Linux Kernel  3.7.1
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
sysctl.c
Go to the documentation of this file.
1 /*
2  * sysctl.c: General linux system control interface
3  *
4  * Begun 24 March 1995, Stephen Tweedie
5  * Added /proc support, Dec 1995
6  * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7  * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8  * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9  * Dynamic registration fixes, Stephen Tweedie.
10  * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11  * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12  * Horn.
13  * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14  * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15  * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16  * Wendling.
17  * The list_for_each() macro wasn't appropriate for the sysctl loop.
18  * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19  */
20 
21 #include <linux/module.h>
22 #include <linux/mm.h>
23 #include <linux/swap.h>
24 #include <linux/slab.h>
25 #include <linux/sysctl.h>
26 #include <linux/bitmap.h>
27 #include <linux/signal.h>
28 #include <linux/printk.h>
29 #include <linux/proc_fs.h>
30 #include <linux/security.h>
31 #include <linux/ctype.h>
32 #include <linux/kmemcheck.h>
33 #include <linux/kmemleak.h>
34 #include <linux/fs.h>
35 #include <linux/init.h>
36 #include <linux/kernel.h>
37 #include <linux/kobject.h>
38 #include <linux/net.h>
39 #include <linux/sysrq.h>
40 #include <linux/highuid.h>
41 #include <linux/writeback.h>
42 #include <linux/ratelimit.h>
43 #include <linux/compaction.h>
44 #include <linux/hugetlb.h>
45 #include <linux/initrd.h>
46 #include <linux/key.h>
47 #include <linux/times.h>
48 #include <linux/limits.h>
49 #include <linux/dcache.h>
50 #include <linux/dnotify.h>
51 #include <linux/syscalls.h>
52 #include <linux/vmstat.h>
53 #include <linux/nfs_fs.h>
54 #include <linux/acpi.h>
55 #include <linux/reboot.h>
56 #include <linux/ftrace.h>
57 #include <linux/perf_event.h>
58 #include <linux/kprobes.h>
59 #include <linux/pipe_fs_i.h>
60 #include <linux/oom.h>
61 #include <linux/kmod.h>
62 #include <linux/capability.h>
63 #include <linux/binfmts.h>
64 
65 #include <asm/uaccess.h>
66 #include <asm/processor.h>
67 
68 #ifdef CONFIG_X86
69 #include <asm/nmi.h>
70 #include <asm/stacktrace.h>
71 #include <asm/io.h>
72 #endif
73 #ifdef CONFIG_SPARC
74 #include <asm/setup.h>
75 #endif
76 #ifdef CONFIG_BSD_PROCESS_ACCT
77 #include <linux/acct.h>
78 #endif
79 #ifdef CONFIG_RT_MUTEXES
80 #include <linux/rtmutex.h>
81 #endif
82 #if defined(CONFIG_PROVE_LOCKING) || defined(CONFIG_LOCK_STAT)
83 #include <linux/lockdep.h>
84 #endif
85 #ifdef CONFIG_CHR_DEV_SG
86 #include <scsi/sg.h>
87 #endif
88 
89 #ifdef CONFIG_LOCKUP_DETECTOR
90 #include <linux/nmi.h>
91 #endif
92 
93 
94 #if defined(CONFIG_SYSCTL)
95 
96 /* External variables not in a header file. */
97 extern int sysctl_overcommit_memory;
98 extern int sysctl_overcommit_ratio;
99 extern int max_threads;
100 extern int suid_dumpable;
101 #ifdef CONFIG_COREDUMP
102 extern int core_uses_pid;
103 extern char core_pattern[];
104 extern unsigned int core_pipe_limit;
105 #endif
106 extern int pid_max;
107 extern int min_free_kbytes;
108 extern int pid_max_min, pid_max_max;
109 extern int sysctl_drop_caches;
110 extern int percpu_pagelist_fraction;
111 extern int compat_log;
112 extern int latencytop_enabled;
114 #ifndef CONFIG_MMU
115 extern int sysctl_nr_trim_pages;
116 #endif
117 #ifdef CONFIG_BLOCK
118 extern int blk_iopoll_enabled;
119 #endif
120 
121 /* Constants used for minimum and maximum */
122 #ifdef CONFIG_LOCKUP_DETECTOR
123 static int sixty = 60;
124 static int neg_one = -1;
125 #endif
126 
127 static int zero;
128 static int __maybe_unused one = 1;
129 static int __maybe_unused two = 2;
130 static int __maybe_unused three = 3;
131 static unsigned long one_ul = 1;
132 static int one_hundred = 100;
133 #ifdef CONFIG_PRINTK
134 static int ten_thousand = 10000;
135 #endif
136 
137 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
138 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
139 
140 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
141 static int maxolduid = 65535;
142 static int minolduid;
143 static int min_percpu_pagelist_fract = 8;
144 
145 static int ngroups_max = NGROUPS_MAX;
146 static const int cap_last_cap = CAP_LAST_CAP;
147 
148 #ifdef CONFIG_INOTIFY_USER
149 #include <linux/inotify.h>
150 #endif
151 #ifdef CONFIG_SPARC
152 #endif
153 
154 #ifdef CONFIG_SPARC64
155 extern int sysctl_tsb_ratio;
156 #endif
157 
158 #ifdef __hppa__
159 extern int pwrsw_enabled;
160 extern int unaligned_enabled;
161 #endif
162 
163 #ifdef CONFIG_IA64
164 extern int no_unaligned_warning;
165 extern int unaligned_dump_stack;
166 #endif
167 
168 #ifdef CONFIG_PROC_SYSCTL
169 static int proc_do_cad_pid(struct ctl_table *table, int write,
170  void __user *buffer, size_t *lenp, loff_t *ppos);
171 static int proc_taint(struct ctl_table *table, int write,
172  void __user *buffer, size_t *lenp, loff_t *ppos);
173 #endif
174 
175 #ifdef CONFIG_PRINTK
176 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
177  void __user *buffer, size_t *lenp, loff_t *ppos);
178 #endif
179 
180 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
181  void __user *buffer, size_t *lenp, loff_t *ppos);
182 #ifdef CONFIG_COREDUMP
183 static int proc_dostring_coredump(struct ctl_table *table, int write,
