io_nmi",
797 		.data		= &panic_on_io_nmi,
798 		.maxlen		= sizeof(int),
799 		.mode		= 0644,
800 		.proc_handler	= proc_dointvec,
801 	},
802 #ifdef CONFIG_DEBUG_STACKOVERFLOW
803 	{
804 		.procname	= "panic_on_stackoverflow",
805 		.data		= &sysctl_panic_on_stackoverflow,
806 		.maxlen		= sizeof(int),
807 		.mode		= 0644,
808 		.proc_handler	= proc_dointvec,
809 	},
810 #endif
811 	{
812 		.procname	= "bootloader_type",
813 		.data		= &bootloader_type,
814 		.maxlen		= sizeof (int),
815 		.mode		= 0444,
816 		.proc_handler	= proc_dointvec,
817 	},
818 	{
819 		.procname	= "bootloader_version",
820 		.data		= &bootloader_version,
821 		.maxlen		= sizeof (int),
822 		.mode		= 0444,
823 		.proc_handler	= proc_dointvec,
824 	},
825 	{
826 		.procname	= "kstack_depth_to_print",
827 		.data		= &kstack_depth_to_print,
828 		.maxlen		= sizeof(int),
829 		.mode		= 0644,
830 		.proc_handler	= proc_dointvec,
831 	},
832 	{
833 		.procname	= "io_delay_type",
834 		.data		= &io_delay_type,
835 		.maxlen		= sizeof(int),
836 		.mode		= 0644,
837 		.proc_handler	= proc_dointvec,
838 	},
839 #endif
840 #if defined(CONFIG_MMU)
841 	{
842 		.procname	= "randomize_va_space",
843 		.data		= &randomize_va_space,
844 		.maxlen		= sizeof(int),
845 		.mode		= 0644,
846 		.proc_handler	= proc_dointvec,
847 	},
848 #endif
849 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
850 	{
851 		.procname	= "spin_retry",
852 		.data		= &spin_retry,
853 		.maxlen		= sizeof (int),
854 		.mode		= 0644,
855 		.proc_handler	= proc_dointvec,
856 	},
857 #endif
858 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
859 	{
860 		.procname	= "acpi_video_flags",
861 		.data		= &acpi_realmode_flags,
862 		.maxlen		= sizeof (unsigned long),
863 		.mode		= 0644,
864 		.proc_handler	= proc_doulongvec_minmax,
865 	},
866 #endif
867 #ifdef CONFIG_IA64
868 	{
869 		.procname	= "ignore-unaligned-usertrap",
870 		.data		= &no_unaligned_warning,
871 		.maxlen		= sizeof (int),
872 	 	.mode		= 0644,
873 		.proc_handler	= proc_dointvec,
874 	},
875 	{
876 		.procname	= "unaligned-dump-stack",
877 		.data		= &unaligned_dump_stack,
878 		.maxlen		= sizeof (int),
879 		.mode		= 0644,
880 		.proc_handler	= proc_dointvec,
881 	},
882 #endif
883 #ifdef CONFIG_DETECT_HUNG_TASK
884 	{
885 		.procname	= "hung_task_panic",
886 		.data		= &sysctl_hung_task_panic,
887 		.maxlen		= sizeof(int),
888 		.mode		= 0644,
889 		.proc_handler	= proc_dointvec_minmax,
890 		.extra1		= &zero,
891 		.extra2		= &one,
892 	},
893 	{
894 		.procname	= "hung_task_check_count",
895 		.data		= &sysctl_hung_task_check_count,
896 		.maxlen		= sizeof(unsigned long),
897 		.mode		= 0644,
898 		.proc_handler	= proc_doulongvec_minmax,
899 	},
900 	{
901 		.procname	= "hung_task_timeout_secs",
902 		.data		= &sysctl_hung_task_timeout_secs,
903 		.maxlen		= sizeof(unsigned long),
904 		.mode		= 0644,
905 		.proc_handler	= proc_dohung_task_timeout_secs,
906 	},
907 	{
908 		.procname	= "hung_task_warnings",
909 		.data		= &sysctl_hung_task_warnings,
910 		.maxlen		= sizeof(unsigned long),
911 		.mode		= 0644,
912 		.proc_handler	= proc_doulongvec_minmax,
913 	},
914 #endif
915 #ifdef CONFIG_COMPAT
916 	{
917 		.procname	= "compat-log",
918 		.data		= &compat_log,
919 		.maxlen		= sizeof (int),
920 	 	.mode		= 0644,
921 		.proc_handler	= proc_dointvec,
922 	},
923 #endif
924 #ifdef CONFIG_RT_MUTEXES
925 	{
926 		.procname	= "max_lock_depth",
927 		.data		= &max_lock_depth,
928 		.maxlen		= sizeof(int),
929 		.mode		= 0644,
930 		.proc_handler	= proc_dointvec,
931 	},
932 #endif
933 	{
934 		.procname	= "poweroff_cmd",
935 		.data		= &poweroff_cmd,
936 		.maxlen		= POWEROFF_CMD_PATH_LEN,
937 		.mode		= 0644,
938 		.proc_handler	= proc_dostring,
939 	},
940 #ifdef CONFIG_KEYS
941 	{
942 		.procname	= "keys",
943 		.mode		= 0555,
944 		.child		= key_sysctls,
945 	},
946 #endif
947 #ifdef CONFIG_RCU_TORTURE_TEST
948 	{
949 		.procname       = "rcutorture_runnable",
950 		.data           = &rcutorture_runnable,
951 		.maxlen         = sizeof(int),
952 		.mode           = 0644,
953 		.proc_handler	= proc_dointvec,
954 	},
955 #endif
956 #ifdef CONFIG_PERF_EVENTS
957 	/*
958 	 * User-space scripts rely on the existence of this file
959 	 * as a feature check for perf_events being enabled.
960 	 *
961 	 * So it's an ABI, do not remove!
