security: remove unused security_sysctl hook
[linux-2.6.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/signal.h>
27 #include <linux/proc_fs.h>
28 #include <linux/security.h>
29 #include <linux/ctype.h>
30 #include <linux/kmemcheck.h>
31 #include <linux/fs.h>
32 #include <linux/init.h>
33 #include <linux/kernel.h>
34 #include <linux/kobject.h>
35 #include <linux/net.h>
36 #include <linux/sysrq.h>
37 #include <linux/highuid.h>
38 #include <linux/writeback.h>
39 #include <linux/ratelimit.h>
40 #include <linux/compaction.h>
41 #include <linux/hugetlb.h>
42 #include <linux/initrd.h>
43 #include <linux/key.h>
44 #include <linux/times.h>
45 #include <linux/limits.h>
46 #include <linux/dcache.h>
47 #include <linux/dnotify.h>
48 #include <linux/syscalls.h>
49 #include <linux/vmstat.h>
50 #include <linux/nfs_fs.h>
51 #include <linux/acpi.h>
52 #include <linux/reboot.h>
53 #include <linux/ftrace.h>
54 #include <linux/perf_event.h>
55 #include <linux/kprobes.h>
56 #include <linux/pipe_fs_i.h>
57 #include <linux/oom.h>
58
59 #include <asm/uaccess.h>
60 #include <asm/processor.h>
61
62 #ifdef CONFIG_X86
63 #include <asm/nmi.h>
64 #include <asm/stacktrace.h>
65 #include <asm/io.h>
66 #endif
67 #ifdef CONFIG_BSD_PROCESS_ACCT
68 #include <linux/acct.h>
69 #endif
70 #ifdef CONFIG_RT_MUTEXES
71 #include <linux/rtmutex.h>
72 #endif
73 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
74 #include <linux/lockdep.h>
75 #endif
76 #ifdef CONFIG_CHR_DEV_SG
77 #include <scsi/sg.h>
78 #endif
79
80 #ifdef CONFIG_LOCKUP_DETECTOR
81 #include <linux/nmi.h>
82 #endif
83
84
85 #if defined(CONFIG_SYSCTL)
86
87 /* External variables not in a header file. */
88 extern int sysctl_overcommit_memory;
89 extern int sysctl_overcommit_ratio;
90 extern int max_threads;
91 extern int core_uses_pid;
92 extern int suid_dumpable;
93 extern char core_pattern[];
94 extern unsigned int core_pipe_limit;
95 extern int pid_max;
96 extern int min_free_kbytes;
97 extern int pid_max_min, pid_max_max;
98 extern int sysctl_drop_caches;
99 extern int percpu_pagelist_fraction;
100 extern int compat_log;
101 extern int latencytop_enabled;
102 extern int sysctl_nr_open_min, sysctl_nr_open_max;
103 #ifndef CONFIG_MMU
104 extern int sysctl_nr_trim_pages;
105 #endif
106 #ifdef CONFIG_BLOCK
107 extern int blk_iopoll_enabled;
108 #endif
109
110 /* Constants used for minimum and  maximum */
111 #ifdef CONFIG_LOCKUP_DETECTOR
112 static int sixty = 60;
113 static int neg_one = -1;
114 #endif
115
116 static int zero;
117 static int __maybe_unused one = 1;
118 static int __maybe_unused two = 2;
119 static unsigned long one_ul = 1;
120 static int one_hundred = 100;
121 #ifdef CONFIG_PRINTK
122 static int ten_thousand = 10000;
123 #endif
124
125 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
126 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
127
128 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
129 static int maxolduid = 65535;
130 static int minolduid;
131 static int min_percpu_pagelist_fract = 8;
132
133 static int ngroups_max = NGROUPS_MAX;
134
135 #ifdef CONFIG_INOTIFY_USER
136 #include <linux/inotify.h>
137 #endif
138 #ifdef CONFIG_SPARC
139 #include <asm/system.h>
140 #endif
141
142 #ifdef CONFIG_SPARC64
143 extern int sysctl_tsb_ratio;
144 #endif
145
146 #ifdef __hppa__
147 extern int pwrsw_enabled;
148 extern int unaligned_enabled;
149 #endif
150
151 #ifdef CONFIG_S390
152 #ifdef CONFIG_MATHEMU
153 extern int sysctl_ieee_emulation_warnings;
154 #endif
155 extern int sysctl_userprocess_debug;
156 extern int spin_retry;
157 #endif
158
159 #ifdef CONFIG_IA64
160 extern int no_unaligned_warning;
161 extern int unaligned_dump_stack;
162 #endif
163
164 #ifdef CONFIG_PROC_SYSCTL
165 static int proc_do_cad_pid(struct ctl_table *table, int write,
166                   void __user *buffer, size_t *lenp, loff_t *ppos);
167 static int proc_taint(struct ctl_table *table, int write,
168                                void __user *buffer, size_t *lenp, loff_t *ppos);
169 #endif
170
171 #ifdef CONFIG_MAGIC_SYSRQ
172 static int __sysrq_enabled; /* Note: sysrq code ises it's own private copy */
173
174 static int sysrq_sysctl_handler(ctl_table *table, int write,
175                                 void __user *buffer, size_t *lenp,
176                                 loff_t *ppos)
177 {
178         int error;
179
180         error = proc_dointvec(table, write, buffer, lenp, ppos);
181         if (error)
182                 return error;
183
184         if (write)
185                 sysrq_toggle_support(__sysrq_enabled);
186
187         return 0;
188 }
189
190 #endif
191
192 static struct ctl_table root_table[];
193 static struct ctl_table_root sysctl_table_root;
194 static struct ctl_table_header root_table_header = {
195         .count = 1,
196         .ctl_table = root_table,
197         .ctl_entry = LIST_HEAD_INIT(sysctl_table_root.default_set.list),
198         .root = &sysctl_table_root,
199         .set = &sysctl_table_root.default_set,
200 };
201 static struct ctl_table_root sysctl_table_root = {
202         .root_list = LIST_HEAD_INIT(sysctl_table_root.root_list),
203         .default_set.list = LIST_HEAD_INIT(root_table_header.ctl_entry),
204 };
205
206 static struct ctl_table kern_table[];
207 static struct ctl_table vm_table[];
208 static struct ctl_table fs_table[];
209 static struct ctl_table debug_table[];
210 static struct ctl_table dev_table[];
211 extern struct ctl_table random_table[];
212 #ifdef CONFIG_EPOLL
213 extern struct ctl_table epoll_table[];
214 #endif
215
216 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
217 int sysctl_legacy_va_layout;
218 #endif
219
220 /* The default sysctl tables: */
221
222 static struct ctl_table root_table[] = {
223         {
224                 .procname       = "kernel",
225                 .mode           = 0555,
226                 .child          = kern_table,
227         },
228         {
229                 .procname       = "vm",
230                 .mode           = 0555,
231                 .child          = vm_table,
232         },
233         {
234                 .procname       = "fs",
235                 .mode           = 0555,
236                 .child          = fs_table,
237         },
238         {
239                 .procname       = "debug",
240                 .mode           = 0555,
241                 .child          = debug_table,
242         },
243         {
244                 .procname       = "dev",
245                 .mode           = 0555,
246                 .child          = dev_table,
247         },
248 /*
249  * NOTE: do not add new entries to this table unless you have read
250  * Documentation/sysctl/ctl_unnumbered.txt
251  */
252         { }
253 };
254
255 #ifdef CONFIG_SCHED_DEBUG
256 static int min_sched_granularity_ns = 100000;           /* 100 usecs */
257 static int max_sched_granularity_ns = NSEC_PER_SEC;     /* 1 second */
258 static int min_wakeup_granularity_ns;                   /* 0 usecs */
259 static int max_wakeup_granularity_ns = NSEC_PER_SEC;    /* 1 second */
260 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
261 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
262 #endif
263
264 #ifdef CONFIG_COMPACTION
265 static int min_extfrag_threshold;
266 static int max_extfrag_threshold = 1000;
267 #endif
268
269 static struct ctl_table kern_table[] = {
270         {
271                 .procname       = "sched_child_runs_first",
272                 .data           = &sysctl_sched_child_runs_first,
273                 .maxlen         = sizeof(unsigned int),
274                 .mode           = 0644,
275                 .proc_handler   = proc_dointvec,
276         },
277 #ifdef CONFIG_SCHED_DEBUG
278         {
279                 .procname       = "sched_min_granularity_ns",
280                 .data           = &sysctl_sched_min_granularity,
281                 .maxlen         = sizeof(unsigned int),
282                 .mode           = 0644,
283                 .proc_handler   = sched_proc_update_handler,
284                 .extra1         = &min_sched_granularity_ns,
285                 .extra2         = &max_sched_granularity_ns,
286         },
287         {
288                 .procname       = "sched_latency_ns",
289                 .data           = &sysctl_sched_latency,
290                 .maxlen         = sizeof(unsigned int),
291                 .mode           = 0644,
292                 .proc_handler   = sched_proc_update_handler,
293                 .extra1         = &min_sched_granularity_ns,
294                 .extra2         = &max_sched_granularity_ns,
295         },
296         {
297                 .procname       = "sched_wakeup_granularity_ns",
298                 .data           = &sysctl_sched_wakeup_granularity,
299                 .maxlen         = sizeof(unsigned int),
300                 .mode           = 0644,
301                 .proc_handler   = sched_proc_update_handler,
302                 .extra1         = &min_wakeup_granularity_ns,
303                 .extra2         = &max_wakeup_granularity_ns,
304         },
305         {
306                 .procname       = "sched_tunable_scaling",
307                 .data           = &sysctl_sched_tunable_scaling,
308                 .maxlen         = sizeof(enum sched_tunable_scaling),
309                 .mode           = 0644,
310                 .proc_handler   = sched_proc_update_handler,
311                 .extra1         = &min_sched_tunable_scaling,
312                 .extra2         = &max_sched_tunable_scaling,
313         },
314         {
315                 .procname       = "sched_migration_cost",
316                 .data           = &sysctl_sched_migration_cost,
317                 .maxlen         = sizeof(unsigned int),
318                 .mode           = 0644,
319                 .proc_handler   = proc_dointvec,
320         },
321         {
322                 .procname       = "sched_nr_migrate",
323                 .data           = &sysctl_sched_nr_migrate,
324                 .maxlen         = sizeof(unsigned int),
325                 .mode           = 0644,
326                 .proc_handler   = proc_dointvec,
327         },
328         {
329                 .procname       = "sched_time_avg",
330                 .data           = &sysctl_sched_time_avg,
331                 .maxlen         = sizeof(unsigned int),
332                 .mode           = 0644,
333                 .proc_handler   = proc_dointvec,
334         },
335         {
336                 .procname       = "sched_shares_window",
337                 .data           = &sysctl_sched_shares_window,
338                 .maxlen         = sizeof(unsigned int),
339                 .mode           = 0644,
340                 .proc_handler   = proc_dointvec,
341         },
342         {
343                 .procname       = "timer_migration",
344                 .data           = &sysctl_timer_migration,
345                 .maxlen         = sizeof(unsigned int),
346                 .mode           = 0644,
347                 .proc_handler   = proc_dointvec_minmax,
348                 .extra1         = &zero,
349                 .extra2         = &one,
350         },
351 #endif
352         {
353                 .procname       = "sched_rt_period_us",
354                 .data           = &sysctl_sched_rt_period,
355                 .maxlen         = sizeof(unsigned int),
356                 .mode           = 0644,
357                 .proc_handler   = sched_rt_handler,
358         },
359         {
360                 .procname       = "sched_rt_runtime_us",
361                 .data           = &sysctl_sched_rt_runtime,
362                 .maxlen         = sizeof(int),
363                 .mode           = 0644,
364                 .proc_handler   = sched_rt_handler,
365         },
366         {
367                 .procname       = "sched_compat_yield",
368                 .data           = &sysctl_sched_compat_yield,
369                 .maxlen         = sizeof(unsigned int),
370                 .mode           = 0644,
371                 .proc_handler   = proc_dointvec,
372         },
373 #ifdef CONFIG_SCHED_AUTOGROUP
374         {
375                 .procname       = "sched_autogroup_enabled",
376                 .data           = &sysctl_sched_autogroup_enabled,
377                 .maxlen         = sizeof(unsigned int),
378                 .mode           = 0644,
379                 .proc_handler   = proc_dointvec,
380                 .extra1         = &zero,
381                 .extra2         = &one,
382         },
383 #endif
384 #ifdef CONFIG_PROVE_LOCKING
385         {
386                 .procname       = "prove_locking",
387                 .data           = &prove_locking,
388                 .maxlen         = sizeof(int),
389                 .mode           = 0644,
390                 .proc_handler   = proc_dointvec,
391         },
392 #endif
393 #ifdef CONFIG_LOCK_STAT
394         {
395                 .procname       = "lock_stat",
396                 .data           = &lock_stat,
397                 .maxlen         = sizeof(int),
398                 .mode           = 0644,
399                 .proc_handler   = proc_dointvec,
400         },
401 #endif
402         {
403                 .procname       = "panic",
404                 .data           = &panic_timeout,
405                 .maxlen         = sizeof(int),
406                 .mode           = 0644,
407                 .proc_handler   = proc_dointvec,
408         },
409         {
410                 .procname       = "core_uses_pid",
411                 .data           = &core_uses_pid,
