• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 // SPDX-License-Identifier: GPL-2.0-only
2 /*
3  * sysctl.c: General linux system control interface
4  *
5  * Begun 24 March 1995, Stephen Tweedie
6  * Added /proc support, Dec 1995
7  * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
8  * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
9  * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
10  * Dynamic registration fixes, Stephen Tweedie.
11  * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
12  * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
13  *  Horn.
14  * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
15  * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
16  * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
17  *  Wendling.
18  * The list_for_each() macro wasn't appropriate for the sysctl loop.
19  *  Removed it and replaced it with older style, 03/23/00, Bill Wendling
20  */
21 
22 #include <linux/module.h>
23 #include <linux/aio.h>
24 #include <linux/mm.h>
25 #include <linux/swap.h>
26 #include <linux/slab.h>
27 #include <linux/sysctl.h>
28 #include <linux/bitmap.h>
29 #include <linux/signal.h>
30 #include <linux/printk.h>
31 #include <linux/proc_fs.h>
32 #include <linux/security.h>
33 #include <linux/ctype.h>
34 #include <linux/kmemleak.h>
35 #include <linux/fs.h>
36 #include <linux/init.h>
37 #include <linux/kernel.h>
38 #include <linux/kobject.h>
39 #include <linux/net.h>
40 #include <linux/sysrq.h>
41 #include <linux/highuid.h>
42 #include <linux/writeback.h>
43 #include <linux/ratelimit.h>
44 #include <linux/compaction.h>
45 #include <linux/hugetlb.h>
46 #include <linux/initrd.h>
47 #include <linux/key.h>
48 #include <linux/times.h>
49 #include <linux/limits.h>
50 #include <linux/dcache.h>
51 #include <linux/dnotify.h>
52 #include <linux/syscalls.h>
53 #include <linux/vmstat.h>
54 #include <linux/nfs_fs.h>
55 #include <linux/acpi.h>
56 #include <linux/reboot.h>
57 #include <linux/ftrace.h>
58 #include <linux/perf_event.h>
59 #include <linux/kprobes.h>
60 #include <linux/pipe_fs_i.h>
61 #include <linux/oom.h>
62 #include <linux/kmod.h>
63 #include <linux/capability.h>
64 #include <linux/binfmts.h>
65 #include <linux/sched/sysctl.h>
66 #include <linux/sched/coredump.h>
67 #include <linux/kexec.h>
68 #include <linux/bpf.h>
69 #include <linux/mount.h>
70 #include <linux/userfaultfd_k.h>
71 
72 #include "../lib/kstrtox.h"
73 
74 #include <linux/uaccess.h>
75 #include <asm/processor.h>
76 
77 #ifdef CONFIG_X86
78 #include <asm/nmi.h>
79 #include <asm/stacktrace.h>
80 #include <asm/io.h>
81 #endif
82 #ifdef CONFIG_SPARC
83 #include <asm/setup.h>
84 #endif
85 #ifdef CONFIG_BSD_PROCESS_ACCT
86 #include <linux/acct.h>
87 #endif
88 #ifdef CONFIG_RT_MUTEXES
89 #include <linux/rtmutex.h>
90 #endif
91 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
92 #include <linux/lockdep.h>
93 #endif
94 #ifdef CONFIG_CHR_DEV_SG
95 #include <scsi/sg.h>
96 #endif
97 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
98 #include <linux/stackleak.h>
99 #endif
100 #ifdef CONFIG_LOCKUP_DETECTOR
101 #include <linux/nmi.h>
102 #endif
103 
104 #if defined(CONFIG_SYSCTL)
105 
106 /* External variables not in a header file. */
107 extern int suid_dumpable;
108 #ifdef CONFIG_COREDUMP
109 extern int core_uses_pid;
110 extern char core_pattern[];
111 extern unsigned int core_pipe_limit;
112 #endif
113 extern int pid_max;
114 extern int extra_free_kbytes;
115 extern int pid_max_min, pid_max_max;
116 extern int percpu_pagelist_fraction;
117 extern int latencytop_enabled;
118 extern unsigned int sysctl_nr_open_min, sysctl_nr_open_max;
119 #ifndef CONFIG_MMU
120 extern int sysctl_nr_trim_pages;
121 #endif
122 
123 /* Constants used for minimum and  maximum */
124 #ifdef CONFIG_LOCKUP_DETECTOR
125 static int sixty = 60;
126 #endif
127 
128 static int __maybe_unused neg_one = -1;
129 static int __maybe_unused two = 2;
130 static int __maybe_unused four = 4;
131 static unsigned long zero_ul;
132 static unsigned long one_ul = 1;
133 static unsigned long long_max = LONG_MAX;
134 static int one_hundred = 100;
135 static int one_thousand = 1000;
136 #ifdef CONFIG_PRINTK
137 static int ten_thousand = 10000;
138 #endif
139 #ifdef CONFIG_PERF_EVENTS
140 static int six_hundred_forty_kb = 640 * 1024;
141 #endif
142 
143 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
144 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
145 
146 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
147 static int maxolduid = 65535;
148 static int minolduid;
149 
150 static int ngroups_max = NGROUPS_MAX;
151 static const int cap_last_cap = CAP_LAST_CAP;
152 
153 /*
154  * This is needed for proc_doulongvec_minmax of sysctl_hung_task_timeout_secs
155  * and hung_task_check_interval_secs
156  */
157 #ifdef CONFIG_DETECT_HUNG_TASK
158 static unsigned long hung_task_timeout_max = (LONG_MAX/HZ);
159 #endif
160 
161 #ifdef CONFIG_INOTIFY_USER
162 #include <linux/inotify.h>
163 #endif
164 #ifdef CONFIG_SPARC
165 #endif
166 
167 #ifdef CONFIG_PARISC
168 extern int pwrsw_enabled;
169 #endif
170 
171 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
172 extern int unaligned_enabled;
173 #endif
174 
175 #ifdef CONFIG_IA64
176 extern int unaligned_dump_stack;
177 #endif
178 
179 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
180 extern int no_unaligned_warning;
181 #endif
182 
183 #ifdef CONFIG_PROC_SYSCTL
184 
185 /**
186  * enum sysctl_writes_mode - supported sysctl write modes
187  *
188  * @SYSCTL_WRITES_LEGACY: each write syscall must fully contain the sysctl value
189  *	to be written, and multiple writes on the same sysctl file descriptor
190  *	will rewrite the sysctl value, regardless of file position. No warning
191  *	is issued when the initial position is not 0.
192  * @SYSCTL_WRITES_WARN: same as above but warn when the initial file position is
193  *	not 0.
194  * @SYSCTL_WRITES_STRICT: writes to numeric sysctl entries must always be at
195  *	file position 0 and the value must be fully contained in the buffer
196  *	sent to the write syscall. If dealing with strings respect the file
197  *	position, but restrict this to the max length of the buffer, anything
198  *	passed the max length will be ignored. Multiple writes will append
199  *	to the buffer.
200  *
201  * These write modes control how current file position affects the behavior of
202  * updating sysctl values through the proc interface on each write.
203  */
204 enum sysctl_writes_mode {
205 	SYSCTL_WRITES_LEGACY		= -1,
206 	SYSCTL_WRITES_WARN		= 0,
207 	SYSCTL_WRITES_STRICT		= 1,
208 };
209 
210 static enum sysctl_writes_mode sysctl_writes_strict = SYSCTL_WRITES_STRICT;
211 
212 static int proc_do_cad_pid(struct ctl_table *table, int write,
213 		  void __user *buffer, size_t *lenp, loff_t *ppos);
214 static int proc_taint(struct ctl_table *table, int write,
215 			       void __user *buffer, size_t *lenp, loff_t *ppos);
216 #endif
217 
218 #ifdef CONFIG_PRINTK
219 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
220 				void __user *buffer, size_t *lenp, loff_t *ppos);
221 #endif
222 
223 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
224 		void __user *buffer, size_t *lenp, loff_t *ppos);
225 #ifdef CONFIG_COREDUMP
226 static int proc_dostring_coredump(struct ctl_table *table, int write,
227 		void __user *buffer, size_t *lenp, loff_t *ppos);
228 #endif
229 static int proc_dopipe_max_size(struct ctl_table *table, int write,
230 		void __user *buffer, size_t *lenp, loff_t *ppos);
231 
232 #ifdef CONFIG_MAGIC_SYSRQ
233 /* Note: sysrq code uses its own private copy */
234 static int __sysrq_enabled = CONFIG_MAGIC_SYSRQ_DEFAULT_ENABLE;
235 
sysrq_sysctl_handler(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)236 static int sysrq_sysctl_handler(struct ctl_table *table, int write,
237 				void __user *buffer, size_t *lenp,
238 				loff_t *ppos)
239 {
240 	int error;
241 
242 	error = proc_dointvec(table, write, buffer, lenp, ppos);
243 	if (error)
244 		return error;
245 
246 	if (write)
247 		sysrq_toggle_support(__sysrq_enabled);
248 
249 	return 0;
250 }
251 
252 #endif
253 
254 #ifdef CONFIG_BPF_SYSCALL
255 
unpriv_ebpf_notify(int new_state)256 void __weak unpriv_ebpf_notify(int new_state)
257 {
258 }
259 
bpf_unpriv_handler(struct ctl_table * table,int write,void * buffer,size_t * lenp,loff_t * ppos)260 static int bpf_unpriv_handler(struct ctl_table *table, int write,
261                              void *buffer, size_t *lenp, loff_t *ppos)
262 {
263 	int ret, unpriv_enable = *(int *)table->data;
264 	bool locked_state = unpriv_enable == 1;
265 	struct ctl_table tmp = *table;
266 
267 	if (write && !capable(CAP_SYS_ADMIN))
268 		return -EPERM;
269 
270 	tmp.data = &unpriv_enable;
271 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
272 	if (write && !ret) {
273 		if (locked_state && unpriv_enable != 1)
274 			return -EPERM;
275 		*(int *)table->data = unpriv_enable;
276 	}
277 
278 	unpriv_ebpf_notify(unpriv_enable);
279 
280 	return ret;
281 }
282 #endif
283 
284 static struct ctl_table kern_table[];
285 static struct ctl_table vm_table[];
286 static struct ctl_table fs_table[];
287 static struct ctl_table debug_table[];
288 static struct ctl_table dev_table[];
289 extern struct ctl_table random_table[];
290 #ifdef CONFIG_EPOLL
291 extern struct ctl_table epoll_table[];
292 #endif
293 
294 #ifdef CONFIG_FW_LOADER_USER_HELPER
295 extern struct ctl_table firmware_config_table[];
296 #endif
297 
298 #if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
299     defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
300 int sysctl_legacy_va_layout;
301 #endif
302 
303 /* The default sysctl tables: */
304 
305 static struct ctl_table sysctl_base_table[] = {
306 	{
307 		.procname	= "kernel",
308 		.mode		= 0555,
309 		.child		= kern_table,
310 	},
311 	{
312 		.procname	= "vm",
313 		.mode		= 0555,
314 		.child		= vm_table,
315 	},
316 	{
317 		.procname	= "fs",
318 		.mode		= 0555,
319 		.child		= fs_table,
320 	},
321 	{
322 		.procname	= "debug",
323 		.mode		= 0555,
324 		.child		= debug_table,
325 	},
326 	{
327 		.procname	= "dev",
328 		.mode		= 0555,
329 		.child		= dev_table,
330 	},
331 	{ }
332 };
333 
334 #ifdef CONFIG_SCHED_DEBUG
335 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
336 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
337 static int min_wakeup_granularity_ns;			/* 0 usecs */
338 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
339 #ifdef CONFIG_SMP
340 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
341 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
342 #endif /* CONFIG_SMP */
343 #endif /* CONFIG_SCHED_DEBUG */
344 
345 #ifdef CONFIG_COMPACTION
346 static int min_extfrag_threshold;
347 static int max_extfrag_threshold = 1000;
348 #endif
349 
350 static struct ctl_table kern_table[] = {
351 	{
352 		.procname	= "sched_child_runs_first",
353 		.data		= &sysctl_sched_child_runs_first,
354 		.maxlen		= sizeof(unsigned int),
355 		.mode		= 0644,
356 		.proc_handler	= proc_dointvec,
357 	},
358 #ifdef CONFIG_SCHED_DEBUG
359 	{
360 		.procname	= "sched_min_granularity_ns",
361 		.data		= &sysctl_sched_min_granularity,
362 		.maxlen		= sizeof(unsigned int),
363 		.mode		= 0644,
364 		.proc_handler	= sched_proc_update_handler,
365 		.extra1		= &min_sched_granularity_ns,
366 		.extra2		= &max_sched_granularity_ns,