962 	 */
963 	{
964 		.procname	= "perf_event_paranoid",
965 		.data		= &sysctl_perf_event_paranoid,
966 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
967 		.mode		= 0644,
968 		.proc_handler	= proc_dointvec,
969 	},
970 	{
971 		.procname	= "perf_event_mlock_kb",
972 		.data		= &sysctl_perf_event_mlock,
973 		.maxlen		= sizeof(sysctl_perf_event_mlock),
974 		.mode		= 0644,
975 		.proc_handler	= proc_dointvec,
976 	},
977 	{
978 		.procname	= "perf_event_max_sample_rate",
979 		.data		= &sysctl_perf_event_sample_rate,
980 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
981 		.mode		= 0644,
982 		.proc_handler	= perf_proc_update_handler,
983 	},
984 #endif
985 #ifdef CONFIG_KMEMCHECK
986 	{
987 		.procname	= "kmemcheck",
988 		.data		= &kmemcheck_enabled,
989 		.maxlen		= sizeof(int),
990 		.mode		= 0644,
991 		.proc_handler	= proc_dointvec,
992 	},
993 #endif
994 #ifdef CONFIG_BLOCK
995 	{
996 		.procname	= "blk_iopoll",
997 		.data		= &blk_iopoll_enabled,
998 		.maxlen		= sizeof(int),
999 		.mode		= 0644,
1000 		.proc_handler	= proc_dointvec,
1001 	},
1002 #endif
1003 	{ }
1004 };
1005 
1006 static struct ctl_table vm_table[] = {
1007 	{
1008 		.procname	= "overcommit_memory",
1009 		.data		= &sysctl_overcommit_memory,
1010 		.maxlen		= sizeof(sysctl_overcommit_memory),
1011 		.mode		= 0644,
1012 		.proc_handler	= proc_dointvec_minmax,
1013 		.extra1		= &zero,
1014 		.extra2		= &two,
1015 	},
1016 	{
1017 		.procname	= "panic_on_oom",
1018 		.data		= &sysctl_panic_on_oom,
1019 		.maxlen		= sizeof(sysctl_panic_on_oom),
1020 		.mode		= 0644,
1021 		.proc_handler	= proc_dointvec_minmax,
1022 		.extra1		= &zero,
1023 		.extra2		= &two,
1024 	},
1025 	{
1026 		.procname	= "oom_kill_allocating_task",
1027 		.data		= &sysctl_oom_kill_allocating_task,
1028 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1029 		.mode		= 0644,
1030 		.proc_handler	= proc_dointvec,
1031 	},
1032 	{
1033 		.procname	= "oom_dump_tasks",
1034 		.data		= &sysctl_oom_dump_tasks,
1035 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1036 		.mode		= 0644,
1037 		.proc_handler	= proc_dointvec,
1038 	},
1039 	{
1040 		.procname	= "overcommit_ratio",
1041 		.data		= &sysctl_overcommit_ratio,
1042 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1043 		.mode		= 0644,
1044 		.proc_handler	= proc_dointvec,
1045 	},
1046 	{
1047 		.procname	= "page-cluster",
1048 		.data		= &page_cluster,
1049 		.maxlen		= sizeof(int),
1050 		.mode		= 0644,
1051 		.proc_handler	= proc_dointvec_minmax,
1052 		.extra1		= &zero,
1053 	},
1054 	{
1055 		.procname	= "dirty_background_ratio",
1056 		.data		= &dirty_background_ratio,
1057 		.maxlen		= sizeof(dirty_background_ratio),
1058 		.mode		= 0644,
1059 		.proc_handler	= dirty_background_ratio_handler,
1060 		.extra1		= &zero,
1061 		.extra2		= &one_hundred,
1062 	},
1063 	{
1064 		.procname	= "dirty_background_bytes",
1065 		.data		= &dirty_background_bytes,
1066 		.maxlen		= sizeof(dirty_background_bytes),
1067 		.mode		= 0644,
1068 		.proc_handler	= dirty_background_bytes_handler,
1069 		.extra1		= &one_ul,
1070 	},
1071 	{
1072 		.procname	= "dirty_ratio",
1073 		.data		= &vm_dirty_ratio,
1074 		.maxlen		= sizeof(vm_dirty_ratio),
1075 		.mode		= 0644,
1076 		.proc_handler	= dirty_ratio_handler,
1077 		.extra1		= &zero,
1078 		.extra2		= &one_hundred,
1079 	},
1080 	{
1081 		.procname	= "dirty_bytes",
1082 		.data		= &vm_dirty_bytes,
1083 		.maxlen		= sizeof(vm_dirty_bytes),
1084 		.mode		= 0644,
1085 		.proc_handler	= dirty_bytes_handler,
1086 		.extra1		= &dirty_bytes_min,
1087 	},
1088 	{
1089 		.procname	= "dirty_writeback_centisecs",
1090 		.data		= &dirty_writeback_interval,
1091 		.maxlen		= sizeof(dirty_writeback_interval),
1092 		.mode		= 0644,
1093 		.proc_handler	= dirty_writeback_centisecs_handler,
1094 	},
1095 	{
1096 		.procname	= "dirty_expire_centisecs",
1097 		.data		= &dirty_expire_interval,
1098 		.maxlen		= sizeof(dirty_expire_interval),
1099 		.mode		= 0644,
1100 		.proc_handler	= proc_dointvec_minmax,
1101 		.extra1		= &zero,
1102 	},
1103 	{
1104 		.procname       = "nr_pdflush_threads",
1105 		.mode           = 0444 /* read-only */,
1106 		.proc_handler   = pdflush_proc_obsolete,
1107 	},
1108 	{
1109 		.procname	= "swappiness",
1110 		.data		= &vm_swappiness,
1111 		.maxlen		= sizeof(vm_swappiness),
1112 		.mode		= 0644,
1113 		.proc_handler	= proc_dointvec_minmax,
1114 		.extra1		= &zero,
1115 		.extra2		= &one_hundred,
1116 	},
1117 #ifdef CONFIG_HUGETLB_PAGE
1118 	{
1119 		.procname	= "nr_hugepages",
1120 		.data		= NULL,
1121 		.maxlen		= sizeof(unsigned long),
1122 		.mode		= 0644,
1123 		.proc_handler	= hugetlb_sysctl_handler,
1124 		.extra1		= (void *)&hugetlb_zero,
1125 		.extra2		= (void *)&hugetlb_infinity,
1126 	},
1127 #ifdef CONFIG_NUMA
1128 	{
1129 		.procname       = "nr_hugepages_mempolicy",
1130 		.data           = NULL,
1131 		.maxlen         = sizeof(unsigned long),
1132 		.mode           = 0644,
1133 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1134 		.extra1		= (void *)&hugetlb_zero,
1135 		.extra2		= (void *)&hugetlb_infinity,
1136 	},
1137 #endif
1138 	 {
1139 		.procname	= "hugetlb_shm_group",
1140 		.data		= &sysctl_hugetlb_shm_group,
1141 		.maxlen		= sizeof(gid_t),
1142 		.mode		= 0644,
1143 		.proc_handler	= proc_dointvec,
1144 	 },
1145 	 {
1146 		.procname	= "hugepages_treat_as_movable",
1147 		.data		= &hugepages_treat_as_movable,
1148 		.maxlen		= sizeof(int),
1149 		.mode		= 0644,
1150 		.proc_handler	= hugetlb_treat_movable_handler,
1151 	},
1152 	{
1153 		.procname	= "nr_overcommit_hugepages",
1154 		.data		= NULL,
1155 		.maxlen		= sizeof(unsigned long),
1156 		.mode		= 0644,
1157 		.proc_handler	= hugetlb_overcommit_handler,
1158 		.extra1		= (void *)&hugetlb_zero,
1159 		.extra2		= (void *)&hugetlb_infinity,
1160 	},
1161 #endif
1162 	{
1163 		.procname	= "lowmem_reserve_ratio",
1164 		.data		= &sysctl_lowmem_reserve_ratio,
1165 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1166 		.mode		= 0644,
1167 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1168 	},
1169 	{
1170 		.procname	= "drop_caches",
1171 		.data		= &sysctl_drop_caches,
1172 		.maxlen		= sizeof(int),
1173 		.mode		= 0644,
1174 		.proc_handler	= drop_caches_sysctl_handler,
1175 		.extra1		= &one,
1176 		.extra2		= &three,
1177 	},
1178 #ifdef CONFIG_COMPACTION
1179 	{
1180 		.procname	= "compact_memory",
