xref: /openbmc/linux/kernel/sysctl.c (revision 8c749ce9)
1 /*
2  * sysctl.c: General linux system control interface
3  *
4  * Begun 24 March 1995, Stephen Tweedie
5  * Added /proc support, Dec 1995
6  * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7  * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8  * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9  * Dynamic registration fixes, Stephen Tweedie.
10  * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11  * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12  *  Horn.
13  * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14  * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15  * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16  *  Wendling.
17  * The list_for_each() macro wasn't appropriate for the sysctl loop.
18  *  Removed it and replaced it with older style, 03/23/00, Bill Wendling
19  */
20 
21 #include <linux/module.h>
22 #include <linux/aio.h>
23 #include <linux/mm.h>
24 #include <linux/swap.h>
25 #include <linux/slab.h>
26 #include <linux/sysctl.h>
27 #include <linux/bitmap.h>
28 #include <linux/signal.h>
29 #include <linux/printk.h>
30 #include <linux/proc_fs.h>
31 #include <linux/security.h>
32 #include <linux/ctype.h>
33 #include <linux/kmemcheck.h>
34 #include <linux/kmemleak.h>
35 #include <linux/fs.h>
36 #include <linux/init.h>
37 #include <linux/kernel.h>
38 #include <linux/kobject.h>
39 #include <linux/net.h>
40 #include <linux/sysrq.h>
41 #include <linux/highuid.h>
42 #include <linux/writeback.h>
43 #include <linux/ratelimit.h>
44 #include <linux/compaction.h>
45 #include <linux/hugetlb.h>
46 #include <linux/initrd.h>
47 #include <linux/key.h>
48 #include <linux/times.h>
49 #include <linux/limits.h>
50 #include <linux/dcache.h>
51 #include <linux/dnotify.h>
52 #include <linux/syscalls.h>
53 #include <linux/vmstat.h>
54 #include <linux/nfs_fs.h>
55 #include <linux/acpi.h>
56 #include <linux/reboot.h>
57 #include <linux/ftrace.h>
58 #include <linux/perf_event.h>
59 #include <linux/kprobes.h>
60 #include <linux/pipe_fs_i.h>
61 #include <linux/oom.h>
62 #include <linux/kmod.h>
63 #include <linux/capability.h>
64 #include <linux/binfmts.h>
65 #include <linux/sched/sysctl.h>
66 #include <linux/kexec.h>
67 #include <linux/bpf.h>
68 
69 #include <asm/uaccess.h>
70 #include <asm/processor.h>
71 
72 #ifdef CONFIG_X86
73 #include <asm/nmi.h>
74 #include <asm/stacktrace.h>
75 #include <asm/io.h>
76 #endif
77 #ifdef CONFIG_SPARC
78 #include <asm/setup.h>
79 #endif
80 #ifdef CONFIG_BSD_PROCESS_ACCT
81 #include <linux/acct.h>
82 #endif
83 #ifdef CONFIG_RT_MUTEXES
84 #include <linux/rtmutex.h>
85 #endif
86 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
87 #include <linux/lockdep.h>
88 #endif
89 #ifdef CONFIG_CHR_DEV_SG
90 #include <scsi/sg.h>
91 #endif
92 
93 #ifdef CONFIG_LOCKUP_DETECTOR
94 #include <linux/nmi.h>
95 #endif
96 
97 #if defined(CONFIG_SYSCTL)
98 
99 /* External variables not in a header file. */
100 extern int suid_dumpable;
101 #ifdef CONFIG_COREDUMP
102 extern int core_uses_pid;
103 extern char core_pattern[];
104 extern unsigned int core_pipe_limit;
105 #endif
106 extern int pid_max;
107 extern int pid_max_min, pid_max_max;
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 
116 /* Constants used for minimum and  maximum */
117 #ifdef CONFIG_LOCKUP_DETECTOR
118 static int sixty = 60;
119 #endif
120 
121 static int __maybe_unused neg_one = -1;
122 
123 static int zero;
124 static int __maybe_unused one = 1;
125 static int __maybe_unused two = 2;
126 static int __maybe_unused four = 4;
127 static unsigned long one_ul = 1;
128 static int one_hundred = 100;
129 #ifdef CONFIG_PRINTK
130 static int ten_thousand = 10000;
131 #endif
132 
133 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
134 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
135 
136 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
137 static int maxolduid = 65535;
138 static int minolduid;
139 
140 static int ngroups_max = NGROUPS_MAX;
141 static const int cap_last_cap = CAP_LAST_CAP;
142 
143 /*this is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs */
144 #ifdef CONFIG_DETECT_HUNG_TASK
145 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
146 #endif
147 
148 #ifdef CONFIG_INOTIFY_USER
149 #include <linux/inotify.h>
150 #endif
151 #ifdef CONFIG_SPARC
152 #endif
153 
154 #ifdef __hppa__
155 extern int pwrsw_enabled;
156 #endif
157 
158 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
159 extern int unaligned_enabled;
160 #endif
161 
162 #ifdef CONFIG_IA64
163 extern int unaligned_dump_stack;
164 #endif
165 
166 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
167 extern int no_unaligned_warning;
168 #endif
169 
170 #ifdef CONFIG_PROC_SYSCTL
171 
172 #define SYSCTL_WRITES_LEGACY	-1
173 #define SYSCTL_WRITES_WARN	 0
174 #define SYSCTL_WRITES_STRICT	 1
175 
176 static int sysctl_writes_strict = SYSCTL_WRITES_STRICT;
177 
178 static int proc_do_cad_pid(struct ctl_table *table, int write,
179 		  void __user *buffer, size_t *lenp, loff_t *ppos);
180 static int proc_taint(struct ctl_table *table, int write,
181 			       void __user *buffer, size_t *lenp, loff_t *ppos);
182 #endif
183 
184 #ifdef CONFIG_PRINTK
185 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
186 				void __user *buffer, size_t *lenp, loff_t *ppos);
187 #endif
188 
189 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
190 		void __user *buffer, size_t *lenp, loff_t *ppos);
191 #ifdef CONFIG_COREDUMP
192 static int proc_dostring_coredump(struct ctl_table *table, int write,
193 		void __user *buffer, size_t *lenp, loff_t *ppos);
194 #endif
195 
196 #ifdef CONFIG_MAGIC_SYSRQ
197 /* Note: sysrq code uses it's own private copy */
198 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
199 
200 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
201 				void __user *buffer, size_t *lenp,
202 				loff_t *ppos)
203 {
204 	int error;
205 
206 	error = proc_dointvec(table, write, buffer, lenp, ppos);
207 	if (error)
208 		return error;
209 
210 	if (write)
211 		sysrq_toggle_support(__sysrq_enabled);
212 
213 	return 0;
214 }
215 
216 #endif
217 
218 static struct ctl_table kern_table[];
219 static struct ctl_table vm_table[];
220 static struct ctl_table fs_table[];
221 static struct ctl_table debug_table[];
222 static struct ctl_table dev_table[];
223 extern struct ctl_table random_table[];
224 #ifdef CONFIG_EPOLL
225 extern struct ctl_table epoll_table[];
226 #endif
227 
228 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
229 int sysctl_legacy_va_layout;
230 #endif
231 
232 /* The default sysctl tables: */
233 
234 static struct ctl_table sysctl_base_table[] = {
235 	{
236 		.procname	= "kernel",
237 		.mode		= 0555,
238 		.child		= kern_table,
239 	},
240 	{
241 		.procname	= "vm",
242 		.mode		= 0555,
243 		.child		= vm_table,
244 	},
245 	{
246 		.procname	= "fs",
247 		.mode		= 0555,
248 		.child		= fs_table,
249 	},
250 	{
251 		.procname	= "debug",
252 		.mode		= 0555,
253 		.child		= debug_table,
254 	},
255 	{
256 		.procname	= "dev",
257 		.mode		= 0555,
258 		.child		= dev_table,
259 	},
260 	{ }
261 };
262 
263 #ifdef CONFIG_SCHED_DEBUG
264 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
265 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
266 static int min_wakeup_granularity_ns;			/* 0 usecs */
267 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
268 #ifdef CONFIG_SMP
269 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
270 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
271 #endif /* CONFIG_SMP */
272 #endif /* CONFIG_SCHED_DEBUG */
273 
274 #ifdef CONFIG_COMPACTION
275 static int min_extfrag_threshold;
276 static int max_extfrag_threshold = 1000;
277 #endif
278 
279 static struct ctl_table kern_table[] = {
280 	{
281 		.procname	= "sched_child_runs_first",
282 		.data		= &sysctl_sched_child_runs_first,
283 		.maxlen		= sizeof(unsigned int),
284 		.mode		= 0644,
285 		.proc_handler	= proc_dointvec,
286 	},
287 #ifdef CONFIG_SCHED_DEBUG
288 	{
289 		.procname	= "sched_min_granularity_ns",
290 		.data		= &sysctl_sched_min_granularity,
291 		.maxlen		= sizeof(unsigned int),
292 		.mode		= 0644,
293 		.proc_handler	= sched_proc_update_handler,
294 		.extra1		= &min_sched_granularity_ns,
295 		.extra2		= &max_sched_granularity_ns,
296 	},
297 	{
298 		.procname	= "sched_latency_ns",
299 		.data		= &sysctl_sched_latency,
300 		.maxlen		= sizeof(unsigned int),
301 		.mode		= 0644,
302 		.proc_handler	= sched_proc_update_handler,
303 		.extra1		= &min_sched_granularity_ns,
304 		.extra2		= &max_sched_granularity_ns,
305 	},
306 	{
307 		.procname	= "sched_wakeup_granularity_ns",
308 		.data		= &sysctl_sched_wakeup_granularity,
309 		.maxlen		= sizeof(unsigned int),
310 		.mode		= 0644,
311 		.proc_handler	= sched_proc_update_handler,
312 		.extra1		= &min_wakeup_granularity_ns,
313 		.extra2		= &max_wakeup_granularity_ns,
314 	},
315 #ifdef CONFIG_SMP
316 	{
317 		.procname	= "sched_tunable_scaling",
318 		.data		= &sysctl_sched_tunable_scaling,
319 		.maxlen		= sizeof(enum sched_tunable_scaling),
320 		.mode		= 0644,
321 		.proc_handler	= sched_proc_update_handler,
322 		.extra1		= &min_sched_tunable_scaling,
323 		.extra2		= &max_sched_tunable_scaling,
324 	},
325 	{
326 		.procname	= "sched_migration_cost_ns",
327 		.data		= &sysctl_sched_migration_cost,
328 		.maxlen		= sizeof(unsigned int),
329 		.mode		= 0644,
330 		.proc_handler	= proc_dointvec,
331 	},
332 	{
333 		.procname	= "sched_nr_migrate",
334 		.data		= &sysctl_sched_nr_migrate,
335 		.maxlen		= sizeof(unsigned int),
336 		.mode		= 0644,
337 		.proc_handler	= proc_dointvec,
338 	},
339 	{
340 		.procname	= "sched_time_avg_ms",
341 		.data		= &sysctl_sched_time_avg,
342 		.maxlen		= sizeof(unsigned int),
343 		.mode		= 0644,
344 		.proc_handler	= proc_dointvec,