367 	},
368 	{
369 		.procname	= "sched_latency_ns",
370 		.data		= &sysctl_sched_latency,
371 		.maxlen		= sizeof(unsigned int),
372 		.mode		= 0644,
373 		.proc_handler	= sched_proc_update_handler,
374 		.extra1		= &min_sched_granularity_ns,
375 		.extra2		= &max_sched_granularity_ns,
376 	},
377 	{
378 		.procname	= "sched_wakeup_granularity_ns",
379 		.data		= &sysctl_sched_wakeup_granularity,
380 		.maxlen		= sizeof(unsigned int),
381 		.mode		= 0644,
382 		.proc_handler	= sched_proc_update_handler,
383 		.extra1		= &min_wakeup_granularity_ns,
384 		.extra2		= &max_wakeup_granularity_ns,
385 	},
386 #ifdef CONFIG_SMP
387 	{
388 		.procname	= "sched_tunable_scaling",
389 		.data		= &sysctl_sched_tunable_scaling,
390 		.maxlen		= sizeof(enum sched_tunable_scaling),
391 		.mode		= 0644,
392 		.proc_handler	= sched_proc_update_handler,
393 		.extra1		= &min_sched_tunable_scaling,
394 		.extra2		= &max_sched_tunable_scaling,
395 	},
396 	{
397 		.procname	= "sched_migration_cost_ns",
398 		.data		= &sysctl_sched_migration_cost,
399 		.maxlen		= sizeof(unsigned int),
400 		.mode		= 0644,
401 		.proc_handler	= proc_dointvec,
402 	},
403 	{
404 		.procname	= "sched_nr_migrate",
405 		.data		= &sysctl_sched_nr_migrate,
406 		.maxlen		= sizeof(unsigned int),
407 		.mode		= 0644,
408 		.proc_handler	= proc_dointvec,
409 	},
410 #ifdef CONFIG_SCHEDSTATS
411 	{
412 		.procname	= "sched_schedstats",
413 		.data		= NULL,
414 		.maxlen		= sizeof(unsigned int),
415 		.mode		= 0644,
416 		.proc_handler	= sysctl_schedstats,
417 		.extra1		= SYSCTL_ZERO,
418 		.extra2		= SYSCTL_ONE,
419 	},
420 #endif /* CONFIG_SCHEDSTATS */
421 #endif /* CONFIG_SMP */
422 #ifdef CONFIG_NUMA_BALANCING
423 	{
424 		.procname	= "numa_balancing_scan_delay_ms",
425 		.data		= &sysctl_numa_balancing_scan_delay,
426 		.maxlen		= sizeof(unsigned int),
427 		.mode		= 0644,
428 		.proc_handler	= proc_dointvec,
429 	},
430 	{
431 		.procname	= "numa_balancing_scan_period_min_ms",
432 		.data		= &sysctl_numa_balancing_scan_period_min,
433 		.maxlen		= sizeof(unsigned int),
434 		.mode		= 0644,
435 		.proc_handler	= proc_dointvec,
436 	},
437 	{
438 		.procname	= "numa_balancing_scan_period_max_ms",
439 		.data		= &sysctl_numa_balancing_scan_period_max,
440 		.maxlen		= sizeof(unsigned int),
441 		.mode		= 0644,
442 		.proc_handler	= proc_dointvec,
443 	},
444 	{
445 		.procname	= "numa_balancing_scan_size_mb",
446 		.data		= &sysctl_numa_balancing_scan_size,
447 		.maxlen		= sizeof(unsigned int),
448 		.mode		= 0644,
449 		.proc_handler	= proc_dointvec_minmax,
450 		.extra1		= SYSCTL_ONE,
451 	},
452 	{
453 		.procname	= "numa_balancing",
454 		.data		= NULL, /* filled in by handler */
455 		.maxlen		= sizeof(unsigned int),
456 		.mode		= 0644,
457 		.proc_handler	= sysctl_numa_balancing,
458 		.extra1		= SYSCTL_ZERO,
459 		.extra2		= SYSCTL_ONE,
460 	},
461 #endif /* CONFIG_NUMA_BALANCING */
462 #endif /* CONFIG_SCHED_DEBUG */
463 	{
464 		.procname	= "sched_rt_period_us",
465 		.data		= &sysctl_sched_rt_period,
466 		.maxlen		= sizeof(unsigned int),
467 		.mode		= 0644,
468 		.proc_handler	= sched_rt_handler,
469 		.extra1		= SYSCTL_ONE,
470 		.extra2		= SYSCTL_INT_MAX,
471 	},
472 	{
473 		.procname	= "sched_rt_runtime_us",
474 		.data		= &sysctl_sched_rt_runtime,
475 		.maxlen		= sizeof(int),
476 		.mode		= 0644,
477 		.proc_handler	= sched_rt_handler,
478 		.extra1		= &neg_one,
479 		.extra2		= SYSCTL_INT_MAX,
480 	},
481 	{
482 		.procname	= "sched_rr_timeslice_ms",
483 		.data		= &sysctl_sched_rr_timeslice,
484 		.maxlen		= sizeof(int),
485 		.mode		= 0644,
486 		.proc_handler	= sched_rr_handler,
487 	},
488 #ifdef CONFIG_UCLAMP_TASK
489 	{
490 		.procname	= "sched_util_clamp_min",
491 		.data		= &sysctl_sched_uclamp_util_min,
492 		.maxlen		= sizeof(unsigned int),
493 		.mode		= 0644,
494 		.proc_handler	= sysctl_sched_uclamp_handler,
495 	},
496 	{
497 		.procname	= "sched_util_clamp_max",
498 		.data		= &sysctl_sched_uclamp_util_max,
499 		.maxlen		= sizeof(unsigned int),
500 		.mode		= 0644,
501 		.proc_handler	= sysctl_sched_uclamp_handler,
502 	},
503 #endif
504 #ifdef CONFIG_SCHED_AUTOGROUP
505 	{
506 		.procname	= "sched_autogroup_enabled",
507 		.data		= &sysctl_sched_autogroup_enabled,
508 		.maxlen		= sizeof(unsigned int),
509 		.mode		= 0644,
510 		.proc_handler	= proc_dointvec_minmax,
511 		.extra1		= SYSCTL_ZERO,
512 		.extra2		= SYSCTL_ONE,
513 	},
514 #endif
515 #ifdef CONFIG_CFS_BANDWIDTH
516 	{
517 		.procname	= "sched_cfs_bandwidth_slice_us",
518 		.data		= &sysctl_sched_cfs_bandwidth_slice,
519 		.maxlen		= sizeof(unsigned int),
520 		.mode		= 0644,
521 		.proc_handler	= proc_dointvec_minmax,
522 		.extra1		= SYSCTL_ONE,
523 	},
524 #endif
525 #if defined(CONFIG_ENERGY_MODEL) && defined(CONFIG_CPU_FREQ_GOV_SCHEDUTIL)
526 	{
527 		.procname	= "sched_energy_aware",
528 		.data		= &sysctl_sched_energy_aware,
529 		.maxlen		= sizeof(unsigned int),
530 		.mode		= 0644,
531 		.proc_handler	= sched_energy_aware_handler,
532 		.extra1		= SYSCTL_ZERO,
533 		.extra2		= SYSCTL_ONE,
534 	},
535 #endif
536 #ifdef CONFIG_PROVE_LOCKING
537 	{
538 		.procname	= "prove_locking",
539 		.data		= &prove_locking,
540 		.maxlen		= sizeof(int),
541 		.mode		= 0644,
542 		.proc_handler	= proc_dointvec,
543 	},
544 #endif
545 #ifdef CONFIG_LOCK_STAT
546 	{
547 		.procname	= "lock_stat",
548 		.data		= &lock_stat,
549 		.maxlen		= sizeof(int),
550 		.mode		= 0644,
551 		.proc_handler	= proc_dointvec,
552 	},
553 #endif
554 	{
555 		.procname	= "panic",
556 		.data		= &panic_timeout,
557 		.maxlen		= sizeof(int),
558 		.mode		= 0644,
559 		.proc_handler	= proc_dointvec,
560 	},
561 #ifdef CONFIG_COREDUMP
562 	{
563 		.procname	= "core_uses_pid",
564 		.data		= &core_uses_pid,
565 		.maxlen		= sizeof(int),
566 		.mode		= 0644,
567 		.proc_handler	= proc_dointvec,
568 	},
569 	{
570 		.procname	= "core_pattern",
571 		.data		= core_pattern,
572 		.maxlen		= CORENAME_MAX_SIZE,
573 		.mode		= 0644,
574 		.proc_handler	= proc_dostring_coredump,
575 	},
576 	{
577 		.procname	= "core_pipe_limit",
578 		.data		= &core_pipe_limit,
579 		.maxlen		= sizeof(unsigned int),
580 		.mode		= 0644,
581 		.proc_handler	= proc_dointvec,
582 	},
583 #endif
584 #ifdef CONFIG_PROC_SYSCTL
585 	{
586 		.procname	= "tainted",
587 		.maxlen 	= sizeof(long),
588 		.mode		= 0644,
589 		.proc_handler	= proc_taint,
590 	},
591 	{
592 		.procname	= "sysctl_writes_strict",
593 		.data		= &sysctl_writes_strict,
594 		.maxlen		= sizeof(int),
595 		.mode		= 0644,
596 		.proc_handler	= proc_dointvec_minmax,
597 		.extra1		= &neg_one,
598 		.extra2		= SYSCTL_ONE,
599 	},
600 #endif
601 #ifdef CONFIG_LATENCYTOP
602 	{
603 		.procname	= "latencytop",
604 		.data		= &latencytop_enabled,
605 		.maxlen		= sizeof(int),
606 		.mode		= 0644,
607 		.proc_handler	= sysctl_latencytop,
608 	},
609 #endif
610 #ifdef CONFIG_BLK_DEV_INITRD
611 	{
612 		.procname	= "real-root-dev",
613 		.data		= &real_root_dev,
614 		.maxlen		= sizeof(int),
615 		.mode		= 0644,
616 		.proc_handler	= proc_dointvec,
617 	},
618 #endif
619 	{
620 		.procname	= "print-fatal-signals",
621 		.data		= &print_fatal_signals,
622 		.maxlen		= sizeof(int),
623 		.mode		= 0644,
624 		.proc_handler	= proc_dointvec,
625 	},
626 #ifdef CONFIG_SPARC
627 	{
628 		.procname	= "reboot-cmd",
629 		.data		= reboot_command,
630 		.maxlen		= 256,
631 		.mode		= 0644,
632 		.proc_handler	= proc_dostring,
633 	},
634 	{
635 		.procname	= "stop-a",
636 		.data		= &stop_a_enabled,
637 		.maxlen		= sizeof (int),
638 		.mode		= 0644,
639 		.proc_handler	= proc_dointvec,
640 	},
641 	{
642 		.procname	= "scons-poweroff",
643 		.data		= &scons_pwroff,
644 		.maxlen		= sizeof (int),
645 		.mode		= 0644,
646 		.proc_handler	= proc_dointvec,
647 	},
648 #endif
649 #ifdef CONFIG_SPARC64
650 	{
651 		.procname	= "tsb-ratio",
652 		.data		= &sysctl_tsb_ratio,
653 		.maxlen		= sizeof (int),
654 		.mode		= 0644,
655 		.proc_handler	= proc_dointvec,
656 	},
657 #endif
658 #ifdef CONFIG_PARISC
659 	{
660 		.procname	= "soft-power",
661 		.data		= &pwrsw_enabled,
662 		.maxlen		= sizeof (int),
663 	 	.mode		= 0644,
664 		.proc_handler	= proc_dointvec,
665 	},
666 #endif
667 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
668 	{
669 		.procname	= "unaligned-trap",
670 		.data		= &unaligned_enabled,
671 		.maxlen		= sizeof (int),
672 		.mode		= 0644,
673 		.proc_handler	= proc_dointvec,
674 	},
675 #endif
676 	{
677 		.procname	= "ctrl-alt-del",
678 		.data		= &C_A_D,
679 		.maxlen		= sizeof(int),
680 		.mode		= 0644,
681 		.proc_handler	= proc_dointvec,
682 	},
683 #ifdef CONFIG_FUNCTION_TRACER
684 	{
685 		.procname	= "ftrace_enabled",
686 		.data		= &ftrace_enabled,
687 		.maxlen		= sizeof(int),
688 		.mode		= 0644,
689 		.proc_handler	= ftrace_enable_sysctl,
690 	},
691 #endif
692 #ifdef CONFIG_STACK_TRACER
693 	{
694 		.procname	= "stack_tracer_enabled",
695 		.data		= &stack_tracer_enabled,
696 		.maxlen		= sizeof(int),
697 		.mode		= 0644,
698 		.proc_handler	= stack_trace_sysctl,
699 	},
700 #endif
701 #ifdef CONFIG_TRACING
702 	{
703 		.procname	= "ftrace_dump_on_oops",
704 		.data		= &ftrace_dump_on_oops,
705 		.maxlen		= sizeof(int),
706 		.mode		= 0644,
707 		.proc_handler	= proc_dointvec,
708 	},
709 	{
710 		.procname	= "traceoff_on_warning",
711 		.data		= &__disable_trace_on_warning,
712 		.maxlen		= sizeof(__disable_trace_on_warning),
713 		.mode		= 0644,
714 		.proc_handler	= proc_dointvec,
715 	},
716 	{
717 		.procname	= "tracepoint_printk",
718 		.data		= &tracepoint_printk,
719 		.maxlen		= sizeof(tracepoint_printk),
720 		.mode		= 0644,
721 		.proc_handler	= tracepoint_printk_sysctl,
722 	},
723 #endif
724 #ifdef CONFIG_KEXEC_CORE
725 	{
726 		.procname	= "kexec_load_disabled",
727 		.data		= &kexec_load_disabled,
728 		.maxlen		= sizeof(int),
729 		.mode		= 0644,
730 		/* only handle a transition from default "0" to "1" */
731 		.proc_handler	= proc_dointvec_minmax,
732 		.extra1		= SYSCTL_ONE,
733 		.extra2		= SYSCTL_ONE,
734 	},
735 #endif
736 #ifdef CONFIG_MODULES
737 	{
738 		.procname	= "modprobe",
739 		.data		= &modprobe_path,
740 		.maxlen		= KMOD_PATH_LEN,
741 		.mode		= 0644,
742 		.proc_handler	= proc_dostring,
743 	},
744 	{
745 		.procname	= "modules_disabled",
746 		.data		= &modules_disabled,
747 		.maxlen		= sizeof(int),
748 		.mode		= 0644,
749 		/* only handle a transition from default "0" to "1" */
750 		.proc_handler	= proc_dointvec_minmax,
751 		.extra1		= SYSCTL_ONE,
752 		.extra2		= SYSCTL_ONE,
753 	},
754 #endif
755 #ifdef CONFIG_UEVENT_HELPER
756 	{
757 		.procname	= "hotplug",
758 		.data		= &uevent_helper,
759 		.maxlen		= UEVENT_HELPER_PATH_LEN,
760 		.mode		= 0644,
761 		.proc_handler	= proc_dostring,
762 	},
763 #endif
764 #ifdef CONFIG_CHR_DEV_SG
765 	{
766 		.procname	= "sg-big-buff",
767 		.data		= &sg_big_buff,
768 		.maxlen		= sizeof (int),
769 		.mode		= 0444,
770 		.proc_handler	= proc_dointvec,
771 	},
772 #endif
773 #ifdef CONFIG_BSD_PROCESS_ACCT
774 	{
775 		.procname	= "acct",
776 		.data		= &acct_parm,
777 		.maxlen		= 3*sizeof(int),
778 		.mode		= 0644,
779 		.proc_handler	= proc_dointvec,
780 	},
781 #endif
782 #ifdef CONFIG_MAGIC_SYSRQ
783 	{
784 		.procname	= "sysrq",
785 		.data		= &__sysrq_enabled,
786 		.maxlen		= sizeof (int),
787 		.mode		= 0644,
788 		.proc_handler	= sysrq_sysctl_handler,
789 	},
790 #endif
791 #ifdef CONFIG_PROC_SYSCTL
