xref: /linux/kernel/sysctl.c (revision 544029862cbb1d7903e19f2e58f48d4884e1201b)
1 /*
2  * sysctl.c: General linux system control interface
3  *
4  * Begun 24 March 1995, Stephen Tweedie
5  * Added /proc support, Dec 1995
6  * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7  * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8  * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9  * Dynamic registration fixes, Stephen Tweedie.
10  * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11  * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12  *  Horn.
13  * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14  * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15  * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16  *  Wendling.
17  * The list_for_each() macro wasn't appropriate for the sysctl loop.
18  *  Removed it and replaced it with older style, 03/23/00, Bill Wendling
19  */
20 
21 #include <linux/module.h>
22 #include <linux/aio.h>
23 #include <linux/mm.h>
24 #include <linux/swap.h>
25 #include <linux/slab.h>
26 #include <linux/sysctl.h>
27 #include <linux/bitmap.h>
28 #include <linux/signal.h>
29 #include <linux/printk.h>
30 #include <linux/proc_fs.h>
31 #include <linux/security.h>
32 #include <linux/ctype.h>
33 #include <linux/kmemleak.h>
34 #include <linux/fs.h>
35 #include <linux/init.h>
36 #include <linux/kernel.h>
37 #include <linux/kobject.h>
38 #include <linux/net.h>
39 #include <linux/sysrq.h>
40 #include <linux/highuid.h>
41 #include <linux/writeback.h>
42 #include <linux/ratelimit.h>
43 #include <linux/compaction.h>
44 #include <linux/hugetlb.h>
45 #include <linux/initrd.h>
46 #include <linux/key.h>
47 #include <linux/times.h>
48 #include <linux/limits.h>
49 #include <linux/dcache.h>
50 #include <linux/dnotify.h>
51 #include <linux/syscalls.h>
52 #include <linux/vmstat.h>
53 #include <linux/nfs_fs.h>
54 #include <linux/acpi.h>
55 #include <linux/reboot.h>
56 #include <linux/ftrace.h>
57 #include <linux/perf_event.h>
58 #include <linux/kprobes.h>
59 #include <linux/pipe_fs_i.h>
60 #include <linux/oom.h>
61 #include <linux/kmod.h>
62 #include <linux/capability.h>
63 #include <linux/binfmts.h>
64 #include <linux/sched/sysctl.h>
65 #include <linux/sched/coredump.h>
66 #include <linux/kexec.h>
67 #include <linux/bpf.h>
68 #include <linux/mount.h>
69 
70 #include <linux/uaccess.h>
71 #include <asm/processor.h>
72 
73 #ifdef CONFIG_X86
74 #include <asm/nmi.h>
75 #include <asm/stacktrace.h>
76 #include <asm/io.h>
77 #endif
78 #ifdef CONFIG_SPARC
79 #include <asm/setup.h>
80 #endif
81 #ifdef CONFIG_BSD_PROCESS_ACCT
82 #include <linux/acct.h>
83 #endif
84 #ifdef CONFIG_RT_MUTEXES
85 #include <linux/rtmutex.h>
86 #endif
87 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
88 #include <linux/lockdep.h>
89 #endif
90 #ifdef CONFIG_CHR_DEV_SG
91 #include <scsi/sg.h>
92 #endif
93 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
94 #include <linux/stackleak.h>
95 #endif
96 #ifdef CONFIG_LOCKUP_DETECTOR
97 #include <linux/nmi.h>
98 #endif
99 
100 #if defined(CONFIG_SYSCTL)
101 
102 /* External variables not in a header file. */
103 extern int suid_dumpable;
104 #ifdef CONFIG_COREDUMP
105 extern int core_uses_pid;
106 extern char core_pattern[];
107 extern unsigned int core_pipe_limit;
108 #endif
109 extern int pid_max;
110 extern int pid_max_min, pid_max_max;
111 extern int percpu_pagelist_fraction;
112 extern int latencytop_enabled;
113 extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
114 #ifndef CONFIG_MMU
115 extern int sysctl_nr_trim_pages;
116 #endif
117 
118 /* Constants used for minimum and  maximum */
119 #ifdef CONFIG_LOCKUP_DETECTOR
120 static int sixty = 60;
121 #endif
122 
123 static int __maybe_unused neg_one = -1;
124 
125 static int zero;
126 static int __maybe_unused one = 1;
127 static int __maybe_unused two = 2;
128 static int __maybe_unused four = 4;
129 static unsigned long one_ul = 1;
130 static int one_hundred = 100;
131 static int one_thousand = 1000;
132 #ifdef CONFIG_PRINTK
133 static int ten_thousand = 10000;
134 #endif
135 #ifdef CONFIG_PERF_EVENTS
136 static int six_hundred_forty_kb = 640 * 1024;
137 #endif
138 
139 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
140 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
141 
142 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
143 static int maxolduid = 65535;
144 static int minolduid;
145 
146 static int ngroups_max = NGROUPS_MAX;
147 static const int cap_last_cap = CAP_LAST_CAP;
148 
149 /*
150  * This is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs
151  * and hung_task_check_interval_secs
152  */
153 #ifdef CONFIG_DETECT_HUNG_TASK
154 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
155 #endif
156 
157 #ifdef CONFIG_INOTIFY_USER
158 #include <linux/inotify.h>
159 #endif
160 #ifdef CONFIG_SPARC
161 #endif
162 
163 #ifdef __hppa__
164 extern int pwrsw_enabled;
165 #endif
166 
167 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
168 extern int unaligned_enabled;
169 #endif
170 
171 #ifdef CONFIG_IA64
172 extern int unaligned_dump_stack;
173 #endif
174 
175 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
176 extern int no_unaligned_warning;
177 #endif
178 
179 #ifdef CONFIG_PROC_SYSCTL
180 
181 /**
182  * enum sysctl_writes_mode - supported sysctl write modes
183  *
184  * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
185  * 	to be written, and multiple writes on the same sysctl file descriptor
186  * 	will rewrite the sysctl value, regardless of file position. No warning
187  * 	is issued when the initial position is not 0.
188  * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
189  * 	not 0.
190  * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
191  * 	file position 0 and the value must be fully contained in the buffer
192  * 	sent to the write syscall. If dealing with strings respect the file
193  * 	position, but restrict this to the max length of the buffer, anything
194  * 	passed the max lenght will be ignored. Multiple writes will append
195  * 	to the buffer.
196  *
197  * These write modes control how current file position affects the behavior of
198  * updating sysctl values through the proc interface on each write.
199  */
200 enum sysctl_writes_mode {
201 	SYSCTL_WRITES_LEGACY		= -1,
202 	SYSCTL_WRITES_WARN		= 0,
203 	SYSCTL_WRITES_STRICT		= 1,
204 };
205 
206 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
207 
208 static int proc_do_cad_pid(struct ctl_table *table, int write,
209 		  void __user *buffer, size_t *lenp, loff_t *ppos);
210 static int proc_taint(struct ctl_table *table, int write,
211 			       void __user *buffer, size_t *lenp, loff_t *ppos);
212 #endif
213 
214 #ifdef CONFIG_PRINTK
215 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
216 				void __user *buffer, size_t *lenp, loff_t *ppos);
217 #endif
218 
219 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
220 		void __user *buffer, size_t *lenp, loff_t *ppos);
221 #ifdef CONFIG_COREDUMP
222 static int proc_dostring_coredump(struct ctl_table *table, int write,
223 		void __user *buffer, size_t *lenp, loff_t *ppos);
224 #endif
225 static int proc_dopipe_max_size(struct ctl_table *table, int write,
226 		void __user *buffer, size_t *lenp, loff_t *ppos);
227 #ifdef CONFIG_BPF_SYSCALL
228 static int proc_dointvec_minmax_bpf_stats(struct ctl_table *table, int write,
229 					  void __user *buffer, size_t *lenp,
230 					  loff_t *ppos);
231 #endif
232 
233 #ifdef CONFIG_MAGIC_SYSRQ
234 /* Note: sysrq code uses its own private copy */
235 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
236 
237 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
238 				void __user *buffer, size_t *lenp,
239 				loff_t *ppos)
240 {
241 	int error;
242 
243 	error = proc_dointvec(table, write, buffer, lenp, ppos);
244 	if (error)
245 		return error;
246 
247 	if (write)
248 		sysrq_toggle_support(__sysrq_enabled);
249 
250 	return 0;
251 }
252 
253 #endif
254 
255 static struct ctl_table kern_table[];
256 static struct ctl_table vm_table[];
257 static struct ctl_table fs_table[];
258 static struct ctl_table debug_table[];
259 static struct ctl_table dev_table[];
260 extern struct ctl_table random_table[];
261 #ifdef CONFIG_EPOLL
262 extern struct ctl_table epoll_table[];
263 #endif
264 
265 #ifdef CONFIG_FW_LOADER_USER_HELPER
266 extern struct ctl_table firmware_config_table[];
267 #endif
268 
269 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
270 int sysctl_legacy_va_layout;
271 #endif
272 
273 /* The default sysctl tables: */
274 
275 static struct ctl_table sysctl_base_table[] = {
276 	{
277 		.procname	= "kernel",
278 		.mode		= 0555,
279 		.child		= kern_table,
280 	},
281 	{
282 		.procname	= "vm",
283 		.mode		= 0555,
284 		.child		= vm_table,
285 	},
286 	{
287 		.procname	= "fs",
288 		.mode		= 0555,
289 		.child		= fs_table,
290 	},
291 	{
292 		.procname	= "debug",
293 		.mode		= 0555,
294 		.child		= debug_table,
295 	},
296 	{
297 		.procname	= "dev",
298 		.mode		= 0555,
299 		.child		= dev_table,
300 	},
301 	{ }
302 };
303 
304 #ifdef CONFIG_SCHED_DEBUG
305 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
306 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
307 static int min_wakeup_granularity_ns;			/* 0 usecs */
308 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
309 #ifdef CONFIG_SMP
310 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
311 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
312 #endif /* CONFIG_SMP */
313 #endif /* CONFIG_SCHED_DEBUG */
314 
315 #ifdef CONFIG_COMPACTION
316 static int min_extfrag_threshold;
317 static int max_extfrag_threshold = 1000;
318 #endif
319 
320 static struct ctl_table kern_table[] = {
321 	{
322 		.procname	= "sched_child_runs_first",
323 		.data		= &sysctl_sched_child_runs_first,
324 		.maxlen		= sizeof(unsigned int),
325 		.mode		= 0644,
326 		.proc_handler	= proc_dointvec,
327 	},
328 #ifdef CONFIG_SCHED_DEBUG
329 	{
330 		.procname	= "sched_min_granularity_ns",
331 		.data		= &sysctl_sched_min_granularity,
332 		.maxlen		= sizeof(unsigned int),
333 		.mode		= 0644,
334 		.proc_handler	= sched_proc_update_handler,
335 		.extra1		= &min_sched_granularity_ns,
336 		.extra2		= &max_sched_granularity_ns,
337 	},
338 	{
339 		.procname	= "sched_latency_ns",
340 		.data		= &sysctl_sched_latency,
341 		.maxlen		= sizeof(unsigned int),
342 		.mode		= 0644,
343 		.proc_handler	= sched_proc_update_handler,
344 		.extra1		= &min_sched_granularity_ns,
345 		.extra2		= &max_sched_granularity_ns,
346 	},
347 	{
348 		.procname	= "sched_wakeup_granularity_ns",