184  void __user *buffer, size_t *lenp, loff_t *ppos);
185 #endif
186 
187 #ifdef CONFIG_MAGIC_SYSRQ
188 /* Note: sysrq code uses it's own private copy */
189 static int __sysrq_enabled = SYSRQ_DEFAULT_ENABLE;
190 
191 static int sysrq_sysctl_handler(ctl_table *table, int write,
192  void __user *buffer, size_t *lenp,
193  loff_t *ppos)
194 {
195  int error;
196 
197  error = proc_dointvec(table, write, buffer, lenp, ppos);
198  if (error)
199  return error;
200 
201  if (write)
202  sysrq_toggle_support(__sysrq_enabled);
203 
204  return 0;
205 }
206 
207 #endif
208 
209 static struct ctl_table kern_table[];
210 static struct ctl_table vm_table[];
211 static struct ctl_table fs_table[];
212 static struct ctl_table debug_table[];
213 static struct ctl_table dev_table[];
214 extern struct ctl_table random_table[];
215 #ifdef CONFIG_EPOLL
216 extern struct ctl_table epoll_table[];
217 #endif
218 
219 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
221 #endif
222 
223 /* The default sysctl tables: */
224 
225 static struct ctl_table sysctl_base_table[] = {
226  {
227  .procname = "kernel",
228  .mode = 0555,
229  .child = kern_table,
230  },
231  {
232  .procname = "vm",
233  .mode = 0555,
234  .child = vm_table,
235  },
236  {
237  .procname = "fs",
238  .mode = 0555,
239  .child = fs_table,
240  },
241  {
242  .procname = "debug",
243  .mode = 0555,
244  .child = debug_table,
245  },
246  {
247  .procname = "dev",
248  .mode = 0555,
249  .child = dev_table,
250  },
251  { }
252 };
253 
254 #ifdef CONFIG_SCHED_DEBUG
255 static int min_sched_granularity_ns = 100000; /* 100 usecs */
256 static int max_sched_granularity_ns = NSEC_PER_SEC; /* 1 second */
257 static int min_wakeup_granularity_ns; /* 0 usecs */
258 static int max_wakeup_granularity_ns = NSEC_PER_SEC; /* 1 second */
259 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
260 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
261 #endif
262 
263 #ifdef CONFIG_COMPACTION
264 static int min_extfrag_threshold;
265 static int max_extfrag_threshold = 1000;
266 #endif
267 
268 static struct ctl_table kern_table[] = {
269  {
270  .procname = "sched_child_runs_first",
272  .maxlen = sizeof(unsigned int),
273  .mode = 0644,
275  },
276 #ifdef CONFIG_SCHED_DEBUG
277  {
278  .procname = "sched_min_granularity_ns",
280  .maxlen = sizeof(unsigned int),
281  .mode = 0644,
282  .proc_handler = sched_proc_update_handler,
283  .extra1 = &min_sched_granularity_ns,
284  .extra2 = &max_sched_granularity_ns,
285  },
286  {
287  .procname = "sched_latency_ns",
288  .data = &sysctl_sched_latency,
289  .maxlen = sizeof(unsigned int),
290  .mode = 0644,
291  .proc_handler = sched_proc_update_handler,
292  .extra1 = &min_sched_granularity_ns,
293  .extra2 = &max_sched_granularity_ns,
294  },
295  {
296  .procname = "sched_wakeup_granularity_ns",
298  .maxlen = sizeof(unsigned int),
299  .mode = 0644,
300  .proc_handler = sched_proc_update_handler,
301  .extra1 = &min_wakeup_granularity_ns,
302  .extra2 = &max_wakeup_granularity_ns,
303  },
304  {
305  .procname = "sched_tunable_scaling",
307  .maxlen = sizeof(enum sched_tunable_scaling),
308  .mode = 0644,
309  .proc_handler = sched_proc_update_handler,
310  .extra1 = &min_sched_tunable_scaling,
311  .extra2 = &max_sched_tunable_scaling,
312  },
313  {
314  .procname = "sched_migration_cost_ns",
316  .maxlen = sizeof(unsigned int),
317  .mode = 0644,
319  },
320  {
321  .procname = "sched_nr_migrate",
322  .data = &sysctl_sched_nr_migrate,
323  .maxlen = sizeof(unsigned int),
324  .mode = 0644,
326  },
327  {
328  .procname = "sched_time_avg_ms",
329  .data = &sysctl_sched_time_avg,
330  .maxlen = sizeof(unsigned int),
331  .mode = 0644,
333  },
334  {
335  .procname = "sched_shares_window_ns",
337  .maxlen = sizeof(unsigned int),
338  .mode = 0644,
340  },
341  {
342  .procname = "timer_migration",
343  .data = &sysctl_timer_migration,
344  .maxlen = sizeof(unsigned int),
345  .mode = 0644,
347  .extra1 = &zero,
348  .extra2 = &one,
349  },
350 #endif
351  {
352  .procname = "sched_rt_period_us",
353  .data = &sysctl_sched_rt_period,
354  .maxlen = sizeof(unsigned int),
355  .mode = 0644,
357  },
358  {
359  .procname = "sched_rt_runtime_us",
360  .data = &sysctl_sched_rt_runtime,
361  .maxlen = sizeof(int),
362  .mode = 0644,
364  },
365 #ifdef CONFIG_SCHED_AUTOGROUP
366  {
367  .procname = "sched_autogroup_enabled",
368  .data = &sysctl_sched_autogroup_enabled,
369  .maxlen = sizeof(unsigned int),
370  .mode = 0644,
372  .extra1 = &zero,
373  .extra2 = &one,
374  },
375 #endif
376 #ifdef CONFIG_CFS_BANDWIDTH
377  {
378  .procname = "sched_cfs_bandwidth_slice_us",
379  .data = &sysctl_sched_cfs_bandwidth_slice,
380  .maxlen = sizeof(unsigned int),
381  .mode = 0644,
383  .extra1 = &one,
384  },
385 #endif
386 #ifdef CONFIG_PROVE_LOCKING
387  {
388  .procname = "prove_locking",
389  .data = &prove_locking,
390  .maxlen = sizeof(int),
391  .mode = 0644,
393  },
394 #endif
395 #ifdef CONFIG_LOCK_STAT
396  {
397  .procname = "lock_stat",
398  .data = &lock_stat,
399  .maxlen = sizeof(int),
400  .mode = 0644,
402  },
403 #endif
404  {
405  .procname = "panic",
406  .data = &panic_timeout,
407  .maxlen = sizeof(int),
408  .mode = 0644,
410  },
411 #ifdef CONFIG_COREDUMP
412  {
413  .procname = "core_uses_pid",
414  .data = &core_uses_pid,
415  .maxlen = sizeof(int),
416  .mode = 0644,
418  },
419  {
420  .procname = "core_pattern",
421  .data = core_pattern,
422  .maxlen = CORENAME_MAX_SIZE,
