]> git.kernelconcepts.de Git - karo-tx-linux.git/blob - kernel/sysctl.c
Merge remote-tracking branch 'asoc/fix/sgtl5000' into asoc-linus
[karo-tx-linux.git] / kernel / sysctl.c
1 /*
2  * sysctl.c: General linux system control interface
3  *
4  * Begun 24 March 1995, Stephen Tweedie
5  * Added /proc support, Dec 1995
6  * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7  * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8  * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9  * Dynamic registration fixes, Stephen Tweedie.
10  * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11  * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12  *  Horn.
13  * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14  * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15  * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16  *  Wendling.
17  * The list_for_each() macro wasn't appropriate for the sysctl loop.
18  *  Removed it and replaced it with older style, 03/23/00, Bill Wendling
19  */
20
21 #include <linux/module.h>
22 #include <linux/mm.h>
23 #include <linux/swap.h>
24 #include <linux/slab.h>
25 #include <linux/sysctl.h>
26 #include <linux/bitmap.h>
27 #include <linux/signal.h>
28 #include <linux/printk.h>
29 #include <linux/proc_fs.h>
30 #include <linux/security.h>
31 #include <linux/ctype.h>
32 #include <linux/kmemcheck.h>
33 #include <linux/kmemleak.h>
34 #include <linux/fs.h>
35 #include <linux/init.h>
36 #include <linux/kernel.h>
37 #include <linux/kobject.h>
38 #include <linux/net.h>
39 #include <linux/sysrq.h>
40 #include <linux/highuid.h>
41 #include <linux/writeback.h>
42 #include <linux/ratelimit.h>
43 #include <linux/compaction.h>
44 #include <linux/hugetlb.h>
45 #include <linux/initrd.h>
46 #include <linux/key.h>
47 #include <linux/times.h>
48 #include <linux/limits.h>
49 #include <linux/dcache.h>
50 #include <linux/dnotify.h>
51 #include <linux/syscalls.h>
52 #include <linux/vmstat.h>
53 #include <linux/nfs_fs.h>
54 #include <linux/acpi.h>
55 #include <linux/reboot.h>
56 #include <linux/ftrace.h>
57 #include <linux/perf_event.h>
58 #include <linux/kprobes.h>
59 #include <linux/pipe_fs_i.h>
60 #include <linux/oom.h>
61 #include <linux/kmod.h>
62 #include <linux/capability.h>
63 #include <linux/binfmts.h>
64 #include <linux/sched/sysctl.h>
65
66 #include <asm/uaccess.h>
67 #include <asm/processor.h>
68
69 #ifdef CONFIG_X86
70 #include <asm/nmi.h>
71 #include <asm/stacktrace.h>
72 #include <asm/io.h>
73 #endif
74 #ifdef CONFIG_SPARC
75 #include <asm/setup.h>
76 #endif
77 #ifdef CONFIG_BSD_PROCESS_ACCT
78 #include <linux/acct.h>
79 #endif
80 #ifdef CONFIG_RT_MUTEXES
81 #include <linux/rtmutex.h>
82 #endif
83 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
84 #include <linux/lockdep.h>
85 #endif
86 #ifdef CONFIG_CHR_DEV_SG
87 #include <scsi/sg.h>
88 #endif
89
90 #ifdef CONFIG_LOCKUP_DETECTOR
91 #include <linux/nmi.h>
92 #endif
93
94
95 #if defined(CONFIG_SYSCTL)
96
97 /* External variables not in a header file. */
98 extern int sysctl_overcommit_memory;
99 extern int sysctl_overcommit_ratio;
100 extern int max_threads;
101 extern int suid_dumpable;
102 #ifdef CONFIG_COREDUMP
103 extern int core_uses_pid;
104 extern char core_pattern[];
105 extern unsigned int core_pipe_limit;
106 #endif
107 extern int pid_max;
108 extern int pid_max_min, pid_max_max;
109 extern int percpu_pagelist_fraction;
110 extern int compat_log;
111 extern int latencytop_enabled;
112 extern int sysctl_nr_open_min, sysctl_nr_open_max;
113 #ifndef CONFIG_MMU
114 extern int sysctl_nr_trim_pages;
115 #endif
116 #ifdef CONFIG_BLOCK
117 extern int blk_iopoll_enabled;
118 #endif
119
120 /* Constants used for minimum and  maximum */
121 #ifdef CONFIG_LOCKUP_DETECTOR
122 static int sixty = 60;
123 #endif
124
125 static int zero;
126 static int __maybe_unused one = 1;
127 static int __maybe_unused two = 2;
128 static int __maybe_unused three = 3;
129 static unsigned long one_ul = 1;
130 static int one_hundred = 100;
131 #ifdef CONFIG_PRINTK
132 static int ten_thousand = 10000;
133 #endif
134
135 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
136 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
137
138 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
139 static int maxolduid = 65535;
140 static int minolduid;
141 static int min_percpu_pagelist_fract = 8;
142
143 static int ngroups_max = NGROUPS_MAX;
144 static const int cap_last_cap = CAP_LAST_CAP;
145
146 #ifdef CONFIG_INOTIFY_USER
147 #include <linux/inotify.h>
148 #endif
149 #ifdef CONFIG_SPARC
150 #endif
151
152 #ifdef CONFIG_SPARC64
153 extern int sysctl_tsb_ratio;
154 #endif
155
156 #ifdef __hppa__
157 extern int pwrsw_enabled;
158 #endif
159
160 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
161 extern int unaligned_enabled;
162 #endif
163
164 #ifdef CONFIG_IA64
165 extern int unaligned_dump_stack;
166 #endif
167
168 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
169 extern int no_unaligned_warning;
170 #endif
171
172 #ifdef CONFIG_PROC_SYSCTL
173 static int proc_do_cad_pid(struct ctl_table *table, int write,
174                   void __user *buffer, size_t *lenp, loff_t *ppos);
175 static int proc_taint(struct ctl_table *table, int write,
176                                void __user *buffer, size_t *lenp, loff_t *ppos);
177 #endif
178
179 #ifdef CONFIG_PRINTK
180 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
181                                 void __user *buffer, size_t *lenp, loff_t *ppos);
182 #endif
183
184 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
185                 void __user *buffer, size_t *lenp, loff_t *ppos);
186 #ifdef CONFIG_COREDUMP
187 static int proc_dostring_coredump(struct ctl_table *table, int write,
188                 void __user *buffer, size_t *lenp, loff_t *ppos);
189 #endif
190
191 #ifdef CONFIG_MAGIC_SYSRQ
192 /* Note: sysrq code uses it's own private copy */
193 static int __sysrq_enabled = SYSRQ_DEFAULT_ENABLE;
194
195 static int sysrq_sysctl_handler(ctl_table *table, int write,
196                                 void __user *buffer, size_t *lenp,
197                                 loff_t *ppos)
198 {
199         int error;
200
201         error = proc_dointvec(table, write, buffer, lenp, ppos);
202         if (error)
203                 return error;
204
205         if (write)
206                 sysrq_toggle_support(__sysrq_enabled);
207
208         return 0;
209 }
210
211 #endif
212
213 static struct ctl_table kern_table[];
214 static struct ctl_table vm_table[];
215 static struct ctl_table fs_table[];
216 static struct ctl_table debug_table[];
217 static struct ctl_table dev_table[];
218 extern struct ctl_table random_table[];
219 #ifdef CONFIG_EPOLL
220 extern struct ctl_table epoll_table[];
221 #endif
222
223 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
224 int sysctl_legacy_va_layout;
225 #endif
226
227 /* The default sysctl tables: */
228
229 static struct ctl_table sysctl_base_table[] = {
230         {
231                 .procname       = "kernel",
232                 .mode           = 0555,
233                 .child          = kern_table,
234         },
235         {
236                 .procname       = "vm",
237                 .mode           = 0555,
238                 .child          = vm_table,
239         },
240         {
241                 .procname       = "fs",
242                 .mode           = 0555,
243                 .child          = fs_table,
244         },
245         {
246                 .procname       = "debug",
247                 .mode           = 0555,
248                 .child          = debug_table,
249         },
250         {
251                 .procname       = "dev",
252                 .mode           = 0555,
253                 .child          = dev_table,
254         },
255         { }
256 };
257
258 #ifdef CONFIG_SCHED_DEBUG
259 static int min_sched_granularity_ns = 100000;           /* 100 usecs */
260 static int max_sched_granularity_ns = NSEC_PER_SEC;     /* 1 second */
261 static int min_wakeup_granularity_ns;                   /* 0 usecs */
262 static int max_wakeup_granularity_ns = NSEC_PER_SEC;    /* 1 second */
263 #ifdef CONFIG_SMP
264 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
265 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
266 #endif /* CONFIG_SMP */
267 #endif /* CONFIG_SCHED_DEBUG */
268
269 #ifdef CONFIG_COMPACTION
270 static int min_extfrag_threshold;
271 static int max_extfrag_threshold = 1000;
272 #endif
273
274 static struct ctl_table kern_table[] = {
275         {
276                 .procname       = "sched_child_runs_first",
277                 .data           = &sysctl_sched_child_runs_first,
278                 .maxlen         = sizeof(unsigned int),
279                 .mode           = 0644,
280                 .proc_handler   = proc_dointvec,
281         },
282 #ifdef CONFIG_SCHED_DEBUG
283         {
284                 .procname       = "sched_min_granularity_ns",
285                 .data           = &sysctl_sched_min_granularity,
286                 .maxlen         = sizeof(unsigned int),
287                 .mode           = 0644,
288                 .proc_handler   = sched_proc_update_handler,
289                 .extra1         = &min_sched_granularity_ns,
290                 .extra2         = &max_sched_granularity_ns,
291         },
292         {
293                 .procname       = "sched_latency_ns",
294                 .data           = &sysctl_sched_latency,
295                 .maxlen         = sizeof(unsigned int),
296                 .mode           = 0644,
297                 .proc_handler   = sched_proc_update_handler,
298                 .extra1         = &min_sched_granularity_ns,
299                 .extra2         = &max_sched_granularity_ns,
300         },
301         {
302                 .procname       = "sched_wakeup_granularity_ns",
303                 .data           = &sysctl_sched_wakeup_granularity,
304                 .maxlen         = sizeof(unsigned int),
305                 .mode           = 0644,
306                 .proc_handler   = sched_proc_update_handler,
307                 .extra1         = &min_wakeup_granularity_ns,
308                 .extra2         = &max_wakeup_granularity_ns,
309         },
310 #ifdef CONFIG_SMP
311         {
312                 .procname       = "sched_tunable_scaling",
313                 .data           = &sysctl_sched_tunable_scaling,
314                 .maxlen         = sizeof(enum sched_tunable_scaling),
315                 .mode           = 0644,
316                 .proc_handler   = sched_proc_update_handler,
317                 .extra1         = &min_sched_tunable_scaling,
318                 .extra2         = &max_sched_tunable_scaling,
319         },
320         {
321                 .procname       = "sched_migration_cost_ns",
322                 .data           = &sysctl_sched_migration_cost,
323                 .maxlen         = sizeof(unsigned int),
324                 .mode           = 0644,
325                 .proc_handler   = proc_dointvec,
326         },
327         {
328                 .procname       = "sched_nr_migrate",
329                 .data           = &sysctl_sched_nr_migrate,
330                 .maxlen         = sizeof(unsigned int),
331                 .mode           = 0644,
332                 .proc_handler   = proc_dointvec,
333         },
334         {
335                 .procname       = "sched_time_avg_ms",
336                 .data           = &sysctl_sched_time_avg,
337                 .maxlen         = sizeof(unsigned int),
338                 .mode           = 0644,
339                 .proc_handler   = proc_dointvec,
340         },
341         {
342                 .procname       = "sched_shares_window_ns",
343                 .data           = &sysctl_sched_shares_window,
344                 .maxlen         = sizeof(unsigned int),
345                 .mode           = 0644,
346                 .proc_handler   = proc_dointvec,
347         },
348         {
349                 .procname       = "timer_migration",
350                 .data           = &sysctl_timer_migration,
351                 .maxlen         = sizeof(unsigned int),
352                 .mode           = 0644,
353                 .proc_handler   = proc_dointvec_minmax,
354                 .extra1         = &zero,
355                 .extra2         = &one,
356         },
357 #endif /* CONFIG_SMP */
358 #ifdef CONFIG_NUMA_BALANCING
359         {
360                 .procname       = "numa_balancing_scan_delay_ms",
361                 .data           = &sysctl_numa_balancing_scan_delay,
362                 .maxlen         = sizeof(unsigned int),
363                 .mode           = 0644,
364                 .proc_handler   = proc_dointvec,
365         },
366         {
367                 .procname       = "numa_balancing_scan_period_min_ms",
368                 .data           = &sysctl_numa_balancing_scan_period_min,
369                 .maxlen         = sizeof(unsigned int),
370                 .mode           = 0644,
371                 .proc_handler   = proc_dointvec,
372         },
373         {