345 	},
346 	{
347 		.procname	= "sched_shares_window_ns",
348 		.data		= &sysctl_sched_shares_window,
349 		.maxlen		= sizeof(unsigned int),
350 		.mode		= 0644,
351 		.proc_handler	= proc_dointvec,
352 	},
353 #endif /* CONFIG_SMP */
354 #ifdef CONFIG_NUMA_BALANCING
355 	{
356 		.procname	= "numa_balancing_scan_delay_ms",
357 		.data		= &sysctl_numa_balancing_scan_delay,
358 		.maxlen		= sizeof(unsigned int),
359 		.mode		= 0644,
360 		.proc_handler	= proc_dointvec,
361 	},
362 	{
363 		.procname	= "numa_balancing_scan_period_min_ms",
364 		.data		= &sysctl_numa_balancing_scan_period_min,
365 		.maxlen		= sizeof(unsigned int),
366 		.mode		= 0644,
367 		.proc_handler	= proc_dointvec,
368 	},
369 	{
370 		.procname	= "numa_balancing_scan_period_max_ms",
371 		.data		= &sysctl_numa_balancing_scan_period_max,
372 		.maxlen		= sizeof(unsigned int),
373 		.mode		= 0644,
374 		.proc_handler	= proc_dointvec,
375 	},
376 	{
377 		.procname	= "numa_balancing_scan_size_mb",
378 		.data		= &sysctl_numa_balancing_scan_size,
379 		.maxlen		= sizeof(unsigned int),
380 		.mode		= 0644,
381 		.proc_handler	= proc_dointvec_minmax,
382 		.extra1		= &one,
383 	},
384 	{
385 		.procname	= "numa_balancing",
386 		.data		= NULL, /* filled in by handler */
387 		.maxlen		= sizeof(unsigned int),
388 		.mode		= 0644,
389 		.proc_handler	= sysctl_numa_balancing,
390 		.extra1		= &zero,
391 		.extra2		= &one,
392 	},
393 #endif /* CONFIG_NUMA_BALANCING */
394 #endif /* CONFIG_SCHED_DEBUG */
395 	{
396 		.procname	= "sched_rt_period_us",
397 		.data		= &sysctl_sched_rt_period,
398 		.maxlen		= sizeof(unsigned int),
399 		.mode		= 0644,
400 		.proc_handler	= sched_rt_handler,
401 	},
402 	{
403 		.procname	= "sched_rt_runtime_us",
404 		.data		= &sysctl_sched_rt_runtime,
405 		.maxlen		= sizeof(int),
406 		.mode		= 0644,
407 		.proc_handler	= sched_rt_handler,
408 	},
409 	{
410 		.procname	= "sched_rr_timeslice_ms",
411 		.data		= &sched_rr_timeslice,
412 		.maxlen		= sizeof(int),
413 		.mode		= 0644,
414 		.proc_handler	= sched_rr_handler,
415 	},
416 #ifdef CONFIG_SCHED_AUTOGROUP
417 	{
418 		.procname	= "sched_autogroup_enabled",
419 		.data		= &sysctl_sched_autogroup_enabled,
420 		.maxlen		= sizeof(unsigned int),
421 		.mode		= 0644,
422 		.proc_handler	= proc_dointvec_minmax,
423 		.extra1		= &zero,
424 		.extra2		= &one,
425 	},
426 #endif
427 #ifdef CONFIG_CFS_BANDWIDTH
428 	{
429 		.procname	= "sched_cfs_bandwidth_slice_us",
430 		.data		= &sysctl_sched_cfs_bandwidth_slice,
431 		.maxlen		= sizeof(unsigned int),
432 		.mode		= 0644,
433 		.proc_handler	= proc_dointvec_minmax,
434 		.extra1		= &one,
435 	},
436 #endif
437 #ifdef CONFIG_PROVE_LOCKING
438 	{
439 		.procname	= "prove_locking",
440 		.data		= &prove_locking,
441 		.maxlen		= sizeof(int),
442 		.mode		= 0644,
443 		.proc_handler	= proc_dointvec,
444 	},
445 #endif
446 #ifdef CONFIG_LOCK_STAT
447 	{
448 		.procname	= "lock_stat",
449 		.data		= &lock_stat,
450 		.maxlen		= sizeof(int),
451 		.mode		= 0644,
452 		.proc_handler	= proc_dointvec,
453 	},
454 #endif
455 	{
456 		.procname	= "panic",
457 		.data		= &panic_timeout,
458 		.maxlen		= sizeof(int),
459 		.mode		= 0644,
460 		.proc_handler	= proc_dointvec,
461 	},
462 #ifdef CONFIG_COREDUMP
463 	{
464 		.procname	= "core_uses_pid",
465 		.data		= &core_uses_pid,
466 		.maxlen		= sizeof(int),
467 		.mode		= 0644,
468 		.proc_handler	= proc_dointvec,
469 	},
470 	{
471 		.procname	= "core_pattern",
472 		.data		= core_pattern,
473 		.maxlen		= CORENAME_MAX_SIZE,
474 		.mode		= 0644,
475 		.proc_handler	= proc_dostring_coredump,
476 	},
477 	{
478 		.procname	= "core_pipe_limit",
479 		.data		= &core_pipe_limit,
480 		.maxlen		= sizeof(unsigned int),
481 		.mode		= 0644,
482 		.proc_handler	= proc_dointvec,
483 	},
484 #endif
485 #ifdef CONFIG_PROC_SYSCTL
486 	{
487 		.procname	= "tainted",
488 		.maxlen 	= sizeof(long),
489 		.mode		= 0644,
490 		.proc_handler	= proc_taint,
491 	},
492 	{
493 		.procname	= "sysctl_writes_strict",
494 		.data		= &sysctl_writes_strict,
495 		.maxlen		= sizeof(int),
496 		.mode		= 0644,
497 		.proc_handler	= proc_dointvec_minmax,
498 		.extra1		= &neg_one,
499 		.extra2		= &one,
500 	},
501 #endif
502 #ifdef CONFIG_LATENCYTOP
503 	{
504 		.procname	= "latencytop",
505 		.data		= &latencytop_enabled,
506 		.maxlen		= sizeof(int),
507 		.mode		= 0644,
508 		.proc_handler	= proc_dointvec,
509 	},
510 #endif
511 #ifdef CONFIG_BLK_DEV_INITRD
512 	{
513 		.procname	= "real-root-dev",
514 		.data		= &real_root_dev,
515 		.maxlen		= sizeof(int),
516 		.mode		= 0644,
517 		.proc_handler	= proc_dointvec,
518 	},
519 #endif
520 	{
521 		.procname	= "print-fatal-signals",
522 		.data		= &print_fatal_signals,
523 		.maxlen		= sizeof(int),
524 		.mode		= 0644,
525 		.proc_handler	= proc_dointvec,
526 	},
527 #ifdef CONFIG_SPARC
528 	{
529 		.procname	= "reboot-cmd",
530 		.data		= reboot_command,
531 		.maxlen		= 256,
532 		.mode		= 0644,
533 		.proc_handler	= proc_dostring,
534 	},
535 	{
536 		.procname	= "stop-a",
537 		.data		= &stop_a_enabled,
538 		.maxlen		= sizeof (int),
539 		.mode		= 0644,
540 		.proc_handler	= proc_dointvec,
541 	},
542 	{
543 		.procname	= "scons-poweroff",
544 		.data		= &scons_pwroff,
545 		.maxlen		= sizeof (int),
546 		.mode		= 0644,
547 		.proc_handler	= proc_dointvec,
548 	},
549 #endif
550 #ifdef CONFIG_SPARC64
551 	{
552 		.procname	= "tsb-ratio",
553 		.data		= &sysctl_tsb_ratio,
554 		.maxlen		= sizeof (int),
555 		.mode		= 0644,
556 		.proc_handler	= proc_dointvec,
557 	},
558 #endif
559 #ifdef __hppa__
560 	{
561 		.procname	= "soft-power",
562 		.data		= &pwrsw_enabled,
563 		.maxlen		= sizeof (int),
564 	 	.mode		= 0644,
565 		.proc_handler	= proc_dointvec,
566 	},
567 #endif
568 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
569 	{
570 		.procname	= "unaligned-trap",
571 		.data		= &unaligned_enabled,
572 		.maxlen		= sizeof (int),
573 		.mode		= 0644,
574 		.proc_handler	= proc_dointvec,
575 	},
576 #endif
577 	{
578 		.procname	= "ctrl-alt-del",
579 		.data		= &C_A_D,
580 		.maxlen		= sizeof(int),
581 		.mode		= 0644,
582 		.proc_handler	= proc_dointvec,
583 	},
584 #ifdef CONFIG_FUNCTION_TRACER
585 	{
586 		.procname	= "ftrace_enabled",
587 		.data		= &ftrace_enabled,
588 		.maxlen		= sizeof(int),
589 		.mode		= 0644,
590 		.proc_handler	= ftrace_enable_sysctl,
591 	},
592 #endif
593 #ifdef CONFIG_STACK_TRACER
594 	{
595 		.procname	= "stack_tracer_enabled",
596 		.data		= &stack_tracer_enabled,
597 		.maxlen		= sizeof(int),
598 		.mode		= 0644,
599 		.proc_handler	= stack_trace_sysctl,
600 	},
601 #endif
602 #ifdef CONFIG_TRACING
603 	{
604 		.procname	= "ftrace_dump_on_oops",
605 		.data		= &ftrace_dump_on_oops,
606 		.maxlen		= sizeof(int),
607 		.mode		= 0644,
608 		.proc_handler	= proc_dointvec,
609 	},
610 	{
611 		.procname	= "traceoff_on_warning",
612 		.data		= &__disable_trace_on_warning,
613 		.maxlen		= sizeof(__disable_trace_on_warning),
614 		.mode		= 0644,
615 		.proc_handler	= proc_dointvec,
616 	},
617 	{
618 		.procname	= "tracepoint_printk",
619 		.data		= &tracepoint_printk,
620 		.maxlen		= sizeof(tracepoint_printk),
621 		.mode		= 0644,
622 		.proc_handler	= proc_dointvec,
623 	},
624 #endif
625 #ifdef CONFIG_KEXEC_CORE
626 	{
627 		.procname	= "kexec_load_disabled",
628 		.data		= &kexec_load_disabled,
629 		.maxlen		= sizeof(int),
630 		.mode		= 0644,
631 		/* only handle a transition from default "0" to "1" */
632 		.proc_handler	= proc_dointvec_minmax,
633 		.extra1		= &one,
634 		.extra2		= &one,
635 	},
636 #endif
637 #ifdef CONFIG_MODULES
638 	{
639 		.procname	= "modprobe",
640 		.data		= &modprobe_path,
641 		.maxlen		= KMOD_PATH_LEN,
642 		.mode		= 0644,
643 		.proc_handler	= proc_dostring,
644 	},
645 	{
646 		.procname	= "modules_disabled",
647 		.data		= &modules_disabled,
648 		.maxlen		= sizeof(int),
649 		.mode		= 0644,
650 		/* only handle a transition from default "0" to "1" */
651 		.proc_handler	= proc_dointvec_minmax,
652 		.extra1		= &one,
653 		.extra2		= &one,
654 	},
655 #endif
656 #ifdef CONFIG_UEVENT_HELPER
657 	{
658 		.procname	= "hotplug",
659 		.data		= &uevent_helper,
660 		.maxlen		= UEVENT_HELPER_PATH_LEN,
661 		.mode		= 0644,
662 		.proc_handler	= proc_dostring,
663 	},
664 #endif
665 #ifdef CONFIG_CHR_DEV_SG
666 	{
667 		.procname	= "sg-big-buff",
668 		.data		= &sg_big_buff,
669 		.maxlen		= sizeof (int),
670 		.mode		= 0444,
671 		.proc_handler	= proc_dointvec,
672 	},
673 #endif
674 #ifdef CONFIG_BSD_PROCESS_ACCT
675 	{
676 		.procname	= "acct",
677 		.data		= &acct_parm,
678 		.maxlen		= 3*sizeof(int),
679 		.mode		= 0644,
680 		.proc_handler	= proc_dointvec,
681 	},
682 #endif
683 #ifdef CONFIG_MAGIC_SYSRQ
684 	{
685 		.procname	= "sysrq",
686 		.data		= &__sysrq_enabled,
687 		.maxlen		= sizeof (int),
688 		.mode		= 0644,
689 		.proc_handler	= sysrq_sysctl_handler,
690 	},
691 #endif
692 #ifdef CONFIG_PROC_SYSCTL
693 	{
694 		.procname	= "cad_pid",
695 		.data		= NULL,
696 		.maxlen		= sizeof (int),
697 		.mode		= 0600,
698 		.proc_handler	= proc_do_cad_pid,
699 	},
700 #endif
701 	{
702 		.procname	= "threads-max",
703 		.data		= NULL,
704 		.maxlen		= sizeof(int),
705 		.mode		= 0644,
706 		.proc_handler	= sysctl_max_threads,
707 	},
708 	{
709 		.procname	= "random",
710 		.mode		= 0555,
711 		.child		= random_table,
712 	},
713 	{
714 		.procname	= "usermodehelper",
715 		.mode		= 0555,
716 		.child		= usermodehelper_table,
717 	},
718 	{
719 		.procname	= "overflowuid",
720 		.data		= &overflowuid,
721 		.maxlen		= sizeof(int),
722 		.mode		= 0644,
723 		.proc_handler	= proc_dointvec_minmax,
724 		.extra1		= &minolduid,
725 		.extra2		= &maxolduid,
726 	},
727 	{
728 		.procname	= "overflowgid",
729 		.data		= &overflowgid,
730 		.maxlen		= sizeof(int),
731 		.mode		= 0644,