423  .mode = 0644,
424  .proc_handler = proc_dostring_coredump,
425  },
426  {
427  .procname = "core_pipe_limit",
428  .data = &core_pipe_limit,
429  .maxlen = sizeof(unsigned int),
430  .mode = 0644,
432  },
433 #endif
434 #ifdef CONFIG_PROC_SYSCTL
435  {
436  .procname = "tainted",
437  .maxlen = sizeof(long),
438  .mode = 0644,
439  .proc_handler = proc_taint,
440  },
441 #endif
442 #ifdef CONFIG_LATENCYTOP
443  {
444  .procname = "latencytop",
445  .data = &latencytop_enabled,
446  .maxlen = sizeof(int),
447  .mode = 0644,
449  },
450 #endif
451 #ifdef CONFIG_BLK_DEV_INITRD
452  {
453  .procname = "real-root-dev",
454  .data = &real_root_dev,
455  .maxlen = sizeof(int),
456  .mode = 0644,
458  },
459 #endif
460  {
461  .procname = "print-fatal-signals",
462  .data = &print_fatal_signals,
463  .maxlen = sizeof(int),
464  .mode = 0644,
466  },
467 #ifdef CONFIG_SPARC
468  {
469  .procname = "reboot-cmd",
470  .data = reboot_command,
471  .maxlen = 256,
472  .mode = 0644,
473  .proc_handler = proc_dostring,
474  },
475  {
476  .procname = "stop-a",
477  .data = &stop_a_enabled,
478  .maxlen = sizeof (int),
479  .mode = 0644,
481  },
482  {
483  .procname = "scons-poweroff",
484  .data = &scons_pwroff,
485  .maxlen = sizeof (int),
486  .mode = 0644,
488  },
489 #endif
490 #ifdef CONFIG_SPARC64
491  {
492  .procname = "tsb-ratio",
493  .data = &sysctl_tsb_ratio,
494  .maxlen = sizeof (int),
495  .mode = 0644,
497  },
498 #endif
499 #ifdef __hppa__
500  {
501  .procname = "soft-power",
502  .data = &pwrsw_enabled,
503  .maxlen = sizeof (int),
504  .mode = 0644,
506  },
507  {
508  .procname = "unaligned-trap",
509  .data = &unaligned_enabled,
510  .maxlen = sizeof (int),
511  .mode = 0644,
513  },
514 #endif
515  {
516  .procname = "ctrl-alt-del",
517  .data = &C_A_D,
518  .maxlen = sizeof(int),
519  .mode = 0644,
521  },
522 #ifdef CONFIG_FUNCTION_TRACER
523  {
524  .procname = "ftrace_enabled",
525  .data = &ftrace_enabled,
526  .maxlen = sizeof(int),
527  .mode = 0644,
529  },
530 #endif
531 #ifdef CONFIG_STACK_TRACER
532  {
533  .procname = "stack_tracer_enabled",
534  .data = &stack_tracer_enabled,
535  .maxlen = sizeof(int),
536  .mode = 0644,
538  },
539 #endif
540 #ifdef CONFIG_TRACING
541  {
542  .procname = "ftrace_dump_on_oops",
543  .data = &ftrace_dump_on_oops,
544  .maxlen = sizeof(int),
545  .mode = 0644,
547  },
548 #endif
549 #ifdef CONFIG_MODULES
550  {
551  .procname = "modprobe",
552  .data = &modprobe_path,
553  .maxlen = KMOD_PATH_LEN,
554  .mode = 0644,
555  .proc_handler = proc_dostring,
556  },
557  {
558  .procname = "modules_disabled",
559  .data = &modules_disabled,
560  .maxlen = sizeof(int),
561  .mode = 0644,
562  /* only handle a transition from default "0" to "1" */
564  .extra1 = &one,
565  .extra2 = &one,
566  },
567 #endif
568 #ifdef CONFIG_HOTPLUG
569  {
570  .procname = "hotplug",
571  .data = &uevent_helper,
572  .maxlen = UEVENT_HELPER_PATH_LEN,
573  .mode = 0644,
574  .proc_handler = proc_dostring,
575  },
576 #endif
577 #ifdef CONFIG_CHR_DEV_SG
578  {
579  .procname = "sg-big-buff",
580  .data = &sg_big_buff,
581  .maxlen = sizeof (int),
582  .mode = 0444,
584  },
585 #endif
586 #ifdef CONFIG_BSD_PROCESS_ACCT
587  {
588  .procname = "acct",
589  .data = &acct_parm,
590  .maxlen = 3*sizeof(int),
591  .mode = 0644,
593  },
594 #endif
595 #ifdef CONFIG_MAGIC_SYSRQ
596  {
597  .procname = "sysrq",
598  .data = &__sysrq_enabled,
599  .maxlen = sizeof (int),
600  .mode = 0644,
601  .proc_handler = sysrq_sysctl_handler,
602  },
603 #endif
604 #ifdef CONFIG_PROC_SYSCTL
605  {
606  .procname = "cad_pid",
607  .data = NULL,
608  .maxlen = sizeof (int),
609  .mode = 0600,
610  .proc_handler = proc_do_cad_pid,
611  },
612 #endif
613  {
614  .procname = "threads-max",
615  .data = &max_threads,
616  .maxlen = sizeof(int),
617  .mode = 0644,
619  },
620  {
621  .procname = "random",
622  .mode = 0555,
623  .child = random_table,
624  },
625  {
626  .procname = "usermodehelper",
627  .mode = 0555,
628  .child = usermodehelper_table,
629  },
630  {
631  .procname = "overflowuid",
632  .data = &overflowuid,
633  .maxlen = sizeof(int),
634  .mode = 0644,
636  .extra1 = &minolduid,
637  .extra2 = &maxolduid,
638  },
639  {
640  .procname = "overflowgid",
641  .data = &overflowgid,
642  .maxlen = sizeof(int),
643  .mode = 0644,
645  .extra1 = &minolduid,
646  .extra2 = &maxolduid,
647  },
648 #ifdef CONFIG_S390
649 #ifdef CONFIG_MATHEMU
650  {
651  .procname = "ieee_emulation_warnings",
653  .maxlen = sizeof(int),
654  .mode = 0644,
656  },
657 #endif
658  {
659  .procname = "userprocess_debug",
660  .data = &show_unhandled_signals,
661  .maxlen = sizeof(int),
662  .mode = 0644,
664  },
665 #endif
666  {
667  .procname = "pid_max",
668  .data = &pid_max,
669  .maxlen = sizeof (int),
670  .mode = 0644,
672  .extra1 = &pid_max_min,
673  .extra2 = &pid_max_max,
674  },
675  {
676  .procname = "panic_on_oops",
677  .data = &panic_on_oops,
678  .maxlen = sizeof(int),
679  .mode = 0644,
681  },
682 #if defined CONFIG_PRINTK
683  {
684  .procname = "printk",
685  .data = &console_loglevel,
686  .maxlen = 4*sizeof(int),
687  .mode = 0644,
689  },
690  {
691  .procname = "printk_ratelimit",
692  .data = &printk_ratelimit_state.interval,
693  .maxlen = sizeof(int),
694  .mode = 0644,
696  },
697  {
698  .procname = "printk_ratelimit_burst",
699  .data = &printk_ratelimit_state.burst,
700  .maxlen = sizeof(int),
701  .mode = 0644,
703  },
704  {
