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

Archive Download this file



interactive