1181 		.data		= &sysctl_compact_memory,
1182 		.maxlen		= sizeof(int),
1183 		.mode		= 0200,
1184 		.proc_handler	= sysctl_compaction_handler,
1185 	},
1186 	{
1187 		.procname	= "extfrag_threshold",
1188 		.data		= &sysctl_extfrag_threshold,
1189 		.maxlen		= sizeof(int),
1190 		.mode		= 0644,
1191 		.proc_handler	= sysctl_extfrag_handler,
1192 		.extra1		= &min_extfrag_threshold,
1193 		.extra2		= &max_extfrag_threshold,
1194 	},
1195 
1196 #endif /* CONFIG_COMPACTION */
1197 	{
1198 		.procname	= "min_free_kbytes",
1199 		.data		= &min_free_kbytes,
1200 		.maxlen		= sizeof(min_free_kbytes),
1201 		.mode		= 0644,
1202 		.proc_handler	= min_free_kbytes_sysctl_handler,
1203 		.extra1		= &zero,
1204 	},
1205 	{
1206 		.procname	= "percpu_pagelist_fraction",
1207 		.data		= &percpu_pagelist_fraction,
1208 		.maxlen		= sizeof(percpu_pagelist_fraction),
1209 		.mode		= 0644,
1210 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1211 		.extra1		= &min_percpu_pagelist_fract,
1212 	},
1213 #ifdef CONFIG_MMU
1214 	{
1215 		.procname	= "max_map_count",
1216 		.data		= &sysctl_max_map_count,
1217 		.maxlen		= sizeof(sysctl_max_map_count),
1218 		.mode		= 0644,
1219 		.proc_handler	= proc_dointvec_minmax,
1220 		.extra1		= &zero,
1221 	},
1222 #else
1223 	{
1224 		.procname	= "nr_trim_pages",
1225 		.data		= &sysctl_nr_trim_pages,
1226 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1227 		.mode		= 0644,
1228 		.proc_handler	= proc_dointvec_minmax,
1229 		.extra1		= &zero,
1230 	},
1231 #endif
1232 	{
1233 		.procname	= "laptop_mode",
1234 		.data		= &laptop_mode,
1235 		.maxlen		= sizeof(laptop_mode),
1236 		.mode		= 0644,
1237 		.proc_handler	= proc_dointvec_jiffies,
1238 	},
1239 	{
1240 		.procname	= "block_dump",
1241 		.data		= &block_dump,
1242 		.maxlen		= sizeof(block_dump),
1243 		.mode		= 0644,
1244 		.proc_handler	= proc_dointvec,
1245 		.extra1		= &zero,
1246 	},
1247 	{
1248 		.procname	= "vfs_cache_pressure",
1249 		.data		= &sysctl_vfs_cache_pressure,
1250 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1251 		.mode		= 0644,
1252 		.proc_handler	= proc_dointvec,
1253 		.extra1		= &zero,
1254 	},
1255 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1256 	{
1257 		.procname	= "legacy_va_layout",
1258 		.data		= &sysctl_legacy_va_layout,
1259 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1260 		.mode		= 0644,
1261 		.proc_handler	= proc_dointvec,
1262 		.extra1		= &zero,
1263 	},
1264 #endif
1265 #ifdef CONFIG_NUMA
1266 	{
1267 		.procname	= "zone_reclaim_mode",
1268 		.data		= &zone_reclaim_mode,
1269 		.maxlen		= sizeof(zone_reclaim_mode),
1270 		.mode		= 0644,
1271 		.proc_handler	= proc_dointvec,
1272 		.extra1		= &zero,
1273 	},
1274 	{
1275 		.procname	= "min_unmapped_ratio",
1276 		.data		= &sysctl_min_unmapped_ratio,
1277 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1278 		.mode		= 0644,
1279 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1280 		.extra1		= &zero,
1281 		.extra2		= &one_hundred,
1282 	},
1283 	{
1284 		.procname	= "min_slab_ratio",
1285 		.data		= &sysctl_min_slab_ratio,
1286 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1287 		.mode		= 0644,
1288 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1289 		.extra1		= &zero,
1290 		.extra2		= &one_hundred,
1291 	},
1292 #endif
1293 #ifdef CONFIG_SMP
1294 	{
1295 		.procname	= "stat_interval",
1296 		.data		= &sysctl_stat_interval,
1297 		.maxlen		= sizeof(sysctl_stat_interval),
1298 		.mode		= 0644,
1299 		.proc_handler	= proc_dointvec_jiffies,
1300 	},
1301 #endif
1302 #ifdef CONFIG_MMU
1303 	{
1304 		.procname	= "mmap_min_addr",
1305 		.data		= &dac_mmap_min_addr,
1306 		.maxlen		= sizeof(unsigned long),
1307 		.mode		= 0644,
1308 		.proc_handler	= mmap_min_addr_handler,
1309 	},
1310 #endif
1311 #ifdef CONFIG_NUMA
1312 	{
1313 		.procname	= "numa_zonelist_order",
1314 		.data		= &numa_zonelist_order,
1315 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1316 		.mode		= 0644,
1317 		.proc_handler	= numa_zonelist_order_handler,
1318 	},
1319 #endif
1320 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1321    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1322 	{
1323 		.procname	= "vdso_enabled",
1324 		.data		= &vdso_enabled,
1325 		.maxlen		= sizeof(vdso_enabled),
1326 		.mode		= 0644,
1327 		.proc_handler	= proc_dointvec,
1328 		.extra1		= &zero,
1329 	},
1330 #endif
1331 #ifdef CONFIG_HIGHMEM
1332 	{
1333 		.procname	= "highmem_is_dirtyable",
1334 		.data		= &vm_highmem_is_dirtyable,
1335 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1336 		.mode		= 0644,
1337 		.proc_handler	= proc_dointvec_minmax,
1338 		.extra1		= &zero,
1339 		.extra2		= &one,
1340 	},
1341 #endif
1342 	{
1343 		.procname	= "scan_unevictable_pages",
1344 		.data		= &scan_unevictable_pages,
1345 		.maxlen		= sizeof(scan_unevictable_pages),
1346 		.mode		= 0644,
1347 		.proc_handler	= scan_unevictable_handler,
1348 	},
1349 #ifdef CONFIG_MEMORY_FAILURE
1350 	{
1351 		.procname	= "memory_failure_early_kill",
1352 		.data		= &sysctl_memory_failure_early_kill,
1353 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1354 		.mode		= 0644,
1355 		.proc_handler	= proc_dointvec_minmax,
1356 		.extra1		= &zero,
1357 		.extra2		= &one,
1358 	},
1359 	{
1360 		.procname	= "memory_failure_recovery",
1361 		.data		= &sysctl_memory_failure_recovery,
1362 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1363 		.mode		= 0644,
1364 		.proc_handler	= proc_dointvec_minmax,
1365 		.extra1		= &zero,
1366 		.extra2		= &one,
1367 	},
1368 #endif
1369 	{ }
1370 };
1371 
1372 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1373 static struct ctl_table binfmt_misc_table[] = {
1374 	{ }
1375 };
1376 #endif
1377 
1378 static struct ctl_table fs_table[] = {
1379 	{
1380 		.procname	= "inode-nr",
1381 		.data		= &inodes_stat,
1382 		.maxlen		= 2*sizeof(int),
1383 		.mode		= 0444,
1384 		.proc_handler	= proc_nr_inodes,
1385 	},
1386 	{
1387 		.procname	= "inode-state",
1388 		.data		= &inodes_stat,
1389 		.maxlen		= 7*sizeof(int),
1390 		.mode		= 0444,
1391 		.proc_handler	= proc_nr_inodes,
1392 	},
1393 	{
1394 		.procname	= "file-nr",
1395 		.data		= &files_stat,
1396 		.maxlen		= sizeof(files_stat),
1397 		.mode		= 0444,
1398 		.proc_handler	= proc_nr_files,
1399 	},
1400 	{
1401 		.procname	= "file-max",
1402 		.data		= &files_stat.max_files,