792 	{
793 		.procname	= "cad_pid",
794 		.data		= NULL,
795 		.maxlen		= sizeof (int),
796 		.mode		= 0600,
797 		.proc_handler	= proc_do_cad_pid,
798 	},
799 #endif
800 	{
801 		.procname	= "threads-max",
802 		.data		= NULL,
803 		.maxlen		= sizeof(int),
804 		.mode		= 0644,
805 		.proc_handler	= sysctl_max_threads,
806 	},
807 	{
808 		.procname	= "random",
809 		.mode		= 0555,
810 		.child		= random_table,
811 	},
812 	{
813 		.procname	= "usermodehelper",
814 		.mode		= 0555,
815 		.child		= usermodehelper_table,
816 	},
817 #ifdef CONFIG_FW_LOADER_USER_HELPER
818 	{
819 		.procname	= "firmware_config",
820 		.mode		= 0555,
821 		.child		= firmware_config_table,
822 	},
823 #endif
824 	{
825 		.procname	= "overflowuid",
826 		.data		= &overflowuid,
827 		.maxlen		= sizeof(int),
828 		.mode		= 0644,
829 		.proc_handler	= proc_dointvec_minmax,
830 		.extra1		= &minolduid,
831 		.extra2		= &maxolduid,
832 	},
833 	{
834 		.procname	= "overflowgid",
835 		.data		= &overflowgid,
836 		.maxlen		= sizeof(int),
837 		.mode		= 0644,
838 		.proc_handler	= proc_dointvec_minmax,
839 		.extra1		= &minolduid,
840 		.extra2		= &maxolduid,
841 	},
842 #ifdef CONFIG_S390
843 #ifdef CONFIG_MATHEMU
844 	{
845 		.procname	= "ieee_emulation_warnings",
846 		.data		= &sysctl_ieee_emulation_warnings,
847 		.maxlen		= sizeof(int),
848 		.mode		= 0644,
849 		.proc_handler	= proc_dointvec,
850 	},
851 #endif
852 	{
853 		.procname	= "userprocess_debug",
854 		.data		= &show_unhandled_signals,
855 		.maxlen		= sizeof(int),
856 		.mode		= 0644,
857 		.proc_handler	= proc_dointvec,
858 	},
859 #endif
860 	{
861 		.procname	= "pid_max",
862 		.data		= &pid_max,
863 		.maxlen		= sizeof (int),
864 		.mode		= 0644,
865 		.proc_handler	= proc_dointvec_minmax,
866 		.extra1		= &pid_max_min,
867 		.extra2		= &pid_max_max,
868 	},
869 	{
870 		.procname	= "panic_on_oops",
871 		.data		= &panic_on_oops,
872 		.maxlen		= sizeof(int),
873 		.mode		= 0644,
874 		.proc_handler	= proc_dointvec,
875 	},
876 	{
877 		.procname	= "panic_print",
878 		.data		= &panic_print,
879 		.maxlen		= sizeof(unsigned long),
880 		.mode		= 0644,
881 		.proc_handler	= proc_doulongvec_minmax,
882 	},
883 #if defined CONFIG_PRINTK
884 	{
885 		.procname	= "printk",
886 		.data		= &console_loglevel,
887 		.maxlen		= 4*sizeof(int),
888 		.mode		= 0644,
889 		.proc_handler	= proc_dointvec,
890 	},
891 	{
892 		.procname	= "printk_ratelimit",
893 		.data		= &printk_ratelimit_state.interval,
894 		.maxlen		= sizeof(int),
895 		.mode		= 0644,
896 		.proc_handler	= proc_dointvec_jiffies,
897 	},
898 	{
899 		.procname	= "printk_ratelimit_burst",
900 		.data		= &printk_ratelimit_state.burst,
901 		.maxlen		= sizeof(int),
902 		.mode		= 0644,
903 		.proc_handler	= proc_dointvec,
904 	},
905 	{
906 		.procname	= "printk_delay",
907 		.data		= &printk_delay_msec,
908 		.maxlen		= sizeof(int),
909 		.mode		= 0644,
910 		.proc_handler	= proc_dointvec_minmax,
911 		.extra1		= SYSCTL_ZERO,
912 		.extra2		= &ten_thousand,
913 	},
914 	{
915 		.procname	= "printk_devkmsg",
916 		.data		= devkmsg_log_str,
917 		.maxlen		= DEVKMSG_STR_MAX_SIZE,
918 		.mode		= 0644,
919 		.proc_handler	= devkmsg_sysctl_set_loglvl,
920 	},
921 	{
922 		.procname	= "dmesg_restrict",
923 		.data		= &dmesg_restrict,
924 		.maxlen		= sizeof(int),
925 		.mode		= 0644,
926 		.proc_handler	= proc_dointvec_minmax_sysadmin,
927 		.extra1		= SYSCTL_ZERO,
928 		.extra2		= SYSCTL_ONE,
929 	},
930 	{
931 		.procname	= "kptr_restrict",
932 		.data		= &kptr_restrict,
933 		.maxlen		= sizeof(int),
934 		.mode		= 0644,
935 		.proc_handler	= proc_dointvec_minmax_sysadmin,
936 		.extra1		= SYSCTL_ZERO,
937 		.extra2		= &two,
938 	},
939 #endif
940 	{
941 		.procname	= "ngroups_max",
942 		.data		= &ngroups_max,
943 		.maxlen		= sizeof (int),
944 		.mode		= 0444,
945 		.proc_handler	= proc_dointvec,
946 	},
947 	{
948 		.procname	= "cap_last_cap",
949 		.data		= (void *)&cap_last_cap,
950 		.maxlen		= sizeof(int),
951 		.mode		= 0444,
952 		.proc_handler	= proc_dointvec,
953 	},
954 #if defined(CONFIG_LOCKUP_DETECTOR)
955 	{
956 		.procname       = "watchdog",
957 		.data		= &watchdog_user_enabled,
958 		.maxlen		= sizeof(int),
959 		.mode		= 0644,
960 		.proc_handler   = proc_watchdog,
961 		.extra1		= SYSCTL_ZERO,
962 		.extra2		= SYSCTL_ONE,
963 	},
964 	{
965 		.procname	= "watchdog_thresh",
966 		.data		= &watchdog_thresh,
967 		.maxlen		= sizeof(int),
968 		.mode		= 0644,
969 		.proc_handler	= proc_watchdog_thresh,
970 		.extra1		= SYSCTL_ZERO,
971 		.extra2		= &sixty,
972 	},
973 	{
974 		.procname       = "nmi_watchdog",
975 		.data		= &nmi_watchdog_user_enabled,
976 		.maxlen		= sizeof(int),
977 		.mode		= NMI_WATCHDOG_SYSCTL_PERM,
978 		.proc_handler   = proc_nmi_watchdog,
979 		.extra1		= SYSCTL_ZERO,
980 		.extra2		= SYSCTL_ONE,
981 	},
982 	{
983 		.procname	= "watchdog_cpumask",
984 		.data		= &watchdog_cpumask_bits,
985 		.maxlen		= NR_CPUS,
986 		.mode		= 0644,
987 		.proc_handler	= proc_watchdog_cpumask,
988 	},
989 #ifdef CONFIG_SOFTLOCKUP_DETECTOR
990 	{
991 		.procname       = "soft_watchdog",
992 		.data		= &soft_watchdog_user_enabled,
993 		.maxlen		= sizeof(int),
994 		.mode		= 0644,
995 		.proc_handler   = proc_soft_watchdog,
996 		.extra1		= SYSCTL_ZERO,
997 		.extra2		= SYSCTL_ONE,
998 	},
999 	{
1000 		.procname	= "softlockup_panic",
1001 		.data		= &softlockup_panic,
1002 		.maxlen		= sizeof(int),
1003 		.mode		= 0644,
1004 		.proc_handler	= proc_dointvec_minmax,
1005 		.extra1		= SYSCTL_ZERO,
1006 		.extra2		= SYSCTL_ONE,
1007 	},
1008 #ifdef CONFIG_SMP
1009 	{
1010 		.procname	= "softlockup_all_cpu_backtrace",
1011 		.data		= &sysctl_softlockup_all_cpu_backtrace,
1012 		.maxlen		= sizeof(int),
1013 		.mode		= 0644,
1014 		.proc_handler	= proc_dointvec_minmax,
1015 		.extra1		= SYSCTL_ZERO,
1016 		.extra2		= SYSCTL_ONE,
1017 	},
1018 #endif /* CONFIG_SMP */
1019 #endif
1020 #ifdef CONFIG_HARDLOCKUP_DETECTOR
1021 	{
1022 		.procname	= "hardlockup_panic",
1023 		.data		= &hardlockup_panic,
1024 		.maxlen		= sizeof(int),
1025 		.mode		= 0644,
1026 		.proc_handler	= proc_dointvec_minmax,
1027 		.extra1		= SYSCTL_ZERO,
1028 		.extra2		= SYSCTL_ONE,
1029 	},
1030 #ifdef CONFIG_SMP
1031 	{
1032 		.procname	= "hardlockup_all_cpu_backtrace",
1033 		.data		= &sysctl_hardlockup_all_cpu_backtrace,
1034 		.maxlen		= sizeof(int),
1035 		.mode		= 0644,
1036 		.proc_handler	= proc_dointvec_minmax,
1037 		.extra1		= SYSCTL_ZERO,
1038 		.extra2		= SYSCTL_ONE,
1039 	},
1040 #endif /* CONFIG_SMP */
1041 #endif
1042 #endif
1043 
1044 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
1045 	{
1046 		.procname       = "unknown_nmi_panic",
1047 		.data           = &unknown_nmi_panic,
1048 		.maxlen         = sizeof (int),
1049 		.mode           = 0644,
1050 		.proc_handler   = proc_dointvec,
1051 	},
1052 #endif
1053 #if defined(CONFIG_X86)
1054 	{
1055 		.procname	= "panic_on_unrecovered_nmi",
1056 		.data		= &panic_on_unrecovered_nmi,
1057 		.maxlen		= sizeof(int),
1058 		.mode		= 0644,
1059 		.proc_handler	= proc_dointvec,
1060 	},
1061 	{
1062 		.procname	= "panic_on_io_nmi",
1063 		.data		= &panic_on_io_nmi,
1064 		.maxlen		= sizeof(int),
1065 		.mode		= 0644,
1066 		.proc_handler	= proc_dointvec,
1067 	},
1068 #ifdef CONFIG_DEBUG_STACKOVERFLOW
1069 	{
1070 		.procname	= "panic_on_stackoverflow",
1071 		.data		= &sysctl_panic_on_stackoverflow,
1072 		.maxlen		= sizeof(int),
1073 		.mode		= 0644,
1074 		.proc_handler	= proc_dointvec,
1075 	},
1076 #endif
1077 	{
1078 		.procname	= "bootloader_type",
1079 		.data		= &bootloader_type,
1080 		.maxlen		= sizeof (int),
1081 		.mode		= 0444,
1082 		.proc_handler	= proc_dointvec,
1083 	},
1084 	{
1085 		.procname	= "bootloader_version",
1086 		.data		= &bootloader_version,
1087 		.maxlen		= sizeof (int),
1088 		.mode		= 0444,
1089 		.proc_handler	= proc_dointvec,
1090 	},
1091 	{
1092 		.procname	= "io_delay_type",
1093 		.data		= &io_delay_type,
1094 		.maxlen		= sizeof(int),
1095 		.mode		= 0644,
1096 		.proc_handler	= proc_dointvec,
1097 	},
1098 #endif
1099 #if defined(CONFIG_MMU)
1100 	{
1101 		.procname	= "randomize_va_space",
1102 		.data		= &randomize_va_space,
1103 		.maxlen		= sizeof(int),
1104 		.mode		= 0644,
1105 		.proc_handler	= proc_dointvec,
1106 	},
1107 #endif
1108 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
1109 	{
1110 		.procname	= "spin_retry",
1111 		.data		= &spin_retry,
1112 		.maxlen		= sizeof (int),
1113 		.mode		= 0644,
1114 		.proc_handler	= proc_dointvec,
1115 	},
1116 #endif
1117 #if	defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
1118 	{
1119 		.procname	= "acpi_video_flags",
1120 		.data		= &acpi_realmode_flags,
1121 		.maxlen		= sizeof (unsigned long),
1122 		.mode		= 0644,
1123 		.proc_handler	= proc_doulongvec_minmax,
1124 	},
1125 #endif
1126 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
1127 	{
1128 		.procname	= "ignore-unaligned-usertrap",
1129 		.data		= &no_unaligned_warning,
1130 		.maxlen		= sizeof (int),
1131 	 	.mode		= 0644,
1132 		.proc_handler	= proc_dointvec,
1133 	},
1134 #endif
1135 #ifdef CONFIG_IA64
1136 	{
1137 		.procname	= "unaligned-dump-stack",
1138 		.data		= &unaligned_dump_stack,
1139 		.maxlen		= sizeof (int),
1140 		.mode		= 0644,
1141 		.proc_handler	= proc_dointvec,
1142 	},
1143 #endif
1144 #ifdef CONFIG_DETECT_HUNG_TASK
1145 	{
1146 		.procname	= "hung_task_panic",
1147 		.data		= &sysctl_hung_task_panic,
1148 		.maxlen		= sizeof(int),
1149 		.mode		= 0644,
1150 		.proc_handler	= proc_dointvec_minmax,
1151 		.extra1		= SYSCTL_ZERO,
1152 		.extra2		= SYSCTL_ONE,
1153 	},
1154 	{
1155 		.procname	= "hung_task_check_count",
1156 		.data		= &sysctl_hung_task_check_count,
1157 		.maxlen		= sizeof(int),
1158 		.mode		= 0644,
1159 		.proc_handler	= proc_dointvec_minmax,
1160 		.extra1		= SYSCTL_ZERO,
1161 	},
1162 	{
1163 		.procname	= "hung_task_timeout_secs",
1164 		.data		= &sysctl_hung_task_timeout_secs,
1165 		.maxlen		= sizeof(unsigned long),
1166 		.mode		= 0644,
1167 		.proc_handler	= proc_dohung_task_timeout_secs,
1168 		.extra2		= &hung_task_timeout_max,
1169 	},
1170 	{
1171 		.procname	= "hung_task_check_interval_secs",
1172 		.data		= &sysctl_hung_task_check_interval_secs,
1173 		.maxlen		= sizeof(unsigned long),
1174 		.mode		= 0644,
1175 		.proc_handler	= proc_dohung_task_timeout_secs,
1176 		.extra2		= &hung_task_timeout_max,
1177 	},
1178 	{
1179 		.procname	= "hung_task_warnings",
1180 		.data		= &sysctl_hung_task_warnings,
1181 		.maxlen		= sizeof(int),
1182 		.mode		= 0644,
1183 		.proc_handler	= proc_dointvec_minmax,
1184 		.extra1		= &neg_one,
1185 	},
1186 #endif
1187 #ifdef CONFIG_RT_MUTEXES
1188 	{
1189 		.procname	= "max_lock_depth",
1190 		.data		= &max_lock_depth,
1191 		.maxlen		= sizeof(int),
1192 		.mode		= 0644,
1193 		.proc_handler	= proc_dointvec,
1194 	},
1195 #endif
1196 	{
1197 		.procname	= "poweroff_cmd",
1198 		.data		= &poweroff_cmd,
1199 		.maxlen		= POWEROFF_CMD_PATH_LEN,
1200 		.mode		= 0644,
1201 		.proc_handler	= proc_dostring,
1202 	},
1203 #ifdef CONFIG_KEYS
1204 	{
1205 		.procname	= "keys",
1206 		.mode		= 0555,
1207 		.child		= key_sysctls,
1208 	},
1209 #endif
1210 #ifdef CONFIG_PERF_EVENTS
1211 	/*
1212 	 * User-space scripts rely on the existence of this file
1213 	 * as a feature check for perf_events being enabled.
1214 	 *
1215 	 * So it's an ABI, do not remove!