374                 .procname       = "numa_balancing_scan_period_reset",
375                 .data           = &sysctl_numa_balancing_scan_period_reset,
376                 .maxlen         = sizeof(unsigned int),
377                 .mode           = 0644,
378                 .proc_handler   = proc_dointvec,
379         },
380         {
381                 .procname       = "numa_balancing_scan_period_max_ms",
382                 .data           = &sysctl_numa_balancing_scan_period_max,
383                 .maxlen         = sizeof(unsigned int),
384                 .mode           = 0644,
385                 .proc_handler   = proc_dointvec,
386         },
387         {
388                 .procname       = "numa_balancing_scan_size_mb",
389                 .data           = &sysctl_numa_balancing_scan_size,
390                 .maxlen         = sizeof(unsigned int),
391                 .mode           = 0644,
392                 .proc_handler   = proc_dointvec,
393         },
394 #endif /* CONFIG_NUMA_BALANCING */
395 #endif /* CONFIG_SCHED_DEBUG */
396         {
397                 .procname       = "sched_rt_period_us",
398                 .data           = &sysctl_sched_rt_period,
399                 .maxlen         = sizeof(unsigned int),
400                 .mode           = 0644,
401                 .proc_handler   = sched_rt_handler,
402         },
403         {
404                 .procname       = "sched_rt_runtime_us",
405                 .data           = &sysctl_sched_rt_runtime,
406                 .maxlen         = sizeof(int),
407                 .mode           = 0644,
408                 .proc_handler   = sched_rt_handler,
409         },
410         {
411                 .procname       = "sched_rr_timeslice_ms",
412                 .data           = &sched_rr_timeslice,
413                 .maxlen         = sizeof(int),
414                 .mode           = 0644,
415                 .proc_handler   = sched_rr_handler,
416         },
417 #ifdef CONFIG_SCHED_AUTOGROUP
418         {
419                 .procname       = "sched_autogroup_enabled",
420                 .data           = &sysctl_sched_autogroup_enabled,
421                 .maxlen         = sizeof(unsigned int),
422                 .mode           = 0644,
423                 .proc_handler   = proc_dointvec_minmax,
424                 .extra1         = &zero,
425                 .extra2         = &one,
426         },
427 #endif
428 #ifdef CONFIG_CFS_BANDWIDTH
429         {
430                 .procname       = "sched_cfs_bandwidth_slice_us",
431                 .data           = &sysctl_sched_cfs_bandwidth_slice,
432                 .maxlen         = sizeof(unsigned int),
433                 .mode           = 0644,
434                 .proc_handler   = proc_dointvec_minmax,
435                 .extra1         = &one,
436         },
437 #endif
438 #ifdef CONFIG_PROVE_LOCKING
439         {
440                 .procname       = "prove_locking",
441                 .data           = &prove_locking,
442                 .maxlen         = sizeof(int),
443                 .mode           = 0644,
444                 .proc_handler   = proc_dointvec,
445         },
446 #endif
447 #ifdef CONFIG_LOCK_STAT
448         {
449                 .procname       = "lock_stat",
450                 .data           = &lock_stat,
451                 .maxlen         = sizeof(int),
452                 .mode           = 0644,
453                 .proc_handler   = proc_dointvec,
454         },
455 #endif
456         {
457                 .procname       = "panic",
458                 .data           = &panic_timeout,
459                 .maxlen         = sizeof(int),
460                 .mode           = 0644,
461                 .proc_handler   = proc_dointvec,
462         },
463 #ifdef CONFIG_COREDUMP
464         {
465                 .procname       = "core_uses_pid",
466                 .data           = &core_uses_pid,
467                 .maxlen         = sizeof(int),
468                 .mode           = 0644,
469                 .proc_handler   = proc_dointvec,
470         },
471         {
472                 .procname       = "core_pattern",
473                 .data           = core_pattern,
474                 .maxlen         = CORENAME_MAX_SIZE,
475                 .mode           = 0644,
476                 .proc_handler   = proc_dostring_coredump,
477         },
478         {
479                 .procname       = "core_pipe_limit",
480                 .data           = &core_pipe_limit,
481                 .maxlen         = sizeof(unsigned int),
482                 .mode           = 0644,
483                 .proc_handler   = proc_dointvec,
484         },
485 #endif
486 #ifdef CONFIG_PROC_SYSCTL
487         {
488                 .procname       = "tainted",
489                 .maxlen         = sizeof(long),
490                 .mode           = 0644,
491                 .proc_handler   = proc_taint,
492         },
493 #endif
494 #ifdef CONFIG_LATENCYTOP
495         {
496                 .procname       = "latencytop",
497                 .data           = &latencytop_enabled,
498                 .maxlen         = sizeof(int),
499                 .mode           = 0644,
500                 .proc_handler   = proc_dointvec,
501         },
502 #endif
503 #ifdef CONFIG_BLK_DEV_INITRD
504         {
505                 .procname       = "real-root-dev",
506                 .data           = &real_root_dev,
507                 .maxlen         = sizeof(int),
508                 .mode           = 0644,
509                 .proc_handler   = proc_dointvec,
510         },
511 #endif
512         {
513                 .procname       = "print-fatal-signals",
514                 .data           = &print_fatal_signals,
515                 .maxlen         = sizeof(int),
516                 .mode           = 0644,
517                 .proc_handler   = proc_dointvec,
518         },
519 #ifdef CONFIG_SPARC
520         {
521                 .procname       = "reboot-cmd",
522                 .data           = reboot_command,
523                 .maxlen         = 256,
524                 .mode           = 0644,
525                 .proc_handler   = proc_dostring,
526         },
527         {
528                 .procname       = "stop-a",
529                 .data           = &stop_a_enabled,
530                 .maxlen         = sizeof (int),
531                 .mode           = 0644,
532                 .proc_handler   = proc_dointvec,
533         },
534         {
535                 .procname       = "scons-poweroff",
536                 .data           = &scons_pwroff,
537                 .maxlen         = sizeof (int),
538                 .mode           = 0644,
539                 .proc_handler   = proc_dointvec,
540         },
541 #endif
542 #ifdef CONFIG_SPARC64
543         {
544                 .procname       = "tsb-ratio",
545                 .data           = &sysctl_tsb_ratio,
546                 .maxlen         = sizeof (int),
547                 .mode           = 0644,
548                 .proc_handler   = proc_dointvec,
549         },
550 #endif
551 #ifdef __hppa__
552         {
553                 .procname       = "soft-power",
554                 .data           = &pwrsw_enabled,
555                 .maxlen         = sizeof (int),
556                 .mode           = 0644,
557                 .proc_handler   = proc_dointvec,
558         },
559 #endif
560 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_ALLOW
561         {
562                 .procname       = "unaligned-trap",
563                 .data           = &unaligned_enabled,
564                 .maxlen         = sizeof (int),
565                 .mode           = 0644,
566                 .proc_handler   = proc_dointvec,
567         },
568 #endif
569         {
570                 .procname       = "ctrl-alt-del",
571                 .data           = &C_A_D,
572                 .maxlen         = sizeof(int),
573                 .mode           = 0644,
574                 .proc_handler   = proc_dointvec,
575         },
576 #ifdef CONFIG_FUNCTION_TRACER
577         {
578                 .procname       = "ftrace_enabled",
579                 .data           = &ftrace_enabled,
580                 .maxlen         = sizeof(int),
581                 .mode           = 0644,
582                 .proc_handler   = ftrace_enable_sysctl,
583         },
584 #endif
585 #ifdef CONFIG_STACK_TRACER
586         {
587                 .procname       = "stack_tracer_enabled",
588                 .data           = &stack_tracer_enabled,
589                 .maxlen         = sizeof(int),
590                 .mode           = 0644,
591                 .proc_handler   = stack_trace_sysctl,
592         },
593 #endif
594 #ifdef CONFIG_TRACING
595         {
596                 .procname       = "ftrace_dump_on_oops",
597                 .data           = &ftrace_dump_on_oops,
598                 .maxlen         = sizeof(int),
599                 .mode           = 0644,
600                 .proc_handler   = proc_dointvec,
601         },
602         {
603                 .procname       = "traceoff_on_warning",
604                 .data           = &__disable_trace_on_warning,
605                 .maxlen         = sizeof(__disable_trace_on_warning),
606                 .mode           = 0644,
607                 .proc_handler   = proc_dointvec,
608         },
609 #endif
610 #ifdef CONFIG_MODULES
611         {
612                 .procname       = "modprobe",
613                 .data           = &modprobe_path,
614                 .maxlen         = KMOD_PATH_LEN,
615                 .mode           = 0644,
616                 .proc_handler   = proc_dostring,
617         },
618         {
619                 .procname       = "modules_disabled",
620                 .data           = &modules_disabled,
621                 .maxlen         = sizeof(int),
622                 .mode           = 0644,
623                 /* only handle a transition from default "0" to "1" */
624                 .proc_handler   = proc_dointvec_minmax,
625                 .extra1         = &one,
626                 .extra2         = &one,
627         },
628 #endif
629
630         {
631                 .procname       = "hotplug",
632                 .data           = &uevent_helper,
633                 .maxlen         = UEVENT_HELPER_PATH_LEN,
634                 .mode           = 0644,
635                 .proc_handler   = proc_dostring,
636         },
637
638 #ifdef CONFIG_CHR_DEV_SG
639         {
640                 .procname       = "sg-big-buff",
641                 .data           = &sg_big_buff,
642                 .maxlen         = sizeof (int),
643                 .mode           = 0444,
644                 .proc_handler   = proc_dointvec,
645         },
646 #endif
647 #ifdef CONFIG_BSD_PROCESS_ACCT
648         {
649                 .procname       = "acct",
650                 .data           = &acct_parm,
651                 .maxlen         = 3*sizeof(int),
652                 .mode           = 0644,
653                 .proc_handler   = proc_dointvec,
654         },
655 #endif
656 #ifdef CONFIG_MAGIC_SYSRQ
657         {
658                 .procname       = "sysrq",
659                 .data           = &__sysrq_enabled,
660                 .maxlen         = sizeof (int),
661                 .mode           = 0644,
662                 .proc_handler   = sysrq_sysctl_handler,
663         },
664 #endif
665 #ifdef CONFIG_PROC_SYSCTL
666         {
667                 .procname       = "cad_pid",
668                 .data           = NULL,
669                 .maxlen         = sizeof (int),
670                 .mode           = 0600,
671                 .proc_handler   = proc_do_cad_pid,
672         },
673 #endif
674         {
675                 .procname       = "threads-max",
676                 .data           = &max_threads,
677                 .maxlen         = sizeof(int),
678                 .mode           = 0644,
679                 .proc_handler   = proc_dointvec,
680         },
681         {
682                 .procname       = "random",
683                 .mode           = 0555,
684                 .child          = random_table,
685         },
686         {
687                 .procname       = "usermodehelper",
688                 .mode           = 0555,
689                 .child          = usermodehelper_table,
690         },
691         {
692                 .procname       = "overflowuid",
693                 .data           = &overflowuid,
694                 .maxlen         = sizeof(int),
695                 .mode           = 0644,
696                 .proc_handler   = proc_dointvec_minmax,
697                 .extra1         = &minolduid,
698                 .extra2         = &maxolduid,
699         },
700         {
701                 .procname       = "overflowgid",
702                 .data           = &overflowgid,
703                 .maxlen         = sizeof(int),
704                 .mode           = 0644,
705                 .proc_handler   = proc_dointvec_minmax,
706                 .extra1         = &minolduid,