732 		.proc_handler	= proc_dointvec_minmax,
733 		.extra1		= &minolduid,
734 		.extra2		= &maxolduid,
735 	},
736 #ifdef CONFIG_S390
737 #ifdef CONFIG_MATHEMU
738 	{
739 		.procname	= "ieee_emulation_warnings",
740 		.data		= &sysctl_ieee_emulation_warnings,
741 		.maxlen		= sizeof(int),
742 		.mode		= 0644,
743 		.proc_handler	= proc_dointvec,
744 	},
745 #endif
746 	{
747 		.procname	= "userprocess_debug",
748 		.data		= &show_unhandled_signals,
749 		.maxlen		= sizeof(int),
750 		.mode		= 0644,
751 		.proc_handler	= proc_dointvec,
752 	},
753 #endif
754 	{
755 		.procname	= "pid_max",
756 		.data		= &pid_max,
757 		.maxlen		= sizeof (int),
758 		.mode		= 0644,
759 		.proc_handler	= proc_dointvec_minmax,
760 		.extra1		= &pid_max_min,
761 		.extra2		= &pid_max_max,
762 	},
763 	{
764 		.procname	= "panic_on_oops",
765 		.data		= &panic_on_oops,
766 		.maxlen		= sizeof(int),
767 		.mode		= 0644,
768 		.proc_handler	= proc_dointvec,
769 	},
770 #if defined CONFIG_PRINTK
771 	{
772 		.procname	= "printk",
773 		.data		= &console_loglevel,
774 		.maxlen		= 4*sizeof(int),
775 		.mode		= 0644,
776 		.proc_handler	= proc_dointvec,
777 	},
778 	{
779 		.procname	= "printk_ratelimit",
780 		.data		= &printk_ratelimit_state.interval,
781 		.maxlen		= sizeof(int),
782 		.mode		= 0644,
783 		.proc_handler	= proc_dointvec_jiffies,
784 	},
785 	{
786 		.procname	= "printk_ratelimit_burst",
787 		.data		= &printk_ratelimit_state.burst,
788 		.maxlen		= sizeof(int),
789 		.mode		= 0644,
790 		.proc_handler	= proc_dointvec,
791 	},
792 	{
793 		.procname	= "printk_delay",
794 		.data		= &printk_delay_msec,
795 		.maxlen		= sizeof(int),
796 		.mode		= 0644,
797 		.proc_handler	= proc_dointvec_minmax,
798 		.extra1		= &zero,
799 		.extra2		= &ten_thousand,
800 	},
801 	{
802 		.procname	= "dmesg_restrict",
803 		.data		= &dmesg_restrict,
804 		.maxlen		= sizeof(int),
805 		.mode		= 0644,
806 		.proc_handler	= proc_dointvec_minmax_sysadmin,
807 		.extra1		= &zero,
808 		.extra2		= &one,
809 	},
810 	{
811 		.procname	= "kptr_restrict",
812 		.data		= &kptr_restrict,
813 		.maxlen		= sizeof(int),
814 		.mode		= 0644,
815 		.proc_handler	= proc_dointvec_minmax_sysadmin,
816 		.extra1		= &zero,
817 		.extra2		= &two,
818 	},
819 #endif
820 	{
821 		.procname	= "ngroups_max",
822 		.data		= &ngroups_max,
823 		.maxlen		= sizeof (int),
824 		.mode		= 0444,
825 		.proc_handler	= proc_dointvec,
826 	},
827 	{
828 		.procname	= "cap_last_cap",
829 		.data		= (void *)&cap_last_cap,
830 		.maxlen		= sizeof(int),
831 		.mode		= 0444,
832 		.proc_handler	= proc_dointvec,
833 	},
834 #if defined(CONFIG_LOCKUP_DETECTOR)
835 	{
836 		.procname       = "watchdog",
837 		.data           = &watchdog_user_enabled,
838 		.maxlen         = sizeof (int),
839 		.mode           = 0644,
840 		.proc_handler   = proc_watchdog,
841 		.extra1		= &zero,
842 		.extra2		= &one,
843 	},
844 	{
845 		.procname	= "watchdog_thresh",
846 		.data		= &watchdog_thresh,
847 		.maxlen		= sizeof(int),
848 		.mode		= 0644,
849 		.proc_handler	= proc_watchdog_thresh,
850 		.extra1		= &zero,
851 		.extra2		= &sixty,
852 	},
853 	{
854 		.procname       = "nmi_watchdog",
855 		.data           = &nmi_watchdog_enabled,
856 		.maxlen         = sizeof (int),
857 		.mode           = 0644,
858 		.proc_handler   = proc_nmi_watchdog,
859 		.extra1		= &zero,
860 #if defined(CONFIG_HAVE_NMI_WATCHDOG) || defined(CONFIG_HARDLOCKUP_DETECTOR)
861 		.extra2		= &one,
862 #else
863 		.extra2		= &zero,
864 #endif
865 	},
866 	{
867 		.procname       = "soft_watchdog",
868 		.data           = &soft_watchdog_enabled,
869 		.maxlen         = sizeof (int),
870 		.mode           = 0644,
871 		.proc_handler   = proc_soft_watchdog,
872 		.extra1		= &zero,
873 		.extra2		= &one,
874 	},
875 	{
876 		.procname	= "watchdog_cpumask",
877 		.data		= &watchdog_cpumask_bits,
878 		.maxlen		= NR_CPUS,
879 		.mode		= 0644,
880 		.proc_handler	= proc_watchdog_cpumask,
881 	},
882 	{
883 		.procname	= "softlockup_panic",
884 		.data		= &softlockup_panic,
885 		.maxlen		= sizeof(int),
886 		.mode		= 0644,
887 		.proc_handler	= proc_dointvec_minmax,
888 		.extra1		= &zero,
889 		.extra2		= &one,
890 	},
891 #ifdef CONFIG_HARDLOCKUP_DETECTOR
892 	{
893 		.procname	= "hardlockup_panic",
894 		.data		= &hardlockup_panic,
895 		.maxlen		= sizeof(int),
896 		.mode		= 0644,
897 		.proc_handler	= proc_dointvec_minmax,
898 		.extra1		= &zero,
899 		.extra2		= &one,
900 	},
901 #endif
902 #ifdef CONFIG_SMP
903 	{
904 		.procname	= "softlockup_all_cpu_backtrace",
905 		.data		= &sysctl_softlockup_all_cpu_backtrace,
906 		.maxlen		= sizeof(int),
907 		.mode		= 0644,
908 		.proc_handler	= proc_dointvec_minmax,
909 		.extra1		= &zero,
910 		.extra2		= &one,
911 	},
912 	{
913 		.procname	= "hardlockup_all_cpu_backtrace",
914 		.data		= &sysctl_hardlockup_all_cpu_backtrace,
915 		.maxlen		= sizeof(int),
916 		.mode		= 0644,
917 		.proc_handler	= proc_dointvec_minmax,
918 		.extra1		= &zero,
919 		.extra2		= &one,
920 	},
921 #endif /* CONFIG_SMP */
922 #endif
923 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
924 	{
925 		.procname       = "unknown_nmi_panic",
926 		.data           = &unknown_nmi_panic,
927 		.maxlen         = sizeof (int),
928 		.mode           = 0644,
929 		.proc_handler   = proc_dointvec,
930 	},
931 #endif
932 #if defined(CONFIG_X86)
933 	{
934 		.procname	= "panic_on_unrecovered_nmi",
935 		.data		= &panic_on_unrecovered_nmi,
936 		.maxlen		= sizeof(int),
937 		.mode		= 0644,
938 		.proc_handler	= proc_dointvec,
939 	},
940 	{
941 		.procname	= "panic_on_io_nmi",
942 		.data		= &panic_on_io_nmi,
943 		.maxlen		= sizeof(int),
944 		.mode		= 0644,
945 		.proc_handler	= proc_dointvec,
946 	},
947 #ifdef CONFIG_DEBUG_STACKOVERFLOW
948 	{
949 		.procname	= "panic_on_stackoverflow",
950 		.data		= &sysctl_panic_on_stackoverflow,
951 		.maxlen		= sizeof(int),
952 		.mode		= 0644,
953 		.proc_handler	= proc_dointvec,
954 	},
955 #endif
956 	{
957 		.procname	= "bootloader_type",
958 		.data		= &bootloader_type,
959 		.maxlen		= sizeof (int),
960 		.mode		= 0444,
961 		.proc_handler	= proc_dointvec,
962 	},
963 	{
964 		.procname	= "bootloader_version",
965 		.data		= &bootloader_version,
966 		.maxlen		= sizeof (int),
967 		.mode		= 0444,
968 		.proc_handler	= proc_dointvec,
969 	},
970 	{
971 		.procname	= "kstack_depth_to_print",
972 		.data		= &kstack_depth_to_print,
973 		.maxlen		= sizeof(int),
974 		.mode		= 0644,
975 		.proc_handler	= proc_dointvec,
976 	},
977 	{
978 		.procname	= "io_delay_type",
979 		.data		= &io_delay_type,
980 		.maxlen		= sizeof(int),
981 		.mode		= 0644,
982 		.proc_handler	= proc_dointvec,
983 	},
984 #endif
985 #if defined(CONFIG_MMU)
986 	{
987 		.procname	= "randomize_va_space",
988 		.data		= &randomize_va_space,
989 		.maxlen		= sizeof(int),
990 		.mode		= 0644,
991 		.proc_handler	= proc_dointvec,
992 	},
993 #endif
994 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
995 	{
996 		.procname	= "spin_retry",
997 		.data		= &spin_retry,
998 		.maxlen		= sizeof (int),
999 		.mode		= 0644,
1000 		.proc_handler	= proc_dointvec,
1001 	},
1002 #endif
1003 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1004 	{
1005 		.procname	= "acpi_video_flags",
1006 		.data		= &acpi_realmode_flags,
1007 		.maxlen		= sizeof (unsigned long),
1008 		.mode		= 0644,
1009 		.proc_handler	= proc_doulongvec_minmax,
1010 	},
1011 #endif
1012 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1013 	{
1014 		.procname	= "ignore-unaligned-usertrap",
1015 		.data		= &no_unaligned_warning,
1016 		.maxlen		= sizeof (int),
1017 	 	.mode		= 0644,
1018 		.proc_handler	= proc_dointvec,
1019 	},
1020 #endif
1021 #ifdef CONFIG_IA64
1022 	{
1023 		.procname	= "unaligned-dump-stack",
1024 		.data		= &unaligned_dump_stack,
1025 		.maxlen		= sizeof (int),
1026 		.mode		= 0644,
1027 		.proc_handler	= proc_dointvec,
1028 	},
1029 #endif
1030 #ifdef CONFIG_DETECT_HUNG_TASK
1031 	{
1032 		.procname	= "hung_task_panic",
1033 		.data		= &sysctl_hung_task_panic,
1034 		.maxlen		= sizeof(int),
1035 		.mode		= 0644,
1036 		.proc_handler	= proc_dointvec_minmax,
1037 		.extra1		= &zero,
1038 		.extra2		= &one,
1039 	},
1040 	{
1041 		.procname	= "hung_task_check_count",
1042 		.data		= &sysctl_hung_task_check_count,
1043 		.maxlen		= sizeof(int),
1044 		.mode		= 0644,
1045 		.proc_handler	= proc_dointvec_minmax,
1046 		.extra1		= &zero,
1047 	},
1048 	{
1049 		.procname	= "hung_task_timeout_secs",
1050 		.data		= &sysctl_hung_task_timeout_secs,
1051 		.maxlen		= sizeof(unsigned long),
1052 		.mode		= 0644,
1053 		.proc_handler	= proc_dohung_task_timeout_secs,
1054 		.extra2		= &hung_task_timeout_max,
1055 	},
1056 	{
1057 		.procname	= "hung_task_warnings",
1058 		.data		= &sysctl_hung_task_warnings,
1059 		.maxlen		= sizeof(int),
1060 		.mode		= 0644,
1061 		.proc_handler	= proc_dointvec_minmax,
1062 		.extra1		= &neg_one,
1063 	},
1064 #endif
1065 #ifdef CONFIG_COMPAT
1066 	{
1067 		.procname	= "compat-log",
1068 		.data		= &compat_log,
1069 		.maxlen		= sizeof (int),
1070 	 	.mode		= 0644,
1071 		.proc_handler	= proc_dointvec,
1072 	},
1073 #endif
1074 #ifdef CONFIG_RT_MUTEXES
1075 	{
1076 		.procname	= "max_lock_depth",
1077 		.data		= &max_lock_depth,
1078 		.maxlen		= sizeof(int),
1079 		.mode		= 0644,
1080 		.proc_handler	= proc_dointvec,
1081 	},
1082 #endif
1083 	{
1084 		.procname	= "poweroff_cmd",
1085 		.data		= &poweroff_cmd,
1086 		.maxlen		= POWEROFF_CMD_PATH_LEN,
1087 		.mode		= 0644,
1088 		.proc_handler	= proc_dostring,
1089 	},
1090 #ifdef CONFIG_KEYS
1091 	{
1092 		.procname	= "keys",
1093 		.mode		= 0555,
1094 		.child		= key_sysctls,
1095 	},
1096 #endif
1097 #ifdef CONFIG_PERF_EVENTS
1098 	/*
1099 	 * User-space scripts rely on the existence of this file
1100 	 * as a feature check for perf_events being enabled.