1216 	 */
1217 	{
1218 		.procname	= "perf_event_paranoid",
1219 		.data		= &sysctl_perf_event_paranoid,
1220 		.maxlen		= sizeof(sysctl_perf_event_paranoid),
1221 		.mode		= 0644,
1222 		.proc_handler	= proc_dointvec,
1223 	},
1224 	{
1225 		.procname	= "perf_event_mlock_kb",
1226 		.data		= &sysctl_perf_event_mlock,
1227 		.maxlen		= sizeof(sysctl_perf_event_mlock),
1228 		.mode		= 0644,
1229 		.proc_handler	= proc_dointvec,
1230 	},
1231 	{
1232 		.procname	= "perf_event_max_sample_rate",
1233 		.data		= &sysctl_perf_event_sample_rate,
1234 		.maxlen		= sizeof(sysctl_perf_event_sample_rate),
1235 		.mode		= 0644,
1236 		.proc_handler	= perf_proc_update_handler,
1237 		.extra1		= SYSCTL_ONE,
1238 	},
1239 	{
1240 		.procname	= "perf_cpu_time_max_percent",
1241 		.data		= &sysctl_perf_cpu_time_max_percent,
1242 		.maxlen		= sizeof(sysctl_perf_cpu_time_max_percent),
1243 		.mode		= 0644,
1244 		.proc_handler	= perf_cpu_time_max_percent_handler,
1245 		.extra1		= SYSCTL_ZERO,
1246 		.extra2		= &one_hundred,
1247 	},
1248 	{
1249 		.procname	= "perf_event_max_stack",
1250 		.data		= &sysctl_perf_event_max_stack,
1251 		.maxlen		= sizeof(sysctl_perf_event_max_stack),
1252 		.mode		= 0644,
1253 		.proc_handler	= perf_event_max_stack_handler,
1254 		.extra1		= SYSCTL_ZERO,
1255 		.extra2		= &six_hundred_forty_kb,
1256 	},
1257 	{
1258 		.procname	= "perf_event_max_contexts_per_stack",
1259 		.data		= &sysctl_perf_event_max_contexts_per_stack,
1260 		.maxlen		= sizeof(sysctl_perf_event_max_contexts_per_stack),
1261 		.mode		= 0644,
1262 		.proc_handler	= perf_event_max_stack_handler,
1263 		.extra1		= SYSCTL_ZERO,
1264 		.extra2		= &one_thousand,
1265 	},
1266 #endif
1267 	{
1268 		.procname	= "panic_on_warn",
1269 		.data		= &panic_on_warn,
1270 		.maxlen		= sizeof(int),
1271 		.mode		= 0644,
1272 		.proc_handler	= proc_dointvec_minmax,
1273 		.extra1		= SYSCTL_ZERO,
1274 		.extra2		= SYSCTL_ONE,
1275 	},
1276 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
1277 	{
1278 		.procname	= "timer_migration",
1279 		.data		= &sysctl_timer_migration,
1280 		.maxlen		= sizeof(unsigned int),
1281 		.mode		= 0644,
1282 		.proc_handler	= timer_migration_handler,
1283 		.extra1		= SYSCTL_ZERO,
1284 		.extra2		= SYSCTL_ONE,
1285 	},
1286 #endif
1287 #ifdef CONFIG_BPF_SYSCALL
1288 	{
1289 		.procname	= "unprivileged_bpf_disabled",
1290 		.data		= &sysctl_unprivileged_bpf_disabled,
1291 		.maxlen		= sizeof(sysctl_unprivileged_bpf_disabled),
1292 		.mode		= 0644,
1293 		.proc_handler	= bpf_unpriv_handler,
1294 		.extra1		= SYSCTL_ZERO,
1295 		.extra2		= &two,
1296 	},
1297 	{
1298 		.procname	= "bpf_stats_enabled",
1299 		.data		= &bpf_stats_enabled_key.key,
1300 		.maxlen		= sizeof(bpf_stats_enabled_key),
1301 		.mode		= 0644,
1302 		.proc_handler	= proc_do_static_key,
1303 	},
1304 #endif
1305 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_PREEMPT_RCU)
1306 	{
1307 		.procname	= "panic_on_rcu_stall",
1308 		.data		= &sysctl_panic_on_rcu_stall,
1309 		.maxlen		= sizeof(sysctl_panic_on_rcu_stall),
1310 		.mode		= 0644,
1311 		.proc_handler	= proc_dointvec_minmax,
1312 		.extra1		= SYSCTL_ZERO,
1313 		.extra2		= SYSCTL_ONE,
1314 	},
1315 #endif
1316 #ifdef CONFIG_STACKLEAK_RUNTIME_DISABLE
1317 	{
1318 		.procname	= "stack_erasing",
1319 		.data		= NULL,
1320 		.maxlen		= sizeof(int),
1321 		.mode		= 0600,
1322 		.proc_handler	= stack_erasing_sysctl,
1323 		.extra1		= SYSCTL_ZERO,
1324 		.extra2		= SYSCTL_ONE,
1325 	},
1326 #endif
1327 	{ }
1328 };
1329 
1330 static struct ctl_table vm_table[] = {
1331 	{
1332 		.procname	= "overcommit_memory",
1333 		.data		= &sysctl_overcommit_memory,
1334 		.maxlen		= sizeof(sysctl_overcommit_memory),
1335 		.mode		= 0644,
1336 		.proc_handler	= proc_dointvec_minmax,
1337 		.extra1		= SYSCTL_ZERO,
1338 		.extra2		= &two,
1339 	},
1340 	{
1341 		.procname	= "panic_on_oom",
1342 		.data		= &sysctl_panic_on_oom,
1343 		.maxlen		= sizeof(sysctl_panic_on_oom),
1344 		.mode		= 0644,
1345 		.proc_handler	= proc_dointvec_minmax,
1346 		.extra1		= SYSCTL_ZERO,
1347 		.extra2		= &two,
1348 	},
1349 	{
1350 		.procname	= "oom_kill_allocating_task",
1351 		.data		= &sysctl_oom_kill_allocating_task,
1352 		.maxlen		= sizeof(sysctl_oom_kill_allocating_task),
1353 		.mode		= 0644,
1354 		.proc_handler	= proc_dointvec,
1355 	},
1356 	{
1357 		.procname	= "oom_dump_tasks",
1358 		.data		= &sysctl_oom_dump_tasks,
1359 		.maxlen		= sizeof(sysctl_oom_dump_tasks),
1360 		.mode		= 0644,
1361 		.proc_handler	= proc_dointvec,
1362 	},
1363 	{
1364 		.procname	= "overcommit_ratio",
1365 		.data		= &sysctl_overcommit_ratio,
1366 		.maxlen		= sizeof(sysctl_overcommit_ratio),
1367 		.mode		= 0644,
1368 		.proc_handler	= overcommit_ratio_handler,
1369 	},
1370 	{
1371 		.procname	= "overcommit_kbytes",
1372 		.data		= &sysctl_overcommit_kbytes,
1373 		.maxlen		= sizeof(sysctl_overcommit_kbytes),
1374 		.mode		= 0644,
1375 		.proc_handler	= overcommit_kbytes_handler,
1376 	},
1377 	{
1378 		.procname	= "page-cluster",
1379 		.data		= &page_cluster,
1380 		.maxlen		= sizeof(int),
1381 		.mode		= 0644,
1382 		.proc_handler	= proc_dointvec_minmax,
1383 		.extra1		= SYSCTL_ZERO,
1384 	},
1385 	{
1386 		.procname	= "dirty_background_ratio",
1387 		.data		= &dirty_background_ratio,
1388 		.maxlen		= sizeof(dirty_background_ratio),
1389 		.mode		= 0644,
1390 		.proc_handler	= dirty_background_ratio_handler,
1391 		.extra1		= SYSCTL_ZERO,
1392 		.extra2		= &one_hundred,
1393 	},
1394 	{
1395 		.procname	= "dirty_background_bytes",
1396 		.data		= &dirty_background_bytes,
1397 		.maxlen		= sizeof(dirty_background_bytes),
1398 		.mode		= 0644,
1399 		.proc_handler	= dirty_background_bytes_handler,
1400 		.extra1		= &one_ul,
1401 	},
1402 	{
1403 		.procname	= "dirty_ratio",
1404 		.data		= &vm_dirty_ratio,
1405 		.maxlen		= sizeof(vm_dirty_ratio),
1406 		.mode		= 0644,
1407 		.proc_handler	= dirty_ratio_handler,
1408 		.extra1		= SYSCTL_ZERO,
1409 		.extra2		= &one_hundred,
1410 	},
1411 	{
1412 		.procname	= "dirty_bytes",
1413 		.data		= &vm_dirty_bytes,
1414 		.maxlen		= sizeof(vm_dirty_bytes),
1415 		.mode		= 0644,
1416 		.proc_handler	= dirty_bytes_handler,
1417 		.extra1		= &dirty_bytes_min,
1418 	},
1419 	{
1420 		.procname	= "dirty_writeback_centisecs",
1421 		.data		= &dirty_writeback_interval,
1422 		.maxlen		= sizeof(dirty_writeback_interval),
1423 		.mode		= 0644,
1424 		.proc_handler	= dirty_writeback_centisecs_handler,
1425 	},
1426 	{
1427 		.procname	= "dirty_expire_centisecs",
1428 		.data		= &dirty_expire_interval,
1429 		.maxlen		= sizeof(dirty_expire_interval),
1430 		.mode		= 0644,
1431 		.proc_handler	= proc_dointvec_minmax,
1432 		.extra1		= SYSCTL_ZERO,
1433 	},
1434 	{
1435 		.procname	= "dirtytime_expire_seconds",
1436 		.data		= &dirtytime_expire_interval,
1437 		.maxlen		= sizeof(dirtytime_expire_interval),
1438 		.mode		= 0644,
1439 		.proc_handler	= dirtytime_interval_handler,
1440 		.extra1		= SYSCTL_ZERO,
1441 	},
1442 	{
1443 		.procname	= "swappiness",
1444 		.data		= &vm_swappiness,
1445 		.maxlen		= sizeof(vm_swappiness),
1446 		.mode		= 0644,
1447 		.proc_handler	= proc_dointvec_minmax,
1448 		.extra1		= SYSCTL_ZERO,
1449 		.extra2		= &one_hundred,
1450 	},
1451 #ifdef CONFIG_NUMA
1452 	{
1453 		.procname	= "numa_stat",
1454 		.data		= &sysctl_vm_numa_stat,
1455 		.maxlen		= sizeof(int),
1456 		.mode		= 0644,
1457 		.proc_handler	= sysctl_vm_numa_stat_handler,
1458 		.extra1		= SYSCTL_ZERO,
1459 		.extra2		= SYSCTL_ONE,
1460 	},
1461 #endif
1462 #ifdef CONFIG_HUGETLB_PAGE
1463 	{
1464 		.procname	= "nr_hugepages",
1465 		.data		= NULL,
1466 		.maxlen		= sizeof(unsigned long),
1467 		.mode		= 0644,
1468 		.proc_handler	= hugetlb_sysctl_handler,
1469 	},
1470 #ifdef CONFIG_NUMA
1471 	{
1472 		.procname       = "nr_hugepages_mempolicy",
1473 		.data           = NULL,
1474 		.maxlen         = sizeof(unsigned long),
1475 		.mode           = 0644,
1476 		.proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1477 	},
1478 #endif
1479 	 {
1480 		.procname	= "hugetlb_shm_group",
1481 		.data		= &sysctl_hugetlb_shm_group,
1482 		.maxlen		= sizeof(gid_t),
1483 		.mode		= 0644,
1484 		.proc_handler	= proc_dointvec,
1485 	 },
1486 	{
1487 		.procname	= "nr_overcommit_hugepages",
1488 		.data		= NULL,
1489 		.maxlen		= sizeof(unsigned long),
1490 		.mode		= 0644,
1491 		.proc_handler	= hugetlb_overcommit_handler,
1492 	},
1493 #endif
1494 	{
1495 		.procname	= "lowmem_reserve_ratio",
1496 		.data		= &sysctl_lowmem_reserve_ratio,
1497 		.maxlen		= sizeof(sysctl_lowmem_reserve_ratio),
1498 		.mode		= 0644,
1499 		.proc_handler	= lowmem_reserve_ratio_sysctl_handler,
1500 	},
1501 	{
1502 		.procname	= "drop_caches",
1503 		.data		= &sysctl_drop_caches,
1504 		.maxlen		= sizeof(int),
1505 		.mode		= 0200,
1506 		.proc_handler	= drop_caches_sysctl_handler,
1507 		.extra1		= SYSCTL_ONE,
1508 		.extra2		= &four,
1509 	},
1510 #ifdef CONFIG_COMPACTION
1511 	{
1512 		.procname	= "compact_memory",
1513 		.data		= &sysctl_compact_memory,
1514 		.maxlen		= sizeof(int),
1515 		.mode		= 0200,
1516 		.proc_handler	= sysctl_compaction_handler,
1517 	},
1518 	{
1519 		.procname	= "extfrag_threshold",
1520 		.data		= &sysctl_extfrag_threshold,
1521 		.maxlen		= sizeof(int),
1522 		.mode		= 0644,
1523 		.proc_handler	= proc_dointvec_minmax,
1524 		.extra1		= &min_extfrag_threshold,
1525 		.extra2		= &max_extfrag_threshold,
1526 	},
1527 	{
1528 		.procname	= "compact_unevictable_allowed",
1529 		.data		= &sysctl_compact_unevictable_allowed,
1530 		.maxlen		= sizeof(int),
1531 		.mode		= 0644,
1532 		.proc_handler	= proc_dointvec,
1533 		.extra1		= SYSCTL_ZERO,
1534 		.extra2		= SYSCTL_ONE,
1535 	},
1536 
1537 #endif /* CONFIG_COMPACTION */
1538 	{
1539 		.procname	= "min_free_kbytes",
1540 		.data		= &min_free_kbytes,
1541 		.maxlen		= sizeof(min_free_kbytes),
1542 		.mode		= 0644,
1543 		.proc_handler	= min_free_kbytes_sysctl_handler,
1544 		.extra1		= SYSCTL_ZERO,
1545 	},
1546 	{
1547 		.procname	= "watermark_boost_factor",
1548 		.data		= &watermark_boost_factor,
1549 		.maxlen		= sizeof(watermark_boost_factor),
1550 		.mode		= 0644,
1551 		.proc_handler	= watermark_boost_factor_sysctl_handler,
1552 		.extra1		= SYSCTL_ZERO,
1553 	},
1554 	{
1555 		.procname	= "watermark_scale_factor",
1556 		.data		= &watermark_scale_factor,
1557 		.maxlen		= sizeof(watermark_scale_factor),
1558 		.mode		= 0644,
1559 		.proc_handler	= watermark_scale_factor_sysctl_handler,
1560 		.extra1		= SYSCTL_ONE,
1561 		.extra2		= &one_thousand,
1562 	},
1563 	{
1564 		.procname	= "extra_free_kbytes",
1565 		.data		= &extra_free_kbytes,
1566 		.maxlen		= sizeof(extra_free_kbytes),
1567 		.mode		= 0644,
1568 		.proc_handler	= min_free_kbytes_sysctl_handler,
1569 		.extra1		= SYSCTL_ZERO,
1570 	},
1571 	{
1572 		.procname	= "percpu_pagelist_fraction",
1573 		.data		= &percpu_pagelist_fraction,
1574 		.maxlen		= sizeof(percpu_pagelist_fraction),
1575 		.mode		= 0644,
1576 		.proc_handler	= percpu_pagelist_fraction_sysctl_handler,
1577 		.extra1		= SYSCTL_ZERO,
1578 	},
1579 	{
1580 		.procname	= "page_lock_unfairness",
1581 		.data		= &sysctl_page_lock_unfairness,
1582 		.maxlen		= sizeof(sysctl_page_lock_unfairness),
1583 		.mode		= 0644,
1584 		.proc_handler	= proc_dointvec_minmax,
1585 		.extra1		= SYSCTL_ZERO,
1586 	},
1587 #ifdef CONFIG_MMU
1588 	{
1589 		.procname	= "max_map_count",
1590 		.data		= &sysctl_max_map_count,
1591 		.maxlen		= sizeof(sysctl_max_map_count),
1592 		.mode		= 0644,
1593 		.proc_handler	= proc_dointvec_minmax,
1594 		.extra1		= SYSCTL_ZERO,
1595 	},
1596 #else
1597 	{
1598 		.procname	= "nr_trim_pages",
1599 		.data		= &sysctl_nr_trim_pages,
1600 		.maxlen		= sizeof(sysctl_nr_trim_pages),
1601 		.mode		= 0644,
1602 		.proc_handler	= proc_dointvec_minmax,
1603 		.extra1		= SYSCTL_ZERO,
1604 	},
1605 #endif
1606 	{
1607 		.procname	= "laptop_mode",
1608 		.data		= &laptop_mode,
1609 		.maxlen		= sizeof(laptop_mode),
1610 		.mode		= 0644,
1611 		.proc_handler	= proc_dointvec_jiffies,
1612 	},
1613 	{
1614 		.procname	= "block_dump",
1615 		.data		= &block_dump,
1616 		.maxlen		= sizeof(block_dump),
1617 		.mode		= 0644,
1618 		.proc_handler	= proc_dointvec_minmax,
1619 		.extra1		= SYSCTL_ZERO,
1620 	},
1621 	{
1622 		.procname	= "vfs_cache_pressure",
1623 		.data		= &sysctl_vfs_cache_pressure,
1624 		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
1625 		.mode		= 0644,
1626 		.proc_handler	= proc_dointvec_minmax,
1627 		.extra1		= SYSCTL_ZERO,
1628 	},
1629 #if defined(HAVE_ARCH_PICK_MMAP_LAYOUT) || \
1630     defined(CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT)
1631 	{
1632 		.procname	= "legacy_va_layout",
1633 		.data		= &sysctl_legacy_va_layout,
1634 		.maxlen		= sizeof(sysctl_legacy_va_layout),
1635 		.mode		= 0644,
1636 		.proc_handler	= proc_dointvec_minmax,
1637 		.extra1		= SYSCTL_ZERO,
1638 	},
1639 #endif
1640 #ifdef CONFIG_NUMA
1641 	{
1642 		.procname	= "zone_reclaim_mode",
1643 		.data		= &node_reclaim_mode,
1644 		.maxlen		= sizeof(node_reclaim_mode),
1645 		.mode		= 0644,
1646 		.proc_handler	= proc_dointvec_minmax,
1647 		.extra1		= SYSCTL_ZERO,
1648 	},
1649 	{
1650 		.procname	= "min_unmapped_ratio",
1651 		.data		= &sysctl_min_unmapped_ratio,