707                 .extra2         = &maxolduid,
708         },
709 #ifdef CONFIG_S390
710 #ifdef CONFIG_MATHEMU
711         {
712                 .procname       = "ieee_emulation_warnings",
713                 .data           = &sysctl_ieee_emulation_warnings,
714                 .maxlen         = sizeof(int),
715                 .mode           = 0644,
716                 .proc_handler   = proc_dointvec,
717         },
718 #endif
719         {
720                 .procname       = "userprocess_debug",
721                 .data           = &show_unhandled_signals,
722                 .maxlen         = sizeof(int),
723                 .mode           = 0644,
724                 .proc_handler   = proc_dointvec,
725         },
726 #endif
727         {
728                 .procname       = "pid_max",
729                 .data           = &pid_max,
730                 .maxlen         = sizeof (int),
731                 .mode           = 0644,
732                 .proc_handler   = proc_dointvec_minmax,
733                 .extra1         = &pid_max_min,
734                 .extra2         = &pid_max_max,
735         },
736         {
737                 .procname       = "panic_on_oops",
738                 .data           = &panic_on_oops,
739                 .maxlen         = sizeof(int),
740                 .mode           = 0644,
741                 .proc_handler   = proc_dointvec,
742         },
743 #if defined CONFIG_PRINTK
744         {
745                 .procname       = "printk",
746                 .data           = &console_loglevel,
747                 .maxlen         = 4*sizeof(int),
748                 .mode           = 0644,
749                 .proc_handler   = proc_dointvec,
750         },
751         {
752                 .procname       = "printk_ratelimit",
753                 .data           = &printk_ratelimit_state.interval,
754                 .maxlen         = sizeof(int),
755                 .mode           = 0644,
756                 .proc_handler   = proc_dointvec_jiffies,
757         },
758         {
759                 .procname       = "printk_ratelimit_burst",
760                 .data           = &printk_ratelimit_state.burst,
761                 .maxlen         = sizeof(int),
762                 .mode           = 0644,
763                 .proc_handler   = proc_dointvec,
764         },
765         {
766                 .procname       = "printk_delay",
767                 .data           = &printk_delay_msec,
768                 .maxlen         = sizeof(int),
769                 .mode           = 0644,
770                 .proc_handler   = proc_dointvec_minmax,
771                 .extra1         = &zero,
772                 .extra2         = &ten_thousand,
773         },
774         {
775                 .procname       = "dmesg_restrict",
776                 .data           = &dmesg_restrict,
777                 .maxlen         = sizeof(int),
778                 .mode           = 0644,
779                 .proc_handler   = proc_dointvec_minmax_sysadmin,
780                 .extra1         = &zero,
781                 .extra2         = &one,
782         },
783         {
784                 .procname       = "kptr_restrict",
785                 .data           = &kptr_restrict,
786                 .maxlen         = sizeof(int),
787                 .mode           = 0644,
788                 .proc_handler   = proc_dointvec_minmax_sysadmin,
789                 .extra1         = &zero,
790                 .extra2         = &two,
791         },
792 #endif
793         {
794                 .procname       = "ngroups_max",
795                 .data           = &ngroups_max,
796                 .maxlen         = sizeof (int),
797                 .mode           = 0444,
798                 .proc_handler   = proc_dointvec,
799         },
800         {
801                 .procname       = "cap_last_cap",
802                 .data           = (void *)&cap_last_cap,
803                 .maxlen         = sizeof(int),
804                 .mode           = 0444,
805                 .proc_handler   = proc_dointvec,
806         },
807 #if defined(CONFIG_LOCKUP_DETECTOR)
808         {
809                 .procname       = "watchdog",
810                 .data           = &watchdog_user_enabled,
811                 .maxlen         = sizeof (int),
812                 .mode           = 0644,
813                 .proc_handler   = proc_dowatchdog,
814                 .extra1         = &zero,
815                 .extra2         = &one,
816         },
817         {
818                 .procname       = "watchdog_thresh",
819                 .data           = &watchdog_thresh,
820                 .maxlen         = sizeof(int),
821                 .mode           = 0644,
822                 .proc_handler   = proc_dowatchdog,
823                 .extra1         = &zero,
824                 .extra2         = &sixty,
825         },
826         {
827                 .procname       = "softlockup_panic",
828                 .data           = &softlockup_panic,
829                 .maxlen         = sizeof(int),
830                 .mode           = 0644,
831                 .proc_handler   = proc_dointvec_minmax,
832                 .extra1         = &zero,
833                 .extra2         = &one,
834         },
835         {
836                 .procname       = "nmi_watchdog",
837                 .data           = &watchdog_user_enabled,
838                 .maxlen         = sizeof (int),
839                 .mode           = 0644,
840                 .proc_handler   = proc_dowatchdog,
841                 .extra1         = &zero,
842                 .extra2         = &one,
843         },
844 #endif
845 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
846         {
847                 .procname       = "unknown_nmi_panic",
848                 .data           = &unknown_nmi_panic,
849                 .maxlen         = sizeof (int),
850                 .mode           = 0644,
851                 .proc_handler   = proc_dointvec,
852         },
853 #endif
854 #if defined(CONFIG_X86)
855         {
856                 .procname       = "panic_on_unrecovered_nmi",
857                 .data           = &panic_on_unrecovered_nmi,
858                 .maxlen         = sizeof(int),
859                 .mode           = 0644,
860                 .proc_handler   = proc_dointvec,
861         },
862         {
863                 .procname       = "panic_on_io_nmi",
864                 .data           = &panic_on_io_nmi,
865                 .maxlen         = sizeof(int),
866                 .mode           = 0644,
867                 .proc_handler   = proc_dointvec,
868         },
869 #ifdef CONFIG_DEBUG_STACKOVERFLOW
870         {
871                 .procname       = "panic_on_stackoverflow",
872                 .data           = &sysctl_panic_on_stackoverflow,
873                 .maxlen         = sizeof(int),
874                 .mode           = 0644,
875                 .proc_handler   = proc_dointvec,
876         },
877 #endif
878         {
879                 .procname       = "bootloader_type",
880                 .data           = &bootloader_type,
881                 .maxlen         = sizeof (int),
882                 .mode           = 0444,
883                 .proc_handler   = proc_dointvec,
884         },
885         {
886                 .procname       = "bootloader_version",
887                 .data           = &bootloader_version,
888                 .maxlen         = sizeof (int),
889                 .mode           = 0444,
890                 .proc_handler   = proc_dointvec,
891         },
892         {
893                 .procname       = "kstack_depth_to_print",
894                 .data           = &kstack_depth_to_print,
895                 .maxlen         = sizeof(int),
896                 .mode           = 0644,
897                 .proc_handler   = proc_dointvec,
898         },
899         {
900                 .procname       = "io_delay_type",
901                 .data           = &io_delay_type,
902                 .maxlen         = sizeof(int),
903                 .mode           = 0644,
904                 .proc_handler   = proc_dointvec,
905         },
906 #endif
907 #if defined(CONFIG_MMU)
908         {
909                 .procname       = "randomize_va_space",
910                 .data           = &randomize_va_space,
911                 .maxlen         = sizeof(int),
912                 .mode           = 0644,
913                 .proc_handler   = proc_dointvec,
914         },
915 #endif
916 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
917         {
918                 .procname       = "spin_retry",
919                 .data           = &spin_retry,
920                 .maxlen         = sizeof (int),
921                 .mode           = 0644,
922                 .proc_handler   = proc_dointvec,
923         },
924 #endif
925 #if     defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
926         {
927                 .procname       = "acpi_video_flags",
928                 .data           = &acpi_realmode_flags,
929                 .maxlen         = sizeof (unsigned long),
930                 .mode           = 0644,
931                 .proc_handler   = proc_doulongvec_minmax,
932         },
933 #endif
934 #ifdef CONFIG_SYSCTL_ARCH_UNALIGN_NO_WARN
935         {
936                 .procname       = "ignore-unaligned-usertrap",
937                 .data           = &no_unaligned_warning,
938                 .maxlen         = sizeof (int),
939                 .mode           = 0644,
940                 .proc_handler   = proc_dointvec,
941         },
942 #endif
943 #ifdef CONFIG_IA64
944         {
945                 .procname       = "unaligned-dump-stack",
946                 .data           = &unaligned_dump_stack,
947                 .maxlen         = sizeof (int),
948                 .mode           = 0644,
949                 .proc_handler   = proc_dointvec,
950         },
951 #endif
952 #ifdef CONFIG_DETECT_HUNG_TASK
953         {
954                 .procname       = "hung_task_panic",
955                 .data           = &sysctl_hung_task_panic,
956                 .maxlen         = sizeof(int),
957                 .mode           = 0644,
958                 .proc_handler   = proc_dointvec_minmax,
959                 .extra1         = &zero,
960                 .extra2         = &one,
961         },
962         {
963                 .procname       = "hung_task_check_count",
964                 .data           = &sysctl_hung_task_check_count,
965                 .maxlen         = sizeof(unsigned long),
966                 .mode           = 0644,
967                 .proc_handler   = proc_doulongvec_minmax,
968         },
969         {
970                 .procname       = "hung_task_timeout_secs",
971                 .data           = &sysctl_hung_task_timeout_secs,
972                 .maxlen         = sizeof(unsigned long),
973                 .mode           = 0644,
974                 .proc_handler   = proc_dohung_task_timeout_secs,
975         },
976         {
977                 .procname       = "hung_task_warnings",
978                 .data           = &sysctl_hung_task_warnings,
979                 .maxlen         = sizeof(unsigned long),
980                 .mode           = 0644,
981                 .proc_handler   = proc_doulongvec_minmax,
982         },
983 #endif
984 #ifdef CONFIG_COMPAT
985         {
986                 .procname       = "compat-log",
987                 .data           = &compat_log,
988                 .maxlen         = sizeof (int),
989                 .mode           = 0644,
990                 .proc_handler   = proc_dointvec,
991         },
992 #endif
993 #ifdef CONFIG_RT_MUTEXES
994         {
995                 .procname       = "max_lock_depth",
996                 .data           = &max_lock_depth,
997                 .maxlen         = sizeof(int),
998                 .mode           = 0644,
999                 .proc_handler   = proc_dointvec,
1000         },
1001 #endif
1002         {
1003                 .procname       = "poweroff_cmd",
1004                 .data           = &poweroff_cmd,
1005                 .maxlen         = POWEROFF_CMD_PATH_LEN,
1006                 .mode           = 0644,
1007                 .proc_handler   = proc_dostring,
1008         },
1009 #ifdef CONFIG_KEYS
1010         {
1011                 .procname       = "keys",
1012                 .mode           = 0555,
1013                 .child          = key_sysctls,
1014         },
1015 #endif
1016 #ifdef CONFIG_RCU_TORTURE_TEST
1017         {
1018                 .procname       = "rcutorture_runnable",
1019                 .data           = &rcutorture_runnable,
1020                 .maxlen         = sizeof(int),
1021                 .mode           = 0644,
1022                 .proc_handler   = proc_dointvec,
1023         },
1024 #endif
1025 #ifdef CONFIG_PERF_EVENTS
1026         /*
1027          * User-space scripts rely on the existence of this file
1028          * as a feature check for perf_events being enabled.