412                 .maxlen         = sizeof(int),
413                 .mode           = 0644,
414                 .proc_handler   = proc_dointvec,
415         },
416         {
417                 .procname       = "core_pattern",
418                 .data           = core_pattern,
419                 .maxlen         = CORENAME_MAX_SIZE,
420                 .mode           = 0644,
421                 .proc_handler   = proc_dostring,
422         },
423         {
424                 .procname       = "core_pipe_limit",
425                 .data           = &core_pipe_limit,
426                 .maxlen         = sizeof(unsigned int),
427                 .mode           = 0644,
428                 .proc_handler   = proc_dointvec,
429         },
430 #ifdef CONFIG_PROC_SYSCTL
431         {
432                 .procname       = "tainted",
433                 .maxlen         = sizeof(long),
434                 .mode           = 0644,
435                 .proc_handler   = proc_taint,
436         },
437 #endif
438 #ifdef CONFIG_LATENCYTOP
439         {
440                 .procname       = "latencytop",
441                 .data           = &latencytop_enabled,
442                 .maxlen         = sizeof(int),
443                 .mode           = 0644,
444                 .proc_handler   = proc_dointvec,
445         },
446 #endif
447 #ifdef CONFIG_BLK_DEV_INITRD
448         {
449                 .procname       = "real-root-dev",
450                 .data           = &real_root_dev,
451                 .maxlen         = sizeof(int),
452                 .mode           = 0644,
453                 .proc_handler   = proc_dointvec,
454         },
455 #endif
456         {
457                 .procname       = "print-fatal-signals",
458                 .data           = &print_fatal_signals,
459                 .maxlen         = sizeof(int),
460                 .mode           = 0644,
461                 .proc_handler   = proc_dointvec,
462         },
463 #ifdef CONFIG_SPARC
464         {
465                 .procname       = "reboot-cmd",
466                 .data           = reboot_command,
467                 .maxlen         = 256,
468                 .mode           = 0644,
469                 .proc_handler   = proc_dostring,
470         },
471         {
472                 .procname       = "stop-a",
473                 .data           = &stop_a_enabled,
474                 .maxlen         = sizeof (int),
475                 .mode           = 0644,
476                 .proc_handler   = proc_dointvec,
477         },
478         {
479                 .procname       = "scons-poweroff",
480                 .data           = &scons_pwroff,
481                 .maxlen         = sizeof (int),
482                 .mode           = 0644,
483                 .proc_handler   = proc_dointvec,
484         },
485 #endif
486 #ifdef CONFIG_SPARC64
487         {
488                 .procname       = "tsb-ratio",
489                 .data           = &sysctl_tsb_ratio,
490                 .maxlen         = sizeof (int),
491                 .mode           = 0644,
492                 .proc_handler   = proc_dointvec,
493         },
494 #endif
495 #ifdef __hppa__
496         {
497                 .procname       = "soft-power",
498                 .data           = &pwrsw_enabled,
499                 .maxlen         = sizeof (int),
500                 .mode           = 0644,
501                 .proc_handler   = proc_dointvec,
502         },
503         {
504                 .procname       = "unaligned-trap",
505                 .data           = &unaligned_enabled,
506                 .maxlen         = sizeof (int),
507                 .mode           = 0644,
508                 .proc_handler   = proc_dointvec,
509         },
510 #endif
511         {
512                 .procname       = "ctrl-alt-del",
513                 .data           = &C_A_D,
514                 .maxlen         = sizeof(int),
515                 .mode           = 0644,
516                 .proc_handler   = proc_dointvec,
517         },
518 #ifdef CONFIG_FUNCTION_TRACER
519         {
520                 .procname       = "ftrace_enabled",
521                 .data           = &ftrace_enabled,
522                 .maxlen         = sizeof(int),
523                 .mode           = 0644,
524                 .proc_handler   = ftrace_enable_sysctl,
525         },
526 #endif
527 #ifdef CONFIG_STACK_TRACER
528         {
529                 .procname       = "stack_tracer_enabled",
530                 .data           = &stack_tracer_enabled,
531                 .maxlen         = sizeof(int),
532                 .mode           = 0644,
533                 .proc_handler   = stack_trace_sysctl,
534         },
535 #endif
536 #ifdef CONFIG_TRACING
537         {
538                 .procname       = "ftrace_dump_on_oops",
539                 .data           = &ftrace_dump_on_oops,
540                 .maxlen         = sizeof(int),
541                 .mode           = 0644,
542                 .proc_handler   = proc_dointvec,
543         },
544 #endif
545 #ifdef CONFIG_MODULES
546         {
547                 .procname       = "modprobe",
548                 .data           = &modprobe_path,
549                 .maxlen         = KMOD_PATH_LEN,
550                 .mode           = 0644,
551                 .proc_handler   = proc_dostring,
552         },
553         {
554                 .procname       = "modules_disabled",
555                 .data           = &modules_disabled,
556                 .maxlen         = sizeof(int),
557                 .mode           = 0644,
558                 /* only handle a transition from default "0" to "1" */
559                 .proc_handler   = proc_dointvec_minmax,
560                 .extra1         = &one,
561                 .extra2         = &one,
562         },
563 #endif
564 #ifdef CONFIG_HOTPLUG
565         {
566                 .procname       = "hotplug",
567                 .data           = &uevent_helper,
568                 .maxlen         = UEVENT_HELPER_PATH_LEN,
569                 .mode           = 0644,
570                 .proc_handler   = proc_dostring,
571         },
572 #endif
573 #ifdef CONFIG_CHR_DEV_SG
574         {
575                 .procname       = "sg-big-buff",
576                 .data           = &sg_big_buff,
577                 .maxlen         = sizeof (int),
578                 .mode           = 0444,
579                 .proc_handler   = proc_dointvec,
580         },
581 #endif
582 #ifdef CONFIG_BSD_PROCESS_ACCT
583         {
584                 .procname       = "acct",
585                 .data           = &acct_parm,
586                 .maxlen         = 3*sizeof(int),
587                 .mode           = 0644,
588                 .proc_handler   = proc_dointvec,
589         },
590 #endif
591 #ifdef CONFIG_MAGIC_SYSRQ
592         {
593                 .procname       = "sysrq",
594                 .data           = &__sysrq_enabled,
595                 .maxlen         = sizeof (int),
596                 .mode           = 0644,
597                 .proc_handler   = sysrq_sysctl_handler,
598         },
599 #endif
600 #ifdef CONFIG_PROC_SYSCTL
601         {
602                 .procname       = "cad_pid",
603                 .data           = NULL,
604                 .maxlen         = sizeof (int),
605                 .mode           = 0600,
606                 .proc_handler   = proc_do_cad_pid,
607         },
608 #endif
609         {
610                 .procname       = "threads-max",
611                 .data           = &max_threads,
612                 .maxlen         = sizeof(int),
613                 .mode           = 0644,
614                 .proc_handler   = proc_dointvec,
615         },
616         {
617                 .procname       = "random",
618                 .mode           = 0555,
619                 .child          = random_table,
620         },
621         {
622                 .procname       = "overflowuid",
623                 .data           = &overflowuid,
624                 .maxlen         = sizeof(int),
625                 .mode           = 0644,
626                 .proc_handler   = proc_dointvec_minmax,
627                 .extra1         = &minolduid,
628                 .extra2         = &maxolduid,
629         },
630         {
631                 .procname       = "overflowgid",
632                 .data           = &overflowgid,
633                 .maxlen         = sizeof(int),
634                 .mode           = 0644,
635                 .proc_handler   = proc_dointvec_minmax,
636                 .extra1         = &minolduid,
637                 .extra2         = &maxolduid,
638         },
639 #ifdef CONFIG_S390
640 #ifdef CONFIG_MATHEMU
641         {
642                 .procname       = "ieee_emulation_warnings",
643                 .data           = &sysctl_ieee_emulation_warnings,
644                 .maxlen         = sizeof(int),
645                 .mode           = 0644,
646                 .proc_handler   = proc_dointvec,
647         },
648 #endif
649         {
650                 .procname       = "userprocess_debug",
651                 .data           = &show_unhandled_signals,
652                 .maxlen         = sizeof(int),
653                 .mode           = 0644,
654                 .proc_handler   = proc_dointvec,
655         },
656 #endif
657         {
658                 .procname       = "pid_max",
659                 .data           = &pid_max,
660                 .maxlen         = sizeof (int),
661                 .mode           = 0644,
662                 .proc_handler   = proc_dointvec_minmax,
663                 .extra1         = &pid_max_min,
664                 .extra2         = &pid_max_max,
665         },
666         {
667                 .procname       = "panic_on_oops",
668                 .data           = &panic_on_oops,
669                 .maxlen         = sizeof(int),
670                 .mode           = 0644,
671                 .proc_handler   = proc_dointvec,
672         },
673 #if defined CONFIG_PRINTK
674         {
675                 .procname       = "printk",
676                 .data           = &console_loglevel,
677                 .maxlen         = 4*sizeof(int),
678                 .mode           = 0644,
679                 .proc_handler   = proc_dointvec,
680         },
681         {
682                 .procname       = "printk_ratelimit",
683                 .data           = &printk_ratelimit_state.interval,
684                 .maxlen         = sizeof(int),
685                 .mode           = 0644,
686                 .proc_handler   = proc_dointvec_jiffies,
687         },
688         {
689                 .procname       = "printk_ratelimit_burst",
690                 .data           = &printk_ratelimit_state.burst,
691                 .maxlen         = sizeof(int),
692                 .mode           = 0644,
693                 .proc_handler   = proc_dointvec,
694         },
695         {
696                 .procname       = "printk_delay",
697                 .data           = &printk_delay_msec,
698                 .maxlen         = sizeof(int),
699                 .mode           = 0644,
700                 .proc_handler   = proc_dointvec_minmax,
701                 .extra1         = &zero,
702                 .extra2         = &ten_thousand,
703         },
704         {
705                 .procname       = "dmesg_restrict",
706                 .data           = &dmesg_restrict,
707                 .maxlen         = sizeof(int),
708                 .mode           = 0644,
709                 .proc_handler   = proc_dointvec_minmax,
710                 .extra1         = &zero,
711                 .extra2         = &one,
712         },
713 #endif
714         {
715                 .procname       = "ngroups_max",
716                 .data           = &ngroups_max,
717                 .maxlen         = sizeof (int),
718                 .mode           = 0444,
719                 .proc_handler   = proc_dointvec,
720         },
721 #if defined(CONFIG_LOCKUP_DETECTOR)
722         {
723                 .procname       = "watchdog",
724                 .data           = &watchdog_enabled,
725                 .maxlen         = sizeof (int),
726                 .mode           = 0644,
727                 .proc_handler   = proc_dowatchdog_enabled,
728         },
729         {
730                 .procname       = "watchdog_thresh",
731                 .data           = &softlockup_thresh,
732                 .maxlen         = sizeof(int),
733                 .mode           = 0644,
734                 .proc_handler   = proc_dowatchdog_thresh,
735                 .extra1         = &neg_one,
736                 .extra2         = &sixty,
737         },
738         {
739                 .procname       = "softlockup_panic",
740                 .data           = &softlockup_panic,
741                 .maxlen         = sizeof(int),
742                 .mode           = 0644,
743                 .proc_handler   = proc_dointvec_minmax,
744                 .extra1         = &zero,
745                 .extra2         = &one,
746         },
747         {
748                 .procname       = "nmi_watchdog",
749                 .data           = &watchdog_enabled,
750                 .maxlen         = sizeof (int),
751                 .mode           = 0644,
752                 .proc_handler   = proc_dowatchdog_enabled,
753         },
754 #endif
755 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
756         {
757                 .procname       = "unknown_nmi_panic",