1101 	 *
1102 	 * So it's an ABI, do not remove!
1103 	 */
1104 	{
1105 		.procname	= "perf_event_paranoid",
1106 		.data		= &sysctl_perf_event_paranoid,
1107 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
1108 		.mode		= 0644,
1109 		.proc_handler	= proc_dointvec,
1110 	},
1111 	{
1112 		.procname	= "perf_event_mlock_kb",
1113 		.data		= &sysctl_perf_event_mlock,
1114 		.maxlen		= sizeof(sysctl_perf_event_mlock),
1115 		.mode		= 0644,
1116 		.proc_handler	= proc_dointvec,
1117 	},
1118 	{
1119 		.procname	= "perf_event_max_sample_rate",
1120 		.data		= &sysctl_perf_event_sample_rate,
1121 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
1122 		.mode		= 0644,
1123 		.proc_handler	= perf_proc_update_handler,
1124 		.extra1		= &one,
1125 	},
1126 	{
1127 		.procname	= "perf_cpu_time_max_percent",
1128 		.data		= &sysctl_perf_cpu_time_max_percent,
1129 		.maxlen		= sizeof(sysctl_perf_cpu_time_max_percent),
1130 		.mode		= 0644,
1131 		.proc_handler	= perf_cpu_time_max_percent_handler,
1132 		.extra1		= &zero,
1133 		.extra2		= &one_hundred,
1134 	},
1135 #endif
1136 #ifdef CONFIG_KMEMCHECK
1137 	{
1138 		.procname	= "kmemcheck",
1139 		.data		= &kmemcheck_enabled,
1140 		.maxlen		= sizeof(int),
1141 		.mode		= 0644,
1142 		.proc_handler	= proc_dointvec,
1143 	},
1144 #endif
1145 	{
1146 		.procname	= "panic_on_warn",
1147 		.data		= &panic_on_warn,
1148 		.maxlen		= sizeof(int),
1149 		.mode		= 0644,
1150 		.proc_handler	= proc_dointvec_minmax,
1151 		.extra1		= &zero,
1152 		.extra2		= &one,
1153 	},
1154 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1155 	{
1156 		.procname	= "timer_migration",
1157 		.data		= &sysctl_timer_migration,
1158 		.maxlen		= sizeof(unsigned int),
1159 		.mode		= 0644,
1160 		.proc_handler	= timer_migration_handler,
1161 	},
1162 #endif
1163 #ifdef CONFIG_BPF_SYSCALL
1164 	{
1165 		.procname	= "unprivileged_bpf_disabled",
1166 		.data		= &sysctl_unprivileged_bpf_disabled,
1167 		.maxlen		= sizeof(sysctl_unprivileged_bpf_disabled),
1168 		.mode		= 0644,
1169 		/* only handle a transition from default "0" to "1" */
1170 		.proc_handler	= proc_dointvec_minmax,
1171 		.extra1		= &one,
1172 		.extra2		= &one,
1173 	},
1174 #endif
1175 	{ }
1176 };
1177 
1178 static struct ctl_table vm_table[] = {
1179 	{
1180 		.procname	= "overcommit_memory",
1181 		.data		= &sysctl_overcommit_memory,
1182 		.maxlen		= sizeof(sysctl_overcommit_memory),
1183 		.mode		= 0644,
1184 		.proc_handler	= proc_dointvec_minmax,
1185 		.extra1		= &zero,
1186 		.extra2		= &two,
1187 	},
1188 	{
1189 		.procname	= "panic_on_oom",
1190 		.data		= &sysctl_panic_on_oom,
1191 		.maxlen		= sizeof(sysctl_panic_on_oom),
1192 		.mode		= 0644,
1193 		.proc_handler	= proc_dointvec_minmax,
1194 		.extra1		= &zero,
1195 		.extra2		= &two,
1196 	},
1197 	{
1198 		.procname	= "oom_kill_allocating_task",
1199 		.data		= &sysctl_oom_kill_allocating_task,
1200 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1201 		.mode		= 0644,
1202 		.proc_handler	= proc_dointvec,
1203 	},
1204 	{
1205 		.procname	= "oom_dump_tasks",
1206 		.data		= &sysctl_oom_dump_tasks,
1207 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1208 		.mode		= 0644,
1209 		.proc_handler	= proc_dointvec,
1210 	},
1211 	{
1212 		.procname	= "overcommit_ratio",
1213 		.data		= &sysctl_overcommit_ratio,
1214 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1215 		.mode		= 0644,
1216 		.proc_handler	= overcommit_ratio_handler,
1217 	},
1218 	{
1219 		.procname	= "overcommit_kbytes",
1220 		.data		= &sysctl_overcommit_kbytes,
1221 		.maxlen		= sizeof(sysctl_overcommit_kbytes),
1222 		.mode		= 0644,
1223 		.proc_handler	= overcommit_kbytes_handler,
1224 	},
1225 	{
1226 		.procname	= "page-cluster",
1227 		.data		= &page_cluster,
1228 		.maxlen		= sizeof(int),
1229 		.mode		= 0644,
1230 		.proc_handler	= proc_dointvec_minmax,
1231 		.extra1		= &zero,
1232 	},
1233 	{
1234 		.procname	= "dirty_background_ratio",
1235 		.data		= &dirty_background_ratio,
1236 		.maxlen		= sizeof(dirty_background_ratio),
1237 		.mode		= 0644,
1238 		.proc_handler	= dirty_background_ratio_handler,
1239 		.extra1		= &zero,
1240 		.extra2		= &one_hundred,
1241 	},
1242 	{
1243 		.procname	= "dirty_background_bytes",
1244 		.data		= &dirty_background_bytes,
1245 		.maxlen		= sizeof(dirty_background_bytes),
1246 		.mode		= 0644,
1247 		.proc_handler	= dirty_background_bytes_handler,
1248 		.extra1		= &one_ul,
1249 	},
1250 	{
1251 		.procname	= "dirty_ratio",
1252 		.data		= &vm_dirty_ratio,
1253 		.maxlen		= sizeof(vm_dirty_ratio),
1254 		.mode		= 0644,
1255 		.proc_handler	= dirty_ratio_handler,
1256 		.extra1		= &zero,
1257 		.extra2		= &one_hundred,
1258 	},
1259 	{
1260 		.procname	= "dirty_bytes",
1261 		.data		= &vm_dirty_bytes,
1262 		.maxlen		= sizeof(vm_dirty_bytes),
1263 		.mode		= 0644,
1264 		.proc_handler	= dirty_bytes_handler,
1265 		.extra1		= &dirty_bytes_min,
1266 	},
1267 	{
1268 		.procname	= "dirty_writeback_centisecs",
1269 		.data		= &dirty_writeback_interval,
1270 		.maxlen		= sizeof(dirty_writeback_interval),
1271 		.mode		= 0644,
1272 		.proc_handler	= dirty_writeback_centisecs_handler,
1273 	},
1274 	{
1275 		.procname	= "dirty_expire_centisecs",
1276 		.data		= &dirty_expire_interval,
1277 		.maxlen		= sizeof(dirty_expire_interval),
1278 		.mode		= 0644,
1279 		.proc_handler	= proc_dointvec_minmax,
1280 		.extra1		= &zero,
1281 	},
1282 	{
1283 		.procname	= "dirtytime_expire_seconds",
1284 		.data		= &dirtytime_expire_interval,
1285 		.maxlen		= sizeof(dirty_expire_interval),
1286 		.mode		= 0644,
1287 		.proc_handler	= dirtytime_interval_handler,
1288 		.extra1		= &zero,
1289 	},
1290 	{
1291 		.procname       = "nr_pdflush_threads",
1292 		.mode           = 0444 /* read-only */,
1293 		.proc_handler   = pdflush_proc_obsolete,
1294 	},
1295 	{
1296 		.procname	= "swappiness",
1297 		.data		= &vm_swappiness,
1298 		.maxlen		= sizeof(vm_swappiness),
1299 		.mode		= 0644,
1300 		.proc_handler	= proc_dointvec_minmax,
1301 		.extra1		= &zero,
1302 		.extra2		= &one_hundred,
1303 	},
1304 #ifdef CONFIG_HUGETLB_PAGE
1305 	{
1306 		.procname	= "nr_hugepages",
1307 		.data		= NULL,
1308 		.maxlen		= sizeof(unsigned long),
1309 		.mode		= 0644,
1310 		.proc_handler	= hugetlb_sysctl_handler,
1311 	},
1312 #ifdef CONFIG_NUMA
1313 	{
1314 		.procname       = "nr_hugepages_mempolicy",
1315 		.data           = NULL,
1316 		.maxlen         = sizeof(unsigned long),
1317 		.mode           = 0644,
1318 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1319 	},
1320 #endif
1321 	 {
1322 		.procname	= "hugetlb_shm_group",
1323 		.data		= &sysctl_hugetlb_shm_group,
1324 		.maxlen		= sizeof(gid_t),
1325 		.mode		= 0644,
1326 		.proc_handler	= proc_dointvec,
1327 	 },
1328 	 {
1329 		.procname	= "hugepages_treat_as_movable",
1330 		.data		= &hugepages_treat_as_movable,
1331 		.maxlen		= sizeof(int),
1332 		.mode		= 0644,
1333 		.proc_handler	= proc_dointvec,
1334 	},
1335 	{
1336 		.procname	= "nr_overcommit_hugepages",
1337 		.data		= NULL,
1338 		.maxlen		= sizeof(unsigned long),
1339 		.mode		= 0644,
1340 		.proc_handler	= hugetlb_overcommit_handler,
1341 	},
1342 #endif
1343 	{
1344 		.procname	= "lowmem_reserve_ratio",
1345 		.data		= &sysctl_lowmem_reserve_ratio,
1346 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1347 		.mode		= 0644,
1348 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1349 	},
1350 	{
1351 		.procname	= "drop_caches",
1352 		.data		= &sysctl_drop_caches,
1353 		.maxlen		= sizeof(int),
1354 		.mode		= 0644,
1355 		.proc_handler	= drop_caches_sysctl_handler,
1356 		.extra1		= &one,
1357 		.extra2		= &four,
1358 	},
1359 #ifdef CONFIG_COMPACTION
1360 	{
1361 		.procname	= "compact_memory",
1362 		.data		= &sysctl_compact_memory,
1363 		.maxlen		= sizeof(int),
1364 		.mode		= 0200,
1365 		.proc_handler	= sysctl_compaction_handler,
1366 	},
1367 	{
1368 		.procname	= "extfrag_threshold",
1369 		.data		= &sysctl_extfrag_threshold,
1370 		.maxlen		= sizeof(int),
1371 		.mode		= 0644,
1372 		.proc_handler	= sysctl_extfrag_handler,
1373 		.extra1		= &min_extfrag_threshold,
1374 		.extra2		= &max_extfrag_threshold,
1375 	},
1376 	{
1377 		.procname	= "compact_unevictable_allowed",
1378 		.data		= &sysctl_compact_unevictable_allowed,
1379 		.maxlen		= sizeof(int),
1380 		.mode		= 0644,
1381 		.proc_handler	= proc_dointvec,
1382 		.extra1		= &zero,
1383 		.extra2		= &one,
1384 	},
1385 
1386 #endif /* CONFIG_COMPACTION */
1387 	{
1388 		.procname	= "min_free_kbytes",
1389 		.data		= &min_free_kbytes,
1390 		.maxlen		= sizeof(min_free_kbytes),
1391 		.mode		= 0644,