1029          *
1030          * So it's an ABI, do not remove!
1031          */
1032         {
1033                 .procname       = "perf_event_paranoid",
1034                 .data           = &sysctl_perf_event_paranoid,
1035                 .maxlen         = sizeof(sysctl_perf_event_paranoid),
1036                 .mode           = 0644,
1037                 .proc_handler   = proc_dointvec,
1038         },
1039         {
1040                 .procname       = "perf_event_mlock_kb",
1041                 .data           = &sysctl_perf_event_mlock,
1042                 .maxlen         = sizeof(sysctl_perf_event_mlock),
1043                 .mode           = 0644,
1044                 .proc_handler   = proc_dointvec,
1045         },
1046         {
1047                 .procname       = "perf_event_max_sample_rate",
1048                 .data           = &sysctl_perf_event_sample_rate,
1049                 .maxlen         = sizeof(sysctl_perf_event_sample_rate),
1050                 .mode           = 0644,
1051                 .proc_handler   = perf_proc_update_handler,
1052         },
1053         {
1054                 .procname       = "perf_cpu_time_max_percent",
1055                 .data           = &sysctl_perf_cpu_time_max_percent,
1056                 .maxlen         = sizeof(sysctl_perf_cpu_time_max_percent),
1057                 .mode           = 0644,
1058                 .proc_handler   = perf_cpu_time_max_percent_handler,
1059                 .extra1         = &zero,
1060                 .extra2         = &one_hundred,
1061         },
1062 #endif
1063 #ifdef CONFIG_KMEMCHECK
1064         {
1065                 .procname       = "kmemcheck",
1066                 .data           = &kmemcheck_enabled,
1067                 .maxlen         = sizeof(int),
1068                 .mode           = 0644,
1069                 .proc_handler   = proc_dointvec,
1070         },
1071 #endif
1072 #ifdef CONFIG_BLOCK
1073         {
1074                 .procname       = "blk_iopoll",
1075                 .data           = &blk_iopoll_enabled,
1076                 .maxlen         = sizeof(int),
1077                 .mode           = 0644,
1078                 .proc_handler   = proc_dointvec,
1079         },
1080 #endif
1081         { }
1082 };
1083
1084 static struct ctl_table vm_table[] = {
1085         {
1086                 .procname       = "overcommit_memory",
1087                 .data           = &sysctl_overcommit_memory,
1088                 .maxlen         = sizeof(sysctl_overcommit_memory),
1089                 .mode           = 0644,
1090                 .proc_handler   = proc_dointvec_minmax,
1091                 .extra1         = &zero,
1092                 .extra2         = &two,
1093         },
1094         {
1095                 .procname       = "panic_on_oom",
1096                 .data           = &sysctl_panic_on_oom,
1097                 .maxlen         = sizeof(sysctl_panic_on_oom),
1098                 .mode           = 0644,
1099                 .proc_handler   = proc_dointvec_minmax,
1100                 .extra1         = &zero,
1101                 .extra2         = &two,
1102         },
1103         {
1104                 .procname       = "oom_kill_allocating_task",
1105                 .data           = &sysctl_oom_kill_allocating_task,
1106                 .maxlen         = sizeof(sysctl_oom_kill_allocating_task),
1107                 .mode           = 0644,
1108                 .proc_handler   = proc_dointvec,
1109         },
1110         {
1111                 .procname       = "oom_dump_tasks",
1112                 .data           = &sysctl_oom_dump_tasks,
1113                 .maxlen         = sizeof(sysctl_oom_dump_tasks),
1114                 .mode           = 0644,
1115                 .proc_handler   = proc_dointvec,
1116         },
1117         {
1118                 .procname       = "overcommit_ratio",
1119                 .data           = &sysctl_overcommit_ratio,
1120                 .maxlen         = sizeof(sysctl_overcommit_ratio),
1121                 .mode           = 0644,
1122                 .proc_handler   = proc_dointvec,
1123         },
1124         {
1125                 .procname       = "page-cluster", 
1126                 .data           = &page_cluster,
1127                 .maxlen         = sizeof(int),
1128                 .mode           = 0644,
1129                 .proc_handler   = proc_dointvec_minmax,
1130                 .extra1         = &zero,
1131         },
1132         {
1133                 .procname       = "dirty_background_ratio",
1134                 .data           = &dirty_background_ratio,
1135                 .maxlen         = sizeof(dirty_background_ratio),
1136                 .mode           = 0644,
1137                 .proc_handler   = dirty_background_ratio_handler,
1138                 .extra1         = &zero,
1139                 .extra2         = &one_hundred,
1140         },
1141         {
1142                 .procname       = "dirty_background_bytes",
1143                 .data           = &dirty_background_bytes,
1144                 .maxlen         = sizeof(dirty_background_bytes),
1145                 .mode           = 0644,
1146                 .proc_handler   = dirty_background_bytes_handler,
1147                 .extra1         = &one_ul,
1148         },
1149         {
1150                 .procname       = "dirty_ratio",
1151                 .data           = &vm_dirty_ratio,
1152                 .maxlen         = sizeof(vm_dirty_ratio),
1153                 .mode           = 0644,
1154                 .proc_handler   = dirty_ratio_handler,
1155                 .extra1         = &zero,
1156                 .extra2         = &one_hundred,
1157         },
1158         {
1159                 .procname       = "dirty_bytes",
1160                 .data           = &vm_dirty_bytes,
1161                 .maxlen         = sizeof(vm_dirty_bytes),
1162                 .mode           = 0644,
1163                 .proc_handler   = dirty_bytes_handler,
1164                 .extra1         = &dirty_bytes_min,
1165         },
1166         {
1167                 .procname       = "dirty_writeback_centisecs",
1168                 .data           = &dirty_writeback_interval,
1169                 .maxlen         = sizeof(dirty_writeback_interval),
1170                 .mode           = 0644,
1171                 .proc_handler   = dirty_writeback_centisecs_handler,
1172         },
1173         {
1174                 .procname       = "dirty_expire_centisecs",
1175                 .data           = &dirty_expire_interval,
1176                 .maxlen         = sizeof(dirty_expire_interval),
1177                 .mode           = 0644,
1178                 .proc_handler   = proc_dointvec_minmax,
1179                 .extra1         = &zero,
1180         },
1181         {
1182                 .procname       = "nr_pdflush_threads",
1183                 .mode           = 0444 /* read-only */,
1184                 .proc_handler   = pdflush_proc_obsolete,
1185         },
1186         {
1187                 .procname       = "swappiness",
1188                 .data           = &vm_swappiness,
1189                 .maxlen         = sizeof(vm_swappiness),
1190                 .mode           = 0644,
1191                 .proc_handler   = proc_dointvec_minmax,
1192                 .extra1         = &zero,
1193                 .extra2         = &one_hundred,
1194         },
1195 #ifdef CONFIG_HUGETLB_PAGE
1196         {
1197                 .procname       = "nr_hugepages",
1198                 .data           = NULL,
1199                 .maxlen         = sizeof(unsigned long),
1200                 .mode           = 0644,
1201                 .proc_handler   = hugetlb_sysctl_handler,
1202                 .extra1         = (void *)&hugetlb_zero,
1203                 .extra2         = (void *)&hugetlb_infinity,
1204         },
1205 #ifdef CONFIG_NUMA
1206         {
1207                 .procname       = "nr_hugepages_mempolicy",
1208                 .data           = NULL,
1209                 .maxlen         = sizeof(unsigned long),
1210                 .mode           = 0644,
1211                 .proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1212                 .extra1         = (void *)&hugetlb_zero,
1213                 .extra2         = (void *)&hugetlb_infinity,
1214         },
1215 #endif
1216          {
1217                 .procname       = "hugetlb_shm_group",
1218                 .data           = &sysctl_hugetlb_shm_group,
1219                 .maxlen         = sizeof(gid_t),
1220                 .mode           = 0644,
1221                 .proc_handler   = proc_dointvec,
1222          },
1223          {
1224                 .procname       = "hugepages_treat_as_movable",
1225                 .data           = &hugepages_treat_as_movable,
1226                 .maxlen         = sizeof(int),
1227                 .mode           = 0644,
1228                 .proc_handler   = hugetlb_treat_movable_handler,
1229         },
1230         {
1231                 .procname       = "nr_overcommit_hugepages",
1232                 .data           = NULL,
1233                 .maxlen         = sizeof(unsigned long),
1234                 .mode           = 0644,
1235                 .proc_handler   = hugetlb_overcommit_handler,
1236                 .extra1         = (void *)&hugetlb_zero,
1237                 .extra2         = (void *)&hugetlb_infinity,
1238         },
1239 #endif
1240         {
1241                 .procname       = "lowmem_reserve_ratio",
1242                 .data           = &sysctl_lowmem_reserve_ratio,
1243                 .maxlen         = sizeof(sysctl_lowmem_reserve_ratio),
1244                 .mode           = 0644,
1245                 .proc_handler   = lowmem_reserve_ratio_sysctl_handler,
1246         },
1247         {
1248                 .procname       = "drop_caches",
1249                 .data           = &sysctl_drop_caches,
1250                 .maxlen         = sizeof(int),
1251                 .mode           = 0644,
1252                 .proc_handler   = drop_caches_sysctl_handler,
1253                 .extra1         = &one,
1254                 .extra2         = &three,
1255         },
1256 #ifdef CONFIG_COMPACTION
1257         {
1258                 .procname       = "compact_memory",
1259                 .data           = &sysctl_compact_memory,
1260                 .maxlen         = sizeof(int),
1261                 .mode           = 0200,
1262                 .proc_handler   = sysctl_compaction_handler,
1263         },
1264         {
1265                 .procname       = "extfrag_threshold",
1266                 .data           = &sysctl_extfrag_threshold,
1267                 .maxlen         = sizeof(int),
1268                 .mode           = 0644,
1269                 .proc_handler   = sysctl_extfrag_handler,
1270                 .extra1         = &min_extfrag_threshold,
1271                 .extra2         = &max_extfrag_threshold,
1272         },
1273
1274 #endif /* CONFIG_COMPACTION */
1275         {
1276                 .procname       = "min_free_kbytes",
1277                 .data           = &min_free_kbytes,