1652 		.maxlen		= sizeof(sysctl_min_unmapped_ratio),
1653 		.mode		= 0644,
1654 		.proc_handler	= sysctl_min_unmapped_ratio_sysctl_handler,
1655 		.extra1		= SYSCTL_ZERO,
1656 		.extra2		= &one_hundred,
1657 	},
1658 	{
1659 		.procname	= "min_slab_ratio",
1660 		.data		= &sysctl_min_slab_ratio,
1661 		.maxlen		= sizeof(sysctl_min_slab_ratio),
1662 		.mode		= 0644,
1663 		.proc_handler	= sysctl_min_slab_ratio_sysctl_handler,
1664 		.extra1		= SYSCTL_ZERO,
1665 		.extra2		= &one_hundred,
1666 	},
1667 #endif
1668 #ifdef CONFIG_SMP
1669 	{
1670 		.procname	= "stat_interval",
1671 		.data		= &sysctl_stat_interval,
1672 		.maxlen		= sizeof(sysctl_stat_interval),
1673 		.mode		= 0644,
1674 		.proc_handler	= proc_dointvec_jiffies,
1675 	},
1676 	{
1677 		.procname	= "stat_refresh",
1678 		.data		= NULL,
1679 		.maxlen		= 0,
1680 		.mode		= 0600,
1681 		.proc_handler	= vmstat_refresh,
1682 	},
1683 #endif
1684 #ifdef CONFIG_MMU
1685 	{
1686 		.procname	= "mmap_min_addr",
1687 		.data		= &dac_mmap_min_addr,
1688 		.maxlen		= sizeof(unsigned long),
1689 		.mode		= 0644,
1690 		.proc_handler	= mmap_min_addr_handler,
1691 	},
1692 #endif
1693 #ifdef CONFIG_NUMA
1694 	{
1695 		.procname	= "numa_zonelist_order",
1696 		.data		= &numa_zonelist_order,
1697 		.maxlen		= NUMA_ZONELIST_ORDER_LEN,
1698 		.mode		= 0644,
1699 		.proc_handler	= numa_zonelist_order_handler,
1700 	},
1701 #endif
1702 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1703    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1704 	{
1705 		.procname	= "vdso_enabled",
1706 #ifdef CONFIG_X86_32
1707 		.data		= &vdso32_enabled,
1708 		.maxlen		= sizeof(vdso32_enabled),
1709 #else
1710 		.data		= &vdso_enabled,
1711 		.maxlen		= sizeof(vdso_enabled),
1712 #endif
1713 		.mode		= 0644,
1714 		.proc_handler	= proc_dointvec,
1715 		.extra1		= SYSCTL_ZERO,
1716 	},
1717 #endif
1718 #ifdef CONFIG_HIGHMEM
1719 	{
1720 		.procname	= "highmem_is_dirtyable",
1721 		.data		= &vm_highmem_is_dirtyable,
1722 		.maxlen		= sizeof(vm_highmem_is_dirtyable),
1723 		.mode		= 0644,
1724 		.proc_handler	= proc_dointvec_minmax,
1725 		.extra1		= SYSCTL_ZERO,
1726 		.extra2		= SYSCTL_ONE,
1727 	},
1728 #endif
1729 #ifdef CONFIG_MEMORY_FAILURE
1730 	{
1731 		.procname	= "memory_failure_early_kill",
1732 		.data		= &sysctl_memory_failure_early_kill,
1733 		.maxlen		= sizeof(sysctl_memory_failure_early_kill),
1734 		.mode		= 0644,
1735 		.proc_handler	= proc_dointvec_minmax,
1736 		.extra1		= SYSCTL_ZERO,
1737 		.extra2		= SYSCTL_ONE,
1738 	},
1739 	{
1740 		.procname	= "memory_failure_recovery",
1741 		.data		= &sysctl_memory_failure_recovery,
1742 		.maxlen		= sizeof(sysctl_memory_failure_recovery),
1743 		.mode		= 0644,
1744 		.proc_handler	= proc_dointvec_minmax,
1745 		.extra1		= SYSCTL_ZERO,
1746 		.extra2		= SYSCTL_ONE,
1747 	},
1748 #endif
1749 	{
1750 		.procname	= "user_reserve_kbytes",
1751 		.data		= &sysctl_user_reserve_kbytes,
1752 		.maxlen		= sizeof(sysctl_user_reserve_kbytes),
1753 		.mode		= 0644,
1754 		.proc_handler	= proc_doulongvec_minmax,
1755 	},
1756 	{
1757 		.procname	= "admin_reserve_kbytes",
1758 		.data		= &sysctl_admin_reserve_kbytes,
1759 		.maxlen		= sizeof(sysctl_admin_reserve_kbytes),
1760 		.mode		= 0644,
1761 		.proc_handler	= proc_doulongvec_minmax,
1762 	},
1763 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
1764 	{
1765 		.procname	= "mmap_rnd_bits",
1766 		.data		= &mmap_rnd_bits,
1767 		.maxlen		= sizeof(mmap_rnd_bits),
1768 		.mode		= 0600,
1769 		.proc_handler	= proc_dointvec_minmax,
1770 		.extra1		= (void *)&mmap_rnd_bits_min,
1771 		.extra2		= (void *)&mmap_rnd_bits_max,
1772 	},
1773 #endif
1774 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
1775 	{
1776 		.procname	= "mmap_rnd_compat_bits",
1777 		.data		= &mmap_rnd_compat_bits,
1778 		.maxlen		= sizeof(mmap_rnd_compat_bits),
1779 		.mode		= 0600,
1780 		.proc_handler	= proc_dointvec_minmax,
1781 		.extra1		= (void *)&mmap_rnd_compat_bits_min,
1782 		.extra2		= (void *)&mmap_rnd_compat_bits_max,
1783 	},
1784 #endif
1785 #ifdef CONFIG_USERFAULTFD
1786 	{
1787 		.procname	= "unprivileged_userfaultfd",
1788 		.data		= &sysctl_unprivileged_userfaultfd,
1789 		.maxlen		= sizeof(sysctl_unprivileged_userfaultfd),
1790 		.mode		= 0644,
1791 		.proc_handler	= proc_dointvec_minmax,
1792 		.extra1		= SYSCTL_ZERO,
1793 		.extra2		= SYSCTL_ONE,
1794 	},
1795 #endif
1796 	{ }
1797 };
1798 
1799 static struct ctl_table fs_table[] = {
1800 	{
1801 		.procname	= "inode-nr",
1802 		.data		= &inodes_stat,
1803 		.maxlen		= 2*sizeof(long),
1804 		.mode		= 0444,
1805 		.proc_handler	= proc_nr_inodes,
1806 	},
1807 	{
1808 		.procname	= "inode-state",
1809 		.data		= &inodes_stat,
1810 		.maxlen		= 7*sizeof(long),
1811 		.mode		= 0444,
1812 		.proc_handler	= proc_nr_inodes,
1813 	},
1814 	{
1815 		.procname	= "file-nr",
1816 		.data		= &files_stat,
1817 		.maxlen		= sizeof(files_stat),
1818 		.mode		= 0444,
1819 		.proc_handler	= proc_nr_files,
1820 	},
1821 	{
1822 		.procname	= "file-max",
1823 		.data		= &files_stat.max_files,
1824 		.maxlen		= sizeof(files_stat.max_files),
1825 		.mode		= 0644,
1826 		.proc_handler	= proc_doulongvec_minmax,
1827 		.extra1		= &zero_ul,
1828 		.extra2		= &long_max,
1829 	},
1830 	{
1831 		.procname	= "nr_open",
1832 		.data		= &sysctl_nr_open,
1833 		.maxlen		= sizeof(unsigned int),
1834 		.mode		= 0644,
1835 		.proc_handler	= proc_dointvec_minmax,
1836 		.extra1		= &sysctl_nr_open_min,
1837 		.extra2		= &sysctl_nr_open_max,
1838 	},
1839 	{
1840 		.procname	= "dentry-state",
1841 		.data		= &dentry_stat,
1842 		.maxlen		= 6*sizeof(long),
1843 		.mode		= 0444,
1844 		.proc_handler	= proc_nr_dentry,
1845 	},
1846 	{
1847 		.procname	= "overflowuid",
1848 		.data		= &fs_overflowuid,
1849 		.maxlen		= sizeof(int),
1850 		.mode		= 0644,
1851 		.proc_handler	= proc_dointvec_minmax,
1852 		.extra1		= &minolduid,
1853 		.extra2		= &maxolduid,
1854 	},
1855 	{
1856 		.procname	= "overflowgid",
1857 		.data		= &fs_overflowgid,
1858 		.maxlen		= sizeof(int),
1859 		.mode		= 0644,
1860 		.proc_handler	= proc_dointvec_minmax,
1861 		.extra1		= &minolduid,
1862 		.extra2		= &maxolduid,
1863 	},
1864 #ifdef CONFIG_FILE_LOCKING
1865 	{
1866 		.procname	= "leases-enable",
1867 		.data		= &leases_enable,
1868 		.maxlen		= sizeof(int),
1869 		.mode		= 0644,
1870 		.proc_handler	= proc_dointvec,
1871 	},
1872 #endif
1873 #ifdef CONFIG_DNOTIFY
1874 	{
1875 		.procname	= "dir-notify-enable",
1876 		.data		= &dir_notify_enable,
1877 		.maxlen		= sizeof(int),
1878 		.mode		= 0644,
1879 		.proc_handler	= proc_dointvec,
1880 	},
1881 #endif
1882 #ifdef CONFIG_MMU
1883 #ifdef CONFIG_FILE_LOCKING
1884 	{
1885 		.procname	= "lease-break-time",
1886 		.data		= &lease_break_time,
1887 		.maxlen		= sizeof(int),
1888 		.mode		= 0644,
1889 		.proc_handler	= proc_dointvec,
1890 	},
1891 #endif
1892 #ifdef CONFIG_AIO
1893 	{
1894 		.procname	= "aio-nr",
1895 		.data		= &aio_nr,
1896 		.maxlen		= sizeof(aio_nr),
1897 		.mode		= 0444,
1898 		.proc_handler	= proc_doulongvec_minmax,
1899 	},
1900 	{
1901 		.procname	= "aio-max-nr",
1902 		.data		= &aio_max_nr,
1903 		.maxlen		= sizeof(aio_max_nr),
1904 		.mode		= 0644,
1905 		.proc_handler	= proc_doulongvec_minmax,
1906 	},
1907 #endif /* CONFIG_AIO */
1908 #ifdef CONFIG_INOTIFY_USER
1909 	{
1910 		.procname	= "inotify",
1911 		.mode		= 0555,
1912 		.child		= inotify_table,
1913 	},
1914 #endif
1915 #ifdef CONFIG_EPOLL
1916 	{
1917 		.procname	= "epoll",
1918 		.mode		= 0555,
1919 		.child		= epoll_table,
1920 	},
1921 #endif
1922 #endif
1923 	{
1924 		.procname	= "protected_symlinks",
1925 		.data		= &sysctl_protected_symlinks,
1926 		.maxlen		= sizeof(int),
1927 		.mode		= 0600,
1928 		.proc_handler	= proc_dointvec_minmax,
1929 		.extra1		= SYSCTL_ZERO,
1930 		.extra2		= SYSCTL_ONE,
1931 	},
1932 	{
1933 		.procname	= "protected_hardlinks",
1934 		.data		= &sysctl_protected_hardlinks,
1935 		.maxlen		= sizeof(int),
1936 		.mode		= 0600,
1937 		.proc_handler	= proc_dointvec_minmax,
1938 		.extra1		= SYSCTL_ZERO,
1939 		.extra2		= SYSCTL_ONE,
1940 	},
1941 	{
1942 		.procname	= "protected_fifos",
1943 		.data		= &sysctl_protected_fifos,
1944 		.maxlen		= sizeof(int),
1945 		.mode		= 0600,
1946 		.proc_handler	= proc_dointvec_minmax,
1947 		.extra1		= SYSCTL_ZERO,
1948 		.extra2		= &two,
1949 	},
1950 	{
1951 		.procname	= "protected_regular",
1952 		.data		= &sysctl_protected_regular,
1953 		.maxlen		= sizeof(int),
1954 		.mode		= 0600,
1955 		.proc_handler	= proc_dointvec_minmax,
1956 		.extra1		= SYSCTL_ZERO,
1957 		.extra2		= &two,
1958 	},
1959 	{
1960 		.procname	= "suid_dumpable",
1961 		.data		= &suid_dumpable,
1962 		.maxlen		= sizeof(int),
1963 		.mode		= 0644,
1964 		.proc_handler	= proc_dointvec_minmax_coredump,
1965 		.extra1		= SYSCTL_ZERO,
1966 		.extra2		= &two,
1967 	},
1968 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1969 	{
1970 		.procname	= "binfmt_misc",
1971 		.mode		= 0555,
1972 		.child		= sysctl_mount_point,
1973 	},
1974 #endif
1975 	{
1976 		.procname	= "pipe-max-size",
1977 		.data		= &pipe_max_size,
1978 		.maxlen		= sizeof(pipe_max_size),
1979 		.mode		= 0644,
1980 		.proc_handler	= proc_dopipe_max_size,
1981 	},
1982 	{
1983 		.procname	= "pipe-user-pages-hard",
1984 		.data		= &pipe_user_pages_hard,
1985 		.maxlen		= sizeof(pipe_user_pages_hard),
1986 		.mode		= 0644,
1987 		.proc_handler	= proc_doulongvec_minmax,
1988 	},
1989 	{
1990 		.procname	= "pipe-user-pages-soft",
1991 		.data		= &pipe_user_pages_soft,
1992 		.maxlen		= sizeof(pipe_user_pages_soft),
1993 		.mode		= 0644,
1994 		.proc_handler	= proc_doulongvec_minmax,
1995 	},
1996 	{
1997 		.procname	= "mount-max",
1998 		.data		= &sysctl_mount_max,
1999 		.maxlen		= sizeof(unsigned int),
2000 		.mode		= 0644,
2001 		.proc_handler	= proc_dointvec_minmax,
2002 		.extra1		= SYSCTL_ONE,
2003 	},
2004 	{ }
2005 };
2006 
2007 static struct ctl_table debug_table[] = {
2008 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
2009 	{
2010 		.procname	= "exception-trace",
2011 		.data		= &show_unhandled_signals,
2012 		.maxlen		= sizeof(int),
2013 		.mode		= 0644,
2014 		.proc_handler	= proc_dointvec
2015 	},
2016 #endif
2017 #if defined(CONFIG_OPTPROBES)
2018 	{
2019 		.procname	= "kprobes-optimization",
2020 		.data		= &sysctl_kprobes_optimization,
2021 		.maxlen		= sizeof(int),
2022 		.mode		= 0644,
2023 		.proc_handler	= proc_kprobes_optimization_handler,
2024 		.extra1		= SYSCTL_ZERO,
2025 		.extra2		= SYSCTL_ONE,
2026 	},
2027 #endif
2028 	{ }
2029 };
2030 
2031 static struct ctl_table dev_table[] = {
2032 	{ }
2033 };
2034 
sysctl_init(void)2035 int __init sysctl_init(void)
2036 {
2037 	struct ctl_table_header *hdr;
2038 
2039 	hdr = register_sysctl_table(sysctl_base_table);
2040 	kmemleak_not_leak(hdr);
2041 	return 0;
2042 }
2043 
2044 #endif /* CONFIG_SYSCTL */
2045 
2046 /*
2047  * /proc/sys support
2048  */
2049 
2050 #ifdef CONFIG_PROC_SYSCTL
2051 
_proc_do_string(char * data,int maxlen,int write,char __user * buffer,size_t * lenp,loff_t * ppos)2052 static int _proc_do_string(char *data, int maxlen, int write,
2053 			   char __user *buffer,
2054 			   size_t *lenp, loff_t *ppos)
2055 {
2056 	size_t len;
2057 	char __user *p;
2058 	char c;
2059 
2060 	if (!data || !maxlen || !*lenp) {
2061 		*lenp = 0;
2062 		return 0;
2063 	}
2064 
2065 	if (write) {
2066 		if (sysctl_writes_strict == SYSCTL_WRITES_STRICT) {
2067 			/* Only continue writes not past the end of buffer. */
2068 			len = strlen(data);
2069 			if (len > maxlen - 1)
2070 				len = maxlen - 1;
2071 
2072 			if (*ppos > len)
2073 				return 0;
2074 			len = *ppos;
2075 		} else {
2076 			/* Start writing from beginning of buffer. */
2077 			len = 0;
2078 		}
2079 
2080 		*ppos += *lenp;
2081 		p = buffer;
2082 		while ((p - buffer) < *lenp && len < maxlen - 1) {
2083 			if (get_user(c, p++))
2084 				return -EFAULT;
2085 			if (c == 0 || c == '\n')
2086 				break;
2087 			data[len++] = c;
2088 		}
2089 		data[len] = 0;
2090 	} else {
2091 		len = strlen(data);
2092 		if (len > maxlen)
2093 			len = maxlen;
2094 
2095 		if (*ppos > len) {
2096 			*lenp = 0;
2097 			return 0;
2098 		}
2099 
2100 		data += *ppos;
2101 		len  -= *ppos;
2102 
2103 		if (len > *lenp)
2104 			len = *lenp;
2105 		if (len)
2106 			if (copy_to_user(buffer, data, len))
2107 				return -EFAULT;
2108 		if (len < *lenp) {
2109 			if (put_user('\n', buffer + len))
2110 				return -EFAULT;
2111 			len++;
2112 		}
2113 		*lenp = len;
2114 		*ppos += len;
2115 	}
2116 	return 0;
2117 }
2118 
warn_sysctl_write(struct ctl_table * table)2119 static void warn_sysctl_write(struct ctl_table *table)
2120 {
2121 	pr_warn_once("%s wrote to %s when file position was not 0!\n"
2122 		"This will not be supported in the future. To silence this\n"
2123 		"warning, set kernel.sysctl_writes_strict = -1\n",
2124 		current->comm, table->procname);
2125 }
2126 
2127 /**
2128  * proc_first_pos_non_zero_ignore - check if first position is allowed
2129  * @ppos: file position
2130  * @table: the sysctl table
2131  *
2132  * Returns true if the first position is non-zero and the sysctl_writes_strict
2133  * mode indicates this is not allowed for numeric input types. String proc
2134  * handlers can ignore the return value.