1392 		.proc_handler	= min_free_kbytes_sysctl_handler,
1393 		.extra1		= &zero,
1394 	},
1395 	{
1396 		.procname	= "percpu_pagelist_fraction",
1397 		.data		= &percpu_pagelist_fraction,
1398 		.maxlen		= sizeof(percpu_pagelist_fraction),
1399 		.mode		= 0644,
1400 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1401 		.extra1		= &zero,
1402 	},
1403 #ifdef CONFIG_MMU
1404 	{
1405 		.procname	= "max_map_count",
1406 		.data		= &sysctl_max_map_count,
1407 		.maxlen		= sizeof(sysctl_max_map_count),
1408 		.mode		= 0644,
1409 		.proc_handler	= proc_dointvec_minmax,
1410 		.extra1		= &zero,
1411 	},
1412 #else
1413 	{
1414 		.procname	= "nr_trim_pages",
1415 		.data		= &sysctl_nr_trim_pages,
1416 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1417 		.mode		= 0644,
1418 		.proc_handler	= proc_dointvec_minmax,
1419 		.extra1		= &zero,
1420 	},
1421 #endif
1422 	{
1423 		.procname	= "laptop_mode",
1424 		.data		= &laptop_mode,
1425 		.maxlen		= sizeof(laptop_mode),
1426 		.mode		= 0644,
1427 		.proc_handler	= proc_dointvec_jiffies,
1428 	},
1429 	{
1430 		.procname	= "block_dump",
1431 		.data		= &block_dump,
1432 		.maxlen		= sizeof(block_dump),
1433 		.mode		= 0644,
1434 		.proc_handler	= proc_dointvec,
1435 		.extra1		= &zero,
1436 	},
1437 	{
1438 		.procname	= "vfs_cache_pressure",
1439 		.data		= &sysctl_vfs_cache_pressure,
1440 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1441 		.mode		= 0644,
1442 		.proc_handler	= proc_dointvec,
1443 		.extra1		= &zero,
1444 	},
1445 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1446 	{
1447 		.procname	= "legacy_va_layout",
1448 		.data		= &sysctl_legacy_va_layout,
1449 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1450 		.mode		= 0644,
1451 		.proc_handler	= proc_dointvec,
1452 		.extra1		= &zero,
1453 	},
1454 #endif
1455 #ifdef CONFIG_NUMA
1456 	{
1457 		.procname	= "zone_reclaim_mode",
1458 		.data		= &zone_reclaim_mode,
1459 		.maxlen		= sizeof(zone_reclaim_mode),
1460 		.mode		= 0644,
1461 		.proc_handler	= proc_dointvec,
1462 		.extra1		= &zero,
1463 	},
1464 	{
1465 		.procname	= "min_unmapped_ratio",
1466 		.data		= &sysctl_min_unmapped_ratio,
1467 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1468 		.mode		= 0644,
1469 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1470 		.extra1		= &zero,
1471 		.extra2		= &one_hundred,
1472 	},
1473 	{
1474 		.procname	= "min_slab_ratio",
1475 		.data		= &sysctl_min_slab_ratio,
1476 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1477 		.mode		= 0644,
1478 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1479 		.extra1		= &zero,
1480 		.extra2		= &one_hundred,
1481 	},
1482 #endif
1483 #ifdef CONFIG_SMP
1484 	{
1485 		.procname	= "stat_interval",
1486 		.data		= &sysctl_stat_interval,
1487 		.maxlen		= sizeof(sysctl_stat_interval),
1488 		.mode		= 0644,
1489 		.proc_handler	= proc_dointvec_jiffies,
1490 	},
1491 #endif
1492 #ifdef CONFIG_MMU
1493 	{
1494 		.procname	= "mmap_min_addr",
1495 		.data		= &dac_mmap_min_addr,
1496 		.maxlen		= sizeof(unsigned long),
1497 		.mode		= 0644,
1498 		.proc_handler	= mmap_min_addr_handler,
1499 	},
1500 #endif
1501 #ifdef CONFIG_NUMA
1502 	{
1503 		.procname	= "numa_zonelist_order",
1504 		.data		= &numa_zonelist_order,
1505 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1506 		.mode		= 0644,
1507 		.proc_handler	= numa_zonelist_order_handler,
1508 	},
1509 #endif
1510 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1511    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1512 	{
1513 		.procname	= "vdso_enabled",
1514 #ifdef CONFIG_X86_32
1515 		.data		= &vdso32_enabled,
1516 		.maxlen		= sizeof(vdso32_enabled),
1517 #else
1518 		.data		= &vdso_enabled,
1519 		.maxlen		= sizeof(vdso_enabled),
1520 #endif
1521 		.mode		= 0644,
1522 		.proc_handler	= proc_dointvec,
1523 		.extra1		= &zero,
1524 	},
1525 #endif
1526 #ifdef CONFIG_HIGHMEM
1527 	{
1528 		.procname	= "highmem_is_dirtyable",
1529 		.data		= &vm_highmem_is_dirtyable,
1530 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1531 		.mode		= 0644,
1532 		.proc_handler	= proc_dointvec_minmax,
1533 		.extra1		= &zero,
1534 		.extra2		= &one,
1535 	},
1536 #endif
1537 #ifdef CONFIG_MEMORY_FAILURE
1538 	{
1539 		.procname	= "memory_failure_early_kill",
1540 		.data		= &sysctl_memory_failure_early_kill,
1541 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1542 		.mode		= 0644,
1543 		.proc_handler	= proc_dointvec_minmax,
1544 		.extra1		= &zero,
1545 		.extra2		= &one,
1546 	},
1547 	{
1548 		.procname	= "memory_failure_recovery",
1549 		.data		= &sysctl_memory_failure_recovery,
1550 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1551 		.mode		= 0644,
1552 		.proc_handler	= proc_dointvec_minmax,
1553 		.extra1		= &zero,
1554 		.extra2		= &one,
1555 	},
1556 #endif
1557 	{
1558 		.procname	= "user_reserve_kbytes",
1559 		.data		= &sysctl_user_reserve_kbytes,
1560 		.maxlen		= sizeof(sysctl_user_reserve_kbytes),
1561 		.mode		= 0644,
1562 		.proc_handler	= proc_doulongvec_minmax,
1563 	},
1564 	{
1565 		.procname	= "admin_reserve_kbytes",
1566 		.data		= &sysctl_admin_reserve_kbytes,
1567 		.maxlen		= sizeof(sysctl_admin_reserve_kbytes),
1568 		.mode		= 0644,
1569 		.proc_handler	= proc_doulongvec_minmax,
1570 	},
1571 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1572 	{
1573 		.procname	= "mmap_rnd_bits",
1574 		.data		= &mmap_rnd_bits,
1575 		.maxlen		= sizeof(mmap_rnd_bits),
1576 		.mode		= 0600,
1577 		.proc_handler	= proc_dointvec_minmax,
1578 		.extra1		= (void *)&mmap_rnd_bits_min,
1579 		.extra2		= (void *)&mmap_rnd_bits_max,
1580 	},
1581 #endif
1582 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1583 	{
1584 		.procname	= "mmap_rnd_compat_bits",
1585 		.data		= &mmap_rnd_compat_bits,
1586 		.maxlen		= sizeof(mmap_rnd_compat_bits),
1587 		.mode		= 0600,
1588 		.proc_handler	= proc_dointvec_minmax,
1589 		.extra1		= (void *)&mmap_rnd_compat_bits_min,
1590 		.extra2		= (void *)&mmap_rnd_compat_bits_max,
1591 	},
1592 #endif
1593 	{ }
1594 };
1595 
1596 static struct ctl_table fs_table[] = {
1597 	{
1598 		.procname	= "inode-nr",
1599 		.data		= &inodes_stat,
1600 		.maxlen		= 2*sizeof(long),
1601 		.mode		= 0444,
1602 		.proc_handler	= proc_nr_inodes,
1603 	},
1604 	{
1605 		.procname	= "inode-state",
1606 		.data		= &inodes_stat,
1607 		.maxlen		= 7*sizeof(long),
1608 		.mode		= 0444,
1609 		.proc_handler	= proc_nr_inodes,
1610 	},
1611 	{
1612 		.procname	= "file-nr",
1613 		.data		= &files_stat,
1614 		.maxlen		= sizeof(files_stat),
1615 		.mode		= 0444,
1616 		.proc_handler	= proc_nr_files,
1617 	},
1618 	{
1619 		.procname	= "file-max",
1620 		.data		= &files_stat.max_files,
1621 		.maxlen		= sizeof(files_stat.max_files),
1622 		.mode		= 0644,
1623 		.proc_handler	= proc_doulongvec_minmax,
1624 	},
1625 	{
1626 		.procname	= "nr_open",
1627 		.data		= &sysctl_nr_open,
1628 		.maxlen		= sizeof(int),
1629 		.mode		= 0644,
1630 		.proc_handler	= proc_dointvec_minmax,
1631 		.extra1		= &sysctl_nr_open_min,
1632 		.extra2		= &sysctl_nr_open_max,
1633 	},
1634 	{
1635 		.procname	= "dentry-state",
1636 		.data		= &dentry_stat,
1637 		.maxlen		= 6*sizeof(long),
1638 		.mode		= 0444,
1639 		.proc_handler	= proc_nr_dentry,
1640 	},
1641 	{
1642 		.procname	= "overflowuid",
1643 		.data		= &fs_overflowuid,
1644 		.maxlen		= sizeof(int),
1645 		.mode		= 0644,
1646 		.proc_handler	= proc_dointvec_minmax,
1647 		.extra1		= &minolduid,
1648 		.extra2		= &maxolduid,
1649 	},
1650 	{
1651 		.procname	= "overflowgid",
1652 		.data		= &fs_overflowgid,
1653 		.maxlen		= sizeof(int),
1654 		.mode		= 0644,
1655 		.proc_handler	= proc_dointvec_minmax,
1656 		.extra1		= &minolduid,
1657 		.extra2		= &maxolduid,
1658 	},
1659 #ifdef CONFIG_FILE_LOCKING
1660 	{
1661 		.procname	= "leases-enable",
1662 		.data		= &leases_enable,
1663 		.maxlen		= sizeof(int),
1664 		.mode		= 0644,
1665 		.proc_handler	= proc_dointvec,
1666 	},
1667 #endif
1668 #ifdef CONFIG_DNOTIFY
1669 	{
1670 		.procname	= "dir-notify-enable",
1671 		.data		= &dir_notify_enable,
1672 		.maxlen		= sizeof(int),
1673 		.mode		= 0644,
1674 		.proc_handler	= proc_dointvec,
1675 	},
1676 #endif
1677 #ifdef CONFIG_MMU
1678 #ifdef CONFIG_FILE_LOCKING
1679 	{
1680 		.procname	= "lease-break-time",
1681 		.data		= &lease_break_time,
1682 		.maxlen		= sizeof(int),
1683 		.mode		= 0644,
1684 		.proc_handler	= proc_dointvec,
1685 	},
1686 #endif
1687 #ifdef CONFIG_AIO
1688 	{
1689 		.procname	= "aio-nr",
1690 		.data		= &aio_nr,
1691 		.maxlen		= sizeof(aio_nr),
1692 		.mode		= 0444,
1693 		.proc_handler	= proc_doulongvec_minmax,
1694 	},
1695 	{
1696 		.procname	= "aio-max-nr",