1278                 .maxlen         = sizeof(min_free_kbytes),
1279                 .mode           = 0644,
1280                 .proc_handler   = min_free_kbytes_sysctl_handler,
1281                 .extra1         = &zero,
1282         },
1283         {
1284                 .procname       = "percpu_pagelist_fraction",
1285                 .data           = &percpu_pagelist_fraction,
1286                 .maxlen         = sizeof(percpu_pagelist_fraction),
1287                 .mode           = 0644,
1288                 .proc_handler   = percpu_pagelist_fraction_sysctl_handler,
1289                 .extra1         = &min_percpu_pagelist_fract,
1290         },
1291 #ifdef CONFIG_MMU
1292         {
1293                 .procname       = "max_map_count",
1294                 .data           = &sysctl_max_map_count,
1295                 .maxlen         = sizeof(sysctl_max_map_count),
1296                 .mode           = 0644,
1297                 .proc_handler   = proc_dointvec_minmax,
1298                 .extra1         = &zero,
1299         },
1300 #else
1301         {
1302                 .procname       = "nr_trim_pages",
1303                 .data           = &sysctl_nr_trim_pages,
1304                 .maxlen         = sizeof(sysctl_nr_trim_pages),
1305                 .mode           = 0644,
1306                 .proc_handler   = proc_dointvec_minmax,
1307                 .extra1         = &zero,
1308         },
1309 #endif
1310         {
1311                 .procname       = "laptop_mode",
1312                 .data           = &laptop_mode,
1313                 .maxlen         = sizeof(laptop_mode),
1314                 .mode           = 0644,
1315                 .proc_handler   = proc_dointvec_jiffies,
1316         },
1317         {
1318                 .procname       = "block_dump",
1319                 .data           = &block_dump,
1320                 .maxlen         = sizeof(block_dump),
1321                 .mode           = 0644,
1322                 .proc_handler   = proc_dointvec,
1323                 .extra1         = &zero,
1324         },
1325         {
1326                 .procname       = "vfs_cache_pressure",
1327                 .data           = &sysctl_vfs_cache_pressure,
1328                 .maxlen         = sizeof(sysctl_vfs_cache_pressure),
1329                 .mode           = 0644,
1330                 .proc_handler   = proc_dointvec,
1331                 .extra1         = &zero,
1332         },
1333 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1334         {
1335                 .procname       = "legacy_va_layout",
1336                 .data           = &sysctl_legacy_va_layout,
1337                 .maxlen         = sizeof(sysctl_legacy_va_layout),
1338                 .mode           = 0644,
1339                 .proc_handler   = proc_dointvec,
1340                 .extra1         = &zero,
1341         },
1342 #endif
1343 #ifdef CONFIG_NUMA
1344         {
1345                 .procname       = "zone_reclaim_mode",
1346                 .data           = &zone_reclaim_mode,
1347                 .maxlen         = sizeof(zone_reclaim_mode),
1348                 .mode           = 0644,
1349                 .proc_handler   = proc_dointvec,
1350                 .extra1         = &zero,
1351         },
1352         {
1353                 .procname       = "min_unmapped_ratio",
1354                 .data           = &sysctl_min_unmapped_ratio,
1355                 .maxlen         = sizeof(sysctl_min_unmapped_ratio),
1356                 .mode           = 0644,
1357                 .proc_handler   = sysctl_min_unmapped_ratio_sysctl_handler,
1358                 .extra1         = &zero,
1359                 .extra2         = &one_hundred,
1360         },
1361         {
1362                 .procname       = "min_slab_ratio",
1363                 .data           = &sysctl_min_slab_ratio,
1364                 .maxlen         = sizeof(sysctl_min_slab_ratio),
1365                 .mode           = 0644,
1366                 .proc_handler   = sysctl_min_slab_ratio_sysctl_handler,
1367                 .extra1         = &zero,
1368                 .extra2         = &one_hundred,
1369         },
1370 #endif
1371 #ifdef CONFIG_SMP
1372         {
1373                 .procname       = "stat_interval",
1374                 .data           = &sysctl_stat_interval,
1375                 .maxlen         = sizeof(sysctl_stat_interval),
1376                 .mode           = 0644,
1377                 .proc_handler   = proc_dointvec_jiffies,
1378         },
1379 #endif
1380 #ifdef CONFIG_MMU
1381         {
1382                 .procname       = "mmap_min_addr",
1383                 .data           = &dac_mmap_min_addr,
1384                 .maxlen         = sizeof(unsigned long),
1385                 .mode           = 0644,
1386                 .proc_handler   = mmap_min_addr_handler,
1387         },
1388 #endif
1389 #ifdef CONFIG_NUMA
1390         {
1391                 .procname       = "numa_zonelist_order",
1392                 .data           = &numa_zonelist_order,
1393                 .maxlen         = NUMA_ZONELIST_ORDER_LEN,
1394                 .mode           = 0644,
1395                 .proc_handler   = numa_zonelist_order_handler,
1396         },
1397 #endif
1398 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1399    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1400         {
1401                 .procname       = "vdso_enabled",
1402                 .data           = &vdso_enabled,
1403                 .maxlen         = sizeof(vdso_enabled),
1404                 .mode           = 0644,
1405                 .proc_handler   = proc_dointvec,
1406                 .extra1         = &zero,
1407         },
1408 #endif
1409 #ifdef CONFIG_HIGHMEM
1410         {
1411                 .procname       = "highmem_is_dirtyable",
1412                 .data           = &vm_highmem_is_dirtyable,
1413                 .maxlen         = sizeof(vm_highmem_is_dirtyable),
1414                 .mode           = 0644,
1415                 .proc_handler   = proc_dointvec_minmax,
1416                 .extra1         = &zero,
1417                 .extra2         = &one,
1418         },
1419 #endif
1420         {
1421                 .procname       = "scan_unevictable_pages",
1422                 .data           = &scan_unevictable_pages,
1423                 .maxlen         = sizeof(scan_unevictable_pages),
1424                 .mode           = 0644,
1425                 .proc_handler   = scan_unevictable_handler,
1426         },
1427 #ifdef CONFIG_MEMORY_FAILURE
1428         {
1429                 .procname       = "memory_failure_early_kill",
1430                 .data           = &sysctl_memory_failure_early_kill,
1431                 .maxlen         = sizeof(sysctl_memory_failure_early_kill),
1432                 .mode           = 0644,
1433                 .proc_handler   = proc_dointvec_minmax,
1434                 .extra1         = &zero,
1435                 .extra2         = &one,
1436         },
1437         {
1438                 .procname       = "memory_failure_recovery",
1439                 .data           = &sysctl_memory_failure_recovery,
1440                 .maxlen         = sizeof(sysctl_memory_failure_recovery),
1441                 .mode           = 0644,
1442                 .proc_handler   = proc_dointvec_minmax,
1443                 .extra1         = &zero,
1444                 .extra2         = &one,
1445         },
1446 #endif
1447         {
1448                 .procname       = "user_reserve_kbytes",
1449                 .data           = &sysctl_user_reserve_kbytes,
1450                 .maxlen         = sizeof(sysctl_user_reserve_kbytes),
1451                 .mode           = 0644,
1452                 .proc_handler   = proc_doulongvec_minmax,
1453         },
1454         {
1455                 .procname       = "admin_reserve_kbytes",
1456                 .data           = &sysctl_admin_reserve_kbytes,
1457                 .maxlen         = sizeof(sysctl_admin_reserve_kbytes),
1458                 .mode           = 0644,
1459                 .proc_handler   = proc_doulongvec_minmax,
1460         },
1461         { }
1462 };
1463
1464 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1465 static struct ctl_table binfmt_misc_table[] = {
1466         { }
1467 };
1468 #endif
1469
1470 static struct ctl_table fs_table[] = {
1471         {
1472                 .procname       = "inode-nr",
1473                 .data           = &inodes_stat,
1474                 .maxlen         = 2*sizeof(int),
1475                 .mode           = 0444,
1476                 .proc_handler   = proc_nr_inodes,
1477         },
1478         {
1479                 .procname       = "inode-state",
1480                 .data           = &inodes_stat,
1481                 .maxlen         = 7*sizeof(int),
1482                 .mode           = 0444,
1483                 .proc_handler   = proc_nr_inodes,
1484         },
1485         {
1486                 .procname       = "file-nr",
1487                 .data           = &files_stat,
1488                 .maxlen         = sizeof(files_stat),
1489                 .mode           = 0444,
1490                 .proc_handler   = proc_nr_files,
1491         },
1492         {
1493                 .procname       = "file-max",
1494                 .data           = &files_stat.max_files,
1495                 .maxlen         = sizeof(files_stat.max_files),
1496                 .mode           = 0644,
1497                 .proc_handler   = proc_doulongvec_minmax,
1498         },
1499         {
1500                 .procname       = "nr_open",
1501                 .data           = &sysctl_nr_open,
1502                 .maxlen         = sizeof(int),
1503                 .mode           = 0644,
1504                 .proc_handler   = proc_dointvec_minmax,
1505                 .extra1         = &sysctl_nr_open_min,
1506                 .extra2         = &sysctl_nr_open_max,
1507         },
1508         {
1509                 .procname       = "dentry-state",
1510                 .data           = &dentry_stat,
1511                 .maxlen         = 6*sizeof(int),
1512                 .mode           = 0444,
1513                 .proc_handler   = proc_nr_dentry,
1514         },
1515         {
1516                 .procname       = "overflowuid",
1517                 .data           = &fs_overflowuid,
1518                 .maxlen         = sizeof(int),
1519                 .mode           = 0644,
1520                 .proc_handler   = proc_dointvec_minmax,
1521                 .extra1         = &minolduid,
1522                 .extra2         = &maxolduid,
1523         },
1524         {
1525                 .procname       = "overflowgid",
1526                 .data           = &fs_overflowgid,
1527                 .maxlen         = sizeof(int),
1528                 .mode           = 0644,
1529                 .proc_handler   = proc_dointvec_minmax,
1530                 .extra1         = &minolduid,
1531                 .extra2         = &maxolduid,
1532         },
1533 #ifdef CONFIG_FILE_LOCKING
1534         {
1535                 .procname       = "leases-enable",