2135  */
proc_first_pos_non_zero_ignore(loff_t * ppos,struct ctl_table * table)2136 static bool proc_first_pos_non_zero_ignore(loff_t *ppos,
2137 					   struct ctl_table *table)
2138 {
2139 	if (!*ppos)
2140 		return false;
2141 
2142 	switch (sysctl_writes_strict) {
2143 	case SYSCTL_WRITES_STRICT:
2144 		return true;
2145 	case SYSCTL_WRITES_WARN:
2146 		warn_sysctl_write(table);
2147 		return false;
2148 	default:
2149 		return false;
2150 	}
2151 }
2152 
2153 /**
2154  * proc_dostring - read a string sysctl
2155  * @table: the sysctl table
2156  * @write: %TRUE if this is a write to the sysctl file
2157  * @buffer: the user buffer
2158  * @lenp: the size of the user buffer
2159  * @ppos: file position
2160  *
2161  * Reads/writes a string from/to the user buffer. If the kernel
2162  * buffer provided is not large enough to hold the string, the
2163  * string is truncated. The copied string is %NULL-terminated.
2164  * If the string is being read by the user process, it is copied
2165  * and a newline '\n' is added. It is truncated if the buffer is
2166  * not large enough.
2167  *
2168  * Returns 0 on success.
2169  */
proc_dostring(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)2170 int proc_dostring(struct ctl_table *table, int write,
2171 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2172 {
2173 	if (write)
2174 		proc_first_pos_non_zero_ignore(ppos, table);
2175 
2176 	return _proc_do_string((char *)(table->data), table->maxlen, write,
2177 			       (char __user *)buffer, lenp, ppos);
2178 }
2179 
proc_skip_spaces(char ** buf,size_t * size)2180 static void proc_skip_spaces(char **buf, size_t *size)
2181 {
2182 	while (*size) {
2183 		if (!isspace(**buf))
2184 			break;
2185 		(*size)--;
2186 		(*buf)++;
2187 	}
2188 }
2189 
proc_skip_char(char ** buf,size_t * size,const char v)2190 static void proc_skip_char(char **buf, size_t *size, const char v)
2191 {
2192 	while (*size) {
2193 		if (**buf != v)
2194 			break;
2195 		(*size)--;
2196 		(*buf)++;
2197 	}
2198 }
2199 
2200 /**
2201  * strtoul_lenient - parse an ASCII formatted integer from a buffer and only
2202  *                   fail on overflow
2203  *
2204  * @cp: kernel buffer containing the string to parse
2205  * @endp: pointer to store the trailing characters
2206  * @base: the base to use
2207  * @res: where the parsed integer will be stored
2208  *
2209  * In case of success 0 is returned and @res will contain the parsed integer,
2210  * @endp will hold any trailing characters.
2211  * This function will fail the parse on overflow. If there wasn't an overflow
2212  * the function will defer the decision what characters count as invalid to the
2213  * caller.
2214  */
strtoul_lenient(const char * cp,char ** endp,unsigned int base,unsigned long * res)2215 static int strtoul_lenient(const char *cp, char **endp, unsigned int base,
2216 			   unsigned long *res)
2217 {
2218 	unsigned long long result;
2219 	unsigned int rv;
2220 
2221 	cp = _parse_integer_fixup_radix(cp, &base);
2222 	rv = _parse_integer(cp, base, &result);
2223 	if ((rv & KSTRTOX_OVERFLOW) || (result != (unsigned long)result))
2224 		return -ERANGE;
2225 
2226 	cp += rv;
2227 
2228 	if (endp)
2229 		*endp = (char *)cp;
2230 
2231 	*res = (unsigned long)result;
2232 	return 0;
2233 }
2234 
2235 #define TMPBUFLEN 22
2236 /**
2237  * proc_get_long - reads an ASCII formatted integer from a user buffer
2238  *
2239  * @buf: a kernel buffer
2240  * @size: size of the kernel buffer
2241  * @val: this is where the number will be stored
2242  * @neg: set to %TRUE if number is negative
2243  * @perm_tr: a vector which contains the allowed trailers
2244  * @perm_tr_len: size of the perm_tr vector
2245  * @tr: pointer to store the trailer character
2246  *
2247  * In case of success %0 is returned and @buf and @size are updated with
2248  * the amount of bytes read. If @tr is non-NULL and a trailing
2249  * character exists (size is non-zero after returning from this
2250  * function), @tr is updated with the trailing character.
2251  */
proc_get_long(char ** buf,size_t * size,unsigned long * val,bool * neg,const char * perm_tr,unsigned perm_tr_len,char * tr)2252 static int proc_get_long(char **buf, size_t *size,
2253 			  unsigned long *val, bool *neg,
2254 			  const char *perm_tr, unsigned perm_tr_len, char *tr)
2255 {
2256 	char *p, tmp[TMPBUFLEN];
2257 	ssize_t len = *size;
2258 
2259 	if (len <= 0)
2260 		return -EINVAL;
2261 
2262 	if (len > TMPBUFLEN - 1)
2263 		len = TMPBUFLEN - 1;
2264 
2265 	memcpy(tmp, *buf, len);
2266 
2267 	tmp[len] = 0;
2268 	p = tmp;
2269 	if (*p == '-' && *size > 1) {
2270 		*neg = true;
2271 		p++;
2272 	} else
2273 		*neg = false;
2274 	if (!isdigit(*p))
2275 		return -EINVAL;
2276 
2277 	if (strtoul_lenient(p, &p, 0, val))
2278 		return -EINVAL;
2279 
2280 	len = p - tmp;
2281 
2282 	/* We don't know if the next char is whitespace thus we may accept
2283 	 * invalid integers (e.g. 1234...a) or two integers instead of one
2284 	 * (e.g. 123...1). So lets not allow such large numbers. */
2285 	if (len == TMPBUFLEN - 1)
2286 		return -EINVAL;
2287 
2288 	if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2289 		return -EINVAL;
2290 
2291 	if (tr && (len < *size))
2292 		*tr = *p;
2293 
2294 	*buf += len;
2295 	*size -= len;
2296 
2297 	return 0;
2298 }
2299 
2300 /**
2301  * proc_put_long - converts an integer to a decimal ASCII formatted string
2302  *
2303  * @buf: the user buffer
2304  * @size: the size of the user buffer
2305  * @val: the integer to be converted
2306  * @neg: sign of the number, %TRUE for negative
2307  *
2308  * In case of success %0 is returned and @buf and @size are updated with
2309  * the amount of bytes written.
2310  */
proc_put_long(void __user ** buf,size_t * size,unsigned long val,bool neg)2311 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2312 			  bool neg)
2313 {
2314 	int len;
2315 	char tmp[TMPBUFLEN], *p = tmp;
2316 
2317 	sprintf(p, "%s%lu", neg ? "-" : "", val);
2318 	len = strlen(tmp);
2319 	if (len > *size)
2320 		len = *size;
2321 	if (copy_to_user(*buf, tmp, len))
2322 		return -EFAULT;
2323 	*size -= len;
2324 	*buf += len;
2325 	return 0;
2326 }
2327 #undef TMPBUFLEN
2328 
proc_put_char(void __user ** buf,size_t * size,char c)2329 static int proc_put_char(void __user **buf, size_t *size, char c)
2330 {
2331 	if (*size) {
2332 		char __user **buffer = (char __user **)buf;
2333 		if (put_user(c, *buffer))
2334 			return -EFAULT;
2335 		(*size)--, (*buffer)++;
2336 		*buf = *buffer;
2337 	}
2338 	return 0;
2339 }
2340 
do_proc_dointvec_conv(bool * negp,unsigned long * lvalp,int * valp,int write,void * data)2341 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2342 				 int *valp,
2343 				 int write, void *data)
2344 {
2345 	if (write) {
2346 		if (*negp) {
2347 			if (*lvalp > (unsigned long) INT_MAX + 1)
2348 				return -EINVAL;
2349 			*valp = -*lvalp;
2350 		} else {
2351 			if (*lvalp > (unsigned long) INT_MAX)
2352 				return -EINVAL;
2353 			*valp = *lvalp;
2354 		}
2355 	} else {
2356 		int val = *valp;
2357 		if (val < 0) {
2358 			*negp = true;
2359 			*lvalp = -(unsigned long)val;
2360 		} else {
2361 			*negp = false;
2362 			*lvalp = (unsigned long)val;
2363 		}
2364 	}
2365 	return 0;
2366 }
2367 
do_proc_douintvec_conv(unsigned long * lvalp,unsigned int * valp,int write,void * data)2368 static int do_proc_douintvec_conv(unsigned long *lvalp,
2369 				  unsigned int *valp,
2370 				  int write, void *data)
2371 {
2372 	if (write) {
2373 		if (*lvalp > UINT_MAX)
2374 			return -EINVAL;
2375 		*valp = *lvalp;
2376 	} else {
2377 		unsigned int val = *valp;
2378 		*lvalp = (unsigned long)val;
2379 	}
2380 	return 0;
2381 }
2382 
2383 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2384 
__do_proc_dointvec(void * tbl_data,struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos,int (* conv)(bool * negp,unsigned long * lvalp,int * valp,int write,void * data),void * data)2385 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2386 		  int write, void __user *buffer,
2387 		  size_t *lenp, loff_t *ppos,
2388 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2389 			      int write, void *data),
2390 		  void *data)
2391 {
2392 	int *i, vleft, first = 1, err = 0;
2393 	size_t left;
2394 	char *kbuf = NULL, *p;
2395 
2396 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2397 		*lenp = 0;
2398 		return 0;
2399 	}
2400 
2401 	i = (int *) tbl_data;
2402 	vleft = table->maxlen / sizeof(*i);
2403 	left = *lenp;
2404 
2405 	if (!conv)
2406 		conv = do_proc_dointvec_conv;
2407 
2408 	if (write) {
2409 		if (proc_first_pos_non_zero_ignore(ppos, table))
2410 			goto out;
2411 
2412 		if (left > PAGE_SIZE - 1)
2413 			left = PAGE_SIZE - 1;
2414 		p = kbuf = memdup_user_nul(buffer, left);
2415 		if (IS_ERR(kbuf))
2416 			return PTR_ERR(kbuf);
2417 	}
2418 
2419 	for (; left && vleft--; i++, first=0) {
2420 		unsigned long lval;
2421 		bool neg;
2422 
2423 		if (write) {
2424 			proc_skip_spaces(&p, &left);
2425 
2426 			if (!left)
2427 				break;
2428 			err = proc_get_long(&p, &left, &lval, &neg,
2429 					     proc_wspace_sep,
2430 					     sizeof(proc_wspace_sep), NULL);
2431 			if (err)
2432 				break;
2433 			if (conv(&neg, &lval, i, 1, data)) {
2434 				err = -EINVAL;
2435 				break;
2436 			}
2437 		} else {
2438 			if (conv(&neg, &lval, i, 0, data)) {
2439 				err = -EINVAL;
2440 				break;
2441 			}
2442 			if (!first)
2443 				err = proc_put_char(&buffer, &left, '\t');
2444 			if (err)
2445 				break;
2446 			err = proc_put_long(&buffer, &left, lval, neg);
2447 			if (err)
2448 				break;
2449 		}
2450 	}
2451 
2452 	if (!write && !first && left && !err)
2453 		err = proc_put_char(&buffer, &left, '\n');
2454 	if (write && !err && left)
2455 		proc_skip_spaces(&p, &left);
2456 	if (write) {
2457 		kfree(kbuf);
2458 		if (first)
2459 			return err ? : -EINVAL;
2460 	}
2461 	*lenp -= left;
2462 out:
2463 	*ppos += *lenp;
2464 	return err;
2465 }
2466 
do_proc_dointvec(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos,int (* conv)(bool * negp,unsigned long * lvalp,int * valp,int write,void * data),void * data)2467 static int do_proc_dointvec(struct ctl_table *table, int write,
2468 		  void __user *buffer, size_t *lenp, loff_t *ppos,
2469 		  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2470 			      int write, void *data),
2471 		  void *data)
2472 {
2473 	return __do_proc_dointvec(table->data, table, write,
2474 			buffer, lenp, ppos, conv, data);
2475 }
2476 
do_proc_douintvec_w(unsigned int * tbl_data,struct ctl_table * table,void __user * buffer,size_t * lenp,loff_t * ppos,int (* conv)(unsigned long * lvalp,unsigned int * valp,int write,void * data),void * data)2477 static int do_proc_douintvec_w(unsigned int *tbl_data,
2478 			       struct ctl_table *table,
2479 			       void __user *buffer,
2480 			       size_t *lenp, loff_t *ppos,
2481 			       int (*conv)(unsigned long *lvalp,
2482 					   unsigned int *valp,
2483 					   int write, void *data),
2484 			       void *data)
2485 {
2486 	unsigned long lval;
2487 	int err = 0;
2488 	size_t left;
2489 	bool neg;
2490 	char *kbuf = NULL, *p;
2491 
2492 	left = *lenp;
2493 
2494 	if (proc_first_pos_non_zero_ignore(ppos, table))
2495 		goto bail_early;
2496 
2497 	if (left > PAGE_SIZE - 1)
2498 		left = PAGE_SIZE - 1;
2499 
2500 	p = kbuf = memdup_user_nul(buffer, left);
2501 	if (IS_ERR(kbuf))
2502 		return -EINVAL;
2503 
2504 	proc_skip_spaces(&p, &left);
2505 	if (!left) {
2506 		err = -EINVAL;
2507 		goto out_free;
2508 	}
2509 
2510 	err = proc_get_long(&p, &left, &lval, &neg,
2511 			     proc_wspace_sep,
2512 			     sizeof(proc_wspace_sep), NULL);
2513 	if (err || neg) {
2514 		err = -EINVAL;
2515 		goto out_free;
2516 	}
2517 
2518 	if (conv(&lval, tbl_data, 1, data)) {
2519 		err = -EINVAL;
2520 		goto out_free;
2521 	}
2522 
2523 	if (!err && left)
2524 		proc_skip_spaces(&p, &left);
2525 
2526 out_free:
2527 	kfree(kbuf);
2528 	if (err)
2529 		return -EINVAL;
2530 
2531 	return 0;
2532 
2533 	/* This is in keeping with old __do_proc_dointvec() */
2534 bail_early:
2535 	*ppos += *lenp;
2536 	return err;
2537 }
2538 
do_proc_douintvec_r(unsigned int * tbl_data,void __user * buffer,size_t * lenp,loff_t * ppos,int (* conv)(unsigned long * lvalp,unsigned int * valp,int write,void * data),void * data)2539 static int do_proc_douintvec_r(unsigned int *tbl_data, void __user *buffer,
2540 			       size_t *lenp, loff_t *ppos,
2541 			       int (*conv)(unsigned long *lvalp,
2542 					   unsigned int *valp,
2543 					   int write, void *data),
2544 			       void *data)
2545 {
2546 	unsigned long lval;
2547 	int err = 0;
2548 	size_t left;
2549 
2550 	left = *lenp;
2551 
2552 	if (conv(&lval, tbl_data, 0, data)) {
2553 		err = -EINVAL;
2554 		goto out;
2555 	}
2556 
2557 	err = proc_put_long(&buffer, &left, lval, false);
2558 	if (err || !left)
2559 		goto out;
2560 
2561 	err = proc_put_char(&buffer, &left, '\n');
2562 
2563 out:
2564 	*lenp -= left;
2565 	*ppos += *lenp;
2566 
2567 	return err;
2568 }
2569 
__do_proc_douintvec(void * tbl_data,struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos,int (* conv)(unsigned long * lvalp,unsigned int * valp,int write,void * data),void * data)2570 static int __do_proc_douintvec(void *tbl_data, struct ctl_table *table,
2571 			       int write, void __user *buffer,
2572 			       size_t *lenp, loff_t *ppos,
2573 			       int (*conv)(unsigned long *lvalp,
2574 					   unsigned int *valp,
2575 					   int write, void *data),
2576 			       void *data)
2577 {
2578 	unsigned int *i, vleft;
2579 
2580 	if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2581 		*lenp = 0;
2582 		return 0;
2583 	}
2584 
2585 	i = (unsigned int *) tbl_data;
2586 	vleft = table->maxlen / sizeof(*i);
2587 
2588 	/*
2589 	 * Arrays are not supported, keep this simple. *Do not* add
2590 	 * support for them.