705  .procname = "printk_delay",
706  .data = &printk_delay_msec,
707  .maxlen = sizeof(int),
708  .mode = 0644,
710  .extra1 = &zero,
711  .extra2 = &ten_thousand,
712  },
713  {
714  .procname = "dmesg_restrict",
715  .data = &dmesg_restrict,
716  .maxlen = sizeof(int),
717  .mode = 0644,
718  .proc_handler = proc_dointvec_minmax_sysadmin,
719  .extra1 = &zero,
720  .extra2 = &one,
721  },
722  {
723  .procname = "kptr_restrict",
724  .data = &kptr_restrict,
725  .maxlen = sizeof(int),
726  .mode = 0644,
727  .proc_handler = proc_dointvec_minmax_sysadmin,
728  .extra1 = &zero,
729  .extra2 = &two,
730  },
731 #endif
732  {
733  .procname = "ngroups_max",
734  .data = &ngroups_max,
735  .maxlen = sizeof (int),
736  .mode = 0444,
738  },
739  {
740  .procname = "cap_last_cap",
741  .data = (void *)&cap_last_cap,
742  .maxlen = sizeof(int),
743  .mode = 0444,
744  .proc_handler = proc_dointvec,
745  },
746 #if defined(CONFIG_LOCKUP_DETECTOR)
747  {
748  .procname = "watchdog",
749  .data = &watchdog_enabled,
750  .maxlen = sizeof (int),
751  .mode = 0644,
752  .proc_handler = proc_dowatchdog,
753  .extra1 = &zero,
754  .extra2 = &one,
755  },
756  {
757  .procname = "watchdog_thresh",
758  .data = &watchdog_thresh,
759  .maxlen = sizeof(int),
760  .mode = 0644,
761  .proc_handler = proc_dowatchdog,
762  .extra1 = &neg_one,
763  .extra2 = &sixty,
764  },
765  {
766  .procname = "softlockup_panic",
767  .data = &softlockup_panic,
768  .maxlen = sizeof(int),
769  .mode = 0644,
771  .extra1 = &zero,
772  .extra2 = &one,
773  },
774  {
775  .procname = "nmi_watchdog",
776  .data = &watchdog_enabled,
777  .maxlen = sizeof (int),
778  .mode = 0644,
779  .proc_handler = proc_dowatchdog,
780  .extra1 = &zero,
781  .extra2 = &one,
782  },
783 #endif
784 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
785  {
786  .procname = "unknown_nmi_panic",
787  .data = &unknown_nmi_panic,
788  .maxlen = sizeof (int),
789  .mode = 0644,
791  },
792 #endif
793 #if defined(CONFIG_X86)
794  {
795  .procname = "panic_on_unrecovered_nmi",
796  .data = &panic_on_unrecovered_nmi,
797  .maxlen = sizeof(int),
798  .mode = 0644,
800  },
801  {
802  .procname = "panic_on_io_nmi",
803  .data = &panic_on_io_nmi,
804  .maxlen = sizeof(int),
805  .mode = 0644,
807  },
808 #ifdef CONFIG_DEBUG_STACKOVERFLOW
809  {
810  .procname = "panic_on_stackoverflow",
812  .maxlen = sizeof(int),
813  .mode = 0644,
815  },
816 #endif
817  {
818  .procname = "bootloader_type",
819  .data = &bootloader_type,
820  .maxlen = sizeof (int),
821  .mode = 0444,
823  },
824  {
825  .procname = "bootloader_version",
826  .data = &bootloader_version,
827  .maxlen = sizeof (int),
828  .mode = 0444,
830  },
831  {
832  .procname = "kstack_depth_to_print",
833  .data = &kstack_depth_to_print,
834  .maxlen = sizeof(int),
835  .mode = 0644,
837  },
838  {
839  .procname = "io_delay_type",
840  .data = &io_delay_type,
841  .maxlen = sizeof(int),
842  .mode = 0644,
844  },
845 #endif
846 #if defined(CONFIG_MMU)
847  {
848  .procname = "randomize_va_space",
849  .data = &randomize_va_space,
850  .maxlen = sizeof(int),
851  .mode = 0644,
853  },
854 #endif
855 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
856  {
857  .procname = "spin_retry",
858  .data = &spin_retry,
859  .maxlen = sizeof (int),
860  .mode = 0644,
862  },
863 #endif
864 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
865  {
866  .procname = "acpi_video_flags",
867  .data = &acpi_realmode_flags,
868  .maxlen = sizeof (unsigned long),
869  .mode = 0644,
871  },
872 #endif
873 #ifdef CONFIG_IA64
874  {
875  .procname = "ignore-unaligned-usertrap",
876  .data = &no_unaligned_warning,
877  .maxlen = sizeof (int),
878  .mode = 0644,
880  },
881  {
882  .procname = "unaligned-dump-stack",
883  .data = &unaligned_dump_stack,
884  .maxlen = sizeof (int),
885  .mode = 0644,
887  },
888 #endif
889 #ifdef CONFIG_DETECT_HUNG_TASK
890  {
891  .procname = "hung_task_panic",
892  .data = &sysctl_hung_task_panic,
893  .maxlen = sizeof(int),
894  .mode = 0644,
896  .extra1 = &zero,
897  .extra2 = &one,
898  },
899  {
900  .procname = "hung_task_check_count",
902  .maxlen = sizeof(unsigned long),
903  .mode = 0644,
905  },
906  {
907  .procname = "hung_task_timeout_secs",
909  .maxlen = sizeof(unsigned long),
910  .mode = 0644,
912  },
913  {
914  .procname = "hung_task_warnings",
915  .data = &sysctl_hung_task_warnings,
916  .maxlen = sizeof(unsigned long),
917  .mode = 0644,
919  },
920 #endif
921 #ifdef CONFIG_COMPAT
922  {
923  .procname = "compat-log",
924  .data = &compat_log,
925  .maxlen = sizeof (int),
926  .mode = 0644,
928  },
929 #endif
930 #ifdef CONFIG_RT_MUTEXES
931  {
932  .procname = "max_lock_depth",
933  .data = &max_lock_depth,
934  .maxlen = sizeof(int),
935  .mode = 0644,
937  },
938 #endif
939  {
940  .procname = "poweroff_cmd",
941  .data = &poweroff_cmd,
942  .maxlen = POWEROFF_CMD_PATH_LEN,
943  .mode = 0644,
944  .proc_handler = proc_dostring,
945  },
946 #ifdef CONFIG_KEYS
947  {
948  .procname = "keys",
949  .mode = 0555,
950  .child = key_sysctls,
951  },
952 #endif
953 #ifdef CONFIG_RCU_TORTURE_TEST
954  {
955  .procname = "rcutorture_runnable",
956  .data = &rcutorture_runnable,
957  .maxlen = sizeof(int),
958  .mode = 0644,
960  },
961 #endif
962 #ifdef CONFIG_PERF_EVENTS
963  /*
964  * User-space scripts rely on the existence of this file
965  * as a feature check for perf_events being enabled.
966  *
967  * So it's an ABI, do not remove!