1536                 .data           = &leases_enable,
1537                 .maxlen         = sizeof(int),
1538                 .mode           = 0644,
1539                 .proc_handler   = proc_dointvec,
1540         },
1541 #endif
1542 #ifdef CONFIG_DNOTIFY
1543         {
1544                 .procname       = "dir-notify-enable",
1545                 .data           = &dir_notify_enable,
1546                 .maxlen         = sizeof(int),
1547                 .mode           = 0644,
1548                 .proc_handler   = proc_dointvec,
1549         },
1550 #endif
1551 #ifdef CONFIG_MMU
1552 #ifdef CONFIG_FILE_LOCKING
1553         {
1554                 .procname       = "lease-break-time",
1555                 .data           = &lease_break_time,
1556                 .maxlen         = sizeof(int),
1557                 .mode           = 0644,
1558                 .proc_handler   = proc_dointvec,
1559         },
1560 #endif
1561 #ifdef CONFIG_AIO
1562         {
1563                 .procname       = "aio-nr",
1564                 .data           = &aio_nr,
1565                 .maxlen         = sizeof(aio_nr),
1566                 .mode           = 0444,
1567                 .proc_handler   = proc_doulongvec_minmax,
1568         },
1569         {
1570                 .procname       = "aio-max-nr",
1571                 .data           = &aio_max_nr,
1572                 .maxlen         = sizeof(aio_max_nr),
1573                 .mode           = 0644,
1574                 .proc_handler   = proc_doulongvec_minmax,
1575         },
1576 #endif /* CONFIG_AIO */
1577 #ifdef CONFIG_INOTIFY_USER
1578         {
1579                 .procname       = "inotify",
1580                 .mode           = 0555,
1581                 .child          = inotify_table,
1582         },
1583 #endif  
1584 #ifdef CONFIG_EPOLL
1585         {
1586                 .procname       = "epoll",
1587                 .mode           = 0555,
1588                 .child          = epoll_table,
1589         },
1590 #endif
1591 #endif
1592         {
1593                 .procname       = "protected_symlinks",
1594                 .data           = &sysctl_protected_symlinks,
1595                 .maxlen         = sizeof(int),
1596                 .mode           = 0600,
1597                 .proc_handler   = proc_dointvec_minmax,
1598                 .extra1         = &zero,
1599                 .extra2         = &one,
1600         },
1601         {
1602                 .procname       = "protected_hardlinks",
1603                 .data           = &sysctl_protected_hardlinks,
1604                 .maxlen         = sizeof(int),
1605                 .mode           = 0600,
1606                 .proc_handler   = proc_dointvec_minmax,
1607                 .extra1         = &zero,
1608                 .extra2         = &one,
1609         },
1610         {
1611                 .procname       = "suid_dumpable",
1612                 .data           = &suid_dumpable,
1613                 .maxlen         = sizeof(int),
1614                 .mode           = 0644,
1615                 .proc_handler   = proc_dointvec_minmax_coredump,
1616                 .extra1         = &zero,
1617                 .extra2         = &two,
1618         },
1619 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1620         {
1621                 .procname       = "binfmt_misc",
1622                 .mode           = 0555,
1623                 .child          = binfmt_misc_table,
1624         },
1625 #endif
1626         {
1627                 .procname       = "pipe-max-size",
1628                 .data           = &pipe_max_size,
1629                 .maxlen         = sizeof(int),
1630                 .mode           = 0644,
1631                 .proc_handler   = &pipe_proc_fn,
1632                 .extra1         = &pipe_min_size,
1633         },
1634         { }
1635 };
1636
1637 static struct ctl_table debug_table[] = {
1638 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1639         {
1640                 .procname       = "exception-trace",
1641                 .data           = &show_unhandled_signals,
1642                 .maxlen         = sizeof(int),
1643                 .mode           = 0644,
1644                 .proc_handler   = proc_dointvec
1645         },
1646 #endif
1647 #if defined(CONFIG_OPTPROBES)
1648         {
1649                 .procname       = "kprobes-optimization",
1650                 .data           = &sysctl_kprobes_optimization,
1651                 .maxlen         = sizeof(int),
1652                 .mode           = 0644,
1653                 .proc_handler   = proc_kprobes_optimization_handler,
1654                 .extra1         = &zero,
1655                 .extra2         = &one,
1656         },
1657 #endif
1658         { }
1659 };
1660
1661 static struct ctl_table dev_table[] = {
1662         { }
1663 };
1664
1665 int __init sysctl_init(void)
1666 {
1667         struct ctl_table_header *hdr;
1668
1669         hdr = register_sysctl_table(sysctl_base_table);
1670         kmemleak_not_leak(hdr);
1671         return 0;
1672 }
1673
1674 #endif /* CONFIG_SYSCTL */
1675
1676 /*
1677  * /proc/sys support
1678  */
1679
1680 #ifdef CONFIG_PROC_SYSCTL
1681
1682 static int _proc_do_string(void* data, int maxlen, int write,
1683                            void __user *buffer,
1684                            size_t *lenp, loff_t *ppos)
1685 {
1686         size_t len;
1687         char __user *p;
1688         char c;
1689
1690         if (!data || !maxlen || !*lenp) {
1691                 *lenp = 0;
1692                 return 0;
1693         }
1694
1695         if (write) {
1696                 len = 0;
1697                 p = buffer;
1698                 while (len < *lenp) {
1699                         if (get_user(c, p++))
1700                                 return -EFAULT;
1701                         if (c == 0 || c == '\n')
1702                                 break;
1703                         len++;
1704                 }
1705                 if (len >= maxlen)
1706                         len = maxlen-1;
1707                 if(copy_from_user(data, buffer, len))
1708                         return -EFAULT;
1709                 ((char *) data)[len] = 0;
1710                 *ppos += *lenp;
1711         } else {
1712                 len = strlen(data);
1713                 if (len > maxlen)
1714                         len = maxlen;
1715
1716                 if (*ppos > len) {
1717                         *lenp = 0;
1718                         return 0;
1719                 }
1720
1721                 data += *ppos;
1722                 len  -= *ppos;
1723
1724                 if (len > *lenp)
1725                         len = *lenp;
1726                 if (len)
1727                         if(copy_to_user(buffer, data, len))
1728                                 return -EFAULT;
1729                 if (len < *lenp) {
1730                         if(put_user('\n', ((char __user *) buffer) + len))
1731                                 return -EFAULT;
1732                         len++;
1733                 }
1734                 *lenp = len;
1735                 *ppos += len;
1736         }
1737         return 0;
1738 }
1739
1740 /**
1741  * proc_dostring - read a string sysctl
1742  * @table: the sysctl table
1743  * @write: %TRUE if this is a write to the sysctl file
1744  * @buffer: the user buffer
1745  * @lenp: the size of the user buffer
1746  * @ppos: file position
1747  *
1748  * Reads/writes a string from/to the user buffer. If the kernel
1749  * buffer provided is not large enough to hold the string, the
1750  * string is truncated. The copied string is %NULL-terminated.
1751  * If the string is being read by the user process, it is copied
1752  * and a newline '\n' is added. It is truncated if the buffer is
1753  * not large enough.
1754  *
1755  * Returns 0 on success.
1756  */
1757 int proc_dostring(struct ctl_table *table, int write,
1758                   void __user *buffer, size_t *lenp, loff_t *ppos)
1759 {
1760         return _proc_do_string(table->data, table->maxlen, write,
1761                                buffer, lenp, ppos);
1762 }
1763
1764 static size_t proc_skip_spaces(char **buf)
1765 {
1766         size_t ret;
1767         char *tmp = skip_spaces(*buf);
1768         ret = tmp - *buf;
1769         *buf = tmp;
1770         return ret;
1771 }
1772
1773 static void proc_skip_char(char **buf, size_t *size, const char v)
1774 {
1775         while (*size) {
1776                 if (**buf != v)
1777                         break;
1778                 (*size)--;
1779                 (*buf)++;
1780         }
1781 }
1782
1783 #define TMPBUFLEN 22
1784 /**
1785  * proc_get_long - reads an ASCII formatted integer from a user buffer
1786  *
1787  * @buf: a kernel buffer
1788  * @size: size of the kernel buffer
1789  * @val: this is where the number will be stored
1790  * @neg: set to %TRUE if number is negative
1791  * @perm_tr: a vector which contains the allowed trailers
1792  * @perm_tr_len: size of the perm_tr vector
1793  * @tr: pointer to store the trailer character
1794  *
1795  * In case of success %0 is returned and @buf and @size are updated with
1796  * the amount of bytes read. If @tr is non-NULL and a trailing
1797  * character exists (size is non-zero after returning from this
1798  * function), @tr is updated with the trailing character.
1799  */
1800 static int proc_get_long(char **buf, size_t *size,
1801                           unsigned long *val, bool *neg,
1802                           const char *perm_tr, unsigned perm_tr_len, char *tr)
1803 {
1804         int len;
1805         char *p, tmp[TMPBUFLEN];
1806
1807         if (!*size)
1808                 return -EINVAL;
1809
1810         len = *size;
1811         if (len > TMPBUFLEN - 1)
1812                 len = TMPBUFLEN - 1;
1813
1814         memcpy(tmp, *buf, len);
1815
1816         tmp[len] = 0;
1817         p = tmp;
1818         if (*p == '-' && *size > 1) {
1819                 *neg = true;
1820                 p++;
1821         } else
1822                 *neg = false;
1823         if (!isdigit(*p))
1824                 return -EINVAL;
1825
1826         *val = simple_strtoul(p, &p, 0);
1827
1828         len = p - tmp;
1829
1830         /* We don't know if the next char is whitespace thus we may accept
1831          * invalid integers (e.g. 1234...a) or two integers instead of one
1832          * (e.g. 123...1). So lets not allow such large numbers. */
1833         if (len == TMPBUFLEN - 1)
1834                 return -EINVAL;
1835
1836         if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
1837                 return -EINVAL;
1838
1839         if (tr && (len < *size))
1840                 *tr = *p;
1841
1842         *buf += len;
1843         *size -= len;
1844
1845         return 0;
1846 }
1847
1848 /**
1849  * proc_put_long - converts an integer to a decimal ASCII formatted string
1850  *
1851  * @buf: the user buffer
1852  * @size: the size of the user buffer
1853  * @val: the integer to be converted
1854  * @neg: sign of the number, %TRUE for negative
1855  *
1856  * In case of success %0 is returned and @buf and @size are updated with
1857  * the amount of bytes written.