2591 	 */
2592 	if (vleft != 1) {
2593 		*lenp = 0;
2594 		return -EINVAL;
2595 	}
2596 
2597 	if (!conv)
2598 		conv = do_proc_douintvec_conv;
2599 
2600 	if (write)
2601 		return do_proc_douintvec_w(i, table, buffer, lenp, ppos,
2602 					   conv, data);
2603 	return do_proc_douintvec_r(i, buffer, lenp, ppos, conv, data);
2604 }
2605 
do_proc_douintvec(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos,int (* conv)(unsigned long * lvalp,unsigned int * valp,int write,void * data),void * data)2606 static int do_proc_douintvec(struct ctl_table *table, int write,
2607 			     void __user *buffer, size_t *lenp, loff_t *ppos,
2608 			     int (*conv)(unsigned long *lvalp,
2609 					 unsigned int *valp,
2610 					 int write, void *data),
2611 			     void *data)
2612 {
2613 	return __do_proc_douintvec(table->data, table, write,
2614 				   buffer, lenp, ppos, conv, data);
2615 }
2616 
2617 /**
2618  * proc_dointvec - read a vector of integers
2619  * @table: the sysctl table
2620  * @write: %TRUE if this is a write to the sysctl file
2621  * @buffer: the user buffer
2622  * @lenp: the size of the user buffer
2623  * @ppos: file position
2624  *
2625  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2626  * values from/to the user buffer, treated as an ASCII string.
2627  *
2628  * Returns 0 on success.
2629  */
proc_dointvec(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)2630 int proc_dointvec(struct ctl_table *table, int write,
2631 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2632 {
2633 	return do_proc_dointvec(table, write, buffer, lenp, ppos, NULL, NULL);
2634 }
2635 
2636 /**
2637  * proc_douintvec - read a vector of unsigned integers
2638  * @table: the sysctl table
2639  * @write: %TRUE if this is a write to the sysctl file
2640  * @buffer: the user buffer
2641  * @lenp: the size of the user buffer
2642  * @ppos: file position
2643  *
2644  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2645  * values from/to the user buffer, treated as an ASCII string.
2646  *
2647  * Returns 0 on success.
2648  */
proc_douintvec(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)2649 int proc_douintvec(struct ctl_table *table, int write,
2650 		     void __user *buffer, size_t *lenp, loff_t *ppos)
2651 {
2652 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
2653 				 do_proc_douintvec_conv, NULL);
2654 }
2655 
2656 /*
2657  * Taint values can only be increased
2658  * This means we can safely use a temporary.
2659  */
proc_taint(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)2660 static int proc_taint(struct ctl_table *table, int write,
2661 			       void __user *buffer, size_t *lenp, loff_t *ppos)
2662 {
2663 	struct ctl_table t;
2664 	unsigned long tmptaint = get_taint();
2665 	int err;
2666 
2667 	if (write && !capable(CAP_SYS_ADMIN))
2668 		return -EPERM;
2669 
2670 	t = *table;
2671 	t.data = &tmptaint;
2672 	err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2673 	if (err < 0)
2674 		return err;
2675 
2676 	if (write) {
2677 		/*
2678 		 * Poor man's atomic or. Not worth adding a primitive
2679 		 * to everyone's atomic.h for this
2680 		 */
2681 		int i;
2682 		for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2683 			if ((tmptaint >> i) & 1)
2684 				add_taint(i, LOCKDEP_STILL_OK);
2685 		}
2686 	}
2687 
2688 	return err;
2689 }
2690 
2691 #ifdef CONFIG_PRINTK
proc_dointvec_minmax_sysadmin(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)2692 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2693 				void __user *buffer, size_t *lenp, loff_t *ppos)
2694 {
2695 	if (write && !capable(CAP_SYS_ADMIN))
2696 		return -EPERM;
2697 
2698 	return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2699 }
2700 #endif
2701 
2702 /**
2703  * struct do_proc_dointvec_minmax_conv_param - proc_dointvec_minmax() range checking structure
2704  * @min: pointer to minimum allowable value
2705  * @max: pointer to maximum allowable value
2706  *
2707  * The do_proc_dointvec_minmax_conv_param structure provides the
2708  * minimum and maximum values for doing range checking for those sysctl
2709  * parameters that use the proc_dointvec_minmax() handler.
2710  */
2711 struct do_proc_dointvec_minmax_conv_param {
2712 	int *min;
2713 	int *max;
2714 };
2715 
do_proc_dointvec_minmax_conv(bool * negp,unsigned long * lvalp,int * valp,int write,void * data)2716 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2717 					int *valp,
2718 					int write, void *data)
2719 {
2720 	int tmp, ret;
2721 	struct do_proc_dointvec_minmax_conv_param *param = data;
2722 	/*
2723 	 * If writing, first do so via a temporary local int so we can
2724 	 * bounds-check it before touching *valp.
2725 	 */
2726 	int *ip = write ? &tmp : valp;
2727 
2728 	ret = do_proc_dointvec_conv(negp, lvalp, ip, write, data);
2729 	if (ret)
2730 		return ret;
2731 
2732 	if (write) {
2733 		if ((param->min && *param->min > tmp) ||
2734 		    (param->max && *param->max < tmp))
2735 			return -EINVAL;
2736 		*valp = tmp;
2737 	}
2738 
2739 	return 0;
2740 }
2741 
2742 /**
2743  * proc_dointvec_minmax - read a vector of integers with min/max values
2744  * @table: the sysctl table
2745  * @write: %TRUE if this is a write to the sysctl file
2746  * @buffer: the user buffer
2747  * @lenp: the size of the user buffer
2748  * @ppos: file position
2749  *
2750  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2751  * values from/to the user buffer, treated as an ASCII string.
2752  *
2753  * This routine will ensure the values are within the range specified by
2754  * table->extra1 (min) and table->extra2 (max).
2755  *
2756  * Returns 0 on success or -EINVAL on write when the range check fails.
2757  */
proc_dointvec_minmax(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)2758 int proc_dointvec_minmax(struct ctl_table *table, int write,
2759 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2760 {
2761 	struct do_proc_dointvec_minmax_conv_param param = {
2762 		.min = (int *) table->extra1,
2763 		.max = (int *) table->extra2,
2764 	};
2765 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
2766 				do_proc_dointvec_minmax_conv, &param);
2767 }
2768 
2769 /**
2770  * struct do_proc_douintvec_minmax_conv_param - proc_douintvec_minmax() range checking structure
2771  * @min: pointer to minimum allowable value
2772  * @max: pointer to maximum allowable value
2773  *
2774  * The do_proc_douintvec_minmax_conv_param structure provides the
2775  * minimum and maximum values for doing range checking for those sysctl
2776  * parameters that use the proc_douintvec_minmax() handler.
2777  */
2778 struct do_proc_douintvec_minmax_conv_param {
2779 	unsigned int *min;
2780 	unsigned int *max;
2781 };
2782 
do_proc_douintvec_minmax_conv(unsigned long * lvalp,unsigned int * valp,int write,void * data)2783 static int do_proc_douintvec_minmax_conv(unsigned long *lvalp,
2784 					 unsigned int *valp,
2785 					 int write, void *data)
2786 {
2787 	int ret;
2788 	unsigned int tmp;
2789 	struct do_proc_douintvec_minmax_conv_param *param = data;
2790 	/* write via temporary local uint for bounds-checking */
2791 	unsigned int *up = write ? &tmp : valp;
2792 
2793 	ret = do_proc_douintvec_conv(lvalp, up, write, data);
2794 	if (ret)
2795 		return ret;
2796 
2797 	if (write) {
2798 		if ((param->min && *param->min > tmp) ||
2799 		    (param->max && *param->max < tmp))
2800 			return -ERANGE;
2801 
2802 		*valp = tmp;
2803 	}
2804 
2805 	return 0;
2806 }
2807 
2808 /**
2809  * proc_douintvec_minmax - read a vector of unsigned ints with min/max values
2810  * @table: the sysctl table
2811  * @write: %TRUE if this is a write to the sysctl file
2812  * @buffer: the user buffer
2813  * @lenp: the size of the user buffer
2814  * @ppos: file position
2815  *
2816  * Reads/writes up to table->maxlen/sizeof(unsigned int) unsigned integer
2817  * values from/to the user buffer, treated as an ASCII string. Negative
2818  * strings are not allowed.
2819  *
2820  * This routine will ensure the values are within the range specified by
2821  * table->extra1 (min) and table->extra2 (max). There is a final sanity
2822  * check for UINT_MAX to avoid having to support wrap around uses from
2823  * userspace.
2824  *
2825  * Returns 0 on success or -ERANGE on write when the range check fails.
2826  */
proc_douintvec_minmax(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)2827 int proc_douintvec_minmax(struct ctl_table *table, int write,
2828 			  void __user *buffer, size_t *lenp, loff_t *ppos)
2829 {
2830 	struct do_proc_douintvec_minmax_conv_param param = {
2831 		.min = (unsigned int *) table->extra1,
2832 		.max = (unsigned int *) table->extra2,
2833 	};
2834 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
2835 				 do_proc_douintvec_minmax_conv, &param);
2836 }
2837 
do_proc_dopipe_max_size_conv(unsigned long * lvalp,unsigned int * valp,int write,void * data)2838 static int do_proc_dopipe_max_size_conv(unsigned long *lvalp,
2839 					unsigned int *valp,
2840 					int write, void *data)
2841 {
2842 	if (write) {
2843 		unsigned int val;
2844 
2845 		val = round_pipe_size(*lvalp);
2846 		if (val == 0)
2847 			return -EINVAL;
2848 
2849 		*valp = val;
2850 	} else {
2851 		unsigned int val = *valp;
2852 		*lvalp = (unsigned long) val;
2853 	}
2854 
2855 	return 0;
2856 }
2857 
proc_dopipe_max_size(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)2858 static int proc_dopipe_max_size(struct ctl_table *table, int write,
2859 				void __user *buffer, size_t *lenp, loff_t *ppos)
2860 {
2861 	return do_proc_douintvec(table, write, buffer, lenp, ppos,
2862 				 do_proc_dopipe_max_size_conv, NULL);
2863 }
2864 
validate_coredump_safety(void)2865 static void validate_coredump_safety(void)
2866 {
2867 #ifdef CONFIG_COREDUMP
2868 	if (suid_dumpable == SUID_DUMP_ROOT &&
2869 	    core_pattern[0] != '/' && core_pattern[0] != '|') {
2870 		printk(KERN_WARNING
2871 "Unsafe core_pattern used with fs.suid_dumpable=2.\n"
2872 "Pipe handler or fully qualified core dump path required.\n"
2873 "Set kernel.core_pattern before fs.suid_dumpable.\n"
2874 		);
2875 	}
2876 #endif
2877 }
2878 
proc_dointvec_minmax_coredump(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)2879 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2880 		void __user *buffer, size_t *lenp, loff_t *ppos)
2881 {
2882 	int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2883 	if (!error)
2884 		validate_coredump_safety();
2885 	return error;
2886 }
2887 
2888 #ifdef CONFIG_COREDUMP
proc_dostring_coredump(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)2889 static int proc_dostring_coredump(struct ctl_table *table, int write,
2890 		  void __user *buffer, size_t *lenp, loff_t *ppos)
2891 {
2892 	int error = proc_dostring(table, write, buffer, lenp, ppos);
2893 	if (!error)
2894 		validate_coredump_safety();
2895 	return error;
2896 }
2897 #endif
2898 
__do_proc_doulongvec_minmax(void * data,struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos,unsigned long convmul,unsigned long convdiv)2899 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2900 				     void __user *buffer,
2901 				     size_t *lenp, loff_t *ppos,
2902 				     unsigned long convmul,
2903 				     unsigned long convdiv)
2904 {
2905 	unsigned long *i, *min, *max;
2906 	int vleft, first = 1, err = 0;
2907 	size_t left;
2908 	char *kbuf = NULL, *p;
2909 
2910 	if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2911 		*lenp = 0;
2912 		return 0;
2913 	}
2914 
2915 	i = (unsigned long *) data;
2916 	min = (unsigned long *) table->extra1;
2917 	max = (unsigned long *) table->extra2;
2918 	vleft = table->maxlen / sizeof(unsigned long);
2919 	left = *lenp;
2920 
2921 	if (write) {
2922 		if (proc_first_pos_non_zero_ignore(ppos, table))
2923 			goto out;
2924 
2925 		if (left > PAGE_SIZE - 1)
2926 			left = PAGE_SIZE - 1;
2927 		p = kbuf = memdup_user_nul(buffer, left);
2928 		if (IS_ERR(kbuf))
2929 			return PTR_ERR(kbuf);
2930 	}
2931 
2932 	for (; left && vleft--; i++, first = 0) {
2933 		unsigned long val;
2934 
2935 		if (write) {
2936 			bool neg;
2937 
2938 			proc_skip_spaces(&p, &left);
2939 			if (!left)
2940 				break;
2941 
2942 			err = proc_get_long(&p, &left, &val, &neg,
2943 					     proc_wspace_sep,
2944 					     sizeof(proc_wspace_sep), NULL);
2945 			if (err)
2946 				break;
2947 			if (neg)
2948 				continue;
2949 			val = convmul * val / convdiv;
2950 			if ((min && val < *min) || (max && val > *max)) {
2951 				err = -EINVAL;
2952 				break;
2953 			}
2954 			*i = val;
2955 		} else {
2956 			val = convdiv * (*i) / convmul;
2957 			if (!first) {
2958 				err = proc_put_char(&buffer, &left, '\t');
2959 				if (err)
2960 					break;
2961 			}
2962 			err = proc_put_long(&buffer, &left, val, false);
2963 			if (err)
2964 				break;
2965 		}
2966 	}
2967 
2968 	if (!write && !first && left && !err)
2969 		err = proc_put_char(&buffer, &left, '\n');
2970 	if (write && !err)
2971 		proc_skip_spaces(&p, &left);
2972 	if (write) {
2973 		kfree(kbuf);
2974 		if (first)
2975 			return err ? : -EINVAL;
2976 	}
2977 	*lenp -= left;
2978 out:
2979 	*ppos += *lenp;
2980 	return err;
2981 }
2982 
do_proc_doulongvec_minmax(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos,unsigned long convmul,unsigned long convdiv)2983 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2984 				     void __user *buffer,
2985 				     size_t *lenp, loff_t *ppos,
2986 				     unsigned long convmul,
2987 				     unsigned long convdiv)
2988 {
2989 	return __do_proc_doulongvec_minmax(table->data, table, write,
2990 			buffer, lenp, ppos, convmul, convdiv);
2991 }
2992 
2993 /**
2994  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2995  * @table: the sysctl table
2996  * @write: %TRUE if this is a write to the sysctl file
2997  * @buffer: the user buffer
2998  * @lenp: the size of the user buffer
2999  * @ppos: file position
3000  *
3001  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
3002  * values from/to the user buffer, treated as an ASCII string.
3003  *
3004  * This routine will ensure the values are within the range specified by
3005  * table->extra1 (min) and table->extra2 (max).
3006  *
3007  * Returns 0 on success.