349 		.data		= &sysctl_sched_wakeup_granularity,
350 		.maxlen		= sizeof(unsigned int),
351 		.mode		= 0644,
352 		.proc_handler	= sched_proc_update_handler,
353 		.extra1		= &min_wakeup_granularity_ns,
354 		.extra2		= &max_wakeup_granularity_ns,
355 	},
356 #ifdef CONFIG_SMP
357 	{
358 		.procname	= "sched_tunable_scaling",
359 		.data		= &sysctl_sched_tunable_scaling,
360 		.maxlen		= sizeof(enum sched_tunable_scaling),
361 		.mode		= 0644,
362 		.proc_handler	= sched_proc_update_handler,
363 		.extra1		= &min_sched_tunable_scaling,
364 		.extra2		= &max_sched_tunable_scaling,
365 	},
366 	{
367 		.procname	= "sched_migration_cost_ns",
368 		.data		= &sysctl_sched_migration_cost,
369 		.maxlen		= sizeof(unsigned int),
370 		.mode		= 0644,
371 		.proc_handler	= proc_dointvec,
372 	},
373 	{
374 		.procname	= "sched_nr_migrate",
375 		.data		= &sysctl_sched_nr_migrate,
376 		.maxlen		= sizeof(unsigned int),
377 		.mode		= 0644,
378 		.proc_handler	= proc_dointvec,
379 	},
380 #ifdef CONFIG_SCHEDSTATS
381 	{
382 		.procname	= "sched_schedstats",
383 		.data		= NULL,
384 		.maxlen		= sizeof(unsigned int),
385 		.mode		= 0644,
386 		.proc_handler	= sysctl_schedstats,
387 		.extra1		= &zero,
388 		.extra2		= &one,
389 	},
390 #endif /* CONFIG_SCHEDSTATS */
391 #endif /* CONFIG_SMP */
392 #ifdef CONFIG_NUMA_BALANCING
393 	{
394 		.procname	= "numa_balancing_scan_delay_ms",
395 		.data		= &sysctl_numa_balancing_scan_delay,
396 		.maxlen		= sizeof(unsigned int),
397 		.mode		= 0644,
398 		.proc_handler	= proc_dointvec,
399 	},
400 	{
401 		.procname	= "numa_balancing_scan_period_min_ms",
402 		.data		= &sysctl_numa_balancing_scan_period_min,
403 		.maxlen		= sizeof(unsigned int),
404 		.mode		= 0644,
405 		.proc_handler	= proc_dointvec,
406 	},
407 	{
408 		.procname	= "numa_balancing_scan_period_max_ms",
409 		.data		= &sysctl_numa_balancing_scan_period_max,
410 		.maxlen		= sizeof(unsigned int),
411 		.mode		= 0644,
412 		.proc_handler	= proc_dointvec,
413 	},
414 	{
415 		.procname	= "numa_balancing_scan_size_mb",
416 		.data		= &sysctl_numa_balancing_scan_size,
417 		.maxlen		= sizeof(unsigned int),
418 		.mode		= 0644,
419 		.proc_handler	= proc_dointvec_minmax,
420 		.extra1		= &one,
421 	},
422 	{
423 		.procname	= "numa_balancing",
424 		.data		= NULL, /* filled in by handler */
425 		.maxlen		= sizeof(unsigned int),
426 		.mode		= 0644,
427 		.proc_handler	= sysctl_numa_balancing,
428 		.extra1		= &zero,
429 		.extra2		= &one,
430 	},
431 #endif /* CONFIG_NUMA_BALANCING */
432 #endif /* CONFIG_SCHED_DEBUG */
433 	{
434 		.procname	= "sched_rt_period_us",
435 		.data		= &sysctl_sched_rt_period,
436 		.maxlen		= sizeof(unsigned int),
437 		.mode		= 0644,
438 		.proc_handler	= sched_rt_handler,
439 	},
440 	{
441 		.procname	= "sched_rt_runtime_us",
442 		.data		= &sysctl_sched_rt_runtime,
443 		.maxlen		= sizeof(int),
444 		.mode		= 0644,
445 		.proc_handler	= sched_rt_handler,
446 	},
447 	{
448 		.procname	= "sched_rr_timeslice_ms",
449 		.data		= &sysctl_sched_rr_timeslice,
450 		.maxlen		= sizeof(int),
451 		.mode		= 0644,
452 		.proc_handler	= sched_rr_handler,
453 	},
454 #ifdef CONFIG_SCHED_AUTOGROUP
455 	{
456 		.procname	= "sched_autogroup_enabled",
457 		.data		= &sysctl_sched_autogroup_enabled,
458 		.maxlen		= sizeof(unsigned int),
459 		.mode		= 0644,
460 		.proc_handler	= proc_dointvec_minmax,
461 		.extra1		= &zero,
462 		.extra2		= &one,
463 	},
464 #endif
465 #ifdef CONFIG_CFS_BANDWIDTH
466 	{
467 		.procname	= "sched_cfs_bandwidth_slice_us",
468 		.data		= &sysctl_sched_cfs_bandwidth_slice,
469 		.maxlen		= sizeof(unsigned int),
470 		.mode		= 0644,
471 		.proc_handler	= proc_dointvec_minmax,
472 		.extra1		= &one,
473 	},
474 #endif
475 #ifdef CONFIG_PROVE_LOCKING
476 	{
477 		.procname	= "prove_locking",
478 		.data		= &prove_locking,
479 		.maxlen		= sizeof(int),
480 		.mode		= 0644,
481 		.proc_handler	= proc_dointvec,
482 	},
483 #endif
484 #ifdef CONFIG_LOCK_STAT
485 	{
486 		.procname	= "lock_stat",
487 		.data		= &lock_stat,
488 		.maxlen		= sizeof(int),
489 		.mode		= 0644,
490 		.proc_handler	= proc_dointvec,
491 	},
492 #endif
493 	{
494 		.procname	= "panic",
495 		.data		= &panic_timeout,
496 		.maxlen		= sizeof(int),
497 		.mode		= 0644,
498 		.proc_handler	= proc_dointvec,
499 	},
500 #ifdef CONFIG_COREDUMP
501 	{
502 		.procname	= "core_uses_pid",
503 		.data		= &core_uses_pid,
504 		.maxlen		= sizeof(int),
505 		.mode		= 0644,
506 		.proc_handler	= proc_dointvec,
507 	},
508 	{
509 		.procname	= "core_pattern",
510 		.data		= core_pattern,
511 		.maxlen		= CORENAME_MAX_SIZE,
512 		.mode		= 0644,
513 		.proc_handler	= proc_dostring_coredump,
514 	},
515 	{
516 		.procname	= "core_pipe_limit",
517 		.data		= &core_pipe_limit,
518 		.maxlen		= sizeof(unsigned int),
519 		.mode		= 0644,
520 		.proc_handler	= proc_dointvec,
521 	},
522 #endif
523 #ifdef CONFIG_PROC_SYSCTL
524 	{
525 		.procname	= "tainted",
526 		.maxlen 	= sizeof(long),
527 		.mode		= 0644,
528 		.proc_handler	= proc_taint,
529 	},
530 	{
531 		.procname	= "sysctl_writes_strict",
532 		.data		= &sysctl_writes_strict,
533 		.maxlen		= sizeof(int),
534 		.mode		= 0644,
535 		.proc_handler	= proc_dointvec_minmax,
536 		.extra1		= &neg_one,
537 		.extra2		= &one,
538 	},
539 #endif
540 #ifdef CONFIG_LATENCYTOP
541 	{
542 		.procname	= "latencytop",
543 		.data		= &latencytop_enabled,
544 		.maxlen		= sizeof(int),
545 		.mode		= 0644,
546 		.proc_handler	= sysctl_latencytop,
547 	},
548 #endif
549 #ifdef CONFIG_BLK_DEV_INITRD
550 	{
551 		.procname	= "real-root-dev",
552 		.data		= &real_root_dev,
553 		.maxlen		= sizeof(int),
554 		.mode		= 0644,
555 		.proc_handler	= proc_dointvec,
556 	},
557 #endif
558 	{
559 		.procname	= "print-fatal-signals",
560 		.data		= &print_fatal_signals,
561 		.maxlen		= sizeof(int),
562 		.mode		= 0644,
563 		.proc_handler	= proc_dointvec,
564 	},
565 #ifdef CONFIG_SPARC
566 	{
567 		.procname	= "reboot-cmd",
568 		.data		= reboot_command,
569 		.maxlen		= 256,
570 		.mode		= 0644,
571 		.proc_handler	= proc_dostring,
572 	},
573 	{
574 		.procname	= "stop-a",
575 		.data		= &stop_a_enabled,
576 		.maxlen		= sizeof (int),
577 		.mode		= 0644,
578 		.proc_handler	= proc_dointvec,
579 	},
580 	{
581 		.procname	= "scons-poweroff",
582 		.data		= &scons_pwroff,
583 		.maxlen		= sizeof (int),
584 		.mode		= 0644,
585 		.proc_handler	= proc_dointvec,
586 	},
587 #endif
588 #ifdef CONFIG_SPARC64
589 	{
590 		.procname	= "tsb-ratio",
591 		.data		= &sysctl_tsb_ratio,
592 		.maxlen		= sizeof (int),
593 		.mode		= 0644,
594 		.proc_handler	= proc_dointvec,
595 	},
596 #endif
597 #ifdef __hppa__
598 	{
599 		.procname	= "soft-power",
600 		.data		= &pwrsw_enabled,
601 		.maxlen		= sizeof (int),
602 	 	.mode		= 0644,
603 		.proc_handler	= proc_dointvec,
604 	},
605 #endif
606 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
607 	{
608 		.procname	= "unaligned-trap",
609 		.data		= &unaligned_enabled,
610 		.maxlen		= sizeof (int),
611 		.mode		= 0644,
612 		.proc_handler	= proc_dointvec,
613 	},
614 #endif
615 	{
616 		.procname	= "ctrl-alt-del",
617 		.data		= &C_A_D,
618 		.maxlen		= sizeof(int),
619 		.mode		= 0644,
620 		.proc_handler	= proc_dointvec,
621 	},
622 #ifdef CONFIG_FUNCTION_TRACER
623 	{
624 		.procname	= "ftrace_enabled",
625 		.data		= &ftrace_enabled,
626 		.maxlen		= sizeof(int),
627 		.mode		= 0644,
628 		.proc_handler	= ftrace_enable_sysctl,
629 	},
630 #endif
631 #ifdef CONFIG_STACK_TRACER
632 	{
633 		.procname	= "stack_tracer_enabled",
634 		.data		= &stack_tracer_enabled,
635 		.maxlen		= sizeof(int),
636 		.mode		= 0644,
637 		.proc_handler	= stack_trace_sysctl,
638 	},
639 #endif
640 #ifdef CONFIG_TRACING
641 	{
642 		.procname	= "ftrace_dump_on_oops",
643 		.data		= &ftrace_dump_on_oops,
644 		.maxlen		= sizeof(int),
645 		.mode		= 0644,
646 		.proc_handler	= proc_dointvec,
647 	},
648 	{
649 		.procname	= "traceoff_on_warning",
650 		.data		= &__disable_trace_on_warning,
651 		.maxlen		= sizeof(__disable_trace_on_warning),
652 		.mode		= 0644,
653 		.proc_handler	= proc_dointvec,
654 	},
655 	{
656 		.procname	= "tracepoint_printk",
657 		.data		= &tracepoint_printk,
658 		.maxlen		= sizeof(tracepoint_printk),
659 		.mode		= 0644,
660 		.proc_handler	= tracepoint_printk_sysctl,
661 	},
662 #endif
663 #ifdef CONFIG_KEXEC_CORE
664 	{
665 		.procname	= "kexec_load_disabled",
666 		.data		= &kexec_load_disabled,
667 		.maxlen		= sizeof(int),
668 		.mode		= 0644,
669 		/* only handle a transition from default "0" to "1" */
670 		.proc_handler	= proc_dointvec_minmax,
671 		.extra1		= &one,
672 		.extra2		= &one,
673 	},
674 #endif
675 #ifdef CONFIG_MODULES
676 	{
677 		.procname	= "modprobe",
678 		.data		= &modprobe_path,
679 		.maxlen		= KMOD_PATH_LEN,
680 		.mode		= 0644,
681 		.proc_handler	= proc_dostring,
682 	},
683 	{
684 		.procname	= "modules_disabled",
685 		.data		= &modules_disabled,
686 		.maxlen		= sizeof(int),
687 		.mode		= 0644,
688 		/* only handle a transition from default "0" to "1" */
689 		.proc_handler	= proc_dointvec_minmax,
690 		.extra1		= &one,
691 		.extra2		= &one,
692 	},
693 #endif
694 #ifdef CONFIG_UEVENT_HELPER
695 	{
696 		.procname	= "hotplug",
697 		.data		= &uevent_helper,
698 		.maxlen		= UEVENT_HELPER_PATH_LEN,
699 		.mode		= 0644,
700 		.proc_handler	= proc_dostring,
701 	},
702 #endif
703 #ifdef CONFIG_CHR_DEV_SG
704 	{
705 		.procname	= "sg-big-buff",
706 		.data		= &sg_big_buff,
707 		.maxlen		= sizeof (int),
708 		.mode		= 0444,
709 		.proc_handler	= proc_dointvec,
710 	},
711 #endif
712 #ifdef CONFIG_BSD_PROCESS_ACCT
713 	{
714 		.procname	= "acct",
715 		.data		= &acct_parm,
716 		.maxlen		= 3*sizeof(int),
717 		.mode		= 0644,
718 		.proc_handler	= proc_dointvec,
719 	},
720 #endif
721 #ifdef CONFIG_MAGIC_SYSRQ
722 	{
723 		.procname	= "sysrq",
724 		.data		= &__sysrq_enabled,
725 		.maxlen		= sizeof (int),
726 		.mode		= 0644,
727 		.proc_handler	= sysrq_sysctl_handler,
728 	},
729 #endif
730 #ifdef CONFIG_PROC_SYSCTL
731 	{
732 		.procname	= "cad_pid",
733 		.data		= NULL,
734 		.maxlen		= sizeof (int),
735 		.mode		= 0600,
736 		.proc_handler	= proc_do_cad_pid,
737 	},
738 #endif
739 	{
740 		.procname	= "threads-max",
741 		.data		= NULL,
742 		.maxlen		= sizeof(int),
743 		.mode		= 0644,
744 		.proc_handler	= sysctl_max_threads,
745 	},
746 	{
747 		.procname	= "random",
748 		.mode		= 0555,
749 		.child		= random_table,
750 	},
751 	{
752 		.procname	= "usermodehelper",
753 		.mode		= 0555,
754 		.child		= usermodehelper_table,
755 	},
756 #ifdef CONFIG_FW_LOADER_USER_HELPER
757 	{