758                 .data           = &unknown_nmi_panic,
759                 .maxlen         = sizeof (int),
760                 .mode           = 0644,
761                 .proc_handler   = proc_dointvec,
762         },
763 #endif
764 #if defined(CONFIG_X86)
765         {
766                 .procname       = "panic_on_unrecovered_nmi",
767                 .data           = &panic_on_unrecovered_nmi,
768                 .maxlen         = sizeof(int),
769                 .mode           = 0644,
770                 .proc_handler   = proc_dointvec,
771         },
772         {
773                 .procname       = "panic_on_io_nmi",
774                 .data           = &panic_on_io_nmi,
775                 .maxlen         = sizeof(int),
776                 .mode           = 0644,
777                 .proc_handler   = proc_dointvec,
778         },
779         {
780                 .procname       = "bootloader_type",
781                 .data           = &bootloader_type,
782                 .maxlen         = sizeof (int),
783                 .mode           = 0444,
784                 .proc_handler   = proc_dointvec,
785         },
786         {
787                 .procname       = "bootloader_version",
788                 .data           = &bootloader_version,
789                 .maxlen         = sizeof (int),
790                 .mode           = 0444,
791                 .proc_handler   = proc_dointvec,
792         },
793         {
794                 .procname       = "kstack_depth_to_print",
795                 .data           = &kstack_depth_to_print,
796                 .maxlen         = sizeof(int),
797                 .mode           = 0644,
798                 .proc_handler   = proc_dointvec,
799         },
800         {
801                 .procname       = "io_delay_type",
802                 .data           = &io_delay_type,
803                 .maxlen         = sizeof(int),
804                 .mode           = 0644,
805                 .proc_handler   = proc_dointvec,
806         },
807 #endif
808 #if defined(CONFIG_MMU)
809         {
810                 .procname       = "randomize_va_space",
811                 .data           = &randomize_va_space,
812                 .maxlen         = sizeof(int),
813                 .mode           = 0644,
814                 .proc_handler   = proc_dointvec,
815         },
816 #endif
817 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
818         {
819                 .procname       = "spin_retry",
820                 .data           = &spin_retry,
821                 .maxlen         = sizeof (int),
822                 .mode           = 0644,
823                 .proc_handler   = proc_dointvec,
824         },
825 #endif
826 #if     defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
827         {
828                 .procname       = "acpi_video_flags",
829                 .data           = &acpi_realmode_flags,
830                 .maxlen         = sizeof (unsigned long),
831                 .mode           = 0644,
832                 .proc_handler   = proc_doulongvec_minmax,
833         },
834 #endif
835 #ifdef CONFIG_IA64
836         {
837                 .procname       = "ignore-unaligned-usertrap",
838                 .data           = &no_unaligned_warning,
839                 .maxlen         = sizeof (int),
840                 .mode           = 0644,
841                 .proc_handler   = proc_dointvec,
842         },
843         {
844                 .procname       = "unaligned-dump-stack",
845                 .data           = &unaligned_dump_stack,
846                 .maxlen         = sizeof (int),
847                 .mode           = 0644,
848                 .proc_handler   = proc_dointvec,
849         },
850 #endif
851 #ifdef CONFIG_DETECT_HUNG_TASK
852         {
853                 .procname       = "hung_task_panic",
854                 .data           = &sysctl_hung_task_panic,
855                 .maxlen         = sizeof(int),
856                 .mode           = 0644,
857                 .proc_handler   = proc_dointvec_minmax,
858                 .extra1         = &zero,
859                 .extra2         = &one,
860         },
861         {
862                 .procname       = "hung_task_check_count",
863                 .data           = &sysctl_hung_task_check_count,
864                 .maxlen         = sizeof(unsigned long),
865                 .mode           = 0644,
866                 .proc_handler   = proc_doulongvec_minmax,
867         },
868         {
869                 .procname       = "hung_task_timeout_secs",
870                 .data           = &sysctl_hung_task_timeout_secs,
871                 .maxlen         = sizeof(unsigned long),
872                 .mode           = 0644,
873                 .proc_handler   = proc_dohung_task_timeout_secs,
874         },
875         {
876                 .procname       = "hung_task_warnings",
877                 .data           = &sysctl_hung_task_warnings,
878                 .maxlen         = sizeof(unsigned long),
879                 .mode           = 0644,
880                 .proc_handler   = proc_doulongvec_minmax,
881         },
882 #endif
883 #ifdef CONFIG_COMPAT
884         {
885                 .procname       = "compat-log",
886                 .data           = &compat_log,
887                 .maxlen         = sizeof (int),
888                 .mode           = 0644,
889                 .proc_handler   = proc_dointvec,
890         },
891 #endif
892 #ifdef CONFIG_RT_MUTEXES
893         {
894                 .procname       = "max_lock_depth",
895                 .data           = &max_lock_depth,
896                 .maxlen         = sizeof(int),
897                 .mode           = 0644,
898                 .proc_handler   = proc_dointvec,
899         },
900 #endif
901         {
902                 .procname       = "poweroff_cmd",
903                 .data           = &poweroff_cmd,
904                 .maxlen         = POWEROFF_CMD_PATH_LEN,
905                 .mode           = 0644,
906                 .proc_handler   = proc_dostring,
907         },
908 #ifdef CONFIG_KEYS
909         {
910                 .procname       = "keys",
911                 .mode           = 0555,
912                 .child          = key_sysctls,
913         },
914 #endif
915 #ifdef CONFIG_RCU_TORTURE_TEST
916         {
917                 .procname       = "rcutorture_runnable",
918                 .data           = &rcutorture_runnable,
919                 .maxlen         = sizeof(int),
920                 .mode           = 0644,
921                 .proc_handler   = proc_dointvec,
922         },
923 #endif
924 #ifdef CONFIG_PERF_EVENTS
925         {
926                 .procname       = "perf_event_paranoid",
927                 .data           = &sysctl_perf_event_paranoid,
928                 .maxlen         = sizeof(sysctl_perf_event_paranoid),
929                 .mode           = 0644,
930                 .proc_handler   = proc_dointvec,
931         },
932         {
933                 .procname       = "perf_event_mlock_kb",
934                 .data           = &sysctl_perf_event_mlock,
935                 .maxlen         = sizeof(sysctl_perf_event_mlock),
936                 .mode           = 0644,
937                 .proc_handler   = proc_dointvec,
938         },
939         {
940                 .procname       = "perf_event_max_sample_rate",
941                 .data           = &sysctl_perf_event_sample_rate,
942                 .maxlen         = sizeof(sysctl_perf_event_sample_rate),
943                 .mode           = 0644,
944                 .proc_handler   = proc_dointvec,
945         },
946 #endif
947 #ifdef CONFIG_KMEMCHECK
948         {
949                 .procname       = "kmemcheck",
950                 .data           = &kmemcheck_enabled,
951                 .maxlen         = sizeof(int),
952                 .mode           = 0644,
953                 .proc_handler   = proc_dointvec,
954         },
955 #endif
956 #ifdef CONFIG_BLOCK
957         {
958                 .procname       = "blk_iopoll",
959                 .data           = &blk_iopoll_enabled,
960                 .maxlen         = sizeof(int),
961                 .mode           = 0644,
962                 .proc_handler   = proc_dointvec,
963         },
964 #endif
965 /*
966  * NOTE: do not add new entries to this table unless you have read
967  * Documentation/sysctl/ctl_unnumbered.txt
968  */
969         { }
970 };
971
972 static struct ctl_table vm_table[] = {
973         {
974                 .procname       = "overcommit_memory",
975                 .data           = &sysctl_overcommit_memory,
976                 .maxlen         = sizeof(sysctl_overcommit_memory),
977                 .mode           = 0644,
978                 .proc_handler   = proc_dointvec,
979         },
980         {
981                 .procname       = "panic_on_oom",
982                 .data           = &sysctl_panic_on_oom,
983                 .maxlen         = sizeof(sysctl_panic_on_oom),
984                 .mode           = 0644,
985                 .proc_handler   = proc_dointvec,
986         },
987         {
988                 .procname       = "oom_kill_allocating_task",
989                 .data           = &sysctl_oom_kill_allocating_task,
990                 .maxlen         = sizeof(sysctl_oom_kill_allocating_task),
991                 .mode           = 0644,
992                 .proc_handler   = proc_dointvec,
993         },
994         {
995                 .procname       = "oom_dump_tasks",
996                 .data           = &sysctl_oom_dump_tasks,
997                 .maxlen         = sizeof(sysctl_oom_dump_tasks),
998                 .mode           = 0644,
999                 .proc_handler   = proc_dointvec,
1000         },
1001         {
1002                 .procname       = "overcommit_ratio",
1003                 .data           = &sysctl_overcommit_ratio,
1004                 .maxlen         = sizeof(sysctl_overcommit_ratio),
1005                 .mode           = 0644,
1006                 .proc_handler   = proc_dointvec,
1007         },
1008         {
1009                 .procname       = "page-cluster", 
1010                 .data           = &page_cluster,
1011                 .maxlen         = sizeof(int),
1012                 .mode           = 0644,
1013                 .proc_handler   = proc_dointvec,
1014         },
1015         {
1016                 .procname       = "dirty_background_ratio",
1017                 .data           = &dirty_background_ratio,
1018                 .maxlen         = sizeof(dirty_background_ratio),
1019                 .mode           = 0644,
1020                 .proc_handler   = dirty_background_ratio_handler,
1021                 .extra1         = &zero,
1022                 .extra2         = &one_hundred,
1023         },
1024         {
1025                 .procname       = "dirty_background_bytes",
1026                 .data           = &dirty_background_bytes,
1027                 .maxlen         = sizeof(dirty_background_bytes),
1028                 .mode           = 0644,
1029                 .proc_handler   = dirty_background_bytes_handler,
1030                 .extra1         = &one_ul,
1031         },
1032         {
1033                 .procname       = "dirty_ratio",
1034                 .data           = &vm_dirty_ratio,
1035                 .maxlen         = sizeof(vm_dirty_ratio),
1036                 .mode           = 0644,
1037                 .proc_handler   = dirty_ratio_handler,
1038                 .extra1         = &zero,
1039                 .extra2         = &one_hundred,
1040         },
1041         {
1042                 .procname       = "dirty_bytes",
1043                 .data           = &vm_dirty_bytes,
1044                 .maxlen         = sizeof(vm_dirty_bytes),
1045                 .mode           = 0644,
1046                 .proc_handler   = dirty_bytes_handler,
1047                 .extra1         = &dirty_bytes_min,
1048         },
1049         {
1050                 .procname       = "dirty_writeback_centisecs",
1051                 .data           = &dirty_writeback_interval,
1052                 .maxlen         = sizeof(dirty_writeback_interval),
1053                 .mode           = 0644,
1054                 .proc_handler   = dirty_writeback_centisecs_handler,
1055         },
1056         {
1057                 .procname       = "dirty_expire_centisecs",
1058                 .data           = &dirty_expire_interval,
1059                 .maxlen         = sizeof(dirty_expire_interval),
1060                 .mode           = 0644,
1061                 .proc_handler   = proc_dointvec,
1062         },
1063         {
1064                 .procname       = "nr_pdflush_threads",
1065                 .data           = &nr_pdflush_threads,
1066                 .maxlen         = sizeof nr_pdflush_threads,
1067                 .mode           = 0444 /* read-only*/,
1068                 .proc_handler   = proc_dointvec,
1069         },
1070         {
1071                 .procname       = "swappiness",
1072                 .data           = &vm_swappiness,
1073                 .maxlen         = sizeof(vm_swappiness),
1074                 .mode           = 0644,
1075                 .proc_handler   = proc_dointvec_minmax,
1076                 .extra1         = &zero,
1077                 .extra2         = &one_hundred,
1078         },
1079 #ifdef CONFIG_HUGETLB_PAGE
1080         {
1081                 .procname       = "nr_hugepages",
1082                 .data           = NULL,