1403 		.maxlen		= sizeof(files_stat.max_files),
1404 		.mode		= 0644,
1405 		.proc_handler	= proc_doulongvec_minmax,
1406 	},
1407 	{
1408 		.procname	= "nr_open",
1409 		.data		= &sysctl_nr_open,
1410 		.maxlen		= sizeof(int),
1411 		.mode		= 0644,
1412 		.proc_handler	= proc_dointvec_minmax,
1413 		.extra1		= &sysctl_nr_open_min,
1414 		.extra2		= &sysctl_nr_open_max,
1415 	},
1416 	{
1417 		.procname	= "dentry-state",
1418 		.data		= &dentry_stat,
1419 		.maxlen		= 6*sizeof(int),
1420 		.mode		= 0444,
1421 		.proc_handler	= proc_nr_dentry,
1422 	},
1423 	{
1424 		.procname	= "overflowuid",
1425 		.data		= &fs_overflowuid,
1426 		.maxlen		= sizeof(int),
1427 		.mode		= 0644,
1428 		.proc_handler	= proc_dointvec_minmax,
1429 		.extra1		= &minolduid,
1430 		.extra2		= &maxolduid,
1431 	},
1432 	{
1433 		.procname	= "overflowgid",
1434 		.data		= &fs_overflowgid,
1435 		.maxlen		= sizeof(int),
1436 		.mode		= 0644,
1437 		.proc_handler	= proc_dointvec_minmax,
1438 		.extra1		= &minolduid,
1439 		.extra2		= &maxolduid,
1440 	},
1441 #ifdef CONFIG_FILE_LOCKING
1442 	{
1443 		.procname	= "leases-enable",
1444 		.data		= &leases_enable,
1445 		.maxlen		= sizeof(int),
1446 		.mode		= 0644,
1447 		.proc_handler	= proc_dointvec,
1448 	},
1449 #endif
1450 #ifdef CONFIG_DNOTIFY
1451 	{
1452 		.procname	= "dir-notify-enable",
1453 		.data		= &dir_notify_enable,
1454 		.maxlen		= sizeof(int),
1455 		.mode		= 0644,
1456 		.proc_handler	= proc_dointvec,
1457 	},
1458 #endif
1459 #ifdef CONFIG_MMU
1460 #ifdef CONFIG_FILE_LOCKING
1461 	{
1462 		.procname	= "lease-break-time",
1463 		.data		= &lease_break_time,
1464 		.maxlen		= sizeof(int),
1465 		.mode		= 0644,
1466 		.proc_handler	= proc_dointvec,
1467 	},
1468 #endif
1469 #ifdef CONFIG_AIO
1470 	{
1471 		.procname	= "aio-nr",
1472 		.data		= &aio_nr,
1473 		.maxlen		= sizeof(aio_nr),
1474 		.mode		= 0444,
1475 		.proc_handler	= proc_doulongvec_minmax,
1476 	},
1477 	{
1478 		.procname	= "aio-max-nr",
1479 		.data		= &aio_max_nr,
1480 		.maxlen		= sizeof(aio_max_nr),
1481 		.mode		= 0644,
1482 		.proc_handler	= proc_doulongvec_minmax,
1483 	},
1484 #endif /* CONFIG_AIO */
1485 #ifdef CONFIG_INOTIFY_USER
1486 	{
1487 		.procname	= "inotify",
1488 		.mode		= 0555,
1489 		.child		= inotify_table,
1490 	},
1491 #endif
1492 #ifdef CONFIG_EPOLL
1493 	{
1494 		.procname	= "epoll",
1495 		.mode		= 0555,
1496 		.child		= epoll_table,
1497 	},
1498 #endif
1499 #endif
1500 	{
1501 		.procname	= "protected_symlinks",
1502 		.data		= &sysctl_protected_symlinks,
1503 		.maxlen		= sizeof(int),
1504 		.mode		= 0600,
1505 		.proc_handler	= proc_dointvec_minmax,
1506 		.extra1		= &zero,
1507 		.extra2		= &one,
1508 	},
1509 	{
1510 		.procname	= "protected_hardlinks",
1511 		.data		= &sysctl_protected_hardlinks,
1512 		.maxlen		= sizeof(int),
1513 		.mode		= 0600,
1514 		.proc_handler	= proc_dointvec_minmax,
1515 		.extra1		= &zero,
1516 		.extra2		= &one,
1517 	},
1518 	{
1519 		.procname	= "suid_dumpable",
1520 		.data		= &suid_dumpable,
1521 		.maxlen		= sizeof(int),
1522 		.mode		= 0644,
1523 		.proc_handler	= proc_dointvec_minmax_coredump,
1524 		.extra1		= &zero,
1525 		.extra2		= &two,
1526 	},
1527 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1528 	{
1529 		.procname	= "binfmt_misc",
1530 		.mode		= 0555,
1531 		.child		= binfmt_misc_table,
1532 	},
1533 #endif
1534 	{
1535 		.procname	= "pipe-max-size",
1536 		.data		= &pipe_max_size,
1537 		.maxlen		= sizeof(int),
1538 		.mode		= 0644,
1539 		.proc_handler	= &pipe_proc_fn,
1540 		.extra1		= &pipe_min_size,
1541 	},
1542 	{ }
1543 };
1544 
1545 static struct ctl_table debug_table[] = {
1546 #if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
1547     defined(CONFIG_S390) || defined(CONFIG_TILE)
1548 	{
1549 		.procname	= "exception-trace",
1550 		.data		= &show_unhandled_signals,
1551 		.maxlen		= sizeof(int),
1552 		.mode		= 0644,
1553 		.proc_handler	= proc_dointvec
1554 	},
1555 #endif
1556 #if defined(CONFIG_OPTPROBES)
1557 	{
1558 		.procname	= "kprobes-optimization",
1559 		.data		= &sysctl_kprobes_optimization,
1560 		.maxlen		= sizeof(int),
1561 		.mode		= 0644,
1562 		.proc_handler	= proc_kprobes_optimization_handler,
1563 		.extra1		= &zero,
1564 		.extra2		= &one,
1565 	},
1566 #endif
1567 	{ }
1568 };
1569 
1570 static struct ctl_table dev_table[] = {
1571 	{ }
1572 };
1573 
1574 int __init sysctl_init(void)
1575 {
1576 	struct ctl_table_header *hdr;
1577 
1578 	hdr = register_sysctl_table(sysctl_base_table);
1579 	kmemleak_not_leak(hdr);
1580 	return 0;
1581 }
1582 
1583 #endif /* CONFIG_SYSCTL */
1584 
1585 /*
1586  * /proc/sys support
1587  */
1588 
1589 #ifdef CONFIG_PROC_SYSCTL
1590 
1591 static int _proc_do_string(void* data, int maxlen, int write,
1592 			   void __user *buffer,
1593 			   size_t *lenp, loff_t *ppos)
1594 {
1595 	size_t len;
1596 	char __user *p;
1597 	char c;
1598 
1599 	if (!data || !maxlen || !*lenp) {
1600 		*lenp = 0;
1601 		return 0;
1602 	}
1603 
1604 	if (write) {
1605 		len = 0;
1606 		p = buffer;
1607 		while (len < *lenp) {
1608 			if (get_user(c, p++))
1609 				return -EFAULT;
1610 			if (c == 0 || c == '\n')
1611 				break;
1612 			len++;
1613 		}
1614 		if (len >= maxlen)
1615 			len = maxlen-1;
1616 		if(copy_from_user(data, buffer, len))
1617 			return -EFAULT;
1618 		((char *) data)[len] = 0;
1619 		*ppos += *lenp;
1620 	} else {
1621 		len = strlen(data);
1622 		if (len > maxlen)
1623 			len = maxlen;
1624 
1625 		if (*ppos > len) {
1626 			*lenp = 0;
1627 			return 0;
1628 		}
1629 
1630 		data += *ppos;
1631 		len  -= *ppos;
1632 
1633 		if (len > *lenp)
1634 			len = *lenp;
1635 		if (len)
1636 			if(copy_to_user(buffer, data, len))
1637 				return -EFAULT;
1638 		if (len < *lenp) {
1639 			if(put_user('\n', ((char __user *) buffer) + len))
1640 				return -EFAULT;
1641 			len++;
1642 		}
1643 		*lenp = len;
1644 		*ppos += len;
1645 	}
1646 	return 0;
1647 }
1648 
1649 /**
1650  * proc_dostring - read a string sysctl
1651  * @table: the sysctl table
1652  * @write: %TRUE if this is a write to the sysctl file
1653  * @buffer: the user buffer
1654  * @lenp: the size of the user buffer
1655  * @ppos: file position
1656  *
1657  * Reads/writes a string from/to the user buffer. If the kernel
1658  * buffer provided is not large enough to hold the string, the
1659  * string is truncated. The copied string is %NULL-terminated.