758 		.procname	= "firmware_config",
759 		.mode		= 0555,
760 		.child		= firmware_config_table,
761 	},
762 #endif
763 	{
764 		.procname	= "overflowuid",
765 		.data		= &overflowuid,
766 		.maxlen		= sizeof(int),
767 		.mode		= 0644,
768 		.proc_handler	= proc_dointvec_minmax,
769 		.extra1		= &minolduid,
770 		.extra2		= &maxolduid,
771 	},
772 	{
773 		.procname	= "overflowgid",
774 		.data		= &overflowgid,
775 		.maxlen		= sizeof(int),
776 		.mode		= 0644,
777 		.proc_handler	= proc_dointvec_minmax,
778 		.extra1		= &minolduid,
779 		.extra2		= &maxolduid,
780 	},
781 #ifdef CONFIG_S390
782 #ifdef CONFIG_MATHEMU
783 	{
784 		.procname	= "ieee_emulation_warnings",
785 		.data		= &sysctl_ieee_emulation_warnings,
786 		.maxlen		= sizeof(int),
787 		.mode		= 0644,
788 		.proc_handler	= proc_dointvec,
789 	},
790 #endif
791 	{
792 		.procname	= "userprocess_debug",
793 		.data		= &show_unhandled_signals,
794 		.maxlen		= sizeof(int),
795 		.mode		= 0644,
796 		.proc_handler	= proc_dointvec,
797 	},
798 #endif
799 	{
800 		.procname	= "pid_max",
801 		.data		= &pid_max,
802 		.maxlen		= sizeof (int),
803 		.mode		= 0644,
804 		.proc_handler	= proc_dointvec_minmax,
805 		.extra1		= &pid_max_min,
806 		.extra2		= &pid_max_max,
807 	},
808 	{
809 		.procname	= "panic_on_oops",
810 		.data		= &panic_on_oops,
811 		.maxlen		= sizeof(int),
812 		.mode		= 0644,
813 		.proc_handler	= proc_dointvec,
814 	},
815 	{
816 		.procname	= "panic_print",
817 		.data		= &panic_print,
818 		.maxlen		= sizeof(unsigned long),
819 		.mode		= 0644,
820 		.proc_handler	= proc_doulongvec_minmax,
821 	},
822 #if defined CONFIG_PRINTK
823 	{
824 		.procname	= "printk",
825 		.data		= &console_loglevel,
826 		.maxlen		= 4*sizeof(int),
827 		.mode		= 0644,
828 		.proc_handler	= proc_dointvec,
829 	},
830 	{
831 		.procname	= "printk_ratelimit",
832 		.data		= &printk_ratelimit_state.interval,
833 		.maxlen		= sizeof(int),
834 		.mode		= 0644,
835 		.proc_handler	= proc_dointvec_jiffies,
836 	},
837 	{
838 		.procname	= "printk_ratelimit_burst",
839 		.data		= &printk_ratelimit_state.burst,
840 		.maxlen		= sizeof(int),
841 		.mode		= 0644,
842 		.proc_handler	= proc_dointvec,
843 	},
844 	{
845 		.procname	= "printk_delay",
846 		.data		= &printk_delay_msec,
847 		.maxlen		= sizeof(int),
848 		.mode		= 0644,
849 		.proc_handler	= proc_dointvec_minmax,
850 		.extra1		= &zero,
851 		.extra2		= &ten_thousand,
852 	},
853 	{
854 		.procname	= "printk_devkmsg",
855 		.data		= devkmsg_log_str,
856 		.maxlen		= DEVKMSG_STR_MAX_SIZE,
857 		.mode		= 0644,
858 		.proc_handler	= devkmsg_sysctl_set_loglvl,
859 	},
860 	{
861 		.procname	= "dmesg_restrict",
862 		.data		= &dmesg_restrict,
863 		.maxlen		= sizeof(int),
864 		.mode		= 0644,
865 		.proc_handler	= proc_dointvec_minmax_sysadmin,
866 		.extra1		= &zero,
867 		.extra2		= &one,
868 	},
869 	{
870 		.procname	= "kptr_restrict",
871 		.data		= &kptr_restrict,
872 		.maxlen		= sizeof(int),
873 		.mode		= 0644,
874 		.proc_handler	= proc_dointvec_minmax_sysadmin,
875 		.extra1		= &zero,
876 		.extra2		= &two,
877 	},
878 #endif
879 	{
880 		.procname	= "ngroups_max",
881 		.data		= &ngroups_max,
882 		.maxlen		= sizeof (int),
883 		.mode		= 0444,
884 		.proc_handler	= proc_dointvec,
885 	},
886 	{
887 		.procname	= "cap_last_cap",
888 		.data		= (void *)&cap_last_cap,
889 		.maxlen		= sizeof(int),
890 		.mode		= 0444,
891 		.proc_handler	= proc_dointvec,
892 	},
893 #if defined(CONFIG_LOCKUP_DETECTOR)
894 	{
895 		.procname       = "watchdog",
896 		.data		= &watchdog_user_enabled,
897 		.maxlen		= sizeof(int),
898 		.mode		= 0644,
899 		.proc_handler   = proc_watchdog,
900 		.extra1		= &zero,
901 		.extra2		= &one,
902 	},
903 	{
904 		.procname	= "watchdog_thresh",
905 		.data		= &watchdog_thresh,
906 		.maxlen		= sizeof(int),
907 		.mode		= 0644,
908 		.proc_handler	= proc_watchdog_thresh,
909 		.extra1		= &zero,
910 		.extra2		= &sixty,
911 	},
912 	{
913 		.procname       = "nmi_watchdog",
914 		.data		= &nmi_watchdog_user_enabled,
915 		.maxlen		= sizeof(int),
916 		.mode		= NMI_WATCHDOG_SYSCTL_PERM,
917 		.proc_handler   = proc_nmi_watchdog,
918 		.extra1		= &zero,
919 		.extra2		= &one,
920 	},
921 	{
922 		.procname	= "watchdog_cpumask",
923 		.data		= &watchdog_cpumask_bits,
924 		.maxlen		= NR_CPUS,
925 		.mode		= 0644,
926 		.proc_handler	= proc_watchdog_cpumask,
927 	},
928 #ifdef CONFIG_SOFTLOCKUP_DETECTOR
929 	{
930 		.procname       = "soft_watchdog",
931 		.data		= &soft_watchdog_user_enabled,
932 		.maxlen		= sizeof(int),
933 		.mode		= 0644,
934 		.proc_handler   = proc_soft_watchdog,
935 		.extra1		= &zero,
936 		.extra2		= &one,
937 	},
938 	{
939 		.procname	= "softlockup_panic",
940 		.data		= &softlockup_panic,
941 		.maxlen		= sizeof(int),
942 		.mode		= 0644,
943 		.proc_handler	= proc_dointvec_minmax,
944 		.extra1		= &zero,
945 		.extra2		= &one,
946 	},
947 #ifdef CONFIG_SMP
948 	{
949 		.procname	= "softlockup_all_cpu_backtrace",
950 		.data		= &sysctl_softlockup_all_cpu_backtrace,
951 		.maxlen		= sizeof(int),
952 		.mode		= 0644,
953 		.proc_handler	= proc_dointvec_minmax,
954 		.extra1		= &zero,
955 		.extra2		= &one,
956 	},
957 #endif /* CONFIG_SMP */
958 #endif
959 #ifdef CONFIG_HARDLOCKUP_DETECTOR
960 	{
961 		.procname	= "hardlockup_panic",
962 		.data		= &hardlockup_panic,
963 		.maxlen		= sizeof(int),
964 		.mode		= 0644,
965 		.proc_handler	= proc_dointvec_minmax,
966 		.extra1		= &zero,
967 		.extra2		= &one,
968 	},
969 #ifdef CONFIG_SMP
970 	{
971 		.procname	= "hardlockup_all_cpu_backtrace",
972 		.data		= &sysctl_hardlockup_all_cpu_backtrace,
973 		.maxlen		= sizeof(int),
974 		.mode		= 0644,
975 		.proc_handler	= proc_dointvec_minmax,
976 		.extra1		= &zero,
977 		.extra2		= &one,
978 	},
979 #endif /* CONFIG_SMP */
980 #endif
981 #endif
982 
983 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
984 	{
985 		.procname       = "unknown_nmi_panic",
986 		.data           = &unknown_nmi_panic,
987 		.maxlen         = sizeof (int),
988 		.mode           = 0644,
989 		.proc_handler   = proc_dointvec,
990 	},
991 #endif
992 #if defined(CONFIG_X86)
993 	{
994 		.procname	= "panic_on_unrecovered_nmi",
995 		.data		= &panic_on_unrecovered_nmi,
996 		.maxlen		= sizeof(int),
997 		.mode		= 0644,
998 		.proc_handler	= proc_dointvec,
999 	},
1000 	{
1001 		.procname	= "panic_on_io_nmi",
1002 		.data		= &panic_on_io_nmi,
1003 		.maxlen		= sizeof(int),
1004 		.mode		= 0644,
1005 		.proc_handler	= proc_dointvec,
1006 	},
1007 #ifdef CONFIG_DEBUG_STACKOVERFLOW
1008 	{
1009 		.procname	= "panic_on_stackoverflow",
1010 		.data		= &sysctl_panic_on_stackoverflow,
1011 		.maxlen		= sizeof(int),
1012 		.mode		= 0644,
1013 		.proc_handler	= proc_dointvec,
1014 	},
1015 #endif
1016 	{
1017 		.procname	= "bootloader_type",
1018 		.data		= &bootloader_type,
1019 		.maxlen		= sizeof (int),
1020 		.mode		= 0444,
1021 		.proc_handler	= proc_dointvec,
1022 	},
1023 	{
1024 		.procname	= "bootloader_version",
1025 		.data		= &bootloader_version,
1026 		.maxlen		= sizeof (int),
1027 		.mode		= 0444,
1028 		.proc_handler	= proc_dointvec,
1029 	},
1030 	{
1031 		.procname	= "io_delay_type",
1032 		.data		= &io_delay_type,
1033 		.maxlen		= sizeof(int),
1034 		.mode		= 0644,
1035 		.proc_handler	= proc_dointvec,
1036 	},
1037 #endif
1038 #if defined(CONFIG_MMU)
1039 	{
1040 		.procname	= "randomize_va_space",
1041 		.data		= &randomize_va_space,
1042 		.maxlen		= sizeof(int),
1043 		.mode		= 0644,
1044 		.proc_handler	= proc_dointvec,
1045 	},
1046 #endif
1047 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1048 	{
1049 		.procname	= "spin_retry",
1050 		.data		= &spin_retry,
1051 		.maxlen		= sizeof (int),
1052 		.mode		= 0644,
1053 		.proc_handler	= proc_dointvec,
1054 	},
1055 #endif
1056 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1057 	{
1058 		.procname	= "acpi_video_flags",
1059 		.data		= &acpi_realmode_flags,
1060 		.maxlen		= sizeof (unsigned long),
1061 		.mode		= 0644,
1062 		.proc_handler	= proc_doulongvec_minmax,
1063 	},
1064 #endif
1065 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1066 	{
1067 		.procname	= "ignore-unaligned-usertrap",
1068 		.data		= &no_unaligned_warning,
1069 		.maxlen		= sizeof (int),
1070 	 	.mode		= 0644,
1071 		.proc_handler	= proc_dointvec,
1072 	},
1073 #endif
1074 #ifdef CONFIG_IA64
1075 	{
1076 		.procname	= "unaligned-dump-stack",
1077 		.data		= &unaligned_dump_stack,
1078 		.maxlen		= sizeof (int),
1079 		.mode		= 0644,
1080 		.proc_handler	= proc_dointvec,
1081 	},
1082 #endif
1083 #ifdef CONFIG_DETECT_HUNG_TASK
1084 	{
1085 		.procname	= "hung_task_panic",
1086 		.data		= &sysctl_hung_task_panic,
1087 		.maxlen		= sizeof(int),
1088 		.mode		= 0644,
1089 		.proc_handler	= proc_dointvec_minmax,
1090 		.extra1		= &zero,
1091 		.extra2		= &one,
1092 	},
1093 	{
1094 		.procname	= "hung_task_check_count",
1095 		.data		= &sysctl_hung_task_check_count,
1096 		.maxlen		= sizeof(int),
1097 		.mode		= 0644,
1098 		.proc_handler	= proc_dointvec_minmax,
1099 		.extra1		= &zero,
1100 	},
1101 	{
1102 		.procname	= "hung_task_timeout_secs",
1103 		.data		= &sysctl_hung_task_timeout_secs,
1104 		.maxlen		= sizeof(unsigned long),
1105 		.mode		= 0644,
1106 		.proc_handler	= proc_dohung_task_timeout_secs,
1107 		.extra2		= &hung_task_timeout_max,
1108 	},
1109 	{
1110 		.procname	= "hung_task_check_interval_secs",
1111 		.data		= &sysctl_hung_task_check_interval_secs,
1112 		.maxlen		= sizeof(unsigned long),
1113 		.mode		= 0644,
1114 		.proc_handler	= proc_dohung_task_timeout_secs,
1115 		.extra2		= &hung_task_timeout_max,
1116 	},
1117 	{
1118 		.procname	= "hung_task_warnings",
1119 		.data		= &sysctl_hung_task_warnings,
1120 		.maxlen		= sizeof(int),
1121 		.mode		= 0644,
1122 		.proc_handler	= proc_dointvec_minmax,
1123 		.extra1		= &neg_one,
1124 	},
1125 #endif
1126 #ifdef CONFIG_RT_MUTEXES
1127 	{
1128 		.procname	= "max_lock_depth",
1129 		.data		= &max_lock_depth,
1130 		.maxlen		= sizeof(int),
1131 		.mode		= 0644,
1132 		.proc_handler	= proc_dointvec,
1133 	},
1134 #endif
1135 	{
1136 		.procname	= "poweroff_cmd",
1137 		.data		= &poweroff_cmd,
1138 		.maxlen		= POWEROFF_CMD_PATH_LEN,
1139 		.mode		= 0644,
1140 		.proc_handler	= proc_dostring,
1141 	},
1142 #ifdef CONFIG_KEYS
1143 	{
1144 		.procname	= "keys",
1145 		.mode		= 0555,
1146 		.child		= key_sysctls,
1147 	},
1148 #endif
1149 #ifdef CONFIG_PERF_EVENTS
1150 	/*
1151 	 * User-space scripts rely on the existence of this file
1152 	 * as a feature check for perf_events being enabled.
1153 	 *
1154 	 * So it's an ABI, do not remove!