1083                 .maxlen         = sizeof(unsigned long),
1084                 .mode           = 0644,
1085                 .proc_handler   = hugetlb_sysctl_handler,
1086                 .extra1         = (void *)&hugetlb_zero,
1087                 .extra2         = (void *)&hugetlb_infinity,
1088         },
1089 #ifdef CONFIG_NUMA
1090         {
1091                 .procname       = "nr_hugepages_mempolicy",
1092                 .data           = NULL,
1093                 .maxlen         = sizeof(unsigned long),
1094                 .mode           = 0644,
1095                 .proc_handler   = &hugetlb_mempolicy_sysctl_handler,
1096                 .extra1         = (void *)&hugetlb_zero,
1097                 .extra2         = (void *)&hugetlb_infinity,
1098         },
1099 #endif
1100          {
1101                 .procname       = "hugetlb_shm_group",
1102                 .data           = &sysctl_hugetlb_shm_group,
1103                 .maxlen         = sizeof(gid_t),
1104                 .mode           = 0644,
1105                 .proc_handler   = proc_dointvec,
1106          },
1107          {
1108                 .procname       = "hugepages_treat_as_movable",
1109                 .data           = &hugepages_treat_as_movable,
1110                 .maxlen         = sizeof(int),
1111                 .mode           = 0644,
1112                 .proc_handler   = hugetlb_treat_movable_handler,
1113         },
1114         {
1115                 .procname       = "nr_overcommit_hugepages",
1116                 .data           = NULL,
1117                 .maxlen         = sizeof(unsigned long),
1118                 .mode           = 0644,
1119                 .proc_handler   = hugetlb_overcommit_handler,
1120                 .extra1         = (void *)&hugetlb_zero,
1121                 .extra2         = (void *)&hugetlb_infinity,
1122         },
1123 #endif
1124         {
1125                 .procname       = "lowmem_reserve_ratio",
1126                 .data           = &sysctl_lowmem_reserve_ratio,
1127                 .maxlen         = sizeof(sysctl_lowmem_reserve_ratio),
1128                 .mode           = 0644,
1129                 .proc_handler   = lowmem_reserve_ratio_sysctl_handler,
1130         },
1131         {
1132                 .procname       = "drop_caches",
1133                 .data           = &sysctl_drop_caches,
1134                 .maxlen         = sizeof(int),
1135                 .mode           = 0644,
1136                 .proc_handler   = drop_caches_sysctl_handler,
1137         },
1138 #ifdef CONFIG_COMPACTION
1139         {
1140                 .procname       = "compact_memory",
1141                 .data           = &sysctl_compact_memory,
1142                 .maxlen         = sizeof(int),
1143                 .mode           = 0200,
1144                 .proc_handler   = sysctl_compaction_handler,
1145         },
1146         {
1147                 .procname       = "extfrag_threshold",
1148                 .data           = &sysctl_extfrag_threshold,
1149                 .maxlen         = sizeof(int),
1150                 .mode           = 0644,
1151                 .proc_handler   = sysctl_extfrag_handler,
1152                 .extra1         = &min_extfrag_threshold,
1153                 .extra2         = &max_extfrag_threshold,
1154         },
1155
1156 #endif /* CONFIG_COMPACTION */
1157         {
1158                 .procname       = "min_free_kbytes",
1159                 .data           = &min_free_kbytes,
1160                 .maxlen         = sizeof(min_free_kbytes),
1161                 .mode           = 0644,
1162                 .proc_handler   = min_free_kbytes_sysctl_handler,
1163                 .extra1         = &zero,
1164         },
1165         {
1166                 .procname       = "percpu_pagelist_fraction",
1167                 .data           = &percpu_pagelist_fraction,
1168                 .maxlen         = sizeof(percpu_pagelist_fraction),
1169                 .mode           = 0644,
1170                 .proc_handler   = percpu_pagelist_fraction_sysctl_handler,
1171                 .extra1         = &min_percpu_pagelist_fract,
1172         },
1173 #ifdef CONFIG_MMU
1174         {
1175                 .procname       = "max_map_count",
1176                 .data           = &sysctl_max_map_count,
1177                 .maxlen         = sizeof(sysctl_max_map_count),
1178                 .mode           = 0644,
1179                 .proc_handler   = proc_dointvec_minmax,
1180                 .extra1         = &zero,
1181         },
1182 #else
1183         {
1184                 .procname       = "nr_trim_pages",
1185                 .data           = &sysctl_nr_trim_pages,
1186                 .maxlen         = sizeof(sysctl_nr_trim_pages),
1187                 .mode           = 0644,
1188                 .proc_handler   = proc_dointvec_minmax,
1189                 .extra1         = &zero,
1190         },
1191 #endif
1192         {
1193                 .procname       = "laptop_mode",
1194                 .data           = &laptop_mode,
1195                 .maxlen         = sizeof(laptop_mode),
1196                 .mode           = 0644,
1197                 .proc_handler   = proc_dointvec_jiffies,
1198         },
1199         {
1200                 .procname       = "block_dump",
1201                 .data           = &block_dump,
1202                 .maxlen         = sizeof(block_dump),
1203                 .mode           = 0644,
1204                 .proc_handler   = proc_dointvec,
1205                 .extra1         = &zero,
1206         },
1207         {
1208                 .procname       = "vfs_cache_pressure",
1209                 .data           = &sysctl_vfs_cache_pressure,
1210                 .maxlen         = sizeof(sysctl_vfs_cache_pressure),
1211                 .mode           = 0644,
1212                 .proc_handler   = proc_dointvec,
1213                 .extra1         = &zero,
1214         },
1215 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
1216         {
1217                 .procname       = "legacy_va_layout",
1218                 .data           = &sysctl_legacy_va_layout,
1219                 .maxlen         = sizeof(sysctl_legacy_va_layout),
1220                 .mode           = 0644,
1221                 .proc_handler   = proc_dointvec,
1222                 .extra1         = &zero,
1223         },
1224 #endif
1225 #ifdef CONFIG_NUMA
1226         {
1227                 .procname       = "zone_reclaim_mode",
1228                 .data           = &zone_reclaim_mode,
1229                 .maxlen         = sizeof(zone_reclaim_mode),
1230                 .mode           = 0644,
1231                 .proc_handler   = proc_dointvec,
1232                 .extra1         = &zero,
1233         },
1234         {
1235                 .procname       = "min_unmapped_ratio",
1236                 .data           = &sysctl_min_unmapped_ratio,
1237                 .maxlen         = sizeof(sysctl_min_unmapped_ratio),
1238                 .mode           = 0644,
1239                 .proc_handler   = sysctl_min_unmapped_ratio_sysctl_handler,
1240                 .extra1         = &zero,
1241                 .extra2         = &one_hundred,
1242         },
1243         {
1244                 .procname       = "min_slab_ratio",
1245                 .data           = &sysctl_min_slab_ratio,
1246                 .maxlen         = sizeof(sysctl_min_slab_ratio),
1247                 .mode           = 0644,
1248                 .proc_handler   = sysctl_min_slab_ratio_sysctl_handler,
1249                 .extra1         = &zero,
1250                 .extra2         = &one_hundred,
1251         },
1252 #endif
1253 #ifdef CONFIG_SMP
1254         {
1255                 .procname       = "stat_interval",
1256                 .data           = &sysctl_stat_interval,
1257                 .maxlen         = sizeof(sysctl_stat_interval),
1258                 .mode           = 0644,
1259                 .proc_handler   = proc_dointvec_jiffies,
1260         },
1261 #endif
1262 #ifdef CONFIG_MMU
1263         {
1264                 .procname       = "mmap_min_addr",
1265                 .data           = &dac_mmap_min_addr,
1266                 .maxlen         = sizeof(unsigned long),
1267                 .mode           = 0644,
1268                 .proc_handler   = mmap_min_addr_handler,
1269         },
1270 #endif
1271 #ifdef CONFIG_NUMA
1272         {
1273                 .procname       = "numa_zonelist_order",
1274                 .data           = &numa_zonelist_order,
1275                 .maxlen         = NUMA_ZONELIST_ORDER_LEN,
1276                 .mode           = 0644,
1277                 .proc_handler   = numa_zonelist_order_handler,
1278         },
1279 #endif
1280 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1281    (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1282         {
1283                 .procname       = "vdso_enabled",
1284                 .data           = &vdso_enabled,
1285                 .maxlen         = sizeof(vdso_enabled),
1286                 .mode           = 0644,
1287                 .proc_handler   = proc_dointvec,
1288                 .extra1         = &zero,
1289         },
1290 #endif
1291 #ifdef CONFIG_HIGHMEM
1292         {
1293                 .procname       = "highmem_is_dirtyable",
1294                 .data           = &vm_highmem_is_dirtyable,
1295                 .maxlen         = sizeof(vm_highmem_is_dirtyable),
1296                 .mode           = 0644,
1297                 .proc_handler   = proc_dointvec_minmax,
1298                 .extra1         = &zero,
1299                 .extra2         = &one,
1300         },
1301 #endif
1302         {
1303                 .procname       = "scan_unevictable_pages",
1304                 .data           = &scan_unevictable_pages,
1305                 .maxlen         = sizeof(scan_unevictable_pages),
1306                 .mode           = 0644,
1307                 .proc_handler   = scan_unevictable_handler,
1308         },
1309 #ifdef CONFIG_MEMORY_FAILURE
1310         {
1311                 .procname       = "memory_failure_early_kill",
1312                 .data           = &sysctl_memory_failure_early_kill,
1313                 .maxlen         = sizeof(sysctl_memory_failure_early_kill),
1314                 .mode           = 0644,
1315                 .proc_handler   = proc_dointvec_minmax,
1316                 .extra1         = &zero,
1317                 .extra2         = &one,
1318         },
1319         {
1320                 .procname       = "memory_failure_recovery",
1321                 .data           = &sysctl_memory_failure_recovery,
1322                 .maxlen         = sizeof(sysctl_memory_failure_recovery),
1323                 .mode           = 0644,
1324                 .proc_handler   = proc_dointvec_minmax,
1325                 .extra1         = &zero,
1326                 .extra2         = &one,
1327         },
1328 #endif
1329
1330 /*
1331  * NOTE: do not add new entries to this table unless you have read
1332  * Documentation/sysctl/ctl_unnumbered.txt
1333  */
1334         { }
1335 };
1336
1337 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1338 static struct ctl_table binfmt_misc_table[] = {
1339         { }
1340 };
1341 #endif
1342
1343 static struct ctl_table fs_table[] = {
1344         {
1345                 .procname       = "inode-nr",
1346                 .data           = &inodes_stat,
1347                 .maxlen         = 2*sizeof(int),
1348                 .mode           = 0444,
1349                 .proc_handler   = proc_nr_inodes,
1350         },
1351         {
1352                 .procname       = "inode-state",
1353                 .data           = &inodes_stat,
1354                 .maxlen         = 7*sizeof(int),
1355                 .mode           = 0444,
1356                 .proc_handler   = proc_nr_inodes,
1357         },
1358         {
1359                 .procname       = "file-nr",
1360                 .data           = &files_stat,
1361                 .maxlen         = sizeof(files_stat),
1362                 .mode           = 0444,
1363                 .proc_handler   = proc_nr_files,
1364         },
1365         {
1366                 .procname       = "file-max",
1367                 .data           = &files_stat.max_files,
1368                 .maxlen         = sizeof(files_stat.max_files),
1369                 .mode           = 0644,
1370                 .proc_handler   = proc_doulongvec_minmax,
1371         },
1372         {
1373                 .procname       = "nr_open",
1374                 .data           = &sysctl_nr_open,
1375                 .maxlen         = sizeof(int),
1376                 .mode           = 0644,
1377                 .proc_handler   = proc_dointvec_minmax,
1378                 .extra1         = &sysctl_nr_open_min,
1379                 .extra2         = &sysctl_nr_open_max,
1380         },
1381         {
1382                 .procname       = "dentry-state",
1383                 .data           = &dentry_stat,
1384                 .maxlen         = 6*sizeof(int),
1385                 .mode           = 0444,
1386                 .proc_handler   = proc_nr_dentry,
1387         },
1388         {
1389                 .procname       = "overflowuid",
1390                 .data           = &fs_overflowuid,
1391                 .maxlen         = sizeof(int),
1392                 .mode           = 0644,
1393                 .proc_handler   = proc_dointvec_minmax,