1858  */
1859 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
1860                           bool neg)
1861 {
1862         int len;
1863         char tmp[TMPBUFLEN], *p = tmp;
1864
1865         sprintf(p, "%s%lu", neg ? "-" : "", val);
1866         len = strlen(tmp);
1867         if (len > *size)
1868                 len = *size;
1869         if (copy_to_user(*buf, tmp, len))
1870                 return -EFAULT;
1871         *size -= len;
1872         *buf += len;
1873         return 0;
1874 }
1875 #undef TMPBUFLEN
1876
1877 static int proc_put_char(void __user **buf, size_t *size, char c)
1878 {
1879         if (*size) {
1880                 char __user **buffer = (char __user **)buf;
1881                 if (put_user(c, *buffer))
1882                         return -EFAULT;
1883                 (*size)--, (*buffer)++;
1884                 *buf = *buffer;
1885         }
1886         return 0;
1887 }
1888
1889 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
1890                                  int *valp,
1891                                  int write, void *data)
1892 {
1893         if (write) {
1894                 *valp = *negp ? -*lvalp : *lvalp;
1895         } else {
1896                 int val = *valp;
1897                 if (val < 0) {
1898                         *negp = true;
1899                         *lvalp = (unsigned long)-val;
1900                 } else {
1901                         *negp = false;
1902                         *lvalp = (unsigned long)val;
1903                 }
1904         }
1905         return 0;
1906 }
1907
1908 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
1909
1910 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1911                   int write, void __user *buffer,
1912                   size_t *lenp, loff_t *ppos,
1913                   int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1914                               int write, void *data),
1915                   void *data)
1916 {
1917         int *i, vleft, first = 1, err = 0;
1918         unsigned long page = 0;
1919         size_t left;
1920         char *kbuf;
1921         
1922         if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
1923                 *lenp = 0;
1924                 return 0;
1925         }
1926         
1927         i = (int *) tbl_data;
1928         vleft = table->maxlen / sizeof(*i);
1929         left = *lenp;
1930
1931         if (!conv)
1932                 conv = do_proc_dointvec_conv;
1933
1934         if (write) {
1935                 if (left > PAGE_SIZE - 1)
1936                         left = PAGE_SIZE - 1;
1937                 page = __get_free_page(GFP_TEMPORARY);
1938                 kbuf = (char *) page;
1939                 if (!kbuf)
1940                         return -ENOMEM;
1941                 if (copy_from_user(kbuf, buffer, left)) {
1942                         err = -EFAULT;
1943                         goto free;
1944                 }
1945                 kbuf[left] = 0;
1946         }
1947
1948         for (; left && vleft--; i++, first=0) {
1949                 unsigned long lval;
1950                 bool neg;
1951
1952                 if (write) {
1953                         left -= proc_skip_spaces(&kbuf);
1954
1955                         if (!left)
1956                                 break;
1957                         err = proc_get_long(&kbuf, &left, &lval, &neg,
1958                                              proc_wspace_sep,
1959                                              sizeof(proc_wspace_sep), NULL);
1960                         if (err)
1961                                 break;
1962                         if (conv(&neg, &lval, i, 1, data)) {
1963                                 err = -EINVAL;
1964                                 break;
1965                         }
1966                 } else {
1967                         if (conv(&neg, &lval, i, 0, data)) {
1968                                 err = -EINVAL;
1969                                 break;
1970                         }
1971                         if (!first)
1972                                 err = proc_put_char(&buffer, &left, '\t');
1973                         if (err)
1974                                 break;
1975                         err = proc_put_long(&buffer, &left, lval, neg);
1976                         if (err)
1977                                 break;
1978                 }
1979         }
1980
1981         if (!write && !first && left && !err)
1982                 err = proc_put_char(&buffer, &left, '\n');
1983         if (write && !err && left)
1984                 left -= proc_skip_spaces(&kbuf);
1985 free:
1986         if (write) {
1987                 free_page(page);
1988                 if (first)
1989                         return err ? : -EINVAL;
1990         }
1991         *lenp -= left;
1992         *ppos += *lenp;
1993         return err;
1994 }
1995
1996 static int do_proc_dointvec(struct ctl_table *table, int write,
1997                   void __user *buffer, size_t *lenp, loff_t *ppos,
1998                   int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1999                               int write, void *data),
2000                   void *data)
2001 {
2002         return __do_proc_dointvec(table->data, table, write,
2003                         buffer, lenp, ppos, conv, data);
2004 }
2005
2006 /**
2007  * proc_dointvec - read a vector of integers
2008  * @table: the sysctl table
2009  * @write: %TRUE if this is a write to the sysctl file
2010  * @buffer: the user buffer
2011  * @lenp: the size of the user buffer
2012  * @ppos: file position
2013  *
2014  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2015  * values from/to the user buffer, treated as an ASCII string. 
2016  *
2017  * Returns 0 on success.
2018  */
2019 int proc_dointvec(struct ctl_table *table, int write,
2020                      void __user *buffer, size_t *lenp, loff_t *ppos)
2021 {
2022     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2023                             NULL,NULL);
2024 }
2025
2026 /*
2027  * Taint values can only be increased
2028  * This means we can safely use a temporary.
2029  */
2030 static int proc_taint(struct ctl_table *table, int write,
2031                                void __user *buffer, size_t *lenp, loff_t *ppos)
2032 {
2033         struct ctl_table t;
2034         unsigned long tmptaint = get_taint();
2035         int err;
2036
2037         if (write && !capable(CAP_SYS_ADMIN))
2038                 return -EPERM;
2039
2040         t = *table;
2041         t.data = &tmptaint;
2042         err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2043         if (err < 0)
2044                 return err;
2045
2046         if (write) {
2047                 /*
2048                  * Poor man's atomic or. Not worth adding a primitive
2049                  * to everyone's atomic.h for this
2050                  */
2051                 int i;
2052                 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2053                         if ((tmptaint >> i) & 1)
2054                                 add_taint(i, LOCKDEP_STILL_OK);
2055                 }
2056         }
2057
2058         return err;
2059 }
2060
2061 #ifdef CONFIG_PRINTK
2062 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
2063                                 void __user *buffer, size_t *lenp, loff_t *ppos)
2064 {
2065         if (write && !capable(CAP_SYS_ADMIN))
2066                 return -EPERM;
2067
2068         return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2069 }
2070 #endif
2071
2072 struct do_proc_dointvec_minmax_conv_param {
2073         int *min;
2074         int *max;
2075 };
2076
2077 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2078                                         int *valp,
2079                                         int write, void *data)
2080 {
2081         struct do_proc_dointvec_minmax_conv_param *param = data;
2082         if (write) {
2083                 int val = *negp ? -*lvalp : *lvalp;
2084                 if ((param->min && *param->min > val) ||
2085                     (param->max && *param->max < val))
2086                         return -EINVAL;
2087                 *valp = val;
2088         } else {
2089                 int val = *valp;
2090                 if (val < 0) {
2091                         *negp = true;
2092                         *lvalp = (unsigned long)-val;
2093                 } else {
2094                         *negp = false;
2095                         *lvalp = (unsigned long)val;
2096                 }
2097         }
2098         return 0;
2099 }
2100
2101 /**
2102  * proc_dointvec_minmax - read a vector of integers with min/max values
2103  * @table: the sysctl table
2104  * @write: %TRUE if this is a write to the sysctl file
2105  * @buffer: the user buffer
2106  * @lenp: the size of the user buffer
2107  * @ppos: file position
2108  *
2109  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2110  * values from/to the user buffer, treated as an ASCII string.
2111  *
2112  * This routine will ensure the values are within the range specified by
2113  * table->extra1 (min) and table->extra2 (max).
2114  *
2115  * Returns 0 on success.
2116  */
2117 int proc_dointvec_minmax(struct ctl_table *table, int write,
2118                   void __user *buffer, size_t *lenp, loff_t *ppos)
2119 {
2120         struct do_proc_dointvec_minmax_conv_param param = {
2121                 .min = (int *) table->extra1,
2122                 .max = (int *) table->extra2,
2123         };
2124         return do_proc_dointvec(table, write, buffer, lenp, ppos,
2125                                 do_proc_dointvec_minmax_conv, &param);
2126 }
2127
2128 static void validate_coredump_safety(void)
2129 {
2130 #ifdef CONFIG_COREDUMP
2131         if (suid_dumpable == SUID_DUMP_ROOT &&
2132             core_pattern[0] != '/' && core_pattern[0] != '|') {
2133                 printk(KERN_WARNING "Unsafe core_pattern used with "\
2134                         "suid_dumpable=2. Pipe handler or fully qualified "\
2135                         "core dump path required.\n");
2136         }
2137 #endif
2138 }
2139
2140 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2141                 void __user *buffer, size_t *lenp, loff_t *ppos)
2142 {
2143         int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2144         if (!error)
2145                 validate_coredump_safety();
2146         return error;
2147 }
2148
2149 #ifdef CONFIG_COREDUMP
2150 static int proc_dostring_coredump(struct ctl_table *table, int write,
2151                   void __user *buffer, size_t *lenp, loff_t *ppos)
2152 {
2153         int error = proc_dostring(table, write, buffer, lenp, ppos);
2154         if (!error)
2155                 validate_coredump_safety();
2156         return error;
2157 }
2158 #endif
2159
2160 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2161                                      void __user *buffer,
2162                                      size_t *lenp, loff_t *ppos,
2163                                      unsigned long convmul,
2164                                      unsigned long convdiv)
2165 {
2166         unsigned long *i, *min, *max;
2167         int vleft, first = 1, err = 0;
2168         unsigned long page = 0;
2169         size_t left;
2170         char *kbuf;
2171
2172         if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2173                 *lenp = 0;
2174                 return 0;
2175         }
2176
2177         i = (unsigned long *) data;
2178         min = (unsigned long *) table->extra1;
2179         max = (unsigned long *) table->extra2;
2180         vleft = table->maxlen / sizeof(unsigned long);
2181         left = *lenp;
2182
2183         if (write) {
2184                 if (left > PAGE_SIZE - 1)
2185                         left = PAGE_SIZE - 1;
2186                 page = __get_free_page(GFP_TEMPORARY);
2187                 kbuf = (char *) page;
2188                 if (!kbuf)
2189                         return -ENOMEM;
2190                 if (copy_from_user(kbuf, buffer, left)) {
2191                         err = -EFAULT;
2192                         goto free;
2193                 }
2194                 kbuf[left] = 0;
2195         }
2196
2197         for (; left && vleft--; i++, first = 0) {
2198                 unsigned long val;
2199
2200                 if (write) {
2201                         bool neg;
2202
2203                         left -= proc_skip_spaces(&kbuf);
2204
2205                         err = proc_get_long(&kbuf, &left, &val, &neg,
2206                                              proc_wspace_sep,
2207                                              sizeof(proc_wspace_sep), NULL);
2208                         if (err)
2209                                 break;
2210                         if (neg)
2211                                 continue;
2212                         if ((min && val < *min) || (max && val > *max))
2213                                 continue;
2214                         *i = val;
2215                 } else {
2216                         val = convdiv * (*i) / convmul;
2217                         if (!first)
2218                                 err = proc_put_char(&buffer, &left, '\t');
2219                         err = proc_put_long(&buffer, &left, val, false);
2220                         if (err)
2221                                 break;
2222                 }
2223         }
2224
2225         if (!write && !first && left && !err)
2226                 err = proc_put_char(&buffer, &left, '\n');
2227         if (write && !err)
2228                 left -= proc_skip_spaces(&kbuf);
2229 free:
2230         if (write) {
2231                 free_page(page);
2232                 if (first)
2233                         return err ? : -EINVAL;
2234         }
2235         *lenp -= left;
2236         *ppos += *lenp;
2237         return err;
2238 }
2239
2240 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2241                                      void __user *buffer,
2242                                      size_t *lenp, loff_t *ppos,
2243                                      unsigned long convmul,
2244                                      unsigned long convdiv)
2245 {
2246         return __do_proc_doulongvec_minmax(table->data, table, write,
2247                         buffer, lenp, ppos, convmul, convdiv);
2248 }
2249
2250 /**
2251  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2252  * @table: the sysctl table
2253  * @write: %TRUE if this is a write to the sysctl file
2254  * @buffer: the user buffer
2255  * @lenp: the size of the user buffer
2256  * @ppos: file position
2257  *
2258  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2259  * values from/to the user buffer, treated as an ASCII string.
2260  *
2261  * This routine will ensure the values are within the range specified by
2262  * table->extra1 (min) and table->extra2 (max).
2263  *
2264  * Returns 0 on success.
2265  */
2266 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2267                            void __user *buffer, size_t *lenp, loff_t *ppos)
2268 {
2269     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2270 }
2271
2272 /**
2273  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2274  * @table: the sysctl table
2275  * @write: %TRUE if this is a write to the sysctl file
2276  * @buffer: the user buffer
2277  * @lenp: the size of the user buffer
2278  * @ppos: file position
2279  *
2280  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2281  * values from/to the user buffer, treated as an ASCII string. The values
2282  * are treated as milliseconds, and converted to jiffies when they are stored.
2283  *
2284  * This routine will ensure the values are within the range specified by
2285  * table->extra1 (min) and table->extra2 (max).
2286  *
2287  * Returns 0 on success.