1155 	 */
1156 	{
1157 		.procname	= "perf_event_paranoid",
1158 		.data		= &sysctl_perf_event_paranoid,
1159 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
1160 		.mode		= 0644,
1161 		.proc_handler	= proc_dointvec,
1162 	},
1163 	{
1164 		.procname	= "perf_event_mlock_kb",
1165 		.data		= &sysctl_perf_event_mlock,
1166 		.maxlen		= sizeof(sysctl_perf_event_mlock),
1167 		.mode		= 0644,
1168 		.proc_handler	= proc_dointvec,
1169 	},
1170 	{
1171 		.procname	= "perf_event_max_sample_rate",
1172 		.data		= &sysctl_perf_event_sample_rate,
1173 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
1174 		.mode		= 0644,
1175 		.proc_handler	= perf_proc_update_handler,
1176 		.extra1		= &one,
1177 	},
1178 	{
1179 		.procname	= "perf_cpu_time_max_percent",
1180 		.data		= &sysctl_perf_cpu_time_max_percent,
1181 		.maxlen		= sizeof(sysctl_perf_cpu_time_max_percent),
1182 		.mode		= 0644,
1183 		.proc_handler	= perf_cpu_time_max_percent_handler,
1184 		.extra1		= &zero,
1185 		.extra2		= &one_hundred,
1186 	},
1187 	{
1188 		.procname	= "perf_event_max_stack",
1189 		.data		= &sysctl_perf_event_max_stack,
1190 		.maxlen		= sizeof(sysctl_perf_event_max_stack),
1191 		.mode		= 0644,
1192 		.proc_handler	= perf_event_max_stack_handler,
1193 		.extra1		= &zero,
1194 		.extra2		= &six_hundred_forty_kb,
1195 	},
1196 	{
1197 		.procname	= "perf_event_max_contexts_per_stack",
1198 		.data		= &sysctl_perf_event_max_contexts_per_stack,
1199 		.maxlen		= sizeof(sysctl_perf_event_max_contexts_per_stack),
1200 		.mode		= 0644,
1201 		.proc_handler	= perf_event_max_stack_handler,
1202 		.extra1		= &zero,
1203 		.extra2		= &one_thousand,
1204 	},
1205 #endif
1206 	{
1207 		.procname	= "panic_on_warn",
1208 		.data		= &panic_on_warn,
1209 		.maxlen		= sizeof(int),
1210 		.mode		= 0644,
1211 		.proc_handler	= proc_dointvec_minmax,
1212 		.extra1		= &zero,
1213 		.extra2		= &one,
1214 	},
1215 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1216 	{
1217 		.procname	= "timer_migration",
1218 		.data		= &sysctl_timer_migration,
1219 		.maxlen		= sizeof(unsigned int),
1220 		.mode		= 0644,
1221 		.proc_handler	= timer_migration_handler,
1222 		.extra1		= &zero,
1223 		.extra2		= &one,
1224 	},
1225 #endif
1226 #ifdef CONFIG_BPF_SYSCALL
1227 	{
1228 		.procname	= "unprivileged_bpf_disabled",
1229 		.data		= &sysctl_unprivileged_bpf_disabled,
1230 		.maxlen		= sizeof(sysctl_unprivileged_bpf_disabled),
1231 		.mode		= 0644,
1232 		/* only handle a transition from default "0" to "1" */
1233 		.proc_handler	= proc_dointvec_minmax,
1234 		.extra1		= &one,
1235 		.extra2		= &one,
1236 	},
1237 	{
1238 		.procname	= "bpf_stats_enabled",
1239 		.data		= &sysctl_bpf_stats_enabled,
1240 		.maxlen		= sizeof(sysctl_bpf_stats_enabled),
1241 		.mode		= 0644,
1242 		.proc_handler	= proc_dointvec_minmax_bpf_stats,
1243 		.extra1		= &zero,
1244 		.extra2		= &one,
1245 	},
1246 #endif
1247 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1248 	{
1249 		.procname	= "panic_on_rcu_stall",
1250 		.data		= &sysctl_panic_on_rcu_stall,
1251 		.maxlen		= sizeof(sysctl_panic_on_rcu_stall),
1252 		.mode		= 0644,
1253 		.proc_handler	= proc_dointvec_minmax,
1254 		.extra1		= &zero,
1255 		.extra2		= &one,
1256 	},
1257 #endif
1258 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
1259 	{
1260 		.procname	= "stack_erasing",
1261 		.data		= NULL,
1262 		.maxlen		= sizeof(int),
1263 		.mode		= 0600,
1264 		.proc_handler	= stack_erasing_sysctl,
1265 		.extra1		= &zero,
1266 		.extra2		= &one,
1267 	},
1268 #endif
1269 	{ }
1270 };
1271 
1272 static struct ctl_table vm_table[] = {
1273 	{
1274 		.procname	= "overcommit_memory",
1275 		.data		= &sysctl_overcommit_memory,
1276 		.maxlen		= sizeof(sysctl_overcommit_memory),
1277 		.mode		= 0644,
1278 		.proc_handler	= proc_dointvec_minmax,
1279 		.extra1		= &zero,
1280 		.extra2		= &two,
1281 	},
1282 	{
1283 		.procname	= "panic_on_oom",
1284 		.data		= &sysctl_panic_on_oom,
1285 		.maxlen		= sizeof(sysctl_panic_on_oom),
1286 		.mode		= 0644,
1287 		.proc_handler	= proc_dointvec_minmax,
1288 		.extra1		= &zero,
1289 		.extra2		= &two,
1290 	},
1291 	{
1292 		.procname	= "oom_kill_allocating_task",
1293 		.data		= &sysctl_oom_kill_allocating_task,
1294 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1295 		.mode		= 0644,
1296 		.proc_handler	= proc_dointvec,
1297 	},
1298 	{
1299 		.procname	= "oom_dump_tasks",
1300 		.data		= &sysctl_oom_dump_tasks,
1301 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1302 		.mode		= 0644,
1303 		.proc_handler	= proc_dointvec,
1304 	},
1305 	{
1306 		.procname	= "overcommit_ratio",
1307 		.data		= &sysctl_overcommit_ratio,
1308 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1309 		.mode		= 0644,
1310 		.proc_handler	= overcommit_ratio_handler,
1311 	},
1312 	{
1313 		.procname	= "overcommit_kbytes",
1314 		.data		= &sysctl_overcommit_kbytes,
1315 		.maxlen		= sizeof(sysctl_overcommit_kbytes),
1316 		.mode		= 0644,
1317 		.proc_handler	= overcommit_kbytes_handler,
1318 	},
1319 	{
1320 		.procname	= "page-cluster",
1321 		.data		= &page_cluster,
1322 		.maxlen		= sizeof(int),
1323 		.mode		= 0644,
1324 		.proc_handler	= proc_dointvec_minmax,
1325 		.extra1		= &zero,
1326 	},
1327 	{
1328 		.procname	= "dirty_background_ratio",
1329 		.data		= &dirty_background_ratio,
1330 		.maxlen		= sizeof(dirty_background_ratio),
1331 		.mode		= 0644,
1332 		.proc_handler	= dirty_background_ratio_handler,
1333 		.extra1		= &zero,
1334 		.extra2		= &one_hundred,
1335 	},
1336 	{
1337 		.procname	= "dirty_background_bytes",
1338 		.data		= &dirty_background_bytes,
1339 		.maxlen		= sizeof(dirty_background_bytes),
1340 		.mode		= 0644,
1341 		.proc_handler	= dirty_background_bytes_handler,
1342 		.extra1		= &one_ul,
1343 	},
1344 	{
1345 		.procname	= "dirty_ratio",
1346 		.data		= &vm_dirty_ratio,
1347 		.maxlen		= sizeof(vm_dirty_ratio),
1348 		.mode		= 0644,
1349 		.proc_handler	= dirty_ratio_handler,
1350 		.extra1		= &zero,
1351 		.extra2		= &one_hundred,
1352 	},
1353 	{
1354 		.procname	= "dirty_bytes",
1355 		.data		= &vm_dirty_bytes,
1356 		.maxlen		= sizeof(vm_dirty_bytes),
1357 		.mode		= 0644,
1358 		.proc_handler	= dirty_bytes_handler,
1359 		.extra1		= &dirty_bytes_min,
1360 	},
1361 	{
1362 		.procname	= "dirty_writeback_centisecs",
1363 		.data		= &dirty_writeback_interval,
1364 		.maxlen		= sizeof(dirty_writeback_interval),
1365 		.mode		= 0644,
1366 		.proc_handler	= dirty_writeback_centisecs_handler,
1367 	},
1368 	{
1369 		.procname	= "dirty_expire_centisecs",
1370 		.data		= &dirty_expire_interval,
1371 		.maxlen		= sizeof(dirty_expire_interval),
1372 		.mode		= 0644,
1373 		.proc_handler	= proc_dointvec_minmax,
1374 		.extra1		= &zero,
1375 	},
1376 	{
1377 		.procname	= "dirtytime_expire_seconds",
1378 		.data		= &dirtytime_expire_interval,
1379 		.maxlen		= sizeof(dirtytime_expire_interval),
1380 		.mode		= 0644,
1381 		.proc_handler	= dirtytime_interval_handler,
1382 		.extra1		= &zero,
1383 	},
1384 	{
1385 		.procname	= "swappiness",
1386 		.data		= &vm_swappiness,
1387 		.maxlen		= sizeof(vm_swappiness),
1388 		.mode		= 0644,
1389 		.proc_handler	= proc_dointvec_minmax,
1390 		.extra1		= &zero,
1391 		.extra2		= &one_hundred,
1392 	},
1393 #ifdef CONFIG_HUGETLB_PAGE
1394 	{
1395 		.procname	= "nr_hugepages",
1396 		.data		= NULL,
1397 		.maxlen		= sizeof(unsigned long),
1398 		.mode		= 0644,
1399 		.proc_handler	= hugetlb_sysctl_handler,
1400 	},
1401 #ifdef CONFIG_NUMA
1402 	{
1403 		.procname       = "nr_hugepages_mempolicy",
1404 		.data           = NULL,
1405 		.maxlen         = sizeof(unsigned long),
1406 		.mode           = 0644,
1407 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1408 	},
1409 	{
1410 		.procname		= "numa_stat",
1411 		.data			= &sysctl_vm_numa_stat,
1412 		.maxlen			= sizeof(int),
1413 		.mode			= 0644,
1414 		.proc_handler	= sysctl_vm_numa_stat_handler,
1415 		.extra1			= &zero,
1416 		.extra2			= &one,
1417 	},
1418 #endif
1419 	 {
1420 		.procname	= "hugetlb_shm_group",
1421 		.data		= &sysctl_hugetlb_shm_group,
1422 		.maxlen		= sizeof(gid_t),
1423 		.mode		= 0644,
1424 		.proc_handler	= proc_dointvec,
1425 	 },
1426 	{
1427 		.procname	= "nr_overcommit_hugepages",
1428 		.data		= NULL,
1429 		.maxlen		= sizeof(unsigned long),
1430 		.mode		= 0644,
1431 		.proc_handler	= hugetlb_overcommit_handler,
1432 	},
1433 #endif
1434 	{
1435 		.procname	= "lowmem_reserve_ratio",
1436 		.data		= &sysctl_lowmem_reserve_ratio,
1437 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1438 		.mode		= 0644,
1439 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1440 	},
1441 	{
1442 		.procname	= "drop_caches",
1443 		.data		= &sysctl_drop_caches,
1444 		.maxlen		= sizeof(int),
1445 		.mode		= 0644,
1446 		.proc_handler	= drop_caches_sysctl_handler,
1447 		.extra1		= &one,
1448 		.extra2		= &four,
1449 	},
1450 #ifdef CONFIG_COMPACTION
1451 	{
1452 		.procname	= "compact_memory",
1453 		.data		= &sysctl_compact_memory,
1454 		.maxlen		= sizeof(int),
1455 		.mode		= 0200,
1456 		.proc_handler	= sysctl_compaction_handler,
1457 	},
1458 	{
1459 		.procname	= "extfrag_threshold",
1460 		.data		= &sysctl_extfrag_threshold,
1461 		.maxlen		= sizeof(int),
1462 		.mode		= 0644,
1463 		.proc_handler	= proc_dointvec_minmax,
1464 		.extra1		= &min_extfrag_threshold,
1465 		.extra2		= &max_extfrag_threshold,
1466 	},
1467 	{
1468 		.procname	= "compact_unevictable_allowed",
1469 		.data		= &sysctl_compact_unevictable_allowed,
1470 		.maxlen		= sizeof(int),
1471 		.mode		= 0644,
1472 		.proc_handler	= proc_dointvec,
1473 		.extra1		= &zero,
1474 		.extra2		= &one,
1475 	},
1476 
1477 #endif /* CONFIG_COMPACTION */
1478 	{
1479 		.procname	= "min_free_kbytes",
1480 		.data		= &min_free_kbytes,
1481 		.maxlen		= sizeof(min_free_kbytes),
1482 		.mode		= 0644,
1483 		.proc_handler	= min_free_kbytes_sysctl_handler,
1484 		.extra1		= &zero,
1485 	},
1486 	{
1487 		.procname	= "watermark_boost_factor",
1488 		.data		= &watermark_boost_factor,
1489 		.maxlen		= sizeof(watermark_boost_factor),
1490 		.mode		= 0644,
1491 		.proc_handler	= watermark_boost_factor_sysctl_handler,
1492 		.extra1		= &zero,
1493 	},
1494 	{
1495 		.procname	= "watermark_scale_factor",
1496 		.data		= &watermark_scale_factor,
1497 		.maxlen		= sizeof(watermark_scale_factor),
1498 		.mode		= 0644,
1499 		.proc_handler	= watermark_scale_factor_sysctl_handler,
1500 		.extra1		= &one,
1501 		.extra2		= &one_thousand,
1502 	},
1503 	{
1504 		.procname	= "percpu_pagelist_fraction",
1505 		.data		= &percpu_pagelist_fraction,
1506 		.maxlen		= sizeof(percpu_pagelist_fraction),
1507 		.mode		= 0644,
1508 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1509 		.extra1		= &zero,
1510 	},
1511 #ifdef CONFIG_MMU
1512 	{
1513 		.procname	= "max_map_count",
1514 		.data		= &sysctl_max_map_count,
1515 		.maxlen		= sizeof(sysctl_max_map_count),
1516 		.mode		= 0644,
1517 		.proc_handler	= proc_dointvec_minmax,
1518 		.extra1		= &zero,
1519 	},
1520 #else
1521 	{
1522 		.procname	= "nr_trim_pages",
1523 		.data		= &sysctl_nr_trim_pages,
1524 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1525 		.mode		= 0644,
1526 		.proc_handler	= proc_dointvec_minmax,
1527 		.extra1		= &zero,
1528 	},
1529 #endif
1530 	{
1531 		.procname	= "laptop_mode",
1532 		.data		= &laptop_mode,
1533 		.maxlen		= sizeof(laptop_mode),
1534 		.mode		= 0644,
1535 		.proc_handler	= proc_dointvec_jiffies,
1536 	},
1537 	{
1538 		.procname	= "block_dump",
1539 		.data		= &block_dump,
1540 		.maxlen		= sizeof(block_dump),
1541 		.mode		= 0644,
1542 		.proc_handler	= proc_dointvec,
1543 		.extra1		= &zero,
1544 	},
1545 	{
1546 		.procname	= "vfs_cache_pressure",
1547 		.data		= &sysctl_vfs_cache_pressure,
1548 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1549 		.mode		= 0644,
1550 		.proc_handler	= proc_dointvec,
1551 		.extra1		= &zero,
1552 	},
1553 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1554 	{
1555 		.procname	= "legacy_va_layout",
1556 		.data		= &sysctl_legacy_va_layout,
1557 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1558 		.mode		= 0644,
1559 		.proc_handler	= proc_dointvec,
1560 		.extra1		= &zero,
1561 	},
1562 #endif
1563 #ifdef CONFIG_NUMA
1564 	{
1565 		.procname	= "zone_reclaim_mode",
1566 		.data		= &node_reclaim_mode,
1567 		.maxlen		= sizeof(node_reclaim_mode),
1568 		.mode		= 0644,
1569 		.proc_handler	= proc_dointvec,
1570 		.extra1		= &zero,
1571 	},
1572 	{
1573 		.procname	= "min_unmapped_ratio",
1574 		.data		= &sysctl_min_unmapped_ratio,