1394                 .extra1         = &minolduid,
1395                 .extra2         = &maxolduid,
1396         },
1397         {
1398                 .procname       = "overflowgid",
1399                 .data           = &fs_overflowgid,
1400                 .maxlen         = sizeof(int),
1401                 .mode           = 0644,
1402                 .proc_handler   = proc_dointvec_minmax,
1403                 .extra1         = &minolduid,
1404                 .extra2         = &maxolduid,
1405         },
1406 #ifdef CONFIG_FILE_LOCKING
1407         {
1408                 .procname       = "leases-enable",
1409                 .data           = &leases_enable,
1410                 .maxlen         = sizeof(int),
1411                 .mode           = 0644,
1412                 .proc_handler   = proc_dointvec,
1413         },
1414 #endif
1415 #ifdef CONFIG_DNOTIFY
1416         {
1417                 .procname       = "dir-notify-enable",
1418                 .data           = &dir_notify_enable,
1419                 .maxlen         = sizeof(int),
1420                 .mode           = 0644,
1421                 .proc_handler   = proc_dointvec,
1422         },
1423 #endif
1424 #ifdef CONFIG_MMU
1425 #ifdef CONFIG_FILE_LOCKING
1426         {
1427                 .procname       = "lease-break-time",
1428                 .data           = &lease_break_time,
1429                 .maxlen         = sizeof(int),
1430                 .mode           = 0644,
1431                 .proc_handler   = proc_dointvec,
1432         },
1433 #endif
1434 #ifdef CONFIG_AIO
1435         {
1436                 .procname       = "aio-nr",
1437                 .data           = &aio_nr,
1438                 .maxlen         = sizeof(aio_nr),
1439                 .mode           = 0444,
1440                 .proc_handler   = proc_doulongvec_minmax,
1441         },
1442         {
1443                 .procname       = "aio-max-nr",
1444                 .data           = &aio_max_nr,
1445                 .maxlen         = sizeof(aio_max_nr),
1446                 .mode           = 0644,
1447                 .proc_handler   = proc_doulongvec_minmax,
1448         },
1449 #endif /* CONFIG_AIO */
1450 #ifdef CONFIG_INOTIFY_USER
1451         {
1452                 .procname       = "inotify",
1453                 .mode           = 0555,
1454                 .child          = inotify_table,
1455         },
1456 #endif  
1457 #ifdef CONFIG_EPOLL
1458         {
1459                 .procname       = "epoll",
1460                 .mode           = 0555,
1461                 .child          = epoll_table,
1462         },
1463 #endif
1464 #endif
1465         {
1466                 .procname       = "suid_dumpable",
1467                 .data           = &suid_dumpable,
1468                 .maxlen         = sizeof(int),
1469                 .mode           = 0644,
1470                 .proc_handler   = proc_dointvec_minmax,
1471                 .extra1         = &zero,
1472                 .extra2         = &two,
1473         },
1474 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1475         {
1476                 .procname       = "binfmt_misc",
1477                 .mode           = 0555,
1478                 .child          = binfmt_misc_table,
1479         },
1480 #endif
1481         {
1482                 .procname       = "pipe-max-size",
1483                 .data           = &pipe_max_size,
1484                 .maxlen         = sizeof(int),
1485                 .mode           = 0644,
1486                 .proc_handler   = &pipe_proc_fn,
1487                 .extra1         = &pipe_min_size,
1488         },
1489 /*
1490  * NOTE: do not add new entries to this table unless you have read
1491  * Documentation/sysctl/ctl_unnumbered.txt
1492  */
1493         { }
1494 };
1495
1496 static struct ctl_table debug_table[] = {
1497 #if defined(CONFIG_X86) || defined(CONFIG_PPC) || defined(CONFIG_SPARC) || \
1498     defined(CONFIG_S390)
1499         {
1500                 .procname       = "exception-trace",
1501                 .data           = &show_unhandled_signals,
1502                 .maxlen         = sizeof(int),
1503                 .mode           = 0644,
1504                 .proc_handler   = proc_dointvec
1505         },
1506 #endif
1507 #if defined(CONFIG_OPTPROBES)
1508         {
1509                 .procname       = "kprobes-optimization",
1510                 .data           = &sysctl_kprobes_optimization,
1511                 .maxlen         = sizeof(int),
1512                 .mode           = 0644,
1513                 .proc_handler   = proc_kprobes_optimization_handler,
1514                 .extra1         = &zero,
1515                 .extra2         = &one,
1516         },
1517 #endif
1518         { }
1519 };
1520
1521 static struct ctl_table dev_table[] = {
1522         { }
1523 };
1524
1525 static DEFINE_SPINLOCK(sysctl_lock);
1526
1527 /* called under sysctl_lock */
1528 static int use_table(struct ctl_table_header *p)
1529 {
1530         if (unlikely(p->unregistering))
1531                 return 0;
1532         p->used++;
1533         return 1;
1534 }
1535
1536 /* called under sysctl_lock */
1537 static void unuse_table(struct ctl_table_header *p)
1538 {
1539         if (!--p->used)
1540                 if (unlikely(p->unregistering))
1541                         complete(p->unregistering);
1542 }
1543
1544 /* called under sysctl_lock, will reacquire if has to wait */
1545 static void start_unregistering(struct ctl_table_header *p)
1546 {
1547         /*
1548          * if p->used is 0, nobody will ever touch that entry again;
1549          * we'll eliminate all paths to it before dropping sysctl_lock
1550          */
1551         if (unlikely(p->used)) {
1552                 struct completion wait;
1553                 init_completion(&wait);
1554                 p->unregistering = &wait;
1555                 spin_unlock(&sysctl_lock);
1556                 wait_for_completion(&wait);
1557                 spin_lock(&sysctl_lock);
1558         } else {
1559                 /* anything non-NULL; we'll never dereference it */
1560                 p->unregistering = ERR_PTR(-EINVAL);
1561         }
1562         /*
1563          * do not remove from the list until nobody holds it; walking the
1564          * list in do_sysctl() relies on that.
1565          */
1566         list_del_init(&p->ctl_entry);
1567 }
1568
1569 void sysctl_head_get(struct ctl_table_header *head)
1570 {
1571         spin_lock(&sysctl_lock);
1572         head->count++;
1573         spin_unlock(&sysctl_lock);
1574 }
1575
1576 void sysctl_head_put(struct ctl_table_header *head)
1577 {
1578         spin_lock(&sysctl_lock);
1579         if (!--head->count)
1580                 kfree(head);
1581         spin_unlock(&sysctl_lock);
1582 }
1583
1584 struct ctl_table_header *sysctl_head_grab(struct ctl_table_header *head)
1585 {
1586         if (!head)
1587                 BUG();
1588         spin_lock(&sysctl_lock);
1589         if (!use_table(head))
1590                 head = ERR_PTR(-ENOENT);
1591         spin_unlock(&sysctl_lock);
1592         return head;
1593 }
1594
1595 void sysctl_head_finish(struct ctl_table_header *head)
1596 {
1597         if (!head)
1598                 return;
1599         spin_lock(&sysctl_lock);
1600         unuse_table(head);
1601         spin_unlock(&sysctl_lock);
1602 }
1603
1604 static struct ctl_table_set *
1605 lookup_header_set(struct ctl_table_root *root, struct nsproxy *namespaces)
1606 {
1607         struct ctl_table_set *set = &root->default_set;
1608         if (root->lookup)
1609                 set = root->lookup(root, namespaces);
1610         return set;
1611 }
1612
1613 static struct list_head *
1614 lookup_header_list(struct ctl_table_root *root, struct nsproxy *namespaces)
1615 {
1616         struct ctl_table_set *set = lookup_header_set(root, namespaces);
1617         return &set->list;
1618 }
1619
1620 struct ctl_table_header *__sysctl_head_next(struct nsproxy *namespaces,
1621                                             struct ctl_table_header *prev)
1622 {
1623         struct ctl_table_root *root;
1624         struct list_head *header_list;
1625         struct ctl_table_header *head;
1626         struct list_head *tmp;
1627
1628         spin_lock(&sysctl_lock);
1629         if (prev) {
1630                 head = prev;
1631                 tmp = &prev->ctl_entry;
1632                 unuse_table(prev);
1633                 goto next;
1634         }
1635         tmp = &root_table_header.ctl_entry;
1636         for (;;) {
1637                 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1638
1639                 if (!use_table(head))
1640                         goto next;
1641                 spin_unlock(&sysctl_lock);
1642                 return head;
1643         next:
1644                 root = head->root;
1645                 tmp = tmp->next;
1646                 header_list = lookup_header_list(root, namespaces);
1647                 if (tmp != header_list)
1648                         continue;
1649
1650                 do {
1651                         root = list_entry(root->root_list.next,
1652                                         struct ctl_table_root, root_list);
1653                         if (root == &sysctl_table_root)
1654                                 goto out;
1655                         header_list = lookup_header_list(root, namespaces);
1656                 } while (list_empty(header_list));
1657                 tmp = header_list->next;
1658         }
1659 out:
1660         spin_unlock(&sysctl_lock);
1661         return NULL;
1662 }
1663
1664 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1665 {
1666         return __sysctl_head_next(current->nsproxy, prev);
1667 }
1668
1669 void register_sysctl_root(struct ctl_table_root *root)
1670 {
1671         spin_lock(&sysctl_lock);
1672         list_add_tail(&root->root_list, &sysctl_table_root.root_list);
1673         spin_unlock(&sysctl_lock);
1674 }
1675
1676 /*
1677  * sysctl_perm does NOT grant the superuser all rights automatically, because
1678  * some sysctl variables are readonly even to root.
1679  */
1680
1681 static int test_perm(int mode, int op)
1682 {
1683         if (!current_euid())
1684                 mode >>= 6;
1685         else if (in_egroup_p(0))
1686                 mode >>= 3;
1687         if ((op & ~mode & (MAY_READ|MAY_WRITE|MAY_EXEC)) == 0)
1688                 return 0;
1689         return -EACCES;
1690 }
1691
1692 int sysctl_perm(struct ctl_table_root *root, struct ctl_table *table, int op)
1693 {
1694         int mode;
1695
1696         if (root->permissions)
1697                 mode = root->permissions(root, current->nsproxy, table);
1698         else
1699                 mode = table->mode;
1700
1701         return test_perm(mode, op);
1702 }
1703
1704 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1705 {
1706         for (; table->procname; table++) {
1707                 table->parent = parent;
1708                 if (table->child)
1709                         sysctl_set_parent(table, table->child);
1710         }
1711 }
1712
1713 static __init int sysctl_init(void)
1714 {
1715         sysctl_set_parent(NULL, root_table);
1716 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1717         sysctl_check_table(current->nsproxy, root_table);
1718 #endif
1719         return 0;
1720 }
1721
1722 core_initcall(sysctl_init);
1723
1724 static struct ctl_table *is_branch_in(struct ctl_table *branch,
1725                                       struct ctl_table *table)
1726 {
1727         struct ctl_table *p;
1728         const char *s = branch->procname;
1729
1730         /* branch should have named subdirectory as its first element */
1731         if (!s || !branch->child)
1732                 return NULL;
1733
1734         /* ... and nothing else */
1735         if (branch[1].procname)
1736                 return NULL;
1737
1738         /* table should contain subdirectory with the same name */
1739         for (p = table; p->procname; p++) {
1740                 if (!p->child)
1741                         continue;
1742                 if (p->procname && strcmp(p->procname, s) == 0)
1743                         return p;
1744         }
1745         return NULL;
1746 }
1747
1748 /* see if attaching q to p would be an improvement */
1749 static void try_attach(struct ctl_table_header *p, struct ctl_table_header *q)
1750 {
1751         struct ctl_table *to = p->ctl_table, *by = q->ctl_table;
1752         struct ctl_table *next;
1753         int is_better = 0;
1754         int not_in_parent = !p->attached_by;
1755
1756         while ((next = is_branch_in(by, to)) != NULL) {
1757                 if (by == q->attached_by)
1758                         is_better = 1;
1759                 if (to == p->attached_by)
1760                         not_in_parent = 1;
1761                 by = by->child;
1762                 to = next->child;
1763         }
1764
1765         if (is_better && not_in_parent) {
1766                 q->attached_by = by;
1767                 q->attached_to = to;
1768                 q->parent = p;
1769         }
1770 }
1771
1772 /**
1773  * __register_sysctl_paths - register a sysctl hierarchy
1774  * @root: List of sysctl headers to register on
1775  * @namespaces: Data to compute which lists of sysctl entries are visible
1776  * @path: The path to the directory the sysctl table is in.
1777  * @table: the top-level table structure
1778  *
1779  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1780  * array. A completely 0 filled entry terminates the table.