968  */
969  {
970  .procname = "perf_event_paranoid",
971  .data = &sysctl_perf_event_paranoid,
972  .maxlen = sizeof(sysctl_perf_event_paranoid),
973  .mode = 0644,
975  },
976  {
977  .procname = "perf_event_mlock_kb",
978  .data = &sysctl_perf_event_mlock,
979  .maxlen = sizeof(sysctl_perf_event_mlock),
980  .mode = 0644,
982  },
983  {
984  .procname = "perf_event_max_sample_rate",
985  .data = &sysctl_perf_event_sample_rate,
986  .maxlen = sizeof(sysctl_perf_event_sample_rate),
987  .mode = 0644,
989  },
990 #endif
991 #ifdef CONFIG_KMEMCHECK
992  {
993  .procname = "kmemcheck",
994  .data = &kmemcheck_enabled,
995  .maxlen = sizeof(int),
996  .mode = 0644,
998  },
999 #endif
1000 #ifdef CONFIG_BLOCK
1001  {
1002  .procname = "blk_iopoll",
1003  .data = &blk_iopoll_enabled,
1004  .maxlen = sizeof(int),
1005  .mode = 0644,
1007  },
1008 #endif
1009  { }
1010 };
1011 
1012 static struct ctl_table vm_table[] = {
1013  {
1014  .procname = "overcommit_memory",
1015  .data = &sysctl_overcommit_memory,
1016  .maxlen = sizeof(sysctl_overcommit_memory),
1017  .mode = 0644,
1019  .extra1 = &zero,
1020  .extra2 = &two,
1021  },
1022  {
1023  .procname = "panic_on_oom",
1024  .data = &sysctl_panic_on_oom,
1025  .maxlen = sizeof(sysctl_panic_on_oom),
1026  .mode = 0644,
1028  .extra1 = &zero,
1029  .extra2 = &two,
1030  },
1031  {
1032  .procname = "oom_kill_allocating_task",
1034  .maxlen = sizeof(sysctl_oom_kill_allocating_task),
1035  .mode = 0644,
1037  },
1038  {
1039  .procname = "oom_dump_tasks",
1040  .data = &sysctl_oom_dump_tasks,
1041  .maxlen = sizeof(sysctl_oom_dump_tasks),
1042  .mode = 0644,
1044  },
1045  {
1046  .procname = "overcommit_ratio",
1047  .data = &sysctl_overcommit_ratio,
1048  .maxlen = sizeof(sysctl_overcommit_ratio),
1049  .mode = 0644,
1051  },
1052  {
1053  .procname = "page-cluster",
1054  .data = &page_cluster,
1055  .maxlen = sizeof(int),
1056  .mode = 0644,
1058  .extra1 = &zero,
1059  },
1060  {
1061  .procname = "dirty_background_ratio",
1062  .data = &dirty_background_ratio,
1063  .maxlen = sizeof(dirty_background_ratio),
1064  .mode = 0644,
1066  .extra1 = &zero,
1067  .extra2 = &one_hundred,
1068  },
1069  {
1070  .procname = "dirty_background_bytes",
1071  .data = &dirty_background_bytes,
1072  .maxlen = sizeof(dirty_background_bytes),
1073  .mode = 0644,
1075  .extra1 = &one_ul,
1076  },
1077  {
1078  .procname = "dirty_ratio",
1079  .data = &vm_dirty_ratio,
1080  .maxlen = sizeof(vm_dirty_ratio),
1081  .mode = 0644,
1083  .extra1 = &zero,
1084  .extra2 = &one_hundred,
1085  },
1086  {
1087  .procname = "dirty_bytes",
1088  .data = &vm_dirty_bytes,
1089  .maxlen = sizeof(vm_dirty_bytes),
1090  .mode = 0644,
1092  .extra1 = &dirty_bytes_min,
1093  },
1094  {
1095  .procname = "dirty_writeback_centisecs",
1096  .data = &dirty_writeback_interval,
1097  .maxlen = sizeof(dirty_writeback_interval),
1098  .mode = 0644,
1100  },
1101  {
1102  .procname = "dirty_expire_centisecs",
1103  .data = &dirty_expire_interval,
1104  .maxlen = sizeof(dirty_expire_interval),
1105  .mode = 0644,
1107  .extra1 = &zero,
1108  },
1109  {
1110  .procname = "nr_pdflush_threads",
1111  .mode = 0444 /* read-only */,
1112  .proc_handler = pdflush_proc_obsolete,
1113  },
1114  {
1115  .procname = "swappiness",
1116  .data = &vm_swappiness,
1117  .maxlen = sizeof(vm_swappiness),
1118  .mode = 0644,
1120  .extra1 = &zero,
1121  .extra2 = &one_hundred,
1122  },
1123 #ifdef CONFIG_HUGETLB_PAGE
1124  {
1125  .procname = "nr_hugepages",
1126  .data = NULL,
1127  .maxlen = sizeof(unsigned long),
1128  .mode = 0644,
1129  .proc_handler = hugetlb_sysctl_handler,
1130  .extra1 = (void *)&hugetlb_zero,
1131  .extra2 = (void *)&hugetlb_infinity,
1132  },
1133 #ifdef CONFIG_NUMA
1134  {
1135  .procname = "nr_hugepages_mempolicy",
1136  .data = NULL,
1137  .maxlen = sizeof(unsigned long),
1138  .mode = 0644,
1139  .proc_handler = &hugetlb_mempolicy_sysctl_handler,
1140  .extra1 = (void *)&hugetlb_zero,
1141  .extra2 = (void *)&hugetlb_infinity,
1142  },
1143 #endif
1144  {
1145  .procname = "hugetlb_shm_group",
1146  .data = &sysctl_hugetlb_shm_group,
1147  .maxlen = sizeof(gid_t),
1148  .mode = 0644,
1150  },
1151  {
1152  .procname = "hugepages_treat_as_movable",
1153  .data = &hugepages_treat_as_movable,
1154  .maxlen = sizeof(int),
1155  .mode = 0644,
1156  .proc_handler = hugetlb_treat_movable_handler,
1157  },
1158  {
1159  .procname = "nr_overcommit_hugepages",
1160  .data = NULL,
1161  .maxlen = sizeof(unsigned long),
1162  .mode = 0644,
1163  .proc_handler = hugetlb_overcommit_handler,
1164  .extra1 = (void *)&hugetlb_zero,
1165  .extra2 = (void *)&hugetlb_infinity,
1166  },
1167 #endif
1168  {
1169  .procname = "lowmem_reserve_ratio",
1170  .data = &sysctl_lowmem_reserve_ratio,
1171  .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
1172  .mode = 0644,
1174  },
1175  {
1176  .procname = "drop_caches",
1177  .data = &sysctl_drop_caches,
1178  .maxlen = sizeof(int),
1179  .mode = 0644,
1181  .extra1 = &one,
1182  .extra2 = &three,
1183  },
1184 #ifdef CONFIG_COMPACTION
1185  {
1186  .procname = "compact_memory",
1187  .data = &sysctl_compact_memory,
1188  .maxlen = sizeof(int),
1189  .mode = 0200,
1190  .proc_handler = sysctl_compaction_handler,
1191  },
1192  {
1193  .procname = "extfrag_threshold",
1194  .data = &sysctl_extfrag_threshold,
1195  .maxlen = sizeof(int),
1196  .mode = 0644,
1197  .proc_handler = sysctl_extfrag_handler,
1198  .extra1 = &min_extfrag_threshold,
1199  .extra2 = &max_extfrag_threshold,
1200  },
1201 
1202 #endif /* CONFIG_COMPACTION */
1203  {
1204  .procname = "min_free_kbytes",
1205  .data = &min_free_kbytes,
1206  .maxlen = sizeof(min_free_kbytes),
1207  .mode = 0644,
1209  .extra1 = &zero,
1210  },
1211  {