1575 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1576 		.mode		= 0644,
1577 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1578 		.extra1		= &zero,
1579 		.extra2		= &one_hundred,
1580 	},
1581 	{
1582 		.procname	= "min_slab_ratio",
1583 		.data		= &sysctl_min_slab_ratio,
1584 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1585 		.mode		= 0644,
1586 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1587 		.extra1		= &zero,
1588 		.extra2		= &one_hundred,
1589 	},
1590 #endif
1591 #ifdef CONFIG_SMP
1592 	{
1593 		.procname	= "stat_interval",
1594 		.data		= &sysctl_stat_interval,
1595 		.maxlen		= sizeof(sysctl_stat_interval),
1596 		.mode		= 0644,
1597 		.proc_handler	= proc_dointvec_jiffies,
1598 	},
1599 	{
1600 		.procname	= "stat_refresh",
1601 		.data		= NULL,
1602 		.maxlen		= 0,
1603 		.mode		= 0600,
1604 		.proc_handler	= vmstat_refresh,
1605 	},
1606 #endif
1607 #ifdef CONFIG_MMU
1608 	{
1609 		.procname	= "mmap_min_addr",
1610 		.data		= &dac_mmap_min_addr,
1611 		.maxlen		= sizeof(unsigned long),
1612 		.mode		= 0644,
1613 		.proc_handler	= mmap_min_addr_handler,
1614 	},
1615 #endif
1616 #ifdef CONFIG_NUMA
1617 	{
1618 		.procname	= "numa_zonelist_order",
1619 		.data		= &numa_zonelist_order,
1620 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1621 		.mode		= 0644,
1622 		.proc_handler	= numa_zonelist_order_handler,
1623 	},
1624 #endif
1625 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1626    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1627 	{
1628 		.procname	= "vdso_enabled",
1629 #ifdef CONFIG_X86_32
1630 		.data		= &vdso32_enabled,
1631 		.maxlen		= sizeof(vdso32_enabled),
1632 #else
1633 		.data		= &vdso_enabled,
1634 		.maxlen		= sizeof(vdso_enabled),
1635 #endif
1636 		.mode		= 0644,
1637 		.proc_handler	= proc_dointvec,
1638 		.extra1		= &zero,
1639 	},
1640 #endif
1641 #ifdef CONFIG_HIGHMEM
1642 	{
1643 		.procname	= "highmem_is_dirtyable",
1644 		.data		= &vm_highmem_is_dirtyable,
1645 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1646 		.mode		= 0644,
1647 		.proc_handler	= proc_dointvec_minmax,
1648 		.extra1		= &zero,
1649 		.extra2		= &one,
1650 	},
1651 #endif
1652 #ifdef CONFIG_MEMORY_FAILURE
1653 	{
1654 		.procname	= "memory_failure_early_kill",
1655 		.data		= &sysctl_memory_failure_early_kill,
1656 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1657 		.mode		= 0644,
1658 		.proc_handler	= proc_dointvec_minmax,
1659 		.extra1		= &zero,
1660 		.extra2		= &one,
1661 	},
1662 	{
1663 		.procname	= "memory_failure_recovery",
1664 		.data		= &sysctl_memory_failure_recovery,
1665 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1666 		.mode		= 0644,
1667 		.proc_handler	= proc_dointvec_minmax,
1668 		.extra1		= &zero,
1669 		.extra2		= &one,
1670 	},
1671 #endif
1672 	{
1673 		.procname	= "user_reserve_kbytes",
1674 		.data		= &sysctl_user_reserve_kbytes,
1675 		.maxlen		= sizeof(sysctl_user_reserve_kbytes),
1676 		.mode		= 0644,
1677 		.proc_handler	= proc_doulongvec_minmax,
1678 	},
1679 	{
1680 		.procname	= "admin_reserve_kbytes",
1681 		.data		= &sysctl_admin_reserve_kbytes,
1682 		.maxlen		= sizeof(sysctl_admin_reserve_kbytes),
1683 		.mode		= 0644,
1684 		.proc_handler	= proc_doulongvec_minmax,
1685 	},
1686 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1687 	{
1688 		.procname	= "mmap_rnd_bits",
1689 		.data		= &mmap_rnd_bits,
1690 		.maxlen		= sizeof(mmap_rnd_bits),
1691 		.mode		= 0600,
1692 		.proc_handler	= proc_dointvec_minmax,
1693 		.extra1		= (void *)&mmap_rnd_bits_min,
1694 		.extra2		= (void *)&mmap_rnd_bits_max,
1695 	},
1696 #endif
1697 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1698 	{
1699 		.procname	= "mmap_rnd_compat_bits",
1700 		.data		= &mmap_rnd_compat_bits,
1701 		.maxlen		= sizeof(mmap_rnd_compat_bits),
1702 		.mode		= 0600,
1703 		.proc_handler	= proc_dointvec_minmax,
1704 		.extra1		= (void *)&mmap_rnd_compat_bits_min,
1705 		.extra2		= (void *)&mmap_rnd_compat_bits_max,
1706 	},
1707 #endif
1708 	{ }
1709 };
1710 
1711 static struct ctl_table fs_table[] = {
1712 	{
1713 		.procname	= "inode-nr",
1714 		.data		= &inodes_stat,
1715 		.maxlen		= 2*sizeof(long),
1716 		.mode		= 0444,
1717 		.proc_handler	= proc_nr_inodes,
1718 	},
1719 	{
1720 		.procname	= "inode-state",
1721 		.data		= &inodes_stat,
1722 		.maxlen		= 7*sizeof(long),
1723 		.mode		= 0444,
1724 		.proc_handler	= proc_nr_inodes,
1725 	},
1726 	{
1727 		.procname	= "file-nr",
1728 		.data		= &files_stat,
1729 		.maxlen		= sizeof(files_stat),
1730 		.mode		= 0444,
1731 		.proc_handler	= proc_nr_files,
1732 	},
1733 	{
1734 		.procname	= "file-max",
1735 		.data		= &files_stat.max_files,
1736 		.maxlen		= sizeof(files_stat.max_files),
1737 		.mode		= 0644,
1738 		.proc_handler	= proc_doulongvec_minmax,
1739 	},
1740 	{
1741 		.procname	= "nr_open",
1742 		.data		= &sysctl_nr_open,
1743 		.maxlen		= sizeof(unsigned int),
1744 		.mode		= 0644,
1745 		.proc_handler	= proc_dointvec_minmax,
1746 		.extra1		= &sysctl_nr_open_min,
1747 		.extra2		= &sysctl_nr_open_max,
1748 	},
1749 	{
1750 		.procname	= "dentry-state",
1751 		.data		= &dentry_stat,
1752 		.maxlen		= 6*sizeof(long),
1753 		.mode		= 0444,
1754 		.proc_handler	= proc_nr_dentry,
1755 	},
1756 	{
1757 		.procname	= "overflowuid",
1758 		.data		= &fs_overflowuid,
1759 		.maxlen		= sizeof(int),
1760 		.mode		= 0644,
1761 		.proc_handler	= proc_dointvec_minmax,
1762 		.extra1		= &minolduid,
1763 		.extra2		= &maxolduid,
1764 	},
1765 	{
1766 		.procname	= "overflowgid",
1767 		.data		= &fs_overflowgid,
1768 		.maxlen		= sizeof(int),
1769 		.mode		= 0644,
1770 		.proc_handler	= proc_dointvec_minmax,
1771 		.extra1		= &minolduid,
1772 		.extra2		= &maxolduid,
1773 	},
1774 #ifdef CONFIG_FILE_LOCKING
1775 	{
1776 		.procname	= "leases-enable",
1777 		.data		= &leases_enable,
1778 		.maxlen		= sizeof(int),
1779 		.mode		= 0644,
1780 		.proc_handler	= proc_dointvec,
1781 	},
1782 #endif
1783 #ifdef CONFIG_DNOTIFY
1784 	{
1785 		.procname	= "dir-notify-enable",
1786 		.data		= &dir_notify_enable,
1787 		.maxlen		= sizeof(int),
1788 		.mode		= 0644,
1789 		.proc_handler	= proc_dointvec,
1790 	},
1791 #endif
1792 #ifdef CONFIG_MMU
1793 #ifdef CONFIG_FILE_LOCKING
1794 	{
1795 		.procname	= "lease-break-time",
1796 		.data		= &lease_break_time,
1797 		.maxlen		= sizeof(int),
1798 		.mode		= 0644,
1799 		.proc_handler	= proc_dointvec,
1800 	},
1801 #endif
1802 #ifdef CONFIG_AIO
1803 	{
1804 		.procname	= "aio-nr",
1805 		.data		= &aio_nr,
1806 		.maxlen		= sizeof(aio_nr),
1807 		.mode		= 0444,
1808 		.proc_handler	= proc_doulongvec_minmax,
1809 	},
1810 	{
1811 		.procname	= "aio-max-nr",
1812 		.data		= &aio_max_nr,
1813 		.maxlen		= sizeof(aio_max_nr),
1814 		.mode		= 0644,
1815 		.proc_handler	= proc_doulongvec_minmax,
1816 	},
1817 #endif /* CONFIG_AIO */
1818 #ifdef CONFIG_INOTIFY_USER
1819 	{
1820 		.procname	= "inotify",
1821 		.mode		= 0555,
1822 		.child		= inotify_table,
1823 	},
1824 #endif
1825 #ifdef CONFIG_EPOLL
1826 	{
1827 		.procname	= "epoll",
1828 		.mode		= 0555,
1829 		.child		= epoll_table,
1830 	},
1831 #endif
1832 #endif
1833 	{
1834 		.procname	= "protected_symlinks",
1835 		.data		= &sysctl_protected_symlinks,
1836 		.maxlen		= sizeof(int),
1837 		.mode		= 0600,
1838 		.proc_handler	= proc_dointvec_minmax,
1839 		.extra1		= &zero,
1840 		.extra2		= &one,
1841 	},
1842 	{
1843 		.procname	= "protected_hardlinks",
1844 		.data		= &sysctl_protected_hardlinks,
1845 		.maxlen		= sizeof(int),
1846 		.mode		= 0600,
1847 		.proc_handler	= proc_dointvec_minmax,
1848 		.extra1		= &zero,
1849 		.extra2		= &one,
1850 	},
1851 	{
1852 		.procname	= "protected_fifos",
1853 		.data		= &sysctl_protected_fifos,
1854 		.maxlen		= sizeof(int),
1855 		.mode		= 0600,
1856 		.proc_handler	= proc_dointvec_minmax,
1857 		.extra1		= &zero,
1858 		.extra2		= &two,
1859 	},
1860 	{
1861 		.procname	= "protected_regular",
1862 		.data		= &sysctl_protected_regular,
1863 		.maxlen		= sizeof(int),
1864 		.mode		= 0600,
1865 		.proc_handler	= proc_dointvec_minmax,
1866 		.extra1		= &zero,
1867 		.extra2		= &two,
1868 	},
1869 	{
1870 		.procname	= "suid_dumpable",
1871 		.data		= &suid_dumpable,
1872 		.maxlen		= sizeof(int),
1873 		.mode		= 0644,
1874 		.proc_handler	= proc_dointvec_minmax_coredump,
1875 		.extra1		= &zero,
1876 		.extra2		= &two,
1877 	},
1878 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1879 	{
1880 		.procname	= "binfmt_misc",
1881 		.mode		= 0555,
1882 		.child		= sysctl_mount_point,
1883 	},
1884 #endif
1885 	{
1886 		.procname	= "pipe-max-size",
1887 		.data		= &pipe_max_size,
1888 		.maxlen		= sizeof(pipe_max_size),
1889 		.mode		= 0644,
1890 		.proc_handler	= proc_dopipe_max_size,
1891 	},
1892 	{
1893 		.procname	= "pipe-user-pages-hard",
1894 		.data		= &pipe_user_pages_hard,
1895 		.maxlen		= sizeof(pipe_user_pages_hard),
1896 		.mode		= 0644,
1897 		.proc_handler	= proc_doulongvec_minmax,
1898 	},
1899 	{
1900 		.procname	= "pipe-user-pages-soft",
1901 		.data		= &pipe_user_pages_soft,
1902 		.maxlen		= sizeof(pipe_user_pages_soft),
1903 		.mode		= 0644,
1904 		.proc_handler	= proc_doulongvec_minmax,
1905 	},
1906 	{
1907 		.procname	= "mount-max",
1908 		.data		= &sysctl_mount_max,
1909 		.maxlen		= sizeof(unsigned int),
1910 		.mode		= 0644,
1911 		.proc_handler	= proc_dointvec_minmax,
1912 		.extra1		= &one,
1913 	},
1914 	{ }
1915 };
1916 
1917 static struct ctl_table debug_table[] = {
1918 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1919 	{
1920 		.procname	= "exception-trace",
1921 		.data		= &show_unhandled_signals,
1922 		.maxlen		= sizeof(int),
1923 		.mode		= 0644,
1924 		.proc_handler	= proc_dointvec
1925 	},
1926 #endif
1927 #if defined(CONFIG_OPTPROBES)
1928 	{
1929 		.procname	= "kprobes-optimization",
1930 		.data		= &sysctl_kprobes_optimization,
1931 		.maxlen		= sizeof(int),
1932 		.mode		= 0644,
1933 		.proc_handler	= proc_kprobes_optimization_handler,
1934 		.extra1		= &zero,
1935 		.extra2		= &one,
1936 	},
1937 #endif
1938 	{ }
1939 };
1940 
1941 static struct ctl_table dev_table[] = {
1942 	{ }
1943 };
1944 
1945 int __init sysctl_init(void)
1946 {
1947 	struct ctl_table_header *hdr;
1948 
1949 	hdr = register_sysctl_table(sysctl_base_table);
1950 	kmemleak_not_leak(hdr);
1951 	return 0;
1952 }
1953 
1954 #endif /* CONFIG_SYSCTL */
1955 
1956 /*
1957  * /proc/sys support
1958  */
1959 
1960 #ifdef CONFIG_PROC_SYSCTL
1961 
1962 static int _proc_do_string(char *data, int maxlen, int write,
1963 			   char __user *buffer,
1964 			   size_t *lenp, loff_t *ppos)
1965 {
1966 	size_t len;
1967 	char __user *p;
1968 	char c;
1969 
1970 	if (!data || !maxlen || !*lenp) {
1971 		*lenp = 0;
1972 		return 0;
1973 	}
1974 
1975 	if (write) {
1976 		if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
1977 			/* Only continue writes not past the end of buffer. */
1978 			len = strlen(data);
1979 			if (len > maxlen - 1)
1980 				len = maxlen - 1;
1981 
1982 			if (*ppos > len)
1983 				return 0;
1984 			len = *ppos;
1985 		} else {
1986 			/* Start writing from beginning of buffer. */
1987 			len = 0;
1988 		}
1989 
1990 		*ppos += *lenp;
1991 		p = buffer;
1992 		while ((p - buffer) < *lenp && len < maxlen - 1) {
1993 			if (get_user(c, p++))
1994 				return -EFAULT;
1995 			if (c == 0 || c == '\n')
1996 				break;
1997 			data[len++] = c;
1998 		}
1999 		data[len] = 0;
2000 	} else {
2001 		len = strlen(data);
2002 		if (len > maxlen)
2003 			len = maxlen;
2004 
2005 		if (*ppos > len) {
2006 			*lenp = 0;
2007 			return 0;
2008 		}
2009 
2010 		data += *ppos;
2011 		len  -= *ppos;
2012 
2013 		if (len > *lenp)
2014 			len = *lenp;
2015 		if (len)
2016 			if (copy_to_user(buffer, data, len))
2017 				return -EFAULT;
2018 		if (len < *lenp) {
2019 			if (put_user('\n', buffer + len))
2020 				return -EFAULT;
2021 			len++;
2022 		}
2023 		*lenp = len;
2024 		*ppos += len;
2025 	}
2026 	return 0;
2027 }
2028 
2029 static void warn_sysctl_write(struct ctl_table *table)
2030 {
2031 	pr_warn_once("%s wrote to %s when file position was not 0!\n"
2032 		"This will not be supported in the future. To silence this\n"
2033 		"warning, set kernel.sysctl_writes_strict = -1\n",
2034 		current->comm, table->procname);
2035 }
2036 
2037 /**
2038  * proc_first_pos_non_zero_ignore - check if first position is allowed
2039  * @ppos: file position
2040  * @table: the sysctl table
2041  *
2042  * Returns true if the first position is non-zero and the sysctl_writes_strict
2043  * mode indicates this is not allowed for numeric input types. String proc
2044  * handlers can ignore the return value.