1781  *
1782  * The members of the &struct ctl_table structure are used as follows:
1783  *
1784  * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1785  *            enter a sysctl file
1786  *
1787  * data - a pointer to data for use by proc_handler
1788  *
1789  * maxlen - the maximum size in bytes of the data
1790  *
1791  * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1792  *
1793  * child - a pointer to the child sysctl table if this entry is a directory, or
1794  *         %NULL.
1795  *
1796  * proc_handler - the text handler routine (described below)
1797  *
1798  * de - for internal use by the sysctl routines
1799  *
1800  * extra1, extra2 - extra pointers usable by the proc handler routines
1801  *
1802  * Leaf nodes in the sysctl tree will be represented by a single file
1803  * under /proc; non-leaf nodes will be represented by directories.
1804  *
1805  * sysctl(2) can automatically manage read and write requests through
1806  * the sysctl table.  The data and maxlen fields of the ctl_table
1807  * struct enable minimal validation of the values being written to be
1808  * performed, and the mode field allows minimal authentication.
1809  *
1810  * There must be a proc_handler routine for any terminal nodes
1811  * mirrored under /proc/sys (non-terminals are handled by a built-in
1812  * directory handler).  Several default handlers are available to
1813  * cover common cases -
1814  *
1815  * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1816  * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(), 
1817  * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1818  *
1819  * It is the handler's job to read the input buffer from user memory
1820  * and process it. The handler should return 0 on success.
1821  *
1822  * This routine returns %NULL on a failure to register, and a pointer
1823  * to the table header on success.
1824  */
1825 struct ctl_table_header *__register_sysctl_paths(
1826         struct ctl_table_root *root,
1827         struct nsproxy *namespaces,
1828         const struct ctl_path *path, struct ctl_table *table)
1829 {
1830         struct ctl_table_header *header;
1831         struct ctl_table *new, **prevp;
1832         unsigned int n, npath;
1833         struct ctl_table_set *set;
1834
1835         /* Count the path components */
1836         for (npath = 0; path[npath].procname; ++npath)
1837                 ;
1838
1839         /*
1840          * For each path component, allocate a 2-element ctl_table array.
1841          * The first array element will be filled with the sysctl entry
1842          * for this, the second will be the sentinel (procname == 0).
1843          *
1844          * We allocate everything in one go so that we don't have to
1845          * worry about freeing additional memory in unregister_sysctl_table.
1846          */
1847         header = kzalloc(sizeof(struct ctl_table_header) +
1848                          (2 * npath * sizeof(struct ctl_table)), GFP_KERNEL);
1849         if (!header)
1850                 return NULL;
1851
1852         new = (struct ctl_table *) (header + 1);
1853
1854         /* Now connect the dots */
1855         prevp = &header->ctl_table;
1856         for (n = 0; n < npath; ++n, ++path) {
1857                 /* Copy the procname */
1858                 new->procname = path->procname;
1859                 new->mode     = 0555;
1860
1861                 *prevp = new;
1862                 prevp = &new->child;
1863
1864                 new += 2;
1865         }
1866         *prevp = table;
1867         header->ctl_table_arg = table;
1868
1869         INIT_LIST_HEAD(&header->ctl_entry);
1870         header->used = 0;
1871         header->unregistering = NULL;
1872         header->root = root;
1873         sysctl_set_parent(NULL, header->ctl_table);
1874         header->count = 1;
1875 #ifdef CONFIG_SYSCTL_SYSCALL_CHECK
1876         if (sysctl_check_table(namespaces, header->ctl_table)) {
1877                 kfree(header);
1878                 return NULL;
1879         }
1880 #endif
1881         spin_lock(&sysctl_lock);
1882         header->set = lookup_header_set(root, namespaces);
1883         header->attached_by = header->ctl_table;
1884         header->attached_to = root_table;
1885         header->parent = &root_table_header;
1886         for (set = header->set; set; set = set->parent) {
1887                 struct ctl_table_header *p;
1888                 list_for_each_entry(p, &set->list, ctl_entry) {
1889                         if (p->unregistering)
1890                                 continue;
1891                         try_attach(p, header);
1892                 }
1893         }
1894         header->parent->count++;
1895         list_add_tail(&header->ctl_entry, &header->set->list);
1896         spin_unlock(&sysctl_lock);
1897
1898         return header;
1899 }
1900
1901 /**
1902  * register_sysctl_table_path - register a sysctl table hierarchy
1903  * @path: The path to the directory the sysctl table is in.
1904  * @table: the top-level table structure
1905  *
1906  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1907  * array. A completely 0 filled entry terminates the table.
1908  *
1909  * See __register_sysctl_paths for more details.
1910  */
1911 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1912                                                 struct ctl_table *table)
1913 {
1914         return __register_sysctl_paths(&sysctl_table_root, current->nsproxy,
1915                                         path, table);
1916 }
1917
1918 /**
1919  * register_sysctl_table - register a sysctl table hierarchy
1920  * @table: the top-level table structure
1921  *
1922  * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1923  * array. A completely 0 filled entry terminates the table.
1924  *
1925  * See register_sysctl_paths for more details.
1926  */
1927 struct ctl_table_header *register_sysctl_table(struct ctl_table *table)
1928 {
1929         static const struct ctl_path null_path[] = { {} };
1930
1931         return register_sysctl_paths(null_path, table);
1932 }
1933
1934 /**
1935  * unregister_sysctl_table - unregister a sysctl table hierarchy
1936  * @header: the header returned from register_sysctl_table
1937  *
1938  * Unregisters the sysctl table and all children. proc entries may not
1939  * actually be removed until they are no longer used by anyone.
1940  */
1941 void unregister_sysctl_table(struct ctl_table_header * header)
1942 {
1943         might_sleep();
1944
1945         if (header == NULL)
1946                 return;
1947
1948         spin_lock(&sysctl_lock);
1949         start_unregistering(header);
1950         if (!--header->parent->count) {
1951                 WARN_ON(1);
1952                 kfree(header->parent);
1953         }
1954         if (!--header->count)
1955                 kfree(header);
1956         spin_unlock(&sysctl_lock);
1957 }
1958
1959 int sysctl_is_seen(struct ctl_table_header *p)
1960 {
1961         struct ctl_table_set *set = p->set;
1962         int res;
1963         spin_lock(&sysctl_lock);
1964         if (p->unregistering)
1965                 res = 0;
1966         else if (!set->is_seen)
1967                 res = 1;
1968         else
1969                 res = set->is_seen(set);
1970         spin_unlock(&sysctl_lock);
1971         return res;
1972 }
1973
1974 void setup_sysctl_set(struct ctl_table_set *p,
1975         struct ctl_table_set *parent,
1976         int (*is_seen)(struct ctl_table_set *))
1977 {
1978         INIT_LIST_HEAD(&p->list);
1979         p->parent = parent ? parent : &sysctl_table_root.default_set;
1980         p->is_seen = is_seen;
1981 }
1982
1983 #else /* !CONFIG_SYSCTL */
1984 struct ctl_table_header *register_sysctl_table(struct ctl_table * table)
1985 {
1986         return NULL;
1987 }
1988
1989 struct ctl_table_header *register_sysctl_paths(const struct ctl_path *path,
1990                                                     struct ctl_table *table)
1991 {
1992         return NULL;
1993 }
1994
1995 void unregister_sysctl_table(struct ctl_table_header * table)
1996 {
1997 }
1998
1999 void setup_sysctl_set(struct ctl_table_set *p,
2000         struct ctl_table_set *parent,
2001         int (*is_seen)(struct ctl_table_set *))
2002 {
2003 }
2004
2005 void sysctl_head_put(struct ctl_table_header *head)
2006 {
2007 }
2008
2009 #endif /* CONFIG_SYSCTL */
2010
2011 /*
2012  * /proc/sys support
2013  */
2014
2015 #ifdef CONFIG_PROC_SYSCTL
2016
2017 static int _proc_do_string(void* data, int maxlen, int write,
2018                            void __user *buffer,
2019                            size_t *lenp, loff_t *ppos)
2020 {
2021         size_t len;
2022         char __user *p;
2023         char c;
2024
2025         if (!data || !maxlen || !*lenp) {
2026                 *lenp = 0;
2027                 return 0;
2028         }
2029
2030         if (write) {
2031                 len = 0;
2032                 p = buffer;
2033                 while (len < *lenp) {
2034                         if (get_user(c, p++))
2035                                 return -EFAULT;
2036                         if (c == 0 || c == '\n')
2037                                 break;
2038                         len++;
2039                 }
2040                 if (len >= maxlen)
2041                         len = maxlen-1;
2042                 if(copy_from_user(data, buffer, len))
2043                         return -EFAULT;
2044                 ((char *) data)[len] = 0;
2045                 *ppos += *lenp;
2046         } else {
2047                 len = strlen(data);
2048                 if (len > maxlen)
2049                         len = maxlen;
2050
2051                 if (*ppos > len) {
2052                         *lenp = 0;
2053                         return 0;
2054                 }
2055
2056                 data += *ppos;
2057                 len  -= *ppos;
2058
2059                 if (len > *lenp)
2060                         len = *lenp;
2061                 if (len)
2062                         if(copy_to_user(buffer, data, len))
2063                                 return -EFAULT;
2064                 if (len < *lenp) {
2065                         if(put_user('\n', ((char __user *) buffer) + len))
2066                                 return -EFAULT;
2067                         len++;
2068                 }
2069                 *lenp = len;
2070                 *ppos += len;
2071         }
2072         return 0;
2073 }
2074
2075 /**
2076  * proc_dostring - read a string sysctl
2077  * @table: the sysctl table
2078  * @write: %TRUE if this is a write to the sysctl file
2079  * @buffer: the user buffer
2080  * @lenp: the size of the user buffer
2081  * @ppos: file position
2082  *
2083  * Reads/writes a string from/to the user buffer. If the kernel
2084  * buffer provided is not large enough to hold the string, the
2085  * string is truncated. The copied string is %NULL-terminated.
2086  * If the string is being read by the user process, it is copied
2087  * and a newline '\n' is added. It is truncated if the buffer is
2088  * not large enough.
2089  *
2090  * Returns 0 on success.
2091  */
2092 int proc_dostring(struct ctl_table *table, int write,
2093                   void __user *buffer, size_t *lenp, loff_t *ppos)
2094 {
2095         return _proc_do_string(table->data, table->maxlen, write,
2096                                buffer, lenp, ppos);
2097 }
2098
2099 static size_t proc_skip_spaces(char **buf)
2100 {
2101         size_t ret;
2102         char *tmp = skip_spaces(*buf);
2103         ret = tmp - *buf;
2104         *buf = tmp;
2105         return ret;
2106 }
2107
2108 static void proc_skip_char(char **buf, size_t *size, const char v)
2109 {
2110         while (*size) {
2111                 if (**buf != v)
2112                         break;
2113                 (*size)--;
2114                 (*buf)++;
2115         }
2116 }
2117
2118 #define TMPBUFLEN 22
2119 /**
2120  * proc_get_long - reads an ASCII formatted integer from a user buffer
2121  *
2122  * @buf: a kernel buffer
2123  * @size: size of the kernel buffer
2124  * @val: this is where the number will be stored
2125  * @neg: set to %TRUE if number is negative
2126  * @perm_tr: a vector which contains the allowed trailers
2127  * @perm_tr_len: size of the perm_tr vector
2128  * @tr: pointer to store the trailer character
2129  *
2130  * In case of success %0 is returned and @buf and @size are updated with
2131  * the amount of bytes read. If @tr is non-NULL and a trailing
2132  * character exists (size is non-zero after returning from this
2133  * function), @tr is updated with the trailing character.