1660  * If the string is being read by the user process, it is copied
1661  * and a newline '\n' is added. It is truncated if the buffer is
1662  * not large enough.
1663  *
1664  * Returns 0 on success.
1665  */
1666 int proc_dostring(struct ctl_table *table, int write,
1667 		  void __user *buffer, size_t *lenp, loff_t *ppos)
1668 {
1669 	return _proc_do_string(table->data, table->maxlen, write,
1670 			       buffer, lenp, ppos);
1671 }
1672 
1673 static size_t proc_skip_spaces(char **buf)
1674 {
1675 	size_t ret;
1676 	char *tmp = skip_spaces(*buf);
1677 	ret = tmp - *buf;
1678 	*buf = tmp;
1679 	return ret;
1680 }
1681 
1682 static void proc_skip_char(char **buf, size_t *size, const char v)
1683 {
1684 	while (*size) {
1685 		if (**buf != v)
1686 			break;
1687 		(*size)--;
1688 		(*buf)++;
1689 	}
1690 }
1691 
1692 #define TMPBUFLEN 22
1693 /**
1694  * proc_get_long - reads an ASCII formatted integer from a user buffer
1695  *
1696  * @buf: a kernel buffer
1697  * @size: size of the kernel buffer
1698  * @val: this is where the number will be stored
1699  * @neg: set to %TRUE if number is negative
1700  * @perm_tr: a vector which contains the allowed trailers
1701  * @perm_tr_len: size of the perm_tr vector
1702  * @tr: pointer to store the trailer character
1703  *
1704  * In case of success %0 is returned and @buf and @size are updated with
1705  * the amount of bytes read. If @tr is non-NULL and a trailing
1706  * character exists (size is non-zero after returning from this
1707  * function), @tr is updated with the trailing character.
1708  */
1709 static int proc_get_long(char **buf, size_t *size,
1710 			  unsigned long *val, bool *neg,
1711 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
1712 {
1713 	int len;
1714 	char *p, tmp[TMPBUFLEN];
1715 
1716 	if (!*size)
1717 		return -EINVAL;
1718 
1719 	len = *size;
1720 	if (len > TMPBUFLEN - 1)
1721 		len = TMPBUFLEN - 1;
1722 
1723 	memcpy(tmp, *buf, len);
1724 
1725 	tmp[len] = 0;
1726 	p = tmp;
1727 	if (*p == '-' && *size > 1) {
1728 		*neg = true;
1729 		p++;
1730 	} else
1731 		*neg = false;
1732 	if (!isdigit(*p))
1733 		return -EINVAL;
1734 
1735 	*val = simple_strtoul(p, &p, 0);
1736 
1737 	len = p - tmp;
1738 
1739 	/* We don't know if the next char is whitespace thus we may accept
1740 	 * invalid integers (e.g. 1234...a) or two integers instead of one
1741 	 * (e.g. 123...1). So lets not allow such large numbers. */
1742 	if (len == TMPBUFLEN - 1)
1743 		return -EINVAL;
1744 
1745 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
1746 		return -EINVAL;
1747 
1748 	if (tr && (len < *size))
1749 		*tr = *p;
1750 
1751 	*buf += len;
1752 	*size -= len;
1753 
1754 	return 0;
1755 }
1756 
1757 /**
1758  * proc_put_long - converts an integer to a decimal ASCII formatted string
1759  *
1760  * @buf: the user buffer
1761  * @size: the size of the user buffer
1762  * @val: the integer to be converted
1763  * @neg: sign of the number, %TRUE for negative
1764  *
1765  * In case of success %0 is returned and @buf and @size are updated with
1766  * the amount of bytes written.