2045  */
2046 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
2047 					   struct ctl_table *table)
2048 {
2049 	if (!*ppos)
2050 		return false;
2051 
2052 	switch (sysctl_writes_strict) {
2053 	case SYSCTL_WRITES_STRICT:
2054 		return true;
2055 	case SYSCTL_WRITES_WARN:
2056 		warn_sysctl_write(table);
2057 		return false;
2058 	default:
2059 		return false;
2060 	}
2061 }
2062 
2063 /**
2064  * proc_dostring - read a string sysctl
2065  * @table: the sysctl table
2066  * @write: %TRUE if this is a write to the sysctl file
2067  * @buffer: the user buffer
2068  * @lenp: the size of the user buffer
2069  * @ppos: file position
2070  *
2071  * Reads/writes a string from/to the user buffer. If the kernel
2072  * buffer provided is not large enough to hold the string, the
2073  * string is truncated. The copied string is %NULL-terminated.
2074  * If the string is being read by the user process, it is copied
2075  * and a newline '\n' is added. It is truncated if the buffer is
2076  * not large enough.
2077  *
2078  * Returns 0 on success.
2079  */
2080 int proc_dostring(struct ctl_table *table, int write,
2081 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2082 {
2083 	if (write)
2084 		proc_first_pos_non_zero_ignore(ppos, table);
2085 
2086 	return _proc_do_string((char *)(table->data), table->maxlen, write,
2087 			       (char __user *)buffer, lenp, ppos);
2088 }
2089 
2090 static size_t proc_skip_spaces(char **buf)
2091 {
2092 	size_t ret;
2093 	char *tmp = skip_spaces(*buf);
2094 	ret = tmp - *buf;
2095 	*buf = tmp;
2096 	return ret;
2097 }
2098 
2099 static void proc_skip_char(char **buf, size_t *size, const char v)
2100 {
2101 	while (*size) {
2102 		if (**buf != v)
2103 			break;
2104 		(*size)--;
2105 		(*buf)++;
2106 	}
2107 }
2108 
2109 #define TMPBUFLEN 22
2110 /**
2111  * proc_get_long - reads an ASCII formatted integer from a user buffer
2112  *
2113  * @buf: a kernel buffer
2114  * @size: size of the kernel buffer
2115  * @val: this is where the number will be stored
2116  * @neg: set to %TRUE if number is negative
2117  * @perm_tr: a vector which contains the allowed trailers
2118  * @perm_tr_len: size of the perm_tr vector
2119  * @tr: pointer to store the trailer character
2120  *
2121  * In case of success %0 is returned and @buf and @size are updated with
2122  * the amount of bytes read. If @tr is non-NULL and a trailing
2123  * character exists (size is non-zero after returning from this
2124  * function), @tr is updated with the trailing character.
2125  */
2126 static int proc_get_long(char **buf, size_t *size,
2127 			  unsigned long *val, bool *neg,
2128 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
2129 {
2130 	int len;
2131 	char *p, tmp[TMPBUFLEN];
2132 
2133 	if (!*size)
2134 		return -EINVAL;
2135 
2136 	len = *size;
2137 	if (len > TMPBUFLEN - 1)
2138 		len = TMPBUFLEN - 1;
2139 
2140 	memcpy(tmp, *buf, len);
2141 
2142 	tmp[len] = 0;
2143 	p = tmp;
2144 	if (*p == '-' && *size > 1) {
2145 		*neg = true;
2146 		p++;
2147 	} else
2148 		*neg = false;
2149 	if (!isdigit(*p))
2150 		return -EINVAL;
2151 
2152 	*val = simple_strtoul(p, &p, 0);
2153 
2154 	len = p - tmp;
2155 
2156 	/* We don't know if the next char is whitespace thus we may accept
2157 	 * invalid integers (e.g. 1234...a) or two integers instead of one
2158 	 * (e.g. 123...1). So lets not allow such large numbers. */
2159 	if (len == TMPBUFLEN - 1)
2160 		return -EINVAL;
2161 
2162 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2163 		return -EINVAL;
2164 
2165 	if (tr && (len < *size))
2166 		*tr = *p;
2167 
2168 	*buf += len;
2169 	*size -= len;
2170 
2171 	return 0;
2172 }
2173 
2174 /**
2175  * proc_put_long - converts an integer to a decimal ASCII formatted string
2176  *
2177  * @buf: the user buffer
2178  * @size: the size of the user buffer
2179  * @val: the integer to be converted
2180  * @neg: sign of the number, %TRUE for negative
2181  *
2182  * In case of success %0 is returned and @buf and @size are updated with
2183  * the amount of bytes written.
2184  */
2185 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2186 			  bool neg)
2187 {
2188 	int len;
2189 	char tmp[TMPBUFLEN], *p = tmp;
2190 
2191 	sprintf(p, "%s%lu", neg ? "-" : "", val);
2192 	len = strlen(tmp);
2193 	if (len > *size)
2194 		len = *size;
2195 	if (copy_to_user(*buf, tmp, len))
2196 		return -EFAULT;
2197 	*size -= len;
2198 	*buf += len;
2199 	return 0;
2200 }
2201 #undef TMPBUFLEN
2202 
2203 static int proc_put_char(void __user **buf, size_t *size, char c)
2204 {
2205 	if (*size) {
2206 		char __user **buffer = (char __user **)buf;
2207 		if (put_user(c, *buffer))
2208 			return -EFAULT;
2209 		(*size)--, (*buffer)++;
2210 		*buf = *buffer;
2211 	}
2212 	return 0;
2213 }
2214 
2215 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2216 				 int *valp,
2217 				 int write, void *data)
2218 {
2219 	if (write) {
2220 		if (*negp) {
2221 			if (*lvalp > (unsigned long) INT_MAX + 1)
2222 				return -EINVAL;
2223 			*valp = -*lvalp;
2224 		} else {
2225 			if (*lvalp > (unsigned long) INT_MAX)
2226 				return -EINVAL;
2227 			*valp = *lvalp;
2228 		}
2229 	} else {
2230 		int val = *valp;
2231 		if (val < 0) {
2232 			*negp = true;
2233 			*lvalp = -(unsigned long)val;
2234 		} else {
2235 			*negp = false;
2236 			*lvalp = (unsigned long)val;
2237 		}
2238 	}
2239 	return 0;
2240 }
2241 
2242 static int do_proc_douintvec_conv(unsigned long *lvalp,
2243 				  unsigned int *valp,
2244 				  int write, void *data)
2245 {
2246 	if (write) {
2247 		if (*lvalp > UINT_MAX)
2248 			return -EINVAL;
2249 		*valp = *lvalp;
2250 	} else {
2251 		unsigned int val = *valp;
2252 		*lvalp = (unsigned long)val;
2253 	}
2254 	return 0;
2255 }
2256 
2257 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2258 
2259 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2260 		  int write, void __user *buffer,
2261 		  size_t *lenp, loff_t *ppos,
2262 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2263 			      int write, void *data),
2264 		  void *data)
2265 {
2266 	int *i, vleft, first = 1, err = 0;
2267 	size_t left;
2268 	char *kbuf = NULL, *p;
2269 
2270 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2271 		*lenp = 0;
2272 		return 0;
2273 	}
2274 
2275 	i = (int *) tbl_data;
2276 	vleft = table->maxlen / sizeof(*i);
2277 	left = *lenp;
2278 
2279 	if (!conv)
2280 		conv = do_proc_dointvec_conv;
2281 
2282 	if (write) {
2283 		if (proc_first_pos_non_zero_ignore(ppos, table))
2284 			goto out;
2285 
2286 		if (left > PAGE_SIZE - 1)
2287 			left = PAGE_SIZE - 1;
2288 		p = kbuf = memdup_user_nul(buffer, left);
2289 		if (IS_ERR(kbuf))
2290 			return PTR_ERR(kbuf);
2291 	}
2292 
2293 	for (; left && vleft--; i++, first=0) {
2294 		unsigned long lval;
2295 		bool neg;
2296 
2297 		if (write) {
2298 			left -= proc_skip_spaces(&p);
2299 
2300 			if (!left)
2301 				break;
2302 			err = proc_get_long(&p, &left, &lval, &neg,
2303 					     proc_wspace_sep,
2304 					     sizeof(proc_wspace_sep), NULL);
2305 			if (err)
2306 				break;
2307 			if (conv(&neg, &lval, i, 1, data)) {
2308 				err = -EINVAL;
2309 				break;
2310 			}
2311 		} else {
2312 			if (conv(&neg, &lval, i, 0, data)) {
2313 				err = -EINVAL;
2314 				break;
2315 			}
2316 			if (!first)
2317 				err = proc_put_char(&buffer, &left, '\t');
2318 			if (err)
2319 				break;
2320 			err = proc_put_long(&buffer, &left, lval, neg);
2321 			if (err)
2322 				break;
2323 		}
2324 	}
2325 
2326 	if (!write && !first && left && !err)
2327 		err = proc_put_char(&buffer, &left, '\n');
2328 	if (write && !err && left)
2329 		left -= proc_skip_spaces(&p);
2330 	if (write) {
2331 		kfree(kbuf);
2332 		if (first)
2333 			return err ? : -EINVAL;
2334 	}
2335 	*lenp -= left;
2336 out:
2337 	*ppos += *lenp;
2338 	return err;
2339 }
2340 
2341 static int do_proc_dointvec(struct ctl_table *table, int write,
2342 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2343 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2344 			      int write, void *data),
2345 		  void *data)
2346 {
2347 	return __do_proc_dointvec(table->data, table, write,
2348 			buffer, lenp, ppos, conv, data);
2349 }
2350 
2351 static int do_proc_douintvec_w(unsigned int *tbl_data,
2352 			       struct ctl_table *table,
2353 			       void __user *buffer,
2354 			       size_t *lenp, loff_t *ppos,
2355 			       int (*conv)(unsigned long *lvalp,
2356 					   unsigned int *valp,
2357 					   int write, void *data),
2358 			       void *data)
2359 {
2360 	unsigned long lval;
2361 	int err = 0;
2362 	size_t left;
2363 	bool neg;
2364 	char *kbuf = NULL, *p;
2365 
2366 	left = *lenp;
2367 
2368 	if (proc_first_pos_non_zero_ignore(ppos, table))
2369 		goto bail_early;
2370 
2371 	if (left > PAGE_SIZE - 1)
2372 		left = PAGE_SIZE - 1;
2373 
2374 	p = kbuf = memdup_user_nul(buffer, left);
2375 	if (IS_ERR(kbuf))
2376 		return -EINVAL;
2377 
2378 	left -= proc_skip_spaces(&p);
2379 	if (!left) {
2380 		err = -EINVAL;
2381 		goto out_free;
2382 	}
2383 
2384 	err = proc_get_long(&p, &left, &lval, &neg,
2385 			     proc_wspace_sep,
2386 			     sizeof(proc_wspace_sep), NULL);
2387 	if (err || neg) {
2388 		err = -EINVAL;
2389 		goto out_free;
2390 	}
2391 
2392 	if (conv(&lval, tbl_data, 1, data)) {
2393 		err = -EINVAL;
2394 		goto out_free;
2395 	}
2396 
2397 	if (!err && left)
2398 		left -= proc_skip_spaces(&p);
2399 
2400 out_free:
2401 	kfree(kbuf);
2402 	if (err)
2403 		return -EINVAL;
2404 
2405 	return 0;
2406 
2407 	/* This is in keeping with old __do_proc_dointvec() */
2408 bail_early:
2409 	*ppos += *lenp;
2410 	return err;
2411 }
2412 
2413 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2414 			       size_t *lenp, loff_t *ppos,
2415 			       int (*conv)(unsigned long *lvalp,
2416 					   unsigned int *valp,
2417 					   int write, void *data),
2418 			       void *data)
2419 {
2420 	unsigned long lval;
2421 	int err = 0;
2422 	size_t left;
2423 
2424 	left = *lenp;
2425 
2426 	if (conv(&lval, tbl_data, 0, data)) {
2427 		err = -EINVAL;
2428 		goto out;
2429 	}
2430 
2431 	err = proc_put_long(&buffer, &left, lval, false);
2432 	if (err || !left)
2433 		goto out;
2434 
2435 	err = proc_put_char(&buffer, &left, '\n');
2436 
2437 out:
2438 	*lenp -= left;
2439 	*ppos += *lenp;
2440 
2441 	return err;
2442 }
2443 
2444 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2445 			       int write, void __user *buffer,
2446 			       size_t *lenp, loff_t *ppos,
2447 			       int (*conv)(unsigned long *lvalp,
2448 					   unsigned int *valp,
2449 					   int write, void *data),
2450 			       void *data)
2451 {
2452 	unsigned int *i, vleft;
2453 
2454 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2455 		*lenp = 0;
2456 		return 0;
2457 	}
2458 
2459 	i = (unsigned int *) tbl_data;
2460 	vleft = table->maxlen / sizeof(*i);
2461 
2462 	/*
2463 	 * Arrays are not supported, keep this simple. *Do not* add
2464 	 * support for them.