1212  .procname = "percpu_pagelist_fraction",
1213  .data = &percpu_pagelist_fraction,
1214  .maxlen = sizeof(percpu_pagelist_fraction),
1215  .mode = 0644,
1217  .extra1 = &min_percpu_pagelist_fract,
1218  },
1219 #ifdef CONFIG_MMU
1220  {
1221  .procname = "max_map_count",
1222  .data = &sysctl_max_map_count,
1223  .maxlen = sizeof(sysctl_max_map_count),
1224  .mode = 0644,
1226  .extra1 = &zero,
1227  },
1228 #else
1229  {
1230  .procname = "nr_trim_pages",
1231  .data = &sysctl_nr_trim_pages,
1232  .maxlen = sizeof(sysctl_nr_trim_pages),
1233  .mode = 0644,
1235  .extra1 = &zero,
1236  },
1237 #endif
1238  {
1239  .procname = "laptop_mode",
1240  .data = &laptop_mode,
1241  .maxlen = sizeof(laptop_mode),
1242  .mode = 0644,
1244  },
1245  {
1246  .procname = "block_dump",
1247  .data = &block_dump,
1248  .maxlen = sizeof(block_dump),
1249  .mode = 0644,
1251  .extra1 = &zero,
1252  },
1253  {
1254  .procname = "vfs_cache_pressure",
1255  .data = &sysctl_vfs_cache_pressure,
1256  .maxlen = sizeof(sysctl_vfs_cache_pressure),
1257  .mode = 0644,
1259  .extra1 = &zero,
1260  },
1262  {
1263  .procname = "legacy_va_layout",
1264  .data = &sysctl_legacy_va_layout,
1265  .maxlen = sizeof(sysctl_legacy_va_layout),
1266  .mode = 0644,
1268  .extra1 = &zero,
1269  },
1270 #endif
1271 #ifdef CONFIG_NUMA
1272  {
1273  .procname = "zone_reclaim_mode",
1274  .data = &zone_reclaim_mode,
1275  .maxlen = sizeof(zone_reclaim_mode),
1276  .mode = 0644,
1278  .extra1 = &zero,
1279  },
1280  {
1281  .procname = "min_unmapped_ratio",
1282  .data = &sysctl_min_unmapped_ratio,
1283  .maxlen = sizeof(sysctl_min_unmapped_ratio),
1284  .mode = 0644,
1286  .extra1 = &zero,
1287  .extra2 = &one_hundred,
1288  },
1289  {
1290  .procname = "min_slab_ratio",
1291  .data = &sysctl_min_slab_ratio,
1292  .maxlen = sizeof(sysctl_min_slab_ratio),
1293  .mode = 0644,
1295  .extra1 = &zero,
1296  .extra2 = &one_hundred,
1297  },
1298 #endif
1299 #ifdef CONFIG_SMP
1300  {
1301  .procname = "stat_interval",
1302  .data = &sysctl_stat_interval,
1303  .maxlen = sizeof(sysctl_stat_interval),
1304  .mode = 0644,
1306  },
1307 #endif
1308 #ifdef CONFIG_MMU
1309  {
1310  .procname = "mmap_min_addr",
1311  .data = &dac_mmap_min_addr,
1312  .maxlen = sizeof(unsigned long),
1313  .mode = 0644,
1315  },
1316 #endif
1317 #ifdef CONFIG_NUMA
1318  {
1319  .procname = "numa_zonelist_order",
1320  .data = &numa_zonelist_order,
1321  .maxlen = NUMA_ZONELIST_ORDER_LEN,
1322  .mode = 0644,
1323  .proc_handler = numa_zonelist_order_handler,
1324  },
1325 #endif
1326 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1327  (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1328  {
1329  .procname = "vdso_enabled",
1330  .data = &vdso_enabled,
1331  .maxlen = sizeof(vdso_enabled),
1332  .mode = 0644,
1334  .extra1 = &zero,
1335  },
1336 #endif
1337 #ifdef CONFIG_HIGHMEM
1338  {
1339  .procname = "highmem_is_dirtyable",
1340  .data = &vm_highmem_is_dirtyable,
1341  .maxlen = sizeof(vm_highmem_is_dirtyable),
1342  .mode = 0644,
1344  .extra1 = &zero,
1345  .extra2 = &one,
1346  },
1347 #endif
1348  {
1349  .procname = "scan_unevictable_pages",
1350  .data = &scan_unevictable_pages,
1351  .maxlen = sizeof(scan_unevictable_pages),
1352  .mode = 0644,
1354  },
1355 #ifdef CONFIG_MEMORY_FAILURE
1356  {
1357  .procname = "memory_failure_early_kill",
1358  .data = &sysctl_memory_failure_early_kill,
1359  .maxlen = sizeof(sysctl_memory_failure_early_kill),
1360  .mode = 0644,
1362  .extra1 = &zero,
1363  .extra2 = &one,
1364  },
1365  {
1366  .procname = "memory_failure_recovery",
1367  .data = &sysctl_memory_failure_recovery,
1368  .maxlen = sizeof(sysctl_memory_failure_recovery),
1369  .mode = 0644,
1371  .extra1 = &zero,
1372  .extra2 = &one,
1373  },
1374 #endif
1375  { }
1376 };
1377 
1378 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1379 static struct ctl_table binfmt_misc_table[] = {
1380  { }
1381 };
1382 #endif
1383 
1384 static struct ctl_table fs_table[] = {
1385  {
1386  .procname = "inode-nr",
1387  .data = &inodes_stat,
1388  .maxlen = 2*sizeof(int),
1389  .mode = 0444,
1391  },
1392  {
1393  .procname = "inode-state",
1394  .data = &inodes_stat,
1395  .maxlen = 7*sizeof(int),
1396  .mode = 0444,
1398  },
1399  {
1400  .procname = "file-nr",
1401  .data = &files_stat,
1402  .maxlen = sizeof(files_stat),
1403  .mode = 0444,
1405  },
1406  {
1407  .procname = "file-max",
1408  .data = &files_stat.max_files,
1409  .maxlen = sizeof(files_stat.max_files),
1410  .mode = 0644,
1412  },
1413  {
1414  .procname = "nr_open",
1415  .data = &sysctl_nr_open,
1416  .maxlen = sizeof(int),
1417  .mode = 0644,
1421  },
1422  {
1423  .procname = "dentry-state",
1424  .data = &dentry_stat,
1425  .maxlen = 6*sizeof(int),
1426  .mode = 0444,
1428  },
1429  {
1430  .procname = "overflowuid",
1431  .data = &fs_overflowuid,
1432  .maxlen = sizeof(int),
1433  .mode = 0644,
1435  .extra1 = &minolduid,
1436  .extra2 = &maxolduid,
1437  },
1438  {
1439  .procname = "overflowgid",
1440  .data = &fs_overflowgid,
1441  .maxlen = sizeof(int),
1442  .mode = 0644,
1444  .extra1 = &minolduid,
1445  .extra2 = &maxolduid,
1446  },
1447 #ifdef CONFIG_FILE_LOCKING
1448  {
1449  .procname = "leases-enable",
1450  .data = &leases_enable,
1451  .maxlen = sizeof(int),
1452  .mode = 0644,
1454  },
1455 #endif
1456 #ifdef CONFIG_DNOTIFY
1457  {
1458  .procname = "dir-notify-enable",
1459  .data = &dir_notify_enable,
1460  .maxlen = sizeof(int),
1461  .mode = 0644,
1463  },
1464 #endif
1465 #ifdef CONFIG_MMU
1466 #ifdef CONFIG_FILE_LOCKING
1467  {
1468  .procname = "lease-break-time",
1469  .data = &lease_break_time,
1470  .maxlen = sizeof(int),
1471  .mode = 0644,
1473  },
1474 #endif
1475 #ifdef CONFIG_AIO
1476  {
1477  .procname = "aio-nr",
1478  .data = &aio_nr,