3008  */
proc_doulongvec_minmax(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3009 int proc_doulongvec_minmax(struct ctl_table *table, int write,
3010 			   void __user *buffer, size_t *lenp, loff_t *ppos)
3011 {
3012     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
3013 }
3014 
3015 /**
3016  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
3017  * @table: the sysctl table
3018  * @write: %TRUE if this is a write to the sysctl file
3019  * @buffer: the user buffer
3020  * @lenp: the size of the user buffer
3021  * @ppos: file position
3022  *
3023  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
3024  * values from/to the user buffer, treated as an ASCII string. The values
3025  * are treated as milliseconds, and converted to jiffies when they are stored.
3026  *
3027  * This routine will ensure the values are within the range specified by
3028  * table->extra1 (min) and table->extra2 (max).
3029  *
3030  * Returns 0 on success.
3031  */
proc_doulongvec_ms_jiffies_minmax(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3032 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3033 				      void __user *buffer,
3034 				      size_t *lenp, loff_t *ppos)
3035 {
3036     return do_proc_doulongvec_minmax(table, write, buffer,
3037 				     lenp, ppos, HZ, 1000l);
3038 }
3039 
3040 
do_proc_dointvec_jiffies_conv(bool * negp,unsigned long * lvalp,int * valp,int write,void * data)3041 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
3042 					 int *valp,
3043 					 int write, void *data)
3044 {
3045 	if (write) {
3046 		if (*lvalp > INT_MAX / HZ)
3047 			return 1;
3048 		*valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
3049 	} else {
3050 		int val = *valp;
3051 		unsigned long lval;
3052 		if (val < 0) {
3053 			*negp = true;
3054 			lval = -(unsigned long)val;
3055 		} else {
3056 			*negp = false;
3057 			lval = (unsigned long)val;
3058 		}
3059 		*lvalp = lval / HZ;
3060 	}
3061 	return 0;
3062 }
3063 
do_proc_dointvec_userhz_jiffies_conv(bool * negp,unsigned long * lvalp,int * valp,int write,void * data)3064 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
3065 						int *valp,
3066 						int write, void *data)
3067 {
3068 	if (write) {
3069 		if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
3070 			return 1;
3071 		*valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
3072 	} else {
3073 		int val = *valp;
3074 		unsigned long lval;
3075 		if (val < 0) {
3076 			*negp = true;
3077 			lval = -(unsigned long)val;
3078 		} else {
3079 			*negp = false;
3080 			lval = (unsigned long)val;
3081 		}
3082 		*lvalp = jiffies_to_clock_t(lval);
3083 	}
3084 	return 0;
3085 }
3086 
do_proc_dointvec_ms_jiffies_conv(bool * negp,unsigned long * lvalp,int * valp,int write,void * data)3087 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
3088 					    int *valp,
3089 					    int write, void *data)
3090 {
3091 	if (write) {
3092 		unsigned long jif = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
3093 
3094 		if (jif > INT_MAX)
3095 			return 1;
3096 		*valp = (int)jif;
3097 	} else {
3098 		int val = *valp;
3099 		unsigned long lval;
3100 		if (val < 0) {
3101 			*negp = true;
3102 			lval = -(unsigned long)val;
3103 		} else {
3104 			*negp = false;
3105 			lval = (unsigned long)val;
3106 		}
3107 		*lvalp = jiffies_to_msecs(lval);
3108 	}
3109 	return 0;
3110 }
3111 
3112 /**
3113  * proc_dointvec_jiffies - read a vector of integers as seconds
3114  * @table: the sysctl table
3115  * @write: %TRUE if this is a write to the sysctl file
3116  * @buffer: the user buffer
3117  * @lenp: the size of the user buffer
3118  * @ppos: file position
3119  *
3120  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3121  * values from/to the user buffer, treated as an ASCII string.
3122  * The values read are assumed to be in seconds, and are converted into
3123  * jiffies.
3124  *
3125  * Returns 0 on success.
3126  */
proc_dointvec_jiffies(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3127 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3128 			  void __user *buffer, size_t *lenp, loff_t *ppos)
3129 {
3130     return do_proc_dointvec(table,write,buffer,lenp,ppos,
3131 		    	    do_proc_dointvec_jiffies_conv,NULL);
3132 }
3133 
3134 /**
3135  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
3136  * @table: the sysctl table
3137  * @write: %TRUE if this is a write to the sysctl file
3138  * @buffer: the user buffer
3139  * @lenp: the size of the user buffer
3140  * @ppos: pointer to the file position
3141  *
3142  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3143  * values from/to the user buffer, treated as an ASCII string.
3144  * The values read are assumed to be in 1/USER_HZ seconds, and
3145  * are converted into jiffies.
3146  *
3147  * Returns 0 on success.
3148  */
proc_dointvec_userhz_jiffies(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3149 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3150 				 void __user *buffer, size_t *lenp, loff_t *ppos)
3151 {
3152     return do_proc_dointvec(table,write,buffer,lenp,ppos,
3153 		    	    do_proc_dointvec_userhz_jiffies_conv,NULL);
3154 }
3155 
3156 /**
3157  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
3158  * @table: the sysctl table
3159  * @write: %TRUE if this is a write to the sysctl file
3160  * @buffer: the user buffer
3161  * @lenp: the size of the user buffer
3162  * @ppos: file position
3163  * @ppos: the current position in the file
3164  *
3165  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
3166  * values from/to the user buffer, treated as an ASCII string.
3167  * The values read are assumed to be in 1/1000 seconds, and
3168  * are converted into jiffies.
3169  *
3170  * Returns 0 on success.
3171  */
proc_dointvec_ms_jiffies(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3172 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3173 			     void __user *buffer, size_t *lenp, loff_t *ppos)
3174 {
3175 	return do_proc_dointvec(table, write, buffer, lenp, ppos,
3176 				do_proc_dointvec_ms_jiffies_conv, NULL);
3177 }
3178 
proc_do_cad_pid(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3179 static int proc_do_cad_pid(struct ctl_table *table, int write,
3180 			   void __user *buffer, size_t *lenp, loff_t *ppos)
3181 {
3182 	struct pid *new_pid;
3183 	pid_t tmp;
3184 	int r;
3185 
3186 	tmp = pid_vnr(cad_pid);
3187 
3188 	r = __do_proc_dointvec(&tmp, table, write, buffer,
3189 			       lenp, ppos, NULL, NULL);
3190 	if (r || !write)
3191 		return r;
3192 
3193 	new_pid = find_get_pid(tmp);
3194 	if (!new_pid)
3195 		return -ESRCH;
3196 
3197 	put_pid(xchg(&cad_pid, new_pid));
3198 	return 0;
3199 }
3200 
3201 /**
3202  * proc_do_large_bitmap - read/write from/to a large bitmap
3203  * @table: the sysctl table
3204  * @write: %TRUE if this is a write to the sysctl file
3205  * @buffer: the user buffer
3206  * @lenp: the size of the user buffer
3207  * @ppos: file position
3208  *
3209  * The bitmap is stored at table->data and the bitmap length (in bits)
3210  * in table->maxlen.
3211  *
3212  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
3213  * large bitmaps may be represented in a compact manner. Writing into
3214  * the file will clear the bitmap then update it with the given input.
3215  *
3216  * Returns 0 on success.
3217  */
proc_do_large_bitmap(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3218 int proc_do_large_bitmap(struct ctl_table *table, int write,
3219 			 void __user *buffer, size_t *lenp, loff_t *ppos)
3220 {
3221 	int err = 0;
3222 	bool first = 1;
3223 	size_t left = *lenp;
3224 	unsigned long bitmap_len = table->maxlen;
3225 	unsigned long *bitmap = *(unsigned long **) table->data;
3226 	unsigned long *tmp_bitmap = NULL;
3227 	char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
3228 
3229 	if (!bitmap || !bitmap_len || !left || (*ppos && !write)) {
3230 		*lenp = 0;
3231 		return 0;
3232 	}
3233 
3234 	if (write) {
3235 		char *kbuf, *p;
3236 		size_t skipped = 0;
3237 
3238 		if (left > PAGE_SIZE - 1) {
3239 			left = PAGE_SIZE - 1;
3240 			/* How much of the buffer we'll skip this pass */
3241 			skipped = *lenp - left;
3242 		}
3243 
3244 		p = kbuf = memdup_user_nul(buffer, left);
3245 		if (IS_ERR(kbuf))
3246 			return PTR_ERR(kbuf);
3247 
3248 		tmp_bitmap = bitmap_zalloc(bitmap_len, GFP_KERNEL);
3249 		if (!tmp_bitmap) {
3250 			kfree(kbuf);
3251 			return -ENOMEM;
3252 		}
3253 		proc_skip_char(&p, &left, '\n');
3254 		while (!err && left) {
3255 			unsigned long val_a, val_b;
3256 			bool neg;
3257 			size_t saved_left;
3258 
3259 			/* In case we stop parsing mid-number, we can reset */
3260 			saved_left = left;
3261 			err = proc_get_long(&p, &left, &val_a, &neg, tr_a,
3262 					     sizeof(tr_a), &c);
3263 			/*
3264 			 * If we consumed the entirety of a truncated buffer or
3265 			 * only one char is left (may be a "-"), then stop here,
3266 			 * reset, & come back for more.
3267 			 */
3268 			if ((left <= 1) && skipped) {
3269 				left = saved_left;
3270 				break;
3271 			}
3272 
3273 			if (err)
3274 				break;
3275 			if (val_a >= bitmap_len || neg) {
3276 				err = -EINVAL;
3277 				break;
3278 			}
3279 
3280 			val_b = val_a;
3281 			if (left) {
3282 				p++;
3283 				left--;
3284 			}
3285 
3286 			if (c == '-') {
3287 				err = proc_get_long(&p, &left, &val_b,
3288 						     &neg, tr_b, sizeof(tr_b),
3289 						     &c);
3290 				/*
3291 				 * If we consumed all of a truncated buffer or
3292 				 * then stop here, reset, & come back for more.
3293 				 */
3294 				if (!left && skipped) {
3295 					left = saved_left;
3296 					break;
3297 				}
3298 
3299 				if (err)
3300 					break;
3301 				if (val_b >= bitmap_len || neg ||
3302 				    val_a > val_b) {
3303 					err = -EINVAL;
3304 					break;
3305 				}
3306 				if (left) {
3307 					p++;
3308 					left--;
3309 				}
3310 			}
3311 
3312 			bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
3313 			first = 0;
3314 			proc_skip_char(&p, &left, '\n');
3315 		}
3316 		kfree(kbuf);
3317 		left += skipped;
3318 	} else {
3319 		unsigned long bit_a, bit_b = 0;
3320 
3321 		while (left) {
3322 			bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
3323 			if (bit_a >= bitmap_len)
3324 				break;
3325 			bit_b = find_next_zero_bit(bitmap, bitmap_len,
3326 						   bit_a + 1) - 1;
3327 
3328 			if (!first) {
3329 				err = proc_put_char(&buffer, &left, ',');
3330 				if (err)
3331 					break;
3332 			}
3333 			err = proc_put_long(&buffer, &left, bit_a, false);
3334 			if (err)
3335 				break;
3336 			if (bit_a != bit_b) {
3337 				err = proc_put_char(&buffer, &left, '-');
3338 				if (err)
3339 					break;
3340 				err = proc_put_long(&buffer, &left, bit_b, false);
3341 				if (err)
3342 					break;
3343 			}
3344 
3345 			first = 0; bit_b++;
3346 		}
3347 		if (!err)
3348 			err = proc_put_char(&buffer, &left, '\n');
3349 	}
3350 
3351 	if (!err) {
3352 		if (write) {
3353 			if (*ppos)
3354 				bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
3355 			else
3356 				bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
3357 		}
3358 		*lenp -= left;
3359 		*ppos += *lenp;
3360 	}
3361 
3362 	bitmap_free(tmp_bitmap);
3363 	return err;
3364 }
3365 
3366 #else /* CONFIG_PROC_SYSCTL */
3367 
proc_dostring(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3368 int proc_dostring(struct ctl_table *table, int write,
3369 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3370 {
3371 	return -ENOSYS;
3372 }
3373 
proc_dointvec(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3374 int proc_dointvec(struct ctl_table *table, int write,
3375 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3376 {
3377 	return -ENOSYS;
3378 }
3379 
proc_douintvec(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3380 int proc_douintvec(struct ctl_table *table, int write,
3381 		  void __user *buffer, size_t *lenp, loff_t *ppos)
3382 {
3383 	return -ENOSYS;
3384 }
3385 
proc_dointvec_minmax(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3386 int proc_dointvec_minmax(struct ctl_table *table, int write,
3387 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3388 {
3389 	return -ENOSYS;
3390 }
3391 
proc_douintvec_minmax(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3392 int proc_douintvec_minmax(struct ctl_table *table, int write,
3393 			  void __user *buffer, size_t *lenp, loff_t *ppos)
3394 {
3395 	return -ENOSYS;
3396 }
3397 
proc_dointvec_jiffies(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3398 int proc_dointvec_jiffies(struct ctl_table *table, int write,
3399 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3400 {
3401 	return -ENOSYS;
3402 }
3403 
proc_dointvec_userhz_jiffies(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3404 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
3405 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3406 {
3407 	return -ENOSYS;
3408 }
3409 
proc_dointvec_ms_jiffies(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3410 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
3411 			     void __user *buffer, size_t *lenp, loff_t *ppos)
3412 {
3413 	return -ENOSYS;
3414 }
3415 
proc_doulongvec_minmax(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3416 int proc_doulongvec_minmax(struct ctl_table *table, int write,
3417 		    void __user *buffer, size_t *lenp, loff_t *ppos)
3418 {
3419 	return -ENOSYS;
3420 }
3421 
proc_doulongvec_ms_jiffies_minmax(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3422 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
3423 				      void __user *buffer,
3424 				      size_t *lenp, loff_t *ppos)
3425 {
3426     return -ENOSYS;
3427 }
3428 
proc_do_large_bitmap(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3429 int proc_do_large_bitmap(struct ctl_table *table, int write,
3430 			 void __user *buffer, size_t *lenp, loff_t *ppos)
3431 {
3432 	return -ENOSYS;
3433 }
3434 
3435 #endif /* CONFIG_PROC_SYSCTL */
3436 
3437 #if defined(CONFIG_SYSCTL)
proc_do_static_key(struct ctl_table * table,int write,void __user * buffer,size_t * lenp,loff_t * ppos)3438 int proc_do_static_key(struct ctl_table *table, int write,
3439 		       void __user *buffer, size_t *lenp,
3440 		       loff_t *ppos)
3441 {
3442 	struct static_key *key = (struct static_key *)table->data;
3443 	static DEFINE_MUTEX(static_key_mutex);
3444 	int val, ret;
3445 	struct ctl_table tmp = {
3446 		.data   = &val,
3447 		.maxlen = sizeof(val),
3448 		.mode   = table->mode,
3449 		.extra1 = SYSCTL_ZERO,
3450 		.extra2 = SYSCTL_ONE,
3451 	};
3452 
3453 	if (write && !capable(CAP_SYS_ADMIN))
3454 		return -EPERM;
3455 
3456 	mutex_lock(&static_key_mutex);
3457 	val = static_key_enabled(key);
3458 	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
3459 	if (write && !ret) {
3460 		if (val)
3461 			static_key_enable(key);
3462 		else
3463 			static_key_disable(key);
3464 	}
3465 	mutex_unlock(&static_key_mutex);
3466 	return ret;
3467 }
3468 #endif
3469 /*
3470  * No sense putting this after each symbol definition, twice,
3471  * exception granted :-)
3472  */
3473 EXPORT_SYMBOL(proc_dointvec);
3474 EXPORT_SYMBOL(proc_douintvec);
3475 EXPORT_SYMBOL(proc_dointvec_jiffies);
3476 EXPORT_SYMBOL(proc_dointvec_minmax);
3477 EXPORT_SYMBOL_GPL(proc_douintvec_minmax);
3478 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
3479 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
3480 EXPORT_SYMBOL(proc_dostring);
3481 EXPORT_SYMBOL(proc_doulongvec_minmax);
3482 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
3483 EXPORT_SYMBOL(proc_do_large_bitmap);
3484