2465 	 */
2466 	if (vleft != 1) {
2467 		*lenp = 0;
2468 		return -EINVAL;
2469 	}
2470 
2471 	if (!conv)
2472 		conv = do_proc_douintvec_conv;
2473 
2474 	if (write)
2475 		return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2476 					   conv, data);
2477 	return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2478 }
2479 
2480 static int do_proc_douintvec(struct ctl_table *table, int write,
2481 			     void __user *buffer, size_t *lenp, loff_t *ppos,
2482 			     int (*conv)(unsigned long *lvalp,
2483 					 unsigned int *valp,
2484 					 int write, void *data),
2485 			     void *data)
2486 {
2487 	return __do_proc_douintvec(table->data, table, write,
2488 				   buffer, lenp, ppos, conv, data);
2489 }
2490 
2491 /**
2492  * proc_dointvec - read a vector of integers
2493  * @table: the sysctl table
2494  * @write: %TRUE if this is a write to the sysctl file
2495  * @buffer: the user buffer
2496  * @lenp: the size of the user buffer
2497  * @ppos: file position
2498  *
2499  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2500  * values from/to the user buffer, treated as an ASCII string.
2501  *
2502  * Returns 0 on success.
2503  */
2504 int proc_dointvec(struct ctl_table *table, int write,
2505 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2506 {
2507 	return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2508 }
2509 
2510 /**
2511  * proc_douintvec - read a vector of unsigned integers
2512  * @table: the sysctl table
2513  * @write: %TRUE if this is a write to the sysctl file
2514  * @buffer: the user buffer
2515  * @lenp: the size of the user buffer
2516  * @ppos: file position
2517  *
2518  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2519  * values from/to the user buffer, treated as an ASCII string.
2520  *
2521  * Returns 0 on success.
2522  */
2523 int proc_douintvec(struct ctl_table *table, int write,
2524 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2525 {
2526 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
2527 				 do_proc_douintvec_conv, NULL);
2528 }
2529 
2530 /*
2531  * Taint values can only be increased
2532  * This means we can safely use a temporary.
2533  */
2534 static int proc_taint(struct ctl_table *table, int write,
2535 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2536 {
2537 	struct ctl_table t;
2538 	unsigned long tmptaint = get_taint();
2539 	int err;
2540 
2541 	if (write && !capable(CAP_SYS_ADMIN))
2542 		return -EPERM;
2543 
2544 	t = *table;
2545 	t.data = &tmptaint;
2546 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2547 	if (err < 0)
2548 		return err;
2549 
2550 	if (write) {
2551 		/*
2552 		 * Poor man's atomic or. Not worth adding a primitive
2553 		 * to everyone's atomic.h for this
2554 		 */
2555 		int i;
2556 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2557 			if ((tmptaint >> i) & 1)
2558 				add_taint(i, LOCKDEP_STILL_OK);
2559 		}
2560 	}
2561 
2562 	return err;
2563 }
2564 
2565 #ifdef CONFIG_PRINTK
2566 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2567 				void __user *buffer, size_t *lenp, loff_t *ppos)
2568 {
2569 	if (write && !capable(CAP_SYS_ADMIN))
2570 		return -EPERM;
2571 
2572 	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2573 }
2574 #endif
2575 
2576 /**
2577  * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
2578  * @min: pointer to minimum allowable value
2579  * @max: pointer to maximum allowable value
2580  *
2581  * The do_proc_dointvec_minmax_conv_param structure provides the
2582  * minimum and maximum values for doing range checking for those sysctl
2583  * parameters that use the proc_dointvec_minmax() handler.
2584  */
2585 struct do_proc_dointvec_minmax_conv_param {
2586 	int *min;
2587 	int *max;
2588 };
2589 
2590 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2591 					int *valp,
2592 					int write, void *data)
2593 {
2594 	struct do_proc_dointvec_minmax_conv_param *param = data;
2595 	if (write) {
2596 		int val = *negp ? -*lvalp : *lvalp;
2597 		if ((param->min && *param->min > val) ||
2598 		    (param->max && *param->max < val))
2599 			return -EINVAL;
2600 		*valp = val;
2601 	} else {
2602 		int val = *valp;
2603 		if (val < 0) {
2604 			*negp = true;
2605 			*lvalp = -(unsigned long)val;
2606 		} else {
2607 			*negp = false;
2608 			*lvalp = (unsigned long)val;
2609 		}
2610 	}
2611 	return 0;
2612 }
2613 
2614 /**
2615  * proc_dointvec_minmax - read a vector of integers with min/max values
2616  * @table: the sysctl table
2617  * @write: %TRUE if this is a write to the sysctl file
2618  * @buffer: the user buffer
2619  * @lenp: the size of the user buffer
2620  * @ppos: file position
2621  *
2622  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2623  * values from/to the user buffer, treated as an ASCII string.
2624  *
2625  * This routine will ensure the values are within the range specified by
2626  * table->extra1 (min) and table->extra2 (max).
2627  *
2628  * Returns 0 on success or -EINVAL on write when the range check fails.
2629  */
2630 int proc_dointvec_minmax(struct ctl_table *table, int write,
2631 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2632 {
2633 	struct do_proc_dointvec_minmax_conv_param param = {
2634 		.min = (int *) table->extra1,
2635 		.max = (int *) table->extra2,
2636 	};
2637 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2638 				do_proc_dointvec_minmax_conv, &param);
2639 }
2640 
2641 /**
2642  * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
2643  * @min: pointer to minimum allowable value
2644  * @max: pointer to maximum allowable value
2645  *
2646  * The do_proc_douintvec_minmax_conv_param structure provides the
2647  * minimum and maximum values for doing range checking for those sysctl
2648  * parameters that use the proc_douintvec_minmax() handler.
2649  */
2650 struct do_proc_douintvec_minmax_conv_param {
2651 	unsigned int *min;
2652 	unsigned int *max;
2653 };
2654 
2655 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2656 					 unsigned int *valp,
2657 					 int write, void *data)
2658 {
2659 	struct do_proc_douintvec_minmax_conv_param *param = data;
2660 
2661 	if (write) {
2662 		unsigned int val = *lvalp;
2663 
2664 		if (*lvalp > UINT_MAX)
2665 			return -EINVAL;
2666 
2667 		if ((param->min && *param->min > val) ||
2668 		    (param->max && *param->max < val))
2669 			return -ERANGE;
2670 
2671 		*valp = val;
2672 	} else {
2673 		unsigned int val = *valp;
2674 		*lvalp = (unsigned long) val;
2675 	}
2676 
2677 	return 0;
2678 }
2679 
2680 /**
2681  * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2682  * @table: the sysctl table
2683  * @write: %TRUE if this is a write to the sysctl file
2684  * @buffer: the user buffer
2685  * @lenp: the size of the user buffer
2686  * @ppos: file position
2687  *
2688  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2689  * values from/to the user buffer, treated as an ASCII string. Negative
2690  * strings are not allowed.
2691  *
2692  * This routine will ensure the values are within the range specified by
2693  * table->extra1 (min) and table->extra2 (max). There is a final sanity
2694  * check for UINT_MAX to avoid having to support wrap around uses from
2695  * userspace.
2696  *
2697  * Returns 0 on success or -ERANGE on write when the range check fails.
2698  */
2699 int proc_douintvec_minmax(struct ctl_table *table, int write,
2700 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2701 {
2702 	struct do_proc_douintvec_minmax_conv_param param = {
2703 		.min = (unsigned int *) table->extra1,
2704 		.max = (unsigned int *) table->extra2,
2705 	};
2706 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
2707 				 do_proc_douintvec_minmax_conv, &param);
2708 }
2709 
2710 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
2711 					unsigned int *valp,
2712 					int write, void *data)
2713 {
2714 	if (write) {
2715 		unsigned int val;
2716 
2717 		val = round_pipe_size(*lvalp);
2718 		if (val == 0)
2719 			return -EINVAL;
2720 
2721 		*valp = val;
2722 	} else {
2723 		unsigned int val = *valp;
2724 		*lvalp = (unsigned long) val;
2725 	}
2726 
2727 	return 0;
2728 }
2729 
2730 static int proc_dopipe_max_size(struct ctl_table *table, int write,
2731 				void __user *buffer, size_t *lenp, loff_t *ppos)
2732 {
2733 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
2734 				 do_proc_dopipe_max_size_conv, NULL);
2735 }
2736 
2737 static void validate_coredump_safety(void)
2738 {
2739 #ifdef CONFIG_COREDUMP
2740 	if (suid_dumpable == SUID_DUMP_ROOT &&
2741 	    core_pattern[0] != '/' && core_pattern[0] != '|') {
2742 		printk(KERN_WARNING
2743 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2744 "Pipe handler or fully qualified core dump path required.\n"
2745 "Set kernel.core_pattern before fs.suid_dumpable.\n"
2746 		);
2747 	}
2748 #endif
2749 }
2750 
2751 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2752 		void __user *buffer, size_t *lenp, loff_t *ppos)
2753 {
2754 	int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2755 	if (!error)
2756 		validate_coredump_safety();
2757 	return error;
2758 }
2759 
2760 #ifdef CONFIG_COREDUMP
2761 static int proc_dostring_coredump(struct ctl_table *table, int write,
2762 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2763 {
2764 	int error = proc_dostring(table, write, buffer, lenp, ppos);
2765 	if (!error)
2766 		validate_coredump_safety();
2767 	return error;
2768 }
2769 #endif
2770 
2771 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2772 				     void __user *buffer,
2773 				     size_t *lenp, loff_t *ppos,
2774 				     unsigned long convmul,
2775 				     unsigned long convdiv)
2776 {
2777 	unsigned long *i, *min, *max;
2778 	int vleft, first = 1, err = 0;
2779 	size_t left;
2780 	char *kbuf = NULL, *p;
2781 
2782 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2783 		*lenp = 0;
2784 		return 0;
2785 	}
2786 
2787 	i = (unsigned long *) data;
2788 	min = (unsigned long *) table->extra1;
2789 	max = (unsigned long *) table->extra2;
2790 	vleft = table->maxlen / sizeof(unsigned long);
2791 	left = *lenp;
2792 
2793 	if (write) {
2794 		if (proc_first_pos_non_zero_ignore(ppos, table))
2795 			goto out;
2796 
2797 		if (left > PAGE_SIZE - 1)
2798 			left = PAGE_SIZE - 1;
2799 		p = kbuf = memdup_user_nul(buffer, left);
2800 		if (IS_ERR(kbuf))
2801 			return PTR_ERR(kbuf);
2802 	}
2803 
2804 	for (; left && vleft--; i++, first = 0) {
2805 		unsigned long val;
2806 
2807 		if (write) {
2808 			bool neg;
2809 
2810 			left -= proc_skip_spaces(&p);
2811 			if (!left)
2812 				break;
2813 
2814 			err = proc_get_long(&p, &left, &val, &neg,
2815 					     proc_wspace_sep,
2816 					     sizeof(proc_wspace_sep), NULL);
2817 			if (err)
2818 				break;
2819 			if (neg)
2820 				continue;
2821 			val = convmul * val / convdiv;
2822 			if ((min && val < *min) || (max && val > *max))
2823 				continue;
2824 			*i = val;
2825 		} else {
2826 			val = convdiv * (*i) / convmul;
2827 			if (!first) {
2828 				err = proc_put_char(&buffer, &left, '\t');
2829 				if (err)
2830 					break;
2831 			}
2832 			err = proc_put_long(&buffer, &left, val, false);
2833 			if (err)
2834 				break;
2835 		}
2836 	}
2837 
2838 	if (!write && !first && left && !err)
2839 		err = proc_put_char(&buffer, &left, '\n');
2840 	if (write && !err)
2841 		left -= proc_skip_spaces(&p);
2842 	if (write) {
2843 		kfree(kbuf);
2844 		if (first)
2845 			return err ? : -EINVAL;
2846 	}
2847 	*lenp -= left;
2848 out:
2849 	*ppos += *lenp;
2850 	return err;
2851 }
2852 
2853 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2854 				     void __user *buffer,
2855 				     size_t *lenp, loff_t *ppos,
2856 				     unsigned long convmul,
2857 				     unsigned long convdiv)
2858 {
2859 	return __do_proc_doulongvec_minmax(table->data, table, write,
2860 			buffer, lenp, ppos, convmul, convdiv);
2861 }
2862 
2863 /**
2864  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2865  * @table: the sysctl table
2866  * @write: %TRUE if this is a write to the sysctl file
2867  * @buffer: the user buffer
2868  * @lenp: the size of the user buffer
2869  * @ppos: file position
2870  *
2871  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2872  * values from/to the user buffer, treated as an ASCII string.