1697 		.data		= &aio_max_nr,
1698 		.maxlen		= sizeof(aio_max_nr),
1699 		.mode		= 0644,
1700 		.proc_handler	= proc_doulongvec_minmax,
1701 	},
1702 #endif /* CONFIG_AIO */
1703 #ifdef CONFIG_INOTIFY_USER
1704 	{
1705 		.procname	= "inotify",
1706 		.mode		= 0555,
1707 		.child		= inotify_table,
1708 	},
1709 #endif
1710 #ifdef CONFIG_EPOLL
1711 	{
1712 		.procname	= "epoll",
1713 		.mode		= 0555,
1714 		.child		= epoll_table,
1715 	},
1716 #endif
1717 #endif
1718 	{
1719 		.procname	= "protected_symlinks",
1720 		.data		= &sysctl_protected_symlinks,
1721 		.maxlen		= sizeof(int),
1722 		.mode		= 0600,
1723 		.proc_handler	= proc_dointvec_minmax,
1724 		.extra1		= &zero,
1725 		.extra2		= &one,
1726 	},
1727 	{
1728 		.procname	= "protected_hardlinks",
1729 		.data		= &sysctl_protected_hardlinks,
1730 		.maxlen		= sizeof(int),
1731 		.mode		= 0600,
1732 		.proc_handler	= proc_dointvec_minmax,
1733 		.extra1		= &zero,
1734 		.extra2		= &one,
1735 	},
1736 	{
1737 		.procname	= "suid_dumpable",
1738 		.data		= &suid_dumpable,
1739 		.maxlen		= sizeof(int),
1740 		.mode		= 0644,
1741 		.proc_handler	= proc_dointvec_minmax_coredump,
1742 		.extra1		= &zero,
1743 		.extra2		= &two,
1744 	},
1745 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1746 	{
1747 		.procname	= "binfmt_misc",
1748 		.mode		= 0555,
1749 		.child		= sysctl_mount_point,
1750 	},
1751 #endif
1752 	{
1753 		.procname	= "pipe-max-size",
1754 		.data		= &pipe_max_size,
1755 		.maxlen		= sizeof(int),
1756 		.mode		= 0644,
1757 		.proc_handler	= &pipe_proc_fn,
1758 		.extra1		= &pipe_min_size,
1759 	},
1760 	{
1761 		.procname	= "pipe-user-pages-hard",
1762 		.data		= &pipe_user_pages_hard,
1763 		.maxlen		= sizeof(pipe_user_pages_hard),
1764 		.mode		= 0644,
1765 		.proc_handler	= proc_doulongvec_minmax,
1766 	},
1767 	{
1768 		.procname	= "pipe-user-pages-soft",
1769 		.data		= &pipe_user_pages_soft,
1770 		.maxlen		= sizeof(pipe_user_pages_soft),
1771 		.mode		= 0644,
1772 		.proc_handler	= proc_doulongvec_minmax,
1773 	},
1774 	{ }
1775 };
1776 
1777 static struct ctl_table debug_table[] = {
1778 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1779 	{
1780 		.procname	= "exception-trace",
1781 		.data		= &show_unhandled_signals,
1782 		.maxlen		= sizeof(int),
1783 		.mode		= 0644,
1784 		.proc_handler	= proc_dointvec
1785 	},
1786 #endif
1787 #if defined(CONFIG_OPTPROBES)
1788 	{
1789 		.procname	= "kprobes-optimization",
1790 		.data		= &sysctl_kprobes_optimization,
1791 		.maxlen		= sizeof(int),
1792 		.mode		= 0644,
1793 		.proc_handler	= proc_kprobes_optimization_handler,
1794 		.extra1		= &zero,
1795 		.extra2		= &one,
1796 	},
1797 #endif
1798 	{ }
1799 };
1800 
1801 static struct ctl_table dev_table[] = {
1802 	{ }
1803 };
1804 
1805 int __init sysctl_init(void)
1806 {
1807 	struct ctl_table_header *hdr;
1808 
1809 	hdr = register_sysctl_table(sysctl_base_table);
1810 	kmemleak_not_leak(hdr);
1811 	return 0;
1812 }
1813 
1814 #endif /* CONFIG_SYSCTL */
1815 
1816 /*
1817  * /proc/sys support
1818  */
1819 
1820 #ifdef CONFIG_PROC_SYSCTL
1821 
1822 static int _proc_do_string(char *data, int maxlen, int write,
1823 			   char __user *buffer,
1824 			   size_t *lenp, loff_t *ppos)
1825 {
1826 	size_t len;
1827 	char __user *p;
1828 	char c;
1829 
1830 	if (!data || !maxlen || !*lenp) {
1831 		*lenp = 0;
1832 		return 0;
1833 	}
1834 
1835 	if (write) {
1836 		if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1837 			/* Only continue writes not past the end of buffer. */
1838 			len = strlen(data);
1839 			if (len > maxlen - 1)
1840 				len = maxlen - 1;
1841 
1842 			if (*ppos > len)
1843 				return 0;
1844 			len = *ppos;
1845 		} else {
1846 			/* Start writing from beginning of buffer. */
1847 			len = 0;
1848 		}
1849 
1850 		*ppos += *lenp;
1851 		p = buffer;
1852 		while ((p - buffer) < *lenp && len < maxlen - 1) {
1853 			if (get_user(c, p++))
1854 				return -EFAULT;
1855 			if (c == 0 || c == '\n')
1856 				break;
1857 			data[len++] = c;
1858 		}
1859 		data[len] = 0;
1860 	} else {
1861 		len = strlen(data);
1862 		if (len > maxlen)
1863 			len = maxlen;
1864 
1865 		if (*ppos > len) {
1866 			*lenp = 0;
1867 			return 0;
1868 		}
1869 
1870 		data += *ppos;
1871 		len  -= *ppos;
1872 
1873 		if (len > *lenp)
1874 			len = *lenp;
1875 		if (len)
1876 			if (copy_to_user(buffer, data, len))
1877 				return -EFAULT;
1878 		if (len < *lenp) {
1879 			if (put_user('\n', buffer + len))
1880 				return -EFAULT;
1881 			len++;
1882 		}
1883 		*lenp = len;
1884 		*ppos += len;
1885 	}
1886 	return 0;
1887 }
1888 
1889 static void warn_sysctl_write(struct ctl_table *table)
1890 {
1891 	pr_warn_once("%s wrote to %s when file position was not 0!\n"
1892 		"This will not be supported in the future. To silence this\n"
1893 		"warning, set kernel.sysctl_writes_strict = -1\n",
1894 		current->comm, table->procname);
1895 }
1896 
1897 /**
1898  * proc_dostring - read a string sysctl
1899  * @table: the sysctl table
1900  * @write: %TRUE if this is a write to the sysctl file
1901  * @buffer: the user buffer
1902  * @lenp: the size of the user buffer
1903  * @ppos: file position
1904  *
1905  * Reads/writes a string from/to the user buffer. If the kernel
1906  * buffer provided is not large enough to hold the string, the
1907  * string is truncated. The copied string is %NULL-terminated.
1908  * If the string is being read by the user process, it is copied
1909  * and a newline '\n' is added. It is truncated if the buffer is
1910  * not large enough.
1911  *
1912  * Returns 0 on success.
1913  */
1914 int proc_dostring(struct ctl_table *table, int write,
1915 		  void __user *buffer, size_t *lenp, loff_t *ppos)
1916 {
1917 	if (write && *ppos && sysctl_writes_strict == SYSCTL_WRITES_WARN)
1918 		warn_sysctl_write(table);
1919 
1920 	return _proc_do_string((char *)(table->data), table->maxlen, write,
1921 			       (char __user *)buffer, lenp, ppos);
1922 }
1923 
1924 static size_t proc_skip_spaces(char **buf)
1925 {
1926 	size_t ret;
1927 	char *tmp = skip_spaces(*buf);
1928 	ret = tmp - *buf;
1929 	*buf = tmp;
1930 	return ret;
1931 }
1932 
1933 static void proc_skip_char(char **buf, size_t *size, const char v)
1934 {
1935 	while (*size) {
1936 		if (**buf != v)
1937 			break;
1938 		(*size)--;
1939 		(*buf)++;
1940 	}
1941 }
1942 
1943 #define TMPBUFLEN 22
1944 /**
1945  * proc_get_long - reads an ASCII formatted integer from a user buffer
1946  *
1947  * @buf: a kernel buffer
1948  * @size: size of the kernel buffer
1949  * @val: this is where the number will be stored
1950  * @neg: set to %TRUE if number is negative
1951  * @perm_tr: a vector which contains the allowed trailers
1952  * @perm_tr_len: size of the perm_tr vector
1953  * @tr: pointer to store the trailer character
1954  *
1955  * In case of success %0 is returned and @buf and @size are updated with
1956  * the amount of bytes read. If @tr is non-NULL and a trailing
1957  * character exists (size is non-zero after returning from this
1958  * function), @tr is updated with the trailing character.
1959  */
1960 static int proc_get_long(char **buf, size_t *size,
1961 			  unsigned long *val, bool *neg,
1962 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
1963 {
1964 	int len;
1965 	char *p, tmp[TMPBUFLEN];
1966 
1967 	if (!*size)
1968 		return -EINVAL;
1969 
1970 	len = *size;
1971 	if (len > TMPBUFLEN - 1)
1972 		len = TMPBUFLEN - 1;
1973 
1974 	memcpy(tmp, *buf, len);
1975 
1976 	tmp[len] = 0;
1977 	p = tmp;
1978 	if (*p == '-' && *size > 1) {
1979 		*neg = true;
1980 		p++;
1981 	} else
1982 		*neg = false;
1983 	if (!isdigit(*p))
1984 		return -EINVAL;
1985 
1986 	*val = simple_strtoul(p, &p, 0);
1987 
1988 	len = p - tmp;
1989 
1990 	/* We don't know if the next char is whitespace thus we may accept
1991 	 * invalid integers (e.g. 1234...a) or two integers instead of one
1992 	 * (e.g. 123...1). So lets not allow such large numbers. */
1993 	if (len == TMPBUFLEN - 1)
1994 		return -EINVAL;
1995 
1996 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
1997 		return -EINVAL;
1998 
1999 	if (tr && (len < *size))
2000 		*tr = *p;
2001 
2002 	*buf += len;
2003 	*size -= len;
2004 
2005 	return 0;
2006 }
2007 
2008 /**
2009  * proc_put_long - converts an integer to a decimal ASCII formatted string
2010  *
2011  * @buf: the user buffer
2012  * @size: the size of the user buffer
2013  * @val: the integer to be converted
2014  * @neg: sign of the number, %TRUE for negative
2015  *
2016  * In case of success %0 is returned and @buf and @size are updated with
2017  * the amount of bytes written.