1767  */
1768 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
1769 			  bool neg)
1770 {
1771 	int len;
1772 	char tmp[TMPBUFLEN], *p = tmp;
1773 
1774 	sprintf(p, "%s%lu", neg ? "-" : "", val);
1775 	len = strlen(tmp);
1776 	if (len > *size)
1777 		len = *size;
1778 	if (copy_to_user(*buf, tmp, len))
1779 		return -EFAULT;
1780 	*size -= len;
1781 	*buf += len;
1782 	return 0;
1783 }
1784 #undef TMPBUFLEN
1785 
1786 static int proc_put_char(void __user **buf, size_t *size, char c)
1787 {
1788 	if (*size) {
1789 		char __user **buffer = (char __user **)buf;
1790 		if (put_user(c, *buffer))
1791 			return -EFAULT;
1792 		(*size)--, (*buffer)++;
1793 		*buf = *buffer;
1794 	}
1795 	return 0;
1796 }
1797 
1798 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
1799 				 int *valp,
1800 				 int write, void *data)
1801 {
1802 	if (write) {
1803 		*valp = *negp ? -*lvalp : *lvalp;
1804 	} else {
1805 		int val = *valp;
1806 		if (val < 0) {
1807 			*negp = true;
1808 			*lvalp = (unsigned long)-val;
1809 		} else {
1810 			*negp = false;
1811 			*lvalp = (unsigned long)val;
1812 		}
1813 	}
1814 	return 0;
1815 }
1816 
1817 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
1818 
1819 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1820 		  int write, void __user *buffer,
1821 		  size_t *lenp, loff_t *ppos,
1822 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1823 			      int write, void *data),
1824 		  void *data)
1825 {
1826 	int *i, vleft, first = 1, err = 0;
1827 	unsigned long page = 0;
1828 	size_t left;
1829 	char *kbuf;
1830 
1831 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
1832 		*lenp = 0;
1833 		return 0;
1834 	}
1835 
1836 	i = (int *) tbl_data;
1837 	vleft = table->maxlen / sizeof(*i);
1838 	left = *lenp;
1839 
1840 	if (!conv)
1841 		conv = do_proc_dointvec_conv;
1842 
1843 	if (write) {
1844 		if (left > PAGE_SIZE - 1)
1845 			left = PAGE_SIZE - 1;
1846 		page = __get_free_page(GFP_TEMPORARY);
1847 		kbuf = (char *) page;
1848 		if (!kbuf)
1849 			return -ENOMEM;
1850 		if (copy_from_user(kbuf, buffer, left)) {
1851 			err = -EFAULT;
1852 			goto free;
1853 		}
1854 		kbuf[left] = 0;
1855 	}
1856 
1857 	for (; left && vleft--; i++, first=0) {
1858 		unsigned long lval;
1859 		bool neg;
1860 
1861 		if (write) {
1862 			left -= proc_skip_spaces(&kbuf);
1863 
1864 			if (!left)
1865 				break;
1866 			err = proc_get_long(&kbuf, &left, &lval, &neg,
1867 					     proc_wspace_sep,
1868 					     sizeof(proc_wspace_sep), NULL);
1869 			if (err)
1870 				break;
1871 			if (conv(&neg, &lval, i, 1, data)) {
1872 				err = -EINVAL;
1873 				break;
1874 			}
1875 		} else {
1876 			if (conv(&neg, &lval, i, 0, data)) {
1877 				err = -EINVAL;
1878 				break;
1879 			}
1880 			if (!first)
1881 				err = proc_put_char(&buffer, &left, '\t');
1882 			if (err)
1883 				break;
1884 			err = proc_put_long(&buffer, &left, lval, neg);
1885 			if (err)
1886 				break;
1887 		}
1888 	}
1889 
1890 	if (!write && !first && left && !err)
1891 		err = proc_put_char(&buffer, &left, '\n');
1892 	if (write && !err && left)
1893 		left -= proc_skip_spaces(&kbuf);
1894 free:
1895 	if (write) {
1896 		free_page(page);
1897 		if (first)
1898 			return err ? : -EINVAL;
1899 	}
1900 	*lenp -= left;
1901 	*ppos += *lenp;
1902 	return err;
1903 }
1904 
1905 static int do_proc_dointvec(struct ctl_table *table, int write,
1906 		  void __user *buffer, size_t *lenp, loff_t *ppos,
1907 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1908 			      int write, void *data),
1909 		  void *data)
1910 {
1911 	return __do_proc_dointvec(table->data, table, write,
1912 			buffer, lenp, ppos, conv, data);
1913 }
1914 
1915 /**
1916  * proc_dointvec - read a vector of integers
1917  * @table: the sysctl table
1918  * @write: %TRUE if this is a write to the sysctl file
1919  * @buffer: the user buffer
1920  * @lenp: the size of the user buffer
1921  * @ppos: file position
1922  *
1923  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1924  * values from/to the user buffer, treated as an ASCII string.
1925  *
1926  * Returns 0 on success.
1927  */
1928 int proc_dointvec(struct ctl_table *table, int write,
1929 		     void __user *buffer, size_t *lenp, loff_t *ppos)
1930 {
1931     return do_proc_dointvec(table,write,buffer,lenp,ppos,
1932 		    	    NULL,NULL);
1933 }
1934 
1935 /*
1936  * Taint values can only be increased
1937  * This means we can safely use a temporary.
1938  */
1939 static int proc_taint(struct ctl_table *table, int write,
1940 			       void __user *buffer, size_t *lenp, loff_t *ppos)
1941 {
1942 	struct ctl_table t;
1943 	unsigned long tmptaint = get_taint();
1944 	int err;
1945 
1946 	if (write && !capable(CAP_SYS_ADMIN))
1947 		return -EPERM;
1948 
1949 	t = *table;
1950 	t.data = &tmptaint;
1951 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
1952 	if (err < 0)
1953 		return err;
1954 
1955 	if (write) {
1956 		/*
1957 		 * Poor man's atomic or. Not worth adding a primitive
1958 		 * to everyone's atomic.h for this
1959 		 */
1960 		int i;
1961 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
1962 			if ((tmptaint >> i) & 1)
1963 				add_taint(i);
1964 		}
1965 	}
1966 
1967 	return err;
1968 }
1969 
1970 #ifdef CONFIG_PRINTK
1971 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
1972 				void __user *buffer, size_t *lenp, loff_t *ppos)
1973 {
1974 	if (write && !capable(CAP_SYS_ADMIN))
1975 		return -EPERM;
1976 
1977 	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
1978 }
1979 #endif
1980 
1981 struct do_proc_dointvec_minmax_conv_param {
1982 	int *min;
1983 	int *max;
1984 };
1985 
1986 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
1987 					int *valp,
1988 					int write, void *data)
1989 {
1990 	struct do_proc_dointvec_minmax_conv_param *param = data;
1991 	if (write) {
1992 		int val = *negp ? -*lvalp : *lvalp;
1993 		if ((param->min && *param->min > val) ||
1994 		    (param->max && *param->max < val))
1995 			return -EINVAL;
1996 		*valp = val;
1997 	} else {
1998 		int val = *valp;
1999 		if (val < 0) {
2000 			*negp = true;
2001 			*lvalp = (unsigned long)-val;
2002 		} else {
2003 			*negp = false;
2004 			*lvalp = (unsigned long)val;
2005 		}
2006 	}
2007 	return 0;
2008 }
2009 
2010 /**
2011  * proc_dointvec_minmax - read a vector of integers with min/max values
2012  * @table: the sysctl table
2013  * @write: %TRUE if this is a write to the sysctl file
2014  * @buffer: the user buffer
2015  * @lenp: the size of the user buffer
2016  * @ppos: file position
2017  *
2018  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2019  * values from/to the user buffer, treated as an ASCII string.
2020  *
2021  * This routine will ensure the values are within the range specified by
2022  * table->extra1 (min) and table->extra2 (max).
2023  *
2024  * Returns 0 on success.