2134  */
2135 static int proc_get_long(char **buf, size_t *size,
2136                           unsigned long *val, bool *neg,
2137                           const char *perm_tr, unsigned perm_tr_len, char *tr)
2138 {
2139         int len;
2140         char *p, tmp[TMPBUFLEN];
2141
2142         if (!*size)
2143                 return -EINVAL;
2144
2145         len = *size;
2146         if (len > TMPBUFLEN - 1)
2147                 len = TMPBUFLEN - 1;
2148
2149         memcpy(tmp, *buf, len);
2150
2151         tmp[len] = 0;
2152         p = tmp;
2153         if (*p == '-' && *size > 1) {
2154                 *neg = true;
2155                 p++;
2156         } else
2157                 *neg = false;
2158         if (!isdigit(*p))
2159                 return -EINVAL;
2160
2161         *val = simple_strtoul(p, &p, 0);
2162
2163         len = p - tmp;
2164
2165         /* We don't know if the next char is whitespace thus we may accept
2166          * invalid integers (e.g. 1234...a) or two integers instead of one
2167          * (e.g. 123...1). So lets not allow such large numbers. */
2168         if (len == TMPBUFLEN - 1)
2169                 return -EINVAL;
2170
2171         if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
2172                 return -EINVAL;
2173
2174         if (tr && (len < *size))
2175                 *tr = *p;
2176
2177         *buf += len;
2178         *size -= len;
2179
2180         return 0;
2181 }
2182
2183 /**
2184  * proc_put_long - converts an integer to a decimal ASCII formatted string
2185  *
2186  * @buf: the user buffer
2187  * @size: the size of the user buffer
2188  * @val: the integer to be converted
2189  * @neg: sign of the number, %TRUE for negative
2190  *
2191  * In case of success %0 is returned and @buf and @size are updated with
2192  * the amount of bytes written.
2193  */
2194 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
2195                           bool neg)
2196 {
2197         int len;
2198         char tmp[TMPBUFLEN], *p = tmp;
2199
2200         sprintf(p, "%s%lu", neg ? "-" : "", val);
2201         len = strlen(tmp);
2202         if (len > *size)
2203                 len = *size;
2204         if (copy_to_user(*buf, tmp, len))
2205                 return -EFAULT;
2206         *size -= len;
2207         *buf += len;
2208         return 0;
2209 }
2210 #undef TMPBUFLEN
2211
2212 static int proc_put_char(void __user **buf, size_t *size, char c)
2213 {
2214         if (*size) {
2215                 char __user **buffer = (char __user **)buf;
2216                 if (put_user(c, *buffer))
2217                         return -EFAULT;
2218                 (*size)--, (*buffer)++;
2219                 *buf = *buffer;
2220         }
2221         return 0;
2222 }
2223
2224 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
2225                                  int *valp,
2226                                  int write, void *data)
2227 {
2228         if (write) {
2229                 *valp = *negp ? -*lvalp : *lvalp;
2230         } else {
2231                 int val = *valp;
2232                 if (val < 0) {
2233                         *negp = true;
2234                         *lvalp = (unsigned long)-val;
2235                 } else {
2236                         *negp = false;
2237                         *lvalp = (unsigned long)val;
2238                 }
2239         }
2240         return 0;
2241 }
2242
2243 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
2244
2245 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
2246                   int write, void __user *buffer,
2247                   size_t *lenp, loff_t *ppos,
2248                   int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2249                               int write, void *data),
2250                   void *data)
2251 {
2252         int *i, vleft, first = 1, err = 0;
2253         unsigned long page = 0;
2254         size_t left;
2255         char *kbuf;
2256         
2257         if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
2258                 *lenp = 0;
2259                 return 0;
2260         }
2261         
2262         i = (int *) tbl_data;
2263         vleft = table->maxlen / sizeof(*i);
2264         left = *lenp;
2265
2266         if (!conv)
2267                 conv = do_proc_dointvec_conv;
2268
2269         if (write) {
2270                 if (left > PAGE_SIZE - 1)
2271                         left = PAGE_SIZE - 1;
2272                 page = __get_free_page(GFP_TEMPORARY);
2273                 kbuf = (char *) page;
2274                 if (!kbuf)
2275                         return -ENOMEM;
2276                 if (copy_from_user(kbuf, buffer, left)) {
2277                         err = -EFAULT;
2278                         goto free;
2279                 }
2280                 kbuf[left] = 0;
2281         }
2282
2283         for (; left && vleft--; i++, first=0) {
2284                 unsigned long lval;
2285                 bool neg;
2286
2287                 if (write) {
2288                         left -= proc_skip_spaces(&kbuf);
2289
2290                         if (!left)
2291                                 break;
2292                         err = proc_get_long(&kbuf, &left, &lval, &neg,
2293                                              proc_wspace_sep,
2294                                              sizeof(proc_wspace_sep), NULL);
2295                         if (err)
2296                                 break;
2297                         if (conv(&neg, &lval, i, 1, data)) {
2298                                 err = -EINVAL;
2299                                 break;
2300                         }
2301                 } else {
2302                         if (conv(&neg, &lval, i, 0, data)) {
2303                                 err = -EINVAL;
2304                                 break;
2305                         }
2306                         if (!first)
2307                                 err = proc_put_char(&buffer, &left, '\t');
2308                         if (err)
2309                                 break;
2310                         err = proc_put_long(&buffer, &left, lval, neg);
2311                         if (err)
2312                                 break;
2313                 }
2314         }
2315
2316         if (!write && !first && left && !err)
2317                 err = proc_put_char(&buffer, &left, '\n');
2318         if (write && !err && left)
2319                 left -= proc_skip_spaces(&kbuf);
2320 free:
2321         if (write) {
2322                 free_page(page);
2323                 if (first)
2324                         return err ? : -EINVAL;
2325         }
2326         *lenp -= left;
2327         *ppos += *lenp;
2328         return err;
2329 }
2330
2331 static int do_proc_dointvec(struct ctl_table *table, int write,
2332                   void __user *buffer, size_t *lenp, loff_t *ppos,
2333                   int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
2334                               int write, void *data),
2335                   void *data)
2336 {
2337         return __do_proc_dointvec(table->data, table, write,
2338                         buffer, lenp, ppos, conv, data);
2339 }
2340
2341 /**
2342  * proc_dointvec - read a vector of integers
2343  * @table: the sysctl table
2344  * @write: %TRUE if this is a write to the sysctl file
2345  * @buffer: the user buffer
2346  * @lenp: the size of the user buffer
2347  * @ppos: file position
2348  *
2349  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2350  * values from/to the user buffer, treated as an ASCII string. 
2351  *
2352  * Returns 0 on success.
2353  */
2354 int proc_dointvec(struct ctl_table *table, int write,
2355                      void __user *buffer, size_t *lenp, loff_t *ppos)
2356 {
2357     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2358                             NULL,NULL);
2359 }
2360
2361 /*
2362  * Taint values can only be increased
2363  * This means we can safely use a temporary.
2364  */
2365 static int proc_taint(struct ctl_table *table, int write,
2366                                void __user *buffer, size_t *lenp, loff_t *ppos)
2367 {
2368         struct ctl_table t;
2369         unsigned long tmptaint = get_taint();
2370         int err;
2371
2372         if (write && !capable(CAP_SYS_ADMIN))
2373                 return -EPERM;
2374
2375         t = *table;
2376         t.data = &tmptaint;
2377         err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
2378         if (err < 0)
2379                 return err;
2380
2381         if (write) {
2382                 /*
2383                  * Poor man's atomic or. Not worth adding a primitive
2384                  * to everyone's atomic.h for this
2385                  */
2386                 int i;
2387                 for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
2388                         if ((tmptaint >> i) & 1)
2389                                 add_taint(i);
2390                 }
2391         }
2392
2393         return err;
2394 }
2395
2396 struct do_proc_dointvec_minmax_conv_param {
2397         int *min;
2398         int *max;
2399 };
2400
2401 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
2402                                         int *valp,
2403                                         int write, void *data)
2404 {
2405         struct do_proc_dointvec_minmax_conv_param *param = data;
2406         if (write) {
2407                 int val = *negp ? -*lvalp : *lvalp;
2408                 if ((param->min && *param->min > val) ||
2409                     (param->max && *param->max < val))
2410                         return -EINVAL;
2411                 *valp = val;
2412         } else {
2413                 int val = *valp;
2414                 if (val < 0) {
2415                         *negp = true;
2416                         *lvalp = (unsigned long)-val;
2417                 } else {
2418                         *negp = false;
2419                         *lvalp = (unsigned long)val;
2420                 }
2421         }
2422         return 0;
2423 }
2424
2425 /**
2426  * proc_dointvec_minmax - read a vector of integers with min/max values
2427  * @table: the sysctl table
2428  * @write: %TRUE if this is a write to the sysctl file
2429  * @buffer: the user buffer
2430  * @lenp: the size of the user buffer
2431  * @ppos: file position
2432  *
2433  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2434  * values from/to the user buffer, treated as an ASCII string.
2435  *
2436  * This routine will ensure the values are within the range specified by
2437  * table->extra1 (min) and table->extra2 (max).
2438  *
2439  * Returns 0 on success.
2440  */
2441 int proc_dointvec_minmax(struct ctl_table *table, int write,
2442                   void __user *buffer, size_t *lenp, loff_t *ppos)
2443 {
2444         struct do_proc_dointvec_minmax_conv_param param = {
2445                 .min = (int *) table->extra1,
2446                 .max = (int *) table->extra2,
2447         };
2448         return do_proc_dointvec(table, write, buffer, lenp, ppos,
2449                                 do_proc_dointvec_minmax_conv, &param);
2450 }
2451
2452 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2453                                      void __user *buffer,
2454                                      size_t *lenp, loff_t *ppos,
2455                                      unsigned long convmul,
2456                                      unsigned long convdiv)
2457 {
2458         unsigned long *i, *min, *max;
2459         int vleft, first = 1, err = 0;
2460         unsigned long page = 0;
2461         size_t left;
2462         char *kbuf;
2463
2464         if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2465                 *lenp = 0;
2466                 return 0;
2467         }
2468
2469         i = (unsigned long *) data;
2470         min = (unsigned long *) table->extra1;
2471         max = (unsigned long *) table->extra2;
2472         vleft = table->maxlen / sizeof(unsigned long);
2473         left = *lenp;
2474
2475         if (write) {
2476                 if (left > PAGE_SIZE - 1)
2477                         left = PAGE_SIZE - 1;
2478                 page = __get_free_page(GFP_TEMPORARY);
2479                 kbuf = (char *) page;
2480                 if (!kbuf)
2481                         return -ENOMEM;
2482                 if (copy_from_user(kbuf, buffer, left)) {
2483                         err = -EFAULT;
2484                         goto free;
2485                 }
2486                 kbuf[left] = 0;
2487         }
2488
2489         for (; left && vleft--; i++, first = 0) {
2490                 unsigned long val;
2491
2492                 if (write) {
2493                         bool neg;
2494
2495                         left -= proc_skip_spaces(&kbuf);
2496
2497                         err = proc_get_long(&kbuf, &left, &val, &neg,
2498                                              proc_wspace_sep,
2499                                              sizeof(proc_wspace_sep), NULL);
2500                         if (err)
2501                                 break;
2502                         if (neg)
2503                                 continue;
2504                         if ((min && val < *min) || (max && val > *max))
2505                                 continue;
2506                         *i = val;
2507                 } else {
2508                         val = convdiv * (*i) / convmul;
2509                         if (!first)
2510                                 err = proc_put_char(&buffer, &left, '\t');
2511                         err = proc_put_long(&buffer, &left, val, false);
2512                         if (err)
2513                                 break;
2514                 }
2515         }
2516
2517         if (!write && !first && left && !err)
2518                 err = proc_put_char(&buffer, &left, '\n');
2519         if (write && !err)
2520                 left -= proc_skip_spaces(&kbuf);
2521 free:
2522         if (write) {
2523                 free_page(page);
2524                 if (first)
2525                         return err ? : -EINVAL;
2526         }
2527         *lenp -= left;
2528         *ppos += *lenp;
2529         return err;
2530 }
2531
2532 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2533                                      void __user *buffer,
2534                                      size_t *lenp, loff_t *ppos,
2535                                      unsigned long convmul,
2536                                      unsigned long convdiv)
2537 {
2538         return __do_proc_doulongvec_minmax(table->data, table, write,
2539                         buffer, lenp, ppos, convmul, convdiv);
2540 }
2541
2542 /**
2543  * proc_doulongvec_minmax - read a vector of long integers with min/max values
2544  * @table: the sysctl table
2545  * @write: %TRUE if this is a write to the sysctl file
2546  * @buffer: the user buffer
2547  * @lenp: the size of the user buffer
2548  * @ppos: file position
2549  *
2550  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2551  * values from/to the user buffer, treated as an ASCII string.
2552  *
2553  * This routine will ensure the values are within the range specified by
2554  * table->extra1 (min) and table->extra2 (max).
2555  *
2556  * Returns 0 on success.