2018  */
2019 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2020 			  bool neg)
2021 {
2022 	int len;
2023 	char tmp[TMPBUFLEN], *p = tmp;
2024 
2025 	sprintf(p, "%s%lu", neg ? "-" : "", val);
2026 	len = strlen(tmp);
2027 	if (len > *size)
2028 		len = *size;
2029 	if (copy_to_user(*buf, tmp, len))
2030 		return -EFAULT;
2031 	*size -= len;
2032 	*buf += len;
2033 	return 0;
2034 }
2035 #undef TMPBUFLEN
2036 
2037 static int proc_put_char(void __user **buf, size_t *size, char c)
2038 {
2039 	if (*size) {
2040 		char __user **buffer = (char __user **)buf;
2041 		if (put_user(c, *buffer))
2042 			return -EFAULT;
2043 		(*size)--, (*buffer)++;
2044 		*buf = *buffer;
2045 	}
2046 	return 0;
2047 }
2048 
2049 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2050 				 int *valp,
2051 				 int write, void *data)
2052 {
2053 	if (write) {
2054 		if (*negp) {
2055 			if (*lvalp > (unsigned long) INT_MAX + 1)
2056 				return -EINVAL;
2057 			*valp = -*lvalp;
2058 		} else {
2059 			if (*lvalp > (unsigned long) INT_MAX)
2060 				return -EINVAL;
2061 			*valp = *lvalp;
2062 		}
2063 	} else {
2064 		int val = *valp;
2065 		if (val < 0) {
2066 			*negp = true;
2067 			*lvalp = -(unsigned long)val;
2068 		} else {
2069 			*negp = false;
2070 			*lvalp = (unsigned long)val;
2071 		}
2072 	}
2073 	return 0;
2074 }
2075 
2076 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2077 
2078 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2079 		  int write, void __user *buffer,
2080 		  size_t *lenp, loff_t *ppos,
2081 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2082 			      int write, void *data),
2083 		  void *data)
2084 {
2085 	int *i, vleft, first = 1, err = 0;
2086 	size_t left;
2087 	char *kbuf = NULL, *p;
2088 
2089 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2090 		*lenp = 0;
2091 		return 0;
2092 	}
2093 
2094 	i = (int *) tbl_data;
2095 	vleft = table->maxlen / sizeof(*i);
2096 	left = *lenp;
2097 
2098 	if (!conv)
2099 		conv = do_proc_dointvec_conv;
2100 
2101 	if (write) {
2102 		if (*ppos) {
2103 			switch (sysctl_writes_strict) {
2104 			case SYSCTL_WRITES_STRICT:
2105 				goto out;
2106 			case SYSCTL_WRITES_WARN:
2107 				warn_sysctl_write(table);
2108 				break;
2109 			default:
2110 				break;
2111 			}
2112 		}
2113 
2114 		if (left > PAGE_SIZE - 1)
2115 			left = PAGE_SIZE - 1;
2116 		p = kbuf = memdup_user_nul(buffer, left);
2117 		if (IS_ERR(kbuf))
2118 			return PTR_ERR(kbuf);
2119 	}
2120 
2121 	for (; left && vleft--; i++, first=0) {
2122 		unsigned long lval;
2123 		bool neg;
2124 
2125 		if (write) {
2126 			left -= proc_skip_spaces(&p);
2127 
2128 			if (!left)
2129 				break;
2130 			err = proc_get_long(&p, &left, &lval, &neg,
2131 					     proc_wspace_sep,
2132 					     sizeof(proc_wspace_sep), NULL);
2133 			if (err)
2134 				break;
2135 			if (conv(&neg, &lval, i, 1, data)) {
2136 				err = -EINVAL;
2137 				break;
2138 			}
2139 		} else {
2140 			if (conv(&neg, &lval, i, 0, data)) {
2141 				err = -EINVAL;
2142 				break;
2143 			}
2144 			if (!first)
2145 				err = proc_put_char(&buffer, &left, '\t');
2146 			if (err)
2147 				break;
2148 			err = proc_put_long(&buffer, &left, lval, neg);
2149 			if (err)
2150 				break;
2151 		}
2152 	}
2153 
2154 	if (!write && !first && left && !err)
2155 		err = proc_put_char(&buffer, &left, '\n');
2156 	if (write && !err && left)
2157 		left -= proc_skip_spaces(&p);
2158 	if (write) {
2159 		kfree(kbuf);
2160 		if (first)
2161 			return err ? : -EINVAL;
2162 	}
2163 	*lenp -= left;
2164 out:
2165 	*ppos += *lenp;
2166 	return err;
2167 }
2168 
2169 static int do_proc_dointvec(struct ctl_table *table, int write,
2170 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2171 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2172 			      int write, void *data),
2173 		  void *data)
2174 {
2175 	return __do_proc_dointvec(table->data, table, write,
2176 			buffer, lenp, ppos, conv, data);
2177 }
2178 
2179 /**
2180  * proc_dointvec - read a vector of integers
2181  * @table: the sysctl table
2182  * @write: %TRUE if this is a write to the sysctl file
2183  * @buffer: the user buffer
2184  * @lenp: the size of the user buffer
2185  * @ppos: file position
2186  *
2187  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2188  * values from/to the user buffer, treated as an ASCII string.
2189  *
2190  * Returns 0 on success.
2191  */
2192 int proc_dointvec(struct ctl_table *table, int write,
2193 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2194 {
2195     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2196 		    	    NULL,NULL);
2197 }
2198 
2199 /*
2200  * Taint values can only be increased
2201  * This means we can safely use a temporary.
2202  */
2203 static int proc_taint(struct ctl_table *table, int write,
2204 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2205 {
2206 	struct ctl_table t;
2207 	unsigned long tmptaint = get_taint();
2208 	int err;
2209 
2210 	if (write && !capable(CAP_SYS_ADMIN))
2211 		return -EPERM;
2212 
2213 	t = *table;
2214 	t.data = &tmptaint;
2215 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2216 	if (err < 0)
2217 		return err;
2218 
2219 	if (write) {
2220 		/*
2221 		 * Poor man's atomic or. Not worth adding a primitive
2222 		 * to everyone's atomic.h for this
2223 		 */
2224 		int i;
2225 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2226 			if ((tmptaint >> i) & 1)
2227 				add_taint(i, LOCKDEP_STILL_OK);
2228 		}
2229 	}
2230 
2231 	return err;
2232 }
2233 
2234 #ifdef CONFIG_PRINTK
2235 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2236 				void __user *buffer, size_t *lenp, loff_t *ppos)
2237 {
2238 	if (write && !capable(CAP_SYS_ADMIN))
2239 		return -EPERM;
2240 
2241 	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2242 }
2243 #endif
2244 
2245 struct do_proc_dointvec_minmax_conv_param {
2246 	int *min;
2247 	int *max;
2248 };
2249 
2250 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2251 					int *valp,
2252 					int write, void *data)
2253 {
2254 	struct do_proc_dointvec_minmax_conv_param *param = data;
2255 	if (write) {
2256 		int val = *negp ? -*lvalp : *lvalp;
2257 		if ((param->min && *param->min > val) ||
2258 		    (param->max && *param->max < val))
2259 			return -EINVAL;
2260 		*valp = val;
2261 	} else {
2262 		int val = *valp;
2263 		if (val < 0) {
2264 			*negp = true;
2265 			*lvalp = -(unsigned long)val;
2266 		} else {
2267 			*negp = false;
2268 			*lvalp = (unsigned long)val;
2269 		}
2270 	}
2271 	return 0;
2272 }
2273 
2274 /**
2275  * proc_dointvec_minmax - read a vector of integers with min/max values
2276  * @table: the sysctl table
2277  * @write: %TRUE if this is a write to the sysctl file
2278  * @buffer: the user buffer
2279  * @lenp: the size of the user buffer
2280  * @ppos: file position
2281  *
2282  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2283  * values from/to the user buffer, treated as an ASCII string.
2284  *
2285  * This routine will ensure the values are within the range specified by
2286  * table->extra1 (min) and table->extra2 (max).
2287  *
2288  * Returns 0 on success.
2289  */
2290 int proc_dointvec_minmax(struct ctl_table *table, int write,
2291 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2292 {
2293 	struct do_proc_dointvec_minmax_conv_param param = {
2294 		.min = (int *) table->extra1,
2295 		.max = (int *) table->extra2,
2296 	};
2297 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2298 				do_proc_dointvec_minmax_conv, &param);
2299 }
2300 
2301 static void validate_coredump_safety(void)
2302 {
2303 #ifdef CONFIG_COREDUMP
2304 	if (suid_dumpable == SUID_DUMP_ROOT &&
2305 	    core_pattern[0] != '/' && core_pattern[0] != '|') {
2306 		printk(KERN_WARNING "Unsafe core_pattern used with "\
2307 			"suid_dumpable=2. Pipe handler or fully qualified "\
2308 			"core dump path required.\n");
2309 	}
2310 #endif
2311 }
2312 
2313 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2314 		void __user *buffer, size_t *lenp, loff_t *ppos)
2315 {
2316 	int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2317 	if (!error)
2318 		validate_coredump_safety();
2319 	return error;
2320 }
2321 
2322 #ifdef CONFIG_COREDUMP
2323 static int proc_dostring_coredump(struct ctl_table *table, int write,
2324 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2325 {
2326 	int error = proc_dostring(table, write, buffer, lenp, ppos);
2327 	if (!error)
2328 		validate_coredump_safety();
2329 	return error;
2330 }
2331 #endif
2332 
2333 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2334 				     void __user *buffer,
2335 				     size_t *lenp, loff_t *ppos,
2336 				     unsigned long convmul,
2337 				     unsigned long convdiv)
2338 {
2339 	unsigned long *i, *min, *max;
2340 	int vleft, first = 1, err = 0;
2341 	size_t left;
2342 	char *kbuf = NULL, *p;
2343 
2344 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2345 		*lenp = 0;
2346 		return 0;
2347 	}
2348 
2349 	i = (unsigned long *) data;
2350 	min = (unsigned long *) table->extra1;
2351 	max = (unsigned long *) table->extra2;
2352 	vleft = table->maxlen / sizeof(unsigned long);
2353 	left = *lenp;
2354 
2355 	if (write) {
2356 		if (*ppos) {
2357 			switch (sysctl_writes_strict) {
2358 			case SYSCTL_WRITES_STRICT:
2359 				goto out;
2360 			case SYSCTL_WRITES_WARN:
2361 				warn_sysctl_write(table);
2362 				break;
2363 			default:
2364 				break;
2365 			}
2366 		}
2367 
2368 		if (left > PAGE_SIZE - 1)
2369 			left = PAGE_SIZE - 1;
2370 		p = kbuf = memdup_user_nul(buffer, left);
2371 		if (IS_ERR(kbuf))
2372 			return PTR_ERR(kbuf);
2373 	}
2374 
2375 	for (; left && vleft--; i++, first = 0) {
2376 		unsigned long val;
2377 
2378 		if (write) {
2379 			bool neg;
2380 
2381 			left -= proc_skip_spaces(&p);
2382 
2383 			err = proc_get_long(&p, &left, &val, &neg,
2384 					     proc_wspace_sep,
2385 					     sizeof(proc_wspace_sep), NULL);
2386 			if (err)
2387 				break;
2388 			if (neg)
2389 				continue;
2390 			if ((min && val < *min) || (max && val > *max))
2391 				continue;
2392 			*i = val;
2393 		} else {
2394 			val = convdiv * (*i) / convmul;
2395 			if (!first) {
2396 				err = proc_put_char(&buffer, &left, '\t');
2397 				if (err)
2398 					break;
2399 			}
2400 			err = proc_put_long(&buffer, &left, val, false);
2401 			if (err)
2402 				break;
2403 		}
2404 	}
2405 
2406 	if (!write && !first && left && !err)
2407 		err = proc_put_char(&buffer, &left, '\n');
2408 	if (write && !err)
2409 		left -= proc_skip_spaces(&p);
2410 	if (write) {
2411 		kfree(kbuf);
2412 		if (first)
2413 			return err ? : -EINVAL;
2414 	}
2415 	*lenp -= left;
2416 out:
2417 	*ppos += *lenp;
2418 	return err;
2419 }
2420 
2421 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2422 				     void __user *buffer,
2423 				     size_t *lenp, loff_t *ppos,
2424 				     unsigned long convmul,
2425 				     unsigned long convdiv)
2426 {
2427 	return __do_proc_doulongvec_minmax(table->data, table, write,
2428 			buffer, lenp, ppos, convmul, convdiv);
2429 }
2430 
2431 /**
2432  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2433  * @table: the sysctl table
2434  * @write: %TRUE if this is a write to the sysctl file
2435  * @buffer: the user buffer
2436  * @lenp: the size of the user buffer
2437  * @ppos: file position
2438  *
2439  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2440  * values from/to the user buffer, treated as an ASCII string.