2288  */
2289 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2290                                       void __user *buffer,
2291                                       size_t *lenp, loff_t *ppos)
2292 {
2293     return do_proc_doulongvec_minmax(table, write, buffer,
2294                                      lenp, ppos, HZ, 1000l);
2295 }
2296
2297
2298 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2299                                          int *valp,
2300                                          int write, void *data)
2301 {
2302         if (write) {
2303                 if (*lvalp > LONG_MAX / HZ)
2304                         return 1;
2305                 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2306         } else {
2307                 int val = *valp;
2308                 unsigned long lval;
2309                 if (val < 0) {
2310                         *negp = true;
2311                         lval = (unsigned long)-val;
2312                 } else {
2313                         *negp = false;
2314                         lval = (unsigned long)val;
2315                 }
2316                 *lvalp = lval / HZ;
2317         }
2318         return 0;
2319 }
2320
2321 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2322                                                 int *valp,
2323                                                 int write, void *data)
2324 {
2325         if (write) {
2326                 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2327                         return 1;
2328                 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2329         } else {
2330                 int val = *valp;
2331                 unsigned long lval;
2332                 if (val < 0) {
2333                         *negp = true;
2334                         lval = (unsigned long)-val;
2335                 } else {
2336                         *negp = false;
2337                         lval = (unsigned long)val;
2338                 }
2339                 *lvalp = jiffies_to_clock_t(lval);
2340         }
2341         return 0;
2342 }
2343
2344 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2345                                             int *valp,
2346                                             int write, void *data)
2347 {
2348         if (write) {
2349                 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2350         } else {
2351                 int val = *valp;
2352                 unsigned long lval;
2353                 if (val < 0) {
2354                         *negp = true;
2355                         lval = (unsigned long)-val;
2356                 } else {
2357                         *negp = false;
2358                         lval = (unsigned long)val;
2359                 }
2360                 *lvalp = jiffies_to_msecs(lval);
2361         }
2362         return 0;
2363 }
2364
2365 /**
2366  * proc_dointvec_jiffies - read a vector of integers as seconds
2367  * @table: the sysctl table
2368  * @write: %TRUE if this is a write to the sysctl file
2369  * @buffer: the user buffer
2370  * @lenp: the size of the user buffer
2371  * @ppos: file position
2372  *
2373  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2374  * values from/to the user buffer, treated as an ASCII string. 
2375  * The values read are assumed to be in seconds, and are converted into
2376  * jiffies.
2377  *
2378  * Returns 0 on success.
2379  */
2380 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2381                           void __user *buffer, size_t *lenp, loff_t *ppos)
2382 {
2383     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2384                             do_proc_dointvec_jiffies_conv,NULL);
2385 }
2386
2387 /**
2388  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2389  * @table: the sysctl table
2390  * @write: %TRUE if this is a write to the sysctl file
2391  * @buffer: the user buffer
2392  * @lenp: the size of the user buffer
2393  * @ppos: pointer to the file position
2394  *
2395  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2396  * values from/to the user buffer, treated as an ASCII string. 
2397  * The values read are assumed to be in 1/USER_HZ seconds, and 
2398  * are converted into jiffies.
2399  *
2400  * Returns 0 on success.
2401  */
2402 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2403                                  void __user *buffer, size_t *lenp, loff_t *ppos)
2404 {
2405     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2406                             do_proc_dointvec_userhz_jiffies_conv,NULL);
2407 }
2408
2409 /**
2410  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2411  * @table: the sysctl table
2412  * @write: %TRUE if this is a write to the sysctl file
2413  * @buffer: the user buffer
2414  * @lenp: the size of the user buffer
2415  * @ppos: file position
2416  * @ppos: the current position in the file
2417  *
2418  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2419  * values from/to the user buffer, treated as an ASCII string. 
2420  * The values read are assumed to be in 1/1000 seconds, and 
2421  * are converted into jiffies.
2422  *
2423  * Returns 0 on success.
2424  */
2425 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2426                              void __user *buffer, size_t *lenp, loff_t *ppos)
2427 {
2428         return do_proc_dointvec(table, write, buffer, lenp, ppos,
2429                                 do_proc_dointvec_ms_jiffies_conv, NULL);
2430 }
2431
2432 static int proc_do_cad_pid(struct ctl_table *table, int write,
2433                            void __user *buffer, size_t *lenp, loff_t *ppos)
2434 {
2435         struct pid *new_pid;
2436         pid_t tmp;
2437         int r;
2438
2439         tmp = pid_vnr(cad_pid);
2440
2441         r = __do_proc_dointvec(&tmp, table, write, buffer,
2442                                lenp, ppos, NULL, NULL);
2443         if (r || !write)
2444                 return r;
2445
2446         new_pid = find_get_pid(tmp);
2447         if (!new_pid)
2448                 return -ESRCH;
2449
2450         put_pid(xchg(&cad_pid, new_pid));
2451         return 0;
2452 }
2453
2454 /**
2455  * proc_do_large_bitmap - read/write from/to a large bitmap
2456  * @table: the sysctl table
2457  * @write: %TRUE if this is a write to the sysctl file
2458  * @buffer: the user buffer
2459  * @lenp: the size of the user buffer
2460  * @ppos: file position
2461  *
2462  * The bitmap is stored at table->data and the bitmap length (in bits)
2463  * in table->maxlen.
2464  *
2465  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2466  * large bitmaps may be represented in a compact manner. Writing into
2467  * the file will clear the bitmap then update it with the given input.
2468  *
2469  * Returns 0 on success.
2470  */
2471 int proc_do_large_bitmap(struct ctl_table *table, int write,
2472                          void __user *buffer, size_t *lenp, loff_t *ppos)
2473 {
2474         int err = 0;
2475         bool first = 1;
2476         size_t left = *lenp;
2477         unsigned long bitmap_len = table->maxlen;
2478         unsigned long *bitmap = (unsigned long *) table->data;
2479         unsigned long *tmp_bitmap = NULL;
2480         char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2481
2482         if (!bitmap_len || !left || (*ppos && !write)) {
2483                 *lenp = 0;
2484                 return 0;
2485         }
2486
2487         if (write) {
2488                 unsigned long page = 0;
2489                 char *kbuf;
2490
2491                 if (left > PAGE_SIZE - 1)
2492                         left = PAGE_SIZE - 1;
2493
2494                 page = __get_free_page(GFP_TEMPORARY);
2495                 kbuf = (char *) page;
2496                 if (!kbuf)
2497                         return -ENOMEM;
2498                 if (copy_from_user(kbuf, buffer, left)) {
2499                         free_page(page);
2500                         return -EFAULT;
2501                 }
2502                 kbuf[left] = 0;
2503
2504                 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2505                                      GFP_KERNEL);
2506                 if (!tmp_bitmap) {
2507                         free_page(page);
2508                         return -ENOMEM;
2509                 }
2510                 proc_skip_char(&kbuf, &left, '\n');
2511                 while (!err && left) {
2512                         unsigned long val_a, val_b;
2513                         bool neg;
2514
2515                         err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2516                                              sizeof(tr_a), &c);
2517                         if (err)
2518                                 break;
2519                         if (val_a >= bitmap_len || neg) {
2520                                 err = -EINVAL;
2521                                 break;
2522                         }
2523
2524                         val_b = val_a;
2525                         if (left) {
2526                                 kbuf++;
2527                                 left--;
2528                         }
2529
2530                         if (c == '-') {
2531                                 err = proc_get_long(&kbuf, &left, &val_b,
2532                                                      &neg, tr_b, sizeof(tr_b),
2533                                                      &c);
2534                                 if (err)
2535                                         break;
2536                                 if (val_b >= bitmap_len || neg ||
2537                                     val_a > val_b) {
2538                                         err = -EINVAL;
2539                                         break;
2540                                 }
2541                                 if (left) {
2542                                         kbuf++;
2543                                         left--;
2544                                 }
2545                         }
2546
2547                         bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2548                         first = 0;
2549                         proc_skip_char(&kbuf, &left, '\n');
2550                 }
2551                 free_page(page);
2552         } else {
2553                 unsigned long bit_a, bit_b = 0;
2554
2555                 while (left) {
2556                         bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2557                         if (bit_a >= bitmap_len)
2558                                 break;
2559                         bit_b = find_next_zero_bit(bitmap, bitmap_len,
2560                                                    bit_a + 1) - 1;
2561
2562                         if (!first) {
2563                                 err = proc_put_char(&buffer, &left, ',');
2564                                 if (err)
2565                                         break;
2566                         }
2567                         err = proc_put_long(&buffer, &left, bit_a, false);
2568                         if (err)
2569                                 break;
2570                         if (bit_a != bit_b) {
2571                                 err = proc_put_char(&buffer, &left, '-');
2572                                 if (err)
2573                                         break;
2574                                 err = proc_put_long(&buffer, &left, bit_b, false);
2575                                 if (err)
2576                                         break;
2577                         }
2578
2579                         first = 0; bit_b++;
2580                 }
2581                 if (!err)
2582                         err = proc_put_char(&buffer, &left, '\n');
2583         }
2584
2585         if (!err) {
2586                 if (write) {
2587                         if (*ppos)
2588                                 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2589                         else
2590                                 bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2591                 }
2592                 kfree(tmp_bitmap);
2593                 *lenp -= left;
2594                 *ppos += *lenp;
2595                 return 0;
2596         } else {
2597                 kfree(tmp_bitmap);
2598                 return err;
2599         }
2600 }
2601
2602 #else /* CONFIG_PROC_SYSCTL */
2603
2604 int proc_dostring(struct ctl_table *table, int write,
2605                   void __user *buffer, size_t *lenp, loff_t *ppos)
2606 {
2607         return -ENOSYS;
2608 }
2609
2610 int proc_dointvec(struct ctl_table *table, int write,
2611                   void __user *buffer, size_t *lenp, loff_t *ppos)
2612 {
2613         return -ENOSYS;
2614 }
2615
2616 int proc_dointvec_minmax(struct ctl_table *table, int write,
2617                     void __user *buffer, size_t *lenp, loff_t *ppos)
2618 {
2619         return -ENOSYS;
2620 }
2621
2622 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2623                     void __user *buffer, size_t *lenp, loff_t *ppos)
2624 {
2625         return -ENOSYS;
2626 }
2627
2628 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2629                     void __user *buffer, size_t *lenp, loff_t *ppos)
2630 {
2631         return -ENOSYS;
2632 }
2633
2634 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2635                              void __user *buffer, size_t *lenp, loff_t *ppos)
2636 {
2637         return -ENOSYS;
2638 }
2639
2640 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2641                     void __user *buffer, size_t *lenp, loff_t *ppos)
2642 {
2643         return -ENOSYS;
2644 }
2645
2646 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2647                                       void __user *buffer,
2648                                       size_t *lenp, loff_t *ppos)
2649 {
2650     return -ENOSYS;
2651 }
2652
2653
2654 #endif /* CONFIG_PROC_SYSCTL */
2655
2656 /*
2657  * No sense putting this after each symbol definition, twice,
2658  * exception granted :-)
2659  */
2660 EXPORT_SYMBOL(proc_dointvec);
2661 EXPORT_SYMBOL(proc_dointvec_jiffies);
2662 EXPORT_SYMBOL(proc_dointvec_minmax);
2663 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2664 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2665 EXPORT_SYMBOL(proc_dostring);
2666 EXPORT_SYMBOL(proc_doulongvec_minmax);
2667 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);