1479  .maxlen = sizeof(aio_nr),
1480  .mode = 0444,
1482  },
1483  {
1484  .procname = "aio-max-nr",
1485  .data = &aio_max_nr,
1486  .maxlen = sizeof(aio_max_nr),
1487  .mode = 0644,
1489  },
1490 #endif /* CONFIG_AIO */
1491 #ifdef CONFIG_INOTIFY_USER
1492  {
1493  .procname = "inotify",
1494  .mode = 0555,
1495  .child = inotify_table,
1496  },
1497 #endif
1498 #ifdef CONFIG_EPOLL
1499  {
1500  .procname = "epoll",
1501  .mode = 0555,
1502  .child = epoll_table,
1503  },
1504 #endif
1505 #endif
1506  {
1507  .procname = "protected_symlinks",
1508  .data = &sysctl_protected_symlinks,
1509  .maxlen = sizeof(int),
1510  .mode = 0600,
1512  .extra1 = &zero,
1513  .extra2 = &one,
1514  },
1515  {
1516  .procname = "protected_hardlinks",
1517  .data = &sysctl_protected_hardlinks,
1518  .maxlen = sizeof(int),
1519  .mode = 0600,
1521  .extra1 = &zero,
1522  .extra2 = &one,
1523  },
1524  {
1525  .procname = "suid_dumpable",
1526  .data = &suid_dumpable,
1527  .maxlen = sizeof(int),
1528  .mode = 0644,
1529  .proc_handler = proc_dointvec_minmax_coredump,
1530  .extra1 = &zero,
1531  .extra2 = &two,
1532  },
1533 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1534  {
1535  .procname = "binfmt_misc",
1536  .mode = 0555,
1537  .child = binfmt_misc_table,
1538  },
1539 #endif
1540  {
1541  .procname = "pipe-max-size",
1542  .data = &pipe_max_size,
1543  .maxlen = sizeof(int),
1544  .mode = 0644,
1546  .extra1 = &pipe_min_size,
1547  },
1548  { }
1549 };
1550 
1551 static struct ctl_table debug_table[] = {
1552 #ifdef CONFIG_SYSCTL_EXCEPTION_TRACE
1553  {
1554  .procname = "exception-trace",
1555  .data = &show_unhandled_signals,
1556  .maxlen = sizeof(int),
1557  .mode = 0644,
1559  },
1560 #endif
1561 #if defined(CONFIG_OPTPROBES)
1562  {
1563  .procname = "kprobes-optimization",
1564  .data = &sysctl_kprobes_optimization,
1565  .maxlen = sizeof(int),
1566  .mode = 0644,
1567  .proc_handler = proc_kprobes_optimization_handler,
1568  .extra1 = &zero,
1569  .extra2 = &one,
1570  },
1571 #endif
1572  { }
1573 };
1574 
1575 static struct ctl_table dev_table[] = {
1576  { }
1577 };
1578 
1579 int __init sysctl_init(void)
1580 {
1581  struct ctl_table_header *hdr;
1582 
1583  hdr = register_sysctl_table(sysctl_base_table);
1584  kmemleak_not_leak(hdr);
1585  return 0;
1586 }
1587 
1588 #endif /* CONFIG_SYSCTL */
1589 
1590 /*
1591  * /proc/sys support
1592  */
1593 
1594 #ifdef CONFIG_PROC_SYSCTL
1595 
1596 static int _proc_do_string(void* data, int maxlen, int write,
1597  void __user *buffer,
1598  size_t *lenp, loff_t *ppos)
1599 {
1600  size_t len;
1601  char __user *p;
1602  char c;
1603 
1604  if (!data || !maxlen || !*lenp) {
1605  *lenp = 0;
1606  return 0;
1607  }
1608 
1609  if (write) {
1610  len = 0;
1611  p = buffer;
1612  while (len < *lenp) {
1613  if (get_user(c, p++))
1614  return -EFAULT;
1615  if (c == 0 || c == '\n')
1616  break;
1617  len++;
1618  }
1619  if (len >= maxlen)
1620  len = maxlen-1;
1621  if(copy_from_user(data, buffer, len))
1622  return -EFAULT;
1623  ((char *) data)[len] = 0;
1624  *ppos += *lenp;
1625  } else {
1626  len = strlen(data);
1627  if (len > maxlen)
1628  len = maxlen;
1629 
1630  if (*ppos > len) {
1631  *lenp = 0;
1632  return 0;
1633  }
1634 
1635  data += *ppos;
1636  len -= *ppos;
1637 
1638  if (len > *lenp)
1639  len = *lenp;
1640  if (len)
1641  if(copy_to_user(buffer, data, len))
1642  return -EFAULT;
1643  if (len < *lenp) {
1644  if(put_user('\n', ((char __user *) buffer) + len))
1645  return -EFAULT;
1646  len++;
1647  }
1648  *lenp = len;
1649  *ppos += len;
1650  }
1651  return 0;
1652 }
1653 
1671 int proc_dostring(struct ctl_table *table, int write,
1672  void __user *buffer, size_t *lenp, loff_t *ppos)
1673 {
1674  return _proc_do_string(table->data, table->maxlen, write,
1675  buffer, lenp, ppos);
1676 }
1677 
1678 static size_t proc_skip_spaces(char **buf)
1679 {
1680  size_t ret;
1681  char *tmp = skip_spaces(*buf);
1682  ret = tmp - *buf;
1683  *buf = tmp;
1684  return ret;
1685 }
1686 
1687 static void proc_skip_char(char **buf, size_t *size, const char v)
1688 {
1689  while (*size) {
1690  if (**buf != v)
1691  break;
1692  (*size)--;
1693  (*buf)++;
1694  }
1695 }
1696 
1697 #define TMPBUFLEN 22
1698 
1714 static int proc_get_long(char **buf, size_t *size,
1715  unsigned long *val, bool *neg,
1716  const char *perm_tr, unsigned perm_tr_len, char *tr)
1717 {
1718  int len;
1719  char *p, tmp[TMPBUFLEN];
1720 
1721  if (!*size)
1722  return -EINVAL;
1723 
1724  len = *size;
1725  if (len > TMPBUFLEN - 1)
1726  len = TMPBUFLEN - 1;
1727 
1728  memcpy(tmp, *buf, len);
1729 
1730  tmp[len] = 0;
1731  p = tmp;
1732  if (*p == '-' && *size > 1) {
1733  *neg = true;
1734  p++;
1735  } else
1736  *neg = false;
1737  if (!isdigit(*p))
1738  return -EINVAL;
1739 
1740  *val = simple_strtoul(p, &p, 0);
1741 
1742  len = p - tmp;
1743 
1744  /* We don't know if the next char is whitespace thus we may accept
1745  * invalid integers (e.g. 1234...a) or two integers instead of one
1746  * (e.g. 123...1). So lets not allow such large numbers. */
1747  if (len == TMPBUFLEN - 1)
1748  return -EINVAL;
1749 
1750  if (len < *size && perm_tr_len && !memchr(perm_tr, *p, perm_tr_len))
1751  return -EINVAL;
1752 
1753  if (tr && (len < *size))
1754  *tr = *p;
1755 
1756  *buf += len;
1757  *size -= len;
1758 
1759  return 0;
1760 }
1761 
1773 static int proc_put_long(void __user **buf, size_t *size, unsigned long val,
1774  bool neg)
1775 {
1776  int len;
1777  char tmp[TMPBUFLEN], *p = tmp;
1778 
1779  sprintf(p, "%s%lu", neg ? "-" : "", val);
1780  len = strlen(tmp);
1781  if (len > *size)
1782  len = *size;
1783  if (copy_to_user(*buf, tmp, len))
1784  return -EFAULT;
1785  *size -= len;
1786  *buf += len;