2557  */
2558 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2559                            void __user *buffer, size_t *lenp, loff_t *ppos)
2560 {
2561     return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2562 }
2563
2564 /**
2565  * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2566  * @table: the sysctl table
2567  * @write: %TRUE if this is a write to the sysctl file
2568  * @buffer: the user buffer
2569  * @lenp: the size of the user buffer
2570  * @ppos: file position
2571  *
2572  * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2573  * values from/to the user buffer, treated as an ASCII string. The values
2574  * are treated as milliseconds, and converted to jiffies when they are stored.
2575  *
2576  * This routine will ensure the values are within the range specified by
2577  * table->extra1 (min) and table->extra2 (max).
2578  *
2579  * Returns 0 on success.
2580  */
2581 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2582                                       void __user *buffer,
2583                                       size_t *lenp, loff_t *ppos)
2584 {
2585     return do_proc_doulongvec_minmax(table, write, buffer,
2586                                      lenp, ppos, HZ, 1000l);
2587 }
2588
2589
2590 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2591                                          int *valp,
2592                                          int write, void *data)
2593 {
2594         if (write) {
2595                 if (*lvalp > LONG_MAX / HZ)
2596                         return 1;
2597                 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2598         } else {
2599                 int val = *valp;
2600                 unsigned long lval;
2601                 if (val < 0) {
2602                         *negp = true;
2603                         lval = (unsigned long)-val;
2604                 } else {
2605                         *negp = false;
2606                         lval = (unsigned long)val;
2607                 }
2608                 *lvalp = lval / HZ;
2609         }
2610         return 0;
2611 }
2612
2613 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2614                                                 int *valp,
2615                                                 int write, void *data)
2616 {
2617         if (write) {
2618                 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2619                         return 1;
2620                 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2621         } else {
2622                 int val = *valp;
2623                 unsigned long lval;
2624                 if (val < 0) {
2625                         *negp = true;
2626                         lval = (unsigned long)-val;
2627                 } else {
2628                         *negp = false;
2629                         lval = (unsigned long)val;
2630                 }
2631                 *lvalp = jiffies_to_clock_t(lval);
2632         }
2633         return 0;
2634 }
2635
2636 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2637                                             int *valp,
2638                                             int write, void *data)
2639 {
2640         if (write) {
2641                 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2642         } else {
2643                 int val = *valp;
2644                 unsigned long lval;
2645                 if (val < 0) {
2646                         *negp = true;
2647                         lval = (unsigned long)-val;
2648                 } else {
2649                         *negp = false;
2650                         lval = (unsigned long)val;
2651                 }
2652                 *lvalp = jiffies_to_msecs(lval);
2653         }
2654         return 0;
2655 }
2656
2657 /**
2658  * proc_dointvec_jiffies - read a vector of integers as seconds
2659  * @table: the sysctl table
2660  * @write: %TRUE if this is a write to the sysctl file
2661  * @buffer: the user buffer
2662  * @lenp: the size of the user buffer
2663  * @ppos: file position
2664  *
2665  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2666  * values from/to the user buffer, treated as an ASCII string. 
2667  * The values read are assumed to be in seconds, and are converted into
2668  * jiffies.
2669  *
2670  * Returns 0 on success.
2671  */
2672 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2673                           void __user *buffer, size_t *lenp, loff_t *ppos)
2674 {
2675     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2676                             do_proc_dointvec_jiffies_conv,NULL);
2677 }
2678
2679 /**
2680  * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2681  * @table: the sysctl table
2682  * @write: %TRUE if this is a write to the sysctl file
2683  * @buffer: the user buffer
2684  * @lenp: the size of the user buffer
2685  * @ppos: pointer to the file position
2686  *
2687  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2688  * values from/to the user buffer, treated as an ASCII string. 
2689  * The values read are assumed to be in 1/USER_HZ seconds, and 
2690  * are converted into jiffies.
2691  *
2692  * Returns 0 on success.
2693  */
2694 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2695                                  void __user *buffer, size_t *lenp, loff_t *ppos)
2696 {
2697     return do_proc_dointvec(table,write,buffer,lenp,ppos,
2698                             do_proc_dointvec_userhz_jiffies_conv,NULL);
2699 }
2700
2701 /**
2702  * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2703  * @table: the sysctl table
2704  * @write: %TRUE if this is a write to the sysctl file
2705  * @buffer: the user buffer
2706  * @lenp: the size of the user buffer
2707  * @ppos: file position
2708  * @ppos: the current position in the file
2709  *
2710  * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2711  * values from/to the user buffer, treated as an ASCII string. 
2712  * The values read are assumed to be in 1/1000 seconds, and 
2713  * are converted into jiffies.
2714  *
2715  * Returns 0 on success.
2716  */
2717 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2718                              void __user *buffer, size_t *lenp, loff_t *ppos)
2719 {
2720         return do_proc_dointvec(table, write, buffer, lenp, ppos,
2721                                 do_proc_dointvec_ms_jiffies_conv, NULL);
2722 }
2723
2724 static int proc_do_cad_pid(struct ctl_table *table, int write,
2725                            void __user *buffer, size_t *lenp, loff_t *ppos)
2726 {
2727         struct pid *new_pid;
2728         pid_t tmp;
2729         int r;
2730
2731         tmp = pid_vnr(cad_pid);
2732
2733         r = __do_proc_dointvec(&tmp, table, write, buffer,
2734                                lenp, ppos, NULL, NULL);
2735         if (r || !write)
2736                 return r;
2737
2738         new_pid = find_get_pid(tmp);
2739         if (!new_pid)
2740                 return -ESRCH;
2741
2742         put_pid(xchg(&cad_pid, new_pid));
2743         return 0;
2744 }
2745
2746 /**
2747  * proc_do_large_bitmap - read/write from/to a large bitmap
2748  * @table: the sysctl table
2749  * @write: %TRUE if this is a write to the sysctl file
2750  * @buffer: the user buffer
2751  * @lenp: the size of the user buffer
2752  * @ppos: file position
2753  *
2754  * The bitmap is stored at table->data and the bitmap length (in bits)
2755  * in table->maxlen.
2756  *
2757  * We use a range comma separated format (e.g. 1,3-4,10-10) so that
2758  * large bitmaps may be represented in a compact manner. Writing into
2759  * the file will clear the bitmap then update it with the given input.
2760  *
2761  * Returns 0 on success.
2762  */
2763 int proc_do_large_bitmap(struct ctl_table *table, int write,
2764                          void __user *buffer, size_t *lenp, loff_t *ppos)
2765 {
2766         int err = 0;
2767         bool first = 1;
2768         size_t left = *lenp;
2769         unsigned long bitmap_len = table->maxlen;
2770         unsigned long *bitmap = (unsigned long *) table->data;
2771         unsigned long *tmp_bitmap = NULL;
2772         char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2773
2774         if (!bitmap_len || !left || (*ppos && !write)) {
2775                 *lenp = 0;
2776                 return 0;
2777         }
2778
2779         if (write) {
2780                 unsigned long page = 0;
2781                 char *kbuf;
2782
2783                 if (left > PAGE_SIZE - 1)
2784                         left = PAGE_SIZE - 1;
2785
2786                 page = __get_free_page(GFP_TEMPORARY);
2787                 kbuf = (char *) page;
2788                 if (!kbuf)
2789                         return -ENOMEM;
2790                 if (copy_from_user(kbuf, buffer, left)) {
2791                         free_page(page);
2792                         return -EFAULT;
2793                 }
2794                 kbuf[left] = 0;
2795
2796                 tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2797                                      GFP_KERNEL);
2798                 if (!tmp_bitmap) {
2799                         free_page(page);
2800                         return -ENOMEM;
2801                 }
2802                 proc_skip_char(&kbuf, &left, '\n');
2803                 while (!err && left) {
2804                         unsigned long val_a, val_b;
2805                         bool neg;
2806
2807                         err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2808                                              sizeof(tr_a), &c);
2809                         if (err)
2810                                 break;
2811                         if (val_a >= bitmap_len || neg) {
2812                                 err = -EINVAL;
2813                                 break;
2814                         }
2815
2816                         val_b = val_a;
2817                         if (left) {
2818                                 kbuf++;
2819                                 left--;
2820                         }
2821
2822                         if (c == '-') {
2823                                 err = proc_get_long(&kbuf, &left, &val_b,
2824                                                      &neg, tr_b, sizeof(tr_b),
2825                                                      &c);
2826                                 if (err)
2827                                         break;
2828                                 if (val_b >= bitmap_len || neg ||
2829                                     val_a > val_b) {
2830                                         err = -EINVAL;
2831                                         break;
2832                                 }
2833                                 if (left) {
2834                                         kbuf++;
2835                                         left--;
2836                                 }
2837                         }
2838
2839                         while (val_a <= val_b)
2840                                 set_bit(val_a++, tmp_bitmap);
2841
2842                         first = 0;
2843                         proc_skip_char(&kbuf, &left, '\n');
2844                 }
2845                 free_page(page);
2846         } else {
2847                 unsigned long bit_a, bit_b = 0;
2848
2849                 while (left) {
2850                         bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2851                         if (bit_a >= bitmap_len)
2852                                 break;
2853                         bit_b = find_next_zero_bit(bitmap, bitmap_len,
2854                                                    bit_a + 1) - 1;
2855
2856                         if (!first) {
2857                                 err = proc_put_char(&buffer, &left, ',');
2858                                 if (err)
2859                                         break;
2860                         }
2861                         err = proc_put_long(&buffer, &left, bit_a, false);
2862                         if (err)
2863                                 break;
2864                         if (bit_a != bit_b) {
2865                                 err = proc_put_char(&buffer, &left, '-');
2866                                 if (err)
2867                                         break;
2868                                 err = proc_put_long(&buffer, &left, bit_b, false);
2869                                 if (err)
2870                                         break;
2871                         }
2872
2873                         first = 0; bit_b++;
2874                 }
2875                 if (!err)
2876                         err = proc_put_char(&buffer, &left, '\n');
2877         }
2878
2879         if (!err) {
2880                 if (write) {
2881                         if (*ppos)
2882                                 bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2883                         else
2884                                 memcpy(bitmap, tmp_bitmap,
2885                                         BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long));
2886                 }
2887                 kfree(tmp_bitmap);
2888                 *lenp -= left;
2889                 *ppos += *lenp;
2890                 return 0;
2891         } else {
2892                 kfree(tmp_bitmap);
2893                 return err;
2894         }
2895 }
2896
2897 #else /* CONFIG_PROC_FS */
2898
2899 int proc_dostring(struct ctl_table *table, int write,
2900                   void __user *buffer, size_t *lenp, loff_t *ppos)
2901 {
2902         return -ENOSYS;
2903 }
2904
2905 int proc_dointvec(struct ctl_table *table, int write,
2906                   void __user *buffer, size_t *lenp, loff_t *ppos)
2907 {
2908         return -ENOSYS;
2909 }
2910
2911 int proc_dointvec_minmax(struct ctl_table *table, int write,
2912                     void __user *buffer, size_t *lenp, loff_t *ppos)
2913 {
2914         return -ENOSYS;
2915 }
2916
2917 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2918                     void __user *buffer, size_t *lenp, loff_t *ppos)
2919 {
2920         return -ENOSYS;
2921 }
2922
2923 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2924                     void __user *buffer, size_t *lenp, loff_t *ppos)
2925 {
2926         return -ENOSYS;
2927 }
2928
2929 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2930                              void __user *buffer, size_t *lenp, loff_t *ppos)
2931 {
2932         return -ENOSYS;
2933 }
2934
2935 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2936                     void __user *buffer, size_t *lenp, loff_t *ppos)
2937 {
2938         return -ENOSYS;
2939 }
2940
2941 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2942                                       void __user *buffer,
2943                                       size_t *lenp, loff_t *ppos)
2944 {
2945     return -ENOSYS;
2946 }
2947
2948
2949 #endif /* CONFIG_PROC_FS */
2950
2951 /*
2952  * No sense putting this after each symbol definition, twice,
2953  * exception granted :-)
2954  */
2955 EXPORT_SYMBOL(proc_dointvec);
2956 EXPORT_SYMBOL(proc_dointvec_jiffies);
2957 EXPORT_SYMBOL(proc_dointvec_minmax);
2958 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2959 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2960 EXPORT_SYMBOL(proc_dostring);
2961 EXPORT_SYMBOL(proc_doulongvec_minmax);
2962 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2963 EXPORT_SYMBOL(register_sysctl_table);
2964 EXPORT_SYMBOL(register_sysctl_paths);
2965 EXPORT_SYMBOL(unregister_sysctl_table);