2441  *
2442  * This routine will ensure the values are within the range specified by
2443  * table->extra1 (min) and table->extra2 (max).
2444  *
2445  * Returns 0 on success.
2446  */
2447 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2448 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2449 {
2450     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2451 }
2452 
2453 /**
2454  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2455  * @table: the sysctl table
2456  * @write: %TRUE if this is a write to the sysctl file
2457  * @buffer: the user buffer
2458  * @lenp: the size of the user buffer
2459  * @ppos: file position
2460  *
2461  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2462  * values from/to the user buffer, treated as an ASCII string. The values
2463  * are treated as milliseconds, and converted to jiffies when they are stored.
2464  *
2465  * This routine will ensure the values are within the range specified by
2466  * table->extra1 (min) and table->extra2 (max).
2467  *
2468  * Returns 0 on success.
2469  */
2470 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2471 				      void __user *buffer,
2472 				      size_t *lenp, loff_t *ppos)
2473 {
2474     return do_proc_doulongvec_minmax(table, write, buffer,
2475 				     lenp, ppos, HZ, 1000l);
2476 }
2477 
2478 
2479 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2480 					 int *valp,
2481 					 int write, void *data)
2482 {
2483 	if (write) {
2484 		if (*lvalp > LONG_MAX / HZ)
2485 			return 1;
2486 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2487 	} else {
2488 		int val = *valp;
2489 		unsigned long lval;
2490 		if (val < 0) {
2491 			*negp = true;
2492 			lval = -(unsigned long)val;
2493 		} else {
2494 			*negp = false;
2495 			lval = (unsigned long)val;
2496 		}
2497 		*lvalp = lval / HZ;
2498 	}
2499 	return 0;
2500 }
2501 
2502 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2503 						int *valp,
2504 						int write, void *data)
2505 {
2506 	if (write) {
2507 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2508 			return 1;
2509 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2510 	} else {
2511 		int val = *valp;
2512 		unsigned long lval;
2513 		if (val < 0) {
2514 			*negp = true;
2515 			lval = -(unsigned long)val;
2516 		} else {
2517 			*negp = false;
2518 			lval = (unsigned long)val;
2519 		}
2520 		*lvalp = jiffies_to_clock_t(lval);
2521 	}
2522 	return 0;
2523 }
2524 
2525 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2526 					    int *valp,
2527 					    int write, void *data)
2528 {
2529 	if (write) {
2530 		unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2531 
2532 		if (jif > INT_MAX)
2533 			return 1;
2534 		*valp = (int)jif;
2535 	} else {
2536 		int val = *valp;
2537 		unsigned long lval;
2538 		if (val < 0) {
2539 			*negp = true;
2540 			lval = -(unsigned long)val;
2541 		} else {
2542 			*negp = false;
2543 			lval = (unsigned long)val;
2544 		}
2545 		*lvalp = jiffies_to_msecs(lval);
2546 	}
2547 	return 0;
2548 }
2549 
2550 /**
2551  * proc_dointvec_jiffies - read a vector of integers as seconds
2552  * @table: the sysctl table
2553  * @write: %TRUE if this is a write to the sysctl file
2554  * @buffer: the user buffer
2555  * @lenp: the size of the user buffer
2556  * @ppos: file position
2557  *
2558  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2559  * values from/to the user buffer, treated as an ASCII string.
2560  * The values read are assumed to be in seconds, and are converted into
2561  * jiffies.
2562  *
2563  * Returns 0 on success.
2564  */
2565 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2566 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2567 {
2568     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2569 		    	    do_proc_dointvec_jiffies_conv,NULL);
2570 }
2571 
2572 /**
2573  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2574  * @table: the sysctl table
2575  * @write: %TRUE if this is a write to the sysctl file
2576  * @buffer: the user buffer
2577  * @lenp: the size of the user buffer
2578  * @ppos: pointer to the file position
2579  *
2580  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2581  * values from/to the user buffer, treated as an ASCII string.
2582  * The values read are assumed to be in 1/USER_HZ seconds, and
2583  * are converted into jiffies.
2584  *
2585  * Returns 0 on success.
2586  */
2587 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2588 				 void __user *buffer, size_t *lenp, loff_t *ppos)
2589 {
2590     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2591 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
2592 }
2593 
2594 /**
2595  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2596  * @table: the sysctl table
2597  * @write: %TRUE if this is a write to the sysctl file
2598  * @buffer: the user buffer
2599  * @lenp: the size of the user buffer
2600  * @ppos: file position
2601  * @ppos: the current position in the file
2602  *
2603  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2604  * values from/to the user buffer, treated as an ASCII string.
2605  * The values read are assumed to be in 1/1000 seconds, and
2606  * are converted into jiffies.
2607  *
2608  * Returns 0 on success.
2609  */
2610 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2611 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2612 {
2613 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2614 				do_proc_dointvec_ms_jiffies_conv, NULL);
2615 }
2616 
2617 static int proc_do_cad_pid(struct ctl_table *table, int write,
2618 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2619 {
2620 	struct pid *new_pid;
2621 	pid_t tmp;
2622 	int r;
2623 
2624 	tmp = pid_vnr(cad_pid);
2625 
2626 	r = __do_proc_dointvec(&tmp, table, write, buffer,
2627 			       lenp, ppos, NULL, NULL);
2628 	if (r || !write)
2629 		return r;
2630 
2631 	new_pid = find_get_pid(tmp);
2632 	if (!new_pid)
2633 		return -ESRCH;
2634 
2635 	put_pid(xchg(&cad_pid, new_pid));
2636 	return 0;
2637 }
2638 
2639 /**
2640  * proc_do_large_bitmap - read/write from/to a large bitmap
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  * The bitmap is stored at table->data and the bitmap length (in bits)
2648  * in table->maxlen.
2649  *
2650  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2651  * large bitmaps may be represented in a compact manner. Writing into
2652  * the file will clear the bitmap then update it with the given input.
2653  *
2654  * Returns 0 on success.
2655  */
2656 int proc_do_large_bitmap(struct ctl_table *table, int write,
2657 			 void __user *buffer, size_t *lenp, loff_t *ppos)
2658 {
2659 	int err = 0;
2660 	bool first = 1;
2661 	size_t left = *lenp;
2662 	unsigned long bitmap_len = table->maxlen;
2663 	unsigned long *bitmap = *(unsigned long **) table->data;
2664 	unsigned long *tmp_bitmap = NULL;
2665 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2666 
2667 	if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
2668 		*lenp = 0;
2669 		return 0;
2670 	}
2671 
2672 	if (write) {
2673 		char *kbuf, *p;
2674 
2675 		if (left > PAGE_SIZE - 1)
2676 			left = PAGE_SIZE - 1;
2677 
2678 		p = kbuf = memdup_user_nul(buffer, left);
2679 		if (IS_ERR(kbuf))
2680 			return PTR_ERR(kbuf);
2681 
2682 		tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2683 				     GFP_KERNEL);
2684 		if (!tmp_bitmap) {
2685 			kfree(kbuf);
2686 			return -ENOMEM;
2687 		}
2688 		proc_skip_char(&p, &left, '\n');
2689 		while (!err && left) {
2690 			unsigned long val_a, val_b;
2691 			bool neg;
2692 
2693 			err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
2694 					     sizeof(tr_a), &c);
2695 			if (err)
2696 				break;
2697 			if (val_a >= bitmap_len || neg) {
2698 				err = -EINVAL;
2699 				break;
2700 			}
2701 
2702 			val_b = val_a;
2703 			if (left) {
2704 				p++;
2705 				left--;
2706 			}
2707 
2708 			if (c == '-') {
2709 				err = proc_get_long(&p, &left, &val_b,
2710 						     &neg, tr_b, sizeof(tr_b),
2711 						     &c);
2712 				if (err)
2713 					break;
2714 				if (val_b >= bitmap_len || neg ||
2715 				    val_a > val_b) {
2716 					err = -EINVAL;
2717 					break;
2718 				}
2719 				if (left) {
2720 					p++;
2721 					left--;
2722 				}
2723 			}
2724 
2725 			bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2726 			first = 0;
2727 			proc_skip_char(&p, &left, '\n');
2728 		}
2729 		kfree(kbuf);
2730 	} else {
2731 		unsigned long bit_a, bit_b = 0;
2732 
2733 		while (left) {
2734 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2735 			if (bit_a >= bitmap_len)
2736 				break;
2737 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
2738 						   bit_a + 1) - 1;
2739 
2740 			if (!first) {
2741 				err = proc_put_char(&buffer, &left, ',');
2742 				if (err)
2743 					break;
2744 			}
2745 			err = proc_put_long(&buffer, &left, bit_a, false);
2746 			if (err)
2747 				break;
2748 			if (bit_a != bit_b) {
2749 				err = proc_put_char(&buffer, &left, '-');
2750 				if (err)
2751 					break;
2752 				err = proc_put_long(&buffer, &left, bit_b, false);
2753 				if (err)
2754 					break;
2755 			}
2756 
2757 			first = 0; bit_b++;
2758 		}
2759 		if (!err)
2760 			err = proc_put_char(&buffer, &left, '\n');
2761 	}
2762 
2763 	if (!err) {
2764 		if (write) {
2765 			if (*ppos)
2766 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2767 			else
2768 				bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2769 		}
2770 		kfree(tmp_bitmap);
2771 		*lenp -= left;
2772 		*ppos += *lenp;
2773 		return 0;
2774 	} else {
2775 		kfree(tmp_bitmap);
2776 		return err;
2777 	}
2778 }
2779 
2780 #else /* CONFIG_PROC_SYSCTL */
2781 
2782 int proc_dostring(struct ctl_table *table, int write,
2783 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2784 {
2785 	return -ENOSYS;
2786 }
2787 
2788 int proc_dointvec(struct ctl_table *table, int write,
2789 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2790 {
2791 	return -ENOSYS;
2792 }
2793 
2794 int proc_dointvec_minmax(struct ctl_table *table, int write,
2795 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2796 {
2797 	return -ENOSYS;
2798 }
2799 
2800 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2801 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2802 {
2803 	return -ENOSYS;
2804 }
2805 
2806 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2807 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2808 {
2809 	return -ENOSYS;
2810 }
2811 
2812 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2813 			     void __user *buffer, size_t *lenp, loff_t *ppos)
2814 {
2815 	return -ENOSYS;
2816 }
2817 
2818 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2819 		    void __user *buffer, size_t *lenp, loff_t *ppos)
2820 {
2821 	return -ENOSYS;
2822 }
2823 
2824 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2825 				      void __user *buffer,
2826 				      size_t *lenp, loff_t *ppos)
2827 {
2828     return -ENOSYS;
2829 }
2830 
2831 
2832 #endif /* CONFIG_PROC_SYSCTL */
2833 
2834 /*
2835  * No sense putting this after each symbol definition, twice,
2836  * exception granted :-)
2837  */
2838 EXPORT_SYMBOL(proc_dointvec);
2839 EXPORT_SYMBOL(proc_dointvec_jiffies);
2840 EXPORT_SYMBOL(proc_dointvec_minmax);
2841 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2842 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2843 EXPORT_SYMBOL(proc_dostring);
2844 EXPORT_SYMBOL(proc_doulongvec_minmax);
2845 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2846