2873  *
2874  * This routine will ensure the values are within the range specified by
2875  * table->extra1 (min) and table->extra2 (max).
2876  *
2877  * Returns 0 on success.
2878  */
2879 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2880 			   void __user *buffer, size_t *lenp, loff_t *ppos)
2881 {
2882     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2883 }
2884 
2885 /**
2886  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2887  * @table: the sysctl table
2888  * @write: %TRUE if this is a write to the sysctl file
2889  * @buffer: the user buffer
2890  * @lenp: the size of the user buffer
2891  * @ppos: file position
2892  *
2893  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2894  * values from/to the user buffer, treated as an ASCII string. The values
2895  * are treated as milliseconds, and converted to jiffies when they are stored.
2896  *
2897  * This routine will ensure the values are within the range specified by
2898  * table->extra1 (min) and table->extra2 (max).
2899  *
2900  * Returns 0 on success.
2901  */
2902 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2903 				      void __user *buffer,
2904 				      size_t *lenp, loff_t *ppos)
2905 {
2906     return do_proc_doulongvec_minmax(table, write, buffer,
2907 				     lenp, ppos, HZ, 1000l);
2908 }
2909 
2910 
2911 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2912 					 int *valp,
2913 					 int write, void *data)
2914 {
2915 	if (write) {
2916 		if (*lvalp > INT_MAX / HZ)
2917 			return 1;
2918 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2919 	} else {
2920 		int val = *valp;
2921 		unsigned long lval;
2922 		if (val < 0) {
2923 			*negp = true;
2924 			lval = -(unsigned long)val;
2925 		} else {
2926 			*negp = false;
2927 			lval = (unsigned long)val;
2928 		}
2929 		*lvalp = lval / HZ;
2930 	}
2931 	return 0;
2932 }
2933 
2934 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2935 						int *valp,
2936 						int write, void *data)
2937 {
2938 	if (write) {
2939 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2940 			return 1;
2941 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2942 	} else {
2943 		int val = *valp;
2944 		unsigned long lval;
2945 		if (val < 0) {
2946 			*negp = true;
2947 			lval = -(unsigned long)val;
2948 		} else {
2949 			*negp = false;
2950 			lval = (unsigned long)val;
2951 		}
2952 		*lvalp = jiffies_to_clock_t(lval);
2953 	}
2954 	return 0;
2955 }
2956 
2957 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2958 					    int *valp,
2959 					    int write, void *data)
2960 {
2961 	if (write) {
2962 		unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2963 
2964 		if (jif > INT_MAX)
2965 			return 1;
2966 		*valp = (int)jif;
2967 	} else {
2968 		int val = *valp;
2969 		unsigned long lval;
2970 		if (val < 0) {
2971 			*negp = true;
2972 			lval = -(unsigned long)val;
2973 		} else {
2974 			*negp = false;
2975 			lval = (unsigned long)val;
2976 		}
2977 		*lvalp = jiffies_to_msecs(lval);
2978 	}
2979 	return 0;
2980 }
2981 
2982 /**
2983  * proc_dointvec_jiffies - read a vector of integers as seconds
2984  * @table: the sysctl table
2985  * @write: %TRUE if this is a write to the sysctl file
2986  * @buffer: the user buffer
2987  * @lenp: the size of the user buffer
2988  * @ppos: file position
2989  *
2990  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2991  * values from/to the user buffer, treated as an ASCII string.
2992  * The values read are assumed to be in seconds, and are converted into
2993  * jiffies.
2994  *
2995  * Returns 0 on success.
2996  */
2997 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2998 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2999 {
3000     return do_proc_dointvec(table,write,buffer,lenp,ppos,
3001 		    	    do_proc_dointvec_jiffies_conv,NULL);
3002 }
3003 
3004 /**
3005  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
3006  * @table: the sysctl table
3007  * @write: %TRUE if this is a write to the sysctl file
3008  * @buffer: the user buffer
3009  * @lenp: the size of the user buffer
3010  * @ppos: pointer to the file position
3011  *
3012  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3013  * values from/to the user buffer, treated as an ASCII string.
3014  * The values read are assumed to be in 1/USER_HZ seconds, and
3015  * are converted into jiffies.
3016  *
3017  * Returns 0 on success.
3018  */
3019 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3020 				 void __user *buffer, size_t *lenp, loff_t *ppos)
3021 {
3022     return do_proc_dointvec(table,write,buffer,lenp,ppos,
3023 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
3024 }
3025 
3026 /**
3027  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
3028  * @table: the sysctl table
3029  * @write: %TRUE if this is a write to the sysctl file
3030  * @buffer: the user buffer
3031  * @lenp: the size of the user buffer
3032  * @ppos: file position
3033  * @ppos: the current position in the file
3034  *
3035  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3036  * values from/to the user buffer, treated as an ASCII string.
3037  * The values read are assumed to be in 1/1000 seconds, and
3038  * are converted into jiffies.
3039  *
3040  * Returns 0 on success.
3041  */
3042 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3043 			     void __user *buffer, size_t *lenp, loff_t *ppos)
3044 {
3045 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
3046 				do_proc_dointvec_ms_jiffies_conv, NULL);
3047 }
3048 
3049 static int proc_do_cad_pid(struct ctl_table *table, int write,
3050 			   void __user *buffer, size_t *lenp, loff_t *ppos)
3051 {
3052 	struct pid *new_pid;
3053 	pid_t tmp;
3054 	int r;
3055 
3056 	tmp = pid_vnr(cad_pid);
3057 
3058 	r = __do_proc_dointvec(&tmp, table, write, buffer,
3059 			       lenp, ppos, NULL, NULL);
3060 	if (r || !write)
3061 		return r;
3062 
3063 	new_pid = find_get_pid(tmp);
3064 	if (!new_pid)
3065 		return -ESRCH;
3066 
3067 	put_pid(xchg(&cad_pid, new_pid));
3068 	return 0;
3069 }
3070 
3071 /**
3072  * proc_do_large_bitmap - read/write from/to a large bitmap
3073  * @table: the sysctl table
3074  * @write: %TRUE if this is a write to the sysctl file
3075  * @buffer: the user buffer
3076  * @lenp: the size of the user buffer
3077  * @ppos: file position
3078  *
3079  * The bitmap is stored at table->data and the bitmap length (in bits)
3080  * in table->maxlen.
3081  *
3082  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
3083  * large bitmaps may be represented in a compact manner. Writing into
3084  * the file will clear the bitmap then update it with the given input.
3085  *
3086  * Returns 0 on success.
3087  */
3088 int proc_do_large_bitmap(struct ctl_table *table, int write,
3089 			 void __user *buffer, size_t *lenp, loff_t *ppos)
3090 {
3091 	int err = 0;
3092 	bool first = 1;
3093 	size_t left = *lenp;
3094 	unsigned long bitmap_len = table->maxlen;
3095 	unsigned long *bitmap = *(unsigned long **) table->data;
3096 	unsigned long *tmp_bitmap = NULL;
3097 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
3098 
3099 	if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
3100 		*lenp = 0;
3101 		return 0;
3102 	}
3103 
3104 	if (write) {
3105 		char *kbuf, *p;
3106 
3107 		if (left > PAGE_SIZE - 1)
3108 			left = PAGE_SIZE - 1;
3109 
3110 		p = kbuf = memdup_user_nul(buffer, left);
3111 		if (IS_ERR(kbuf))
3112 			return PTR_ERR(kbuf);
3113 
3114 		tmp_bitmap = kcalloc(BITS_TO_LONGS(bitmap_len),
3115 				     sizeof(unsigned long),
3116 				     GFP_KERNEL);
3117 		if (!tmp_bitmap) {
3118 			kfree(kbuf);
3119 			return -ENOMEM;
3120 		}
3121 		proc_skip_char(&p, &left, '\n');
3122 		while (!err && left) {
3123 			unsigned long val_a, val_b;
3124 			bool neg;
3125 
3126 			err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3127 					     sizeof(tr_a), &c);
3128 			if (err)
3129 				break;
3130 			if (val_a >= bitmap_len || neg) {
3131 				err = -EINVAL;
3132 				break;
3133 			}
3134 
3135 			val_b = val_a;
3136 			if (left) {
3137 				p++;
3138 				left--;
3139 			}
3140 
3141 			if (c == '-') {
3142 				err = proc_get_long(&p, &left, &val_b,
3143 						     &neg, tr_b, sizeof(tr_b),
3144 						     &c);
3145 				if (err)
3146 					break;
3147 				if (val_b >= bitmap_len || neg ||
3148 				    val_a > val_b) {
3149 					err = -EINVAL;
3150 					break;
3151 				}
3152 				if (left) {
3153 					p++;
3154 					left--;
3155 				}
3156 			}
3157 
3158 			bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3159 			first = 0;
3160 			proc_skip_char(&p, &left, '\n');
3161 		}
3162 		kfree(kbuf);
3163 	} else {
3164 		unsigned long bit_a, bit_b = 0;
3165 
3166 		while (left) {
3167 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3168 			if (bit_a >= bitmap_len)
3169 				break;
3170 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
3171 						   bit_a + 1) - 1;
3172 
3173 			if (!first) {
3174 				err = proc_put_char(&buffer, &left, ',');
3175 				if (err)
3176 					break;
3177 			}
3178 			err = proc_put_long(&buffer, &left, bit_a, false);
3179 			if (err)
3180 				break;
3181 			if (bit_a != bit_b) {
3182 				err = proc_put_char(&buffer, &left, '-');
3183 				if (err)
3184 					break;
3185 				err = proc_put_long(&buffer, &left, bit_b, false);
3186 				if (err)
3187 					break;
3188 			}
3189 
3190 			first = 0; bit_b++;
3191 		}
3192 		if (!err)
3193 			err = proc_put_char(&buffer, &left, '\n');
3194 	}
3195 
3196 	if (!err) {
3197 		if (write) {
3198 			if (*ppos)
3199 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3200 			else
3201 				bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3202 		}
3203 		*lenp -= left;
3204 		*ppos += *lenp;
3205 	}
3206 
3207 	kfree(tmp_bitmap);
3208 	return err;
3209 }
3210 
3211 #else /* CONFIG_PROC_SYSCTL */
3212 
3213 int proc_dostring(struct ctl_table *table, int write,
3214 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3215 {
3216 	return -ENOSYS;
3217 }
3218 
3219 int proc_dointvec(struct ctl_table *table, int write,
3220 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3221 {
3222 	return -ENOSYS;
3223 }
3224 
3225 int proc_douintvec(struct ctl_table *table, int write,
3226 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3227 {
3228 	return -ENOSYS;
3229 }
3230 
3231 int proc_dointvec_minmax(struct ctl_table *table, int write,
3232 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3233 {
3234 	return -ENOSYS;
3235 }
3236 
3237 int proc_douintvec_minmax(struct ctl_table *table, int write,
3238 			  void __user *buffer, size_t *lenp, loff_t *ppos)
3239 {
3240 	return -ENOSYS;
3241 }
3242 
3243 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3244 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3245 {
3246 	return -ENOSYS;
3247 }
3248 
3249 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3250 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3251 {
3252 	return -ENOSYS;
3253 }
3254 
3255 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3256 			     void __user *buffer, size_t *lenp, loff_t *ppos)
3257 {
3258 	return -ENOSYS;
3259 }
3260 
3261 int proc_doulongvec_minmax(struct ctl_table *table, int write,
3262 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3263 {
3264 	return -ENOSYS;
3265 }
3266 
3267 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3268 				      void __user *buffer,
3269 				      size_t *lenp, loff_t *ppos)
3270 {
3271     return -ENOSYS;
3272 }
3273 
3274 
3275 #endif /* CONFIG_PROC_SYSCTL */
3276 
3277 #ifdef CONFIG_BPF_SYSCALL
3278 static int proc_dointvec_minmax_bpf_stats(struct ctl_table *table, int write,
3279 					  void __user *buffer, size_t *lenp,
3280 					  loff_t *ppos)
3281 {
3282 	int ret, bpf_stats = *(int *)table->data;
3283 	struct ctl_table tmp = *table;
3284 
3285 	if (write && !capable(CAP_SYS_ADMIN))
3286 		return -EPERM;
3287 
3288 	tmp.data = &bpf_stats;
3289 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
3290 	if (write && !ret) {
3291 		*(int *)table->data = bpf_stats;
3292 		if (bpf_stats)
3293 			static_branch_enable(&bpf_stats_enabled_key);
3294 		else
3295 			static_branch_disable(&bpf_stats_enabled_key);
3296 	}
3297 	return ret;
3298 }
3299 #endif
3300 /*
3301  * No sense putting this after each symbol definition, twice,
3302  * exception granted :-)
3303  */
3304 EXPORT_SYMBOL(proc_dointvec);
3305 EXPORT_SYMBOL(proc_douintvec);
3306 EXPORT_SYMBOL(proc_dointvec_jiffies);
3307 EXPORT_SYMBOL(proc_dointvec_minmax);
3308 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3309 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3310 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3311 EXPORT_SYMBOL(proc_dostring);
3312 EXPORT_SYMBOL(proc_doulongvec_minmax);
3313 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3314