2025  */
2026 int proc_dointvec_minmax(struct ctl_table *table, int write,
2027 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2028 {
2029 	struct do_proc_dointvec_minmax_conv_param param = {
2030 		.min = (int *) table->extra1,
2031 		.max = (int *) table->extra2,
2032 	};
2033 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2034 				do_proc_dointvec_minmax_conv, &param);
2035 }
2036 
2037 static void validate_coredump_safety(void)
2038 {
2039 	if (suid_dumpable == SUID_DUMPABLE_SAFE &&
2040 	    core_pattern[0] != '/' && core_pattern[0] != '|') {
2041 		printk(KERN_WARNING "Unsafe core_pattern used with "\
2042 			"suid_dumpable=2. Pipe handler or fully qualified "\
2043 			"core dump path required.\n");
2044 	}
2045 }
2046 
2047 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2048 		void __user *buffer, size_t *lenp, loff_t *ppos)
2049 {
2050 	int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2051 	if (!error)
2052 		validate_coredump_safety();
2053 	return error;
2054 }
2055 
2056 static int proc_dostring_coredump(struct ctl_table *table, int write,
2057 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2058 {
2059 	int error = proc_dostring(table, write, buffer, lenp, ppos);
2060 	if (!error)
2061 		validate_coredump_safety();
2062 	return error;
2063 }
2064 
2065 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2066 				     void __user *buffer,
2067 				     size_t *lenp, loff_t *ppos,
2068 				     unsigned long convmul,
2069 				     unsigned long convdiv)
2070 {
2071 	unsigned long *i, *min, *max;
2072 	int vleft, first = 1, err = 0;
2073 	unsigned long page = 0;
2074 	size_t left;
2075 	char *kbuf;
2076 
2077 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2078 		*lenp = 0;
2079 		return 0;
2080 	}
2081 
2082 	i = (unsigned long *) data;
2083 	min = (unsigned long *) table->extra1;
2084 	max = (unsigned long *) table->extra2;
2085 	vleft = table->maxlen / sizeof(unsigned long);
2086 	left = *lenp;
2087 
2088 	if (write) {
2089 		if (left > PAGE_SIZE - 1)
2090 			left = PAGE_SIZE - 1;
2091 		page = __get_free_page(GFP_TEMPORARY);
2092 		kbuf = (char *) page;
2093 		if (!kbuf)
2094 			return -ENOMEM;
2095 		if (copy_from_user(kbuf, buffer, left)) {
2096 			err = -EFAULT;
2097 			goto free;
2098 		}
2099 		kbuf[left] = 0;
2100 	}
2101 
2102 	for (; left && vleft--; i++, first = 0) {
2103 		unsigned long val;
2104 
2105 		if (write) {
2106 			bool neg;
2107 
2108 			left -= proc_skip_spaces(&kbuf);
2109 
2110 			err = proc_get_long(&kbuf, &left, &val, &neg,
2111 					     proc_wspace_sep,
2112 					     sizeof(proc_wspace_sep), NULL);
2113 			if (err)
2114 				break;
2115 			if (neg)
2116 				continue;
2117 			if ((min && val < *min) || (max && val > *max))
2118 				continue;
2119 			*i = val;
2120 		} else {
2121 			val = convdiv * (*i) / convmul;
2122 			if (!first)
2123 				err = proc_put_char(&buffer, &left, '\t');
2124 			err = proc_put_long(&buffer, &left, val, false);
2125 			if (err)
2126 				break;
2127 		}
2128 	}
2129 
2130 	if (!write && !first && left && !err)
2131 		err = proc_put_char(&buffer, &left, '\n');
2132 	if (write && !err)
2133 		left -= proc_skip_spaces(&kbuf);
2134 free:
2135 	if (write) {
2136 		free_page(page);
2137 		if (first)
2138 			return err ? : -EINVAL;
2139 	}
2140 	*lenp -= left;
2141 	*ppos += *lenp;
2142 	return err;
2143 }
2144 
2145 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2146 				     void __user *buffer,
2147 				     size_t *lenp, loff_t *ppos,
2148 				     unsigned long convmul,
2149 				     unsigned long convdiv)
2150 {
2151 	return __do_proc_doulongvec_minmax(table->data, table, write,
2152 			buffer, lenp, ppos, convmul, convdiv);
2153 }
2154 
2155 /**
2156  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2157  * @table: the sysctl table
2158  * @write: %TRUE if this is a write to the sysctl file
2159  * @buffer: the user buffer
2160  * @lenp: the size of the user buffer
2161  * @ppos: file position
2162  *
2163  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2164  * values from/to the user buffer, treated as an ASCII string.
2165  *
2166  * This routine will ensure the values are within the range specified by
2167  * table->extra1 (min) and table->extra2 (max).
2168  *
2169  * Returns 0 on success.
2170  */
2171 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2172 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2173 {
2174     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2175 }
2176 
2177 /**
2178  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2179  * @table: the sysctl table
2180  * @write: %TRUE if this is a write to the sysctl file
2181  * @buffer: the user buffer
2182  * @lenp: the size of the user buffer
2183  * @ppos: file position
2184  *
2185  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2186  * values from/to the user buffer, treated as an ASCII string. The values
2187  * are treated as milliseconds, and converted to jiffies when they are stored.
2188  *
2189  * This routine will ensure the values are within the range specified by
2190  * table->extra1 (min) and table->extra2 (max).
2191  *
2192  * Returns 0 on success.
2193  */
2194 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2195 				      void __user *buffer,
2196 				      size_t *lenp, loff_t *ppos)
2197 {
2198     return do_proc_doulongvec_minmax(table, write, buffer,
2199 				     lenp, ppos, HZ, 1000l);
2200 }
2201 
2202 
2203 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2204 					 int *valp,
2205 					 int write, void *data)
2206 {
2207 	if (write) {
2208 		if (*lvalp > LONG_MAX / HZ)
2209 			return 1;
2210 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2211 	} else {
2212 		int val = *valp;
2213 		unsigned long lval;
2214 		if (val < 0) {
2215 			*negp = true;
2216 			lval = (unsigned long)-val;
2217 		} else {
2218 			*negp = false;
2219 			lval = (unsigned long)val;
2220 		}
2221 		*lvalp = lval / HZ;
2222 	}
2223 	return 0;
2224 }
2225 
2226 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2227 						int *valp,
2228 						int write, void *data)
2229 {
2230 	if (write) {
2231 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2232 			return 1;
2233 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2234 	} else {
2235 		int val = *valp;
2236 		unsigned long lval;
2237 		if (val < 0) {
2238 			*negp = true;
2239 			lval = (unsigned long)-val;
2240 		} else {
2241 			*negp = false;
2242 			lval = (unsigned long)val;
2243 		}
2244 		*lvalp = jiffies_to_clock_t(lval);
2245 	}
2246 	return 0;
2247 }
2248 
2249 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2250 					    int *valp,
2251 					    int write, void *data)
2252 {
2253 	if (write) {
2254 		*valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2255 	} else {
2256 		int val = *valp;
2257 		unsigned long lval;
2258 		if (val < 0) {
2259 			*negp = true;
2260 			lval = (unsigned long)-val;
2261 		} else {
2262 			*negp = false;
2263 			lval = (unsigned long)val;
2264 		}
2265 		*lvalp = jiffies_to_msecs(lval);
2266 	}
2267 	return 0;
2268 }
2269 
2270 /**
2271  * proc_dointvec_jiffies - read a vector of integers as seconds
2272  * @table: the sysctl table
2273  * @write: %TRUE if this is a write to the sysctl file
2274  * @buffer: the user buffer
2275  * @lenp: the size of the user buffer
2276  * @ppos: file position
2277  *
2278  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2279  * values from/to the user buffer, treated as an ASCII string.
2280  * The values read are assumed to be in seconds, and are converted into
2281  * jiffies.
2282  *
2283  * Returns 0 on success.
2284  */
2285 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2286 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2287 {
2288     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2289 		    	    do_proc_dointvec_jiffies_conv,NULL);
2290 }
2291 
2292 /**
2293  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2294  * @table: the sysctl table
2295  * @write: %TRUE if this is a write to the sysctl file
2296  * @buffer: the user buffer
2297  * @lenp: the size of the user buffer
2298  * @ppos: pointer to the file position
2299  *
2300  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2301  * values from/to the user buffer, treated as an ASCII string.
2302  * The values read are assumed to be in 1/USER_HZ seconds, and
2303  * are converted into jiffies.
2304  *
2305  * Returns 0 on success.