1787  return 0;
1788 }
1789 #undef TMPBUFLEN
1790 
1791 static int proc_put_char(void __user **buf, size_t *size, char c)
1792 {
1793  if (*size) {
1794  char __user **buffer = (char __user **)buf;
1795  if (put_user(c, *buffer))
1796  return -EFAULT;
1797  (*size)--, (*buffer)++;
1798  *buf = *buffer;
1799  }
1800  return 0;
1801 }
1802 
1803 static int do_proc_dointvec_conv(bool *negp, unsigned long *lvalp,
1804  int *valp,
1805  int write, void *data)
1806 {
1807  if (write) {
1808  *valp = *negp ? -*lvalp : *lvalp;
1809  } else {
1810  int val = *valp;
1811  if (val < 0) {
1812  *negp = true;
1813  *lvalp = (unsigned long)-val;
1814  } else {
1815  *negp = false;
1816  *lvalp = (unsigned long)val;
1817  }
1818  }
1819  return 0;
1820 }
1821 
1822 static const char proc_wspace_sep[] = { ' ', '\t', '\n' };
1823 
1824 static int __do_proc_dointvec(void *tbl_data, struct ctl_table *table,
1825  int write, void __user *buffer,
1826  size_t *lenp, loff_t *ppos,
1827  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1828  int write, void *data),
1829  void *data)
1830 {
1831  int *i, vleft, first = 1, err = 0;
1832  unsigned long page = 0;
1833  size_t left;
1834  char *kbuf;
1835 
1836  if (!tbl_data || !table->maxlen || !*lenp || (*ppos && !write)) {
1837  *lenp = 0;
1838  return 0;
1839  }
1840 
1841  i = (int *) tbl_data;
1842  vleft = table->maxlen / sizeof(*i);
1843  left = *lenp;
1844 
1845  if (!conv)
1846  conv = do_proc_dointvec_conv;
1847 
1848  if (write) {
1849  if (left > PAGE_SIZE - 1)
1850  left = PAGE_SIZE - 1;
1852  kbuf = (char *) page;
1853  if (!kbuf)
1854  return -ENOMEM;
1855  if (copy_from_user(kbuf, buffer, left)) {
1856  err = -EFAULT;
1857  goto free;
1858  }
1859  kbuf[left] = 0;
1860  }
1861 
1862  for (; left && vleft--; i++, first=0) {
1863  unsigned long lval;
1864  bool neg;
1865 
1866  if (write) {
1867  left -= proc_skip_spaces(&kbuf);
1868 
1869  if (!left)
1870  break;
1871  err = proc_get_long(&kbuf, &left, &lval, &neg,
1872  proc_wspace_sep,
1873  sizeof(proc_wspace_sep), NULL);
1874  if (err)
1875  break;
1876  if (conv(&neg, &lval, i, 1, data)) {
1877  err = -EINVAL;
1878  break;
1879  }
1880  } else {
1881  if (conv(&neg, &lval, i, 0, data)) {
1882  err = -EINVAL;
1883  break;
1884  }
1885  if (!first)
1886  err = proc_put_char(&buffer, &left, '\t');
1887  if (err)
1888  break;
1889  err = proc_put_long(&buffer, &left, lval, neg);
1890  if (err)
1891  break;
1892  }
1893  }
1894 
1895  if (!write && !first && left && !err)
1896  err = proc_put_char(&buffer, &left, '\n');
1897  if (write && !err && left)
1898  left -= proc_skip_spaces(&kbuf);
1899 free:
1900  if (write) {
1901  free_page(page);
1902  if (first)
1903  return err ? : -EINVAL;
1904  }
1905  *lenp -= left;
1906  *ppos += *lenp;
1907  return err;
1908 }
1909 
1910 static int do_proc_dointvec(struct ctl_table *table, int write,
1911  void __user *buffer, size_t *lenp, loff_t *ppos,
1912  int (*conv)(bool *negp, unsigned long *lvalp, int *valp,
1913  int write, void *data),
1914  void *data)
1915 {
1916  return __do_proc_dointvec(table->data, table, write,
1917  buffer, lenp, ppos, conv, data);
1918 }
1919 
1933 int proc_dointvec(struct ctl_table *table, int write,
1934  void __user *buffer, size_t *lenp, loff_t *ppos)
1935 {
1936  return do_proc_dointvec(table,write,buffer,lenp,ppos,
1937  NULL,NULL);
1938 }
1939 
1940 /*
1941  * Taint values can only be increased
1942  * This means we can safely use a temporary.
1943  */
1944 static int proc_taint(struct ctl_table *table, int write,
1945  void __user *buffer, size_t *lenp, loff_t *ppos)
1946 {
1947  struct ctl_table t;
1948  unsigned long tmptaint = get_taint();
1949  int err;
1950 
1951  if (write && !capable(CAP_SYS_ADMIN))
1952  return -EPERM;
1953 
1954  t = *table;
1955  t.data = &tmptaint;
1956  err = proc_doulongvec_minmax(&t, write, buffer, lenp, ppos);
1957  if (err < 0)
1958  return err;
1959 
1960  if (write) {
1961  /*
1962  * Poor man's atomic or. Not worth adding a primitive
1963  * to everyone's atomic.h for this
1964  */
1965  int i;
1966  for (i = 0; i < BITS_PER_LONG && tmptaint >> i; i++) {
1967  if ((tmptaint >> i) & 1)
1968  add_taint(i);
1969  }
1970  }
1971 
1972  return err;
1973 }
1974 
1975 #ifdef CONFIG_PRINTK
1976 static int proc_dointvec_minmax_sysadmin(struct ctl_table *table, int write,
1977  void __user *buffer, size_t *lenp, loff_t *ppos)
1978 {
1979  if (write && !capable(CAP_SYS_ADMIN))
1980  return -EPERM;
1981 
1982  return proc_dointvec_minmax(table, write, buffer, lenp, ppos);
1983 }
1984 #endif
1985 
1986 struct do_proc_dointvec_minmax_conv_param {
1987  int *min;
1988  int *max;
1989 };
1990 
1991 static int do_proc_dointvec_minmax_conv(bool *negp, unsigned long *lvalp,
1992  int *valp,
1993  int write, void *data)
1994 {
1995  struct do_proc_dointvec_minmax_conv_param *param = data;
1996  if (write) {
1997  int val = *negp ? -*lvalp : *lvalp;
1998  if ((param->min && *param->min > val) ||
1999  (param->max && *param->max < val))
2000  return -EINVAL;
2001  *valp = val;
2002  } else {
2003  int val = *valp;
2004  if (val < 0) {
2005  *negp = true;
2006  *lvalp = (unsigned long)-val;
2007  } else {
2008  *negp = false;
2009  *lvalp = (unsigned long)val;
2010  }
2011  }
2012  return 0;
2013 }
2014 
2031 int proc_dointvec_minmax(struct ctl_table *table, int write,
2032  void __user *buffer, size_t *lenp, loff_t *ppos)
2033 {
2034  struct do_proc_dointvec_minmax_conv_param param = {
2035  .min = (int *) table->extra1,
2036  .max = (int *) table->extra2,
2037  };
2038  return do_proc_dointvec(table, write, buffer, lenp, ppos,
2039  do_proc_dointvec_minmax_conv, &param);
2040 }
2041 
2042 static void validate_coredump_safety(void)
2043 {
2044 #ifdef CONFIG_COREDUMP
2046  core_