2306  */
2307 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2308 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2309 {
2310     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2311 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2312 }
2313 
2314 /**
2315  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2316  * @table: the sysctl table
2317  * @write: %TRUE if this is a write to the sysctl file
2318  * @buffer: the user buffer
2319  * @lenp: the size of the user buffer
2320  * @ppos: file position
2321  * @ppos: the current position in the file
2322  *
2323  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2324  * values from/to the user buffer, treated as an ASCII string.
2325  * The values read are assumed to be in 1/1000 seconds, and
2326  * are converted into jiffies.
2327  *
2328  * Returns 0 on success.
2329  */
2330 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2331 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2332 {
2333 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2334 				do_proc_dointvec_ms_jiffies_conv, NULL);
2335 }
2336 
2337 static int proc_do_cad_pid(struct ctl_table *table, int write,
2338 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2339 {
2340 	struct pid *new_pid;
2341 	pid_t tmp;
2342 	int r;
2343 
2344 	tmp = pid_vnr(cad_pid);
2345 
2346 	r = __do_proc_dointvec(&tmp, table, write, buffer,
2347 			       lenp, ppos, NULL, NULL);
2348 	if (r || !write)
2349 		return r;
2350 
2351 	new_pid = find_get_pid(tmp);
2352 	if (!new_pid)
2353 		return -ESRCH;
2354 
2355 	put_pid(xchg(&cad_pid, new_pid));
2356 	return 0;
2357 }
2358 
2359 /**
2360  * proc_do_large_bitmap - read/write from/to a large bitmap
2361  * @table: the sysctl table
2362  * @write: %TRUE if this is a write to the sysctl file
2363  * @buffer: the user buffer
2364  * @lenp: the size of the user buffer
2365  * @ppos: file position
2366  *
2367  * The bitmap is stored at table->data and the bitmap length (in bits)
2368  * in table->maxlen.
2369  *
2370  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2371  * large bitmaps may be represented in a compact manner. Writing into
2372  * the file will clear the bitmap then update it with the given input.
2373  *
2374  * Returns 0 on success.
2375  */
2376 int proc_do_large_bitmap(struct ctl_table *table, int write,
2377 			 void __user *buffer, size_t *lenp, loff_t *ppos)
2378 {
2379 	int err = 0;
2380 	bool first = 1;
2381 	size_t left = *lenp;
2382 	unsigned long bitmap_len = table->maxlen;
2383 	unsigned long *bitmap = (unsigned long *) table->data;
2384 	unsigned long *tmp_bitmap = NULL;
2385 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2386 
2387 	if (!bitmap_len || !left || (*ppos && !write)) {
2388 		*lenp = 0;
2389 		return 0;
2390 	}
2391 
2392 	if (write) {
2393 		unsigned long page = 0;
2394 		char *kbuf;
2395 
2396 		if (left > PAGE_SIZE - 1)
2397 			left = PAGE_SIZE - 1;
2398 
2399 		page = __get_free_page(GFP_TEMPORARY);
2400 		kbuf = (char *) page;
2401 		if (!kbuf)
2402 			return -ENOMEM;
2403 		if (copy_from_user(kbuf, buffer, left)) {
2404 			free_page(page);
2405 			return -EFAULT;
2406                 }
2407 		kbuf[left] = 0;
2408 
2409 		tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2410 				     GFP_KERNEL);
2411 		if (!tmp_bitmap) {
2412 			free_page(page);
2413 			return -ENOMEM;
2414 		}
2415 		proc_skip_char(&kbuf, &left, '\n');
2416 		while (!err && left) {
2417 			unsigned long val_a, val_b;
2418 			bool neg;
2419 
2420 			err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2421 					     sizeof(tr_a), &c);
2422 			if (err)
2423 				break;
2424 			if (val_a >= bitmap_len || neg) {
2425 				err = -EINVAL;
2426 				break;
2427 			}
2428 
2429 			val_b = val_a;
2430 			if (left) {
2431 				kbuf++;
2432 				left--;
2433 			}
2434 
2435 			if (c == '-') {
2436 				err = proc_get_long(&kbuf, &left, &val_b,
2437 						     &neg, tr_b, sizeof(tr_b),
2438 						     &c);
2439 				if (err)
2440 					break;
2441 				if (val_b >= bitmap_len || neg ||
2442 				    val_a > val_b) {
2443 					err = -EINVAL;
2444 					break;
2445 				}
2446 				if (left) {
2447 					kbuf++;
2448 					left--;
2449 				}
2450 			}
2451 
2452 			bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2453 			first = 0;
2454 			proc_skip_char(&kbuf, &left, '\n');
2455 		}
2456 		free_page(page);
2457 	} else {
2458 		unsigned long bit_a, bit_b = 0;
2459 
2460 		while (left) {
2461 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2462 			if (bit_a >= bitmap_len)
2463 				break;
2464 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
2465 						   bit_a + 1) - 1;
2466 
2467 			if (!first) {
2468 				err = proc_put_char(&buffer, &left, ',');
2469 				if (err)
2470 					break;
2471 			}
2472 			err = proc_put_long(&buffer, &left, bit_a, false);
2473 			if (err)
2474 				break;
2475 			if (bit_a != bit_b) {
2476 				err = proc_put_char(&buffer, &left, '-');
2477 				if (err)
2478 					break;
2479 				err = proc_put_long(&buffer, &left, bit_b, false);
2480 				if (err)
2481 					break;
2482 			}
2483 
2484 			first = 0; bit_b++;
2485 		}
2486 		if (!err)
2487 			err = proc_put_char(&buffer, &left, '\n');
2488 	}
2489 
2490 	if (!err) {
2491 		if (write) {
2492 			if (*ppos)
2493 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2494 			else
2495 				bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2496 		}
2497 		kfree(tmp_bitmap);
2498 		*lenp -= left;
2499 		*ppos += *lenp;
2500 		return 0;
2501 	} else {
2502 		kfree(tmp_bitmap);
2503 		return err;
2504 	}
2505 }
2506 
2507 #else /* CONFIG_PROC_SYSCTL */
2508 
2509 int proc_dostring(struct ctl_table *table, int write,
2510 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2511 {
2512 	return -ENOSYS;
2513 }
2514 
2515 int proc_dointvec(struct ctl_table *table, int write,
2516 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2517 {
2518 	return -ENOSYS;
2519 }
2520 
2521 int proc_dointvec_minmax(struct ctl_table *table, int write,
2522 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2523 {
2524 	return -ENOSYS;
2525 }
2526 
2527 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2528 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2529 {
2530 	return -ENOSYS;
2531 }
2532 
2533 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2534 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2535 {
2536 	return -ENOSYS;
2537 }
2538 
2539 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2540 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2541 {
2542 	return -ENOSYS;
2543 }
2544 
2545 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2546 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2547 {
2548 	return -ENOSYS;
2549 }
2550 
2551 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2552 				      void __user *buffer,
2553 				      size_t *lenp, loff_t *ppos)
2554 {
2555     return -ENOSYS;
2556 }
2557 
2558 
2559 #endif /* CONFIG_PROC_SYSCTL */
2560 
2561 /*
2562  * No sense putting this after each symbol definition, twice,
2563  * exception granted :-)
2564  */
2565 EXPORT_SYMBOL(proc_dointvec);
2566 EXPORT_SYMBOL(proc_dointvec_jiffies);
2567 EXPORT_SYMBOL(proc_dointvec_minmax);
2568 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2569 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2570 EXPORT_SYMBOL(proc_dostring);
2571 EXPORT_SYMBOL(proc_doulongvec_minmax);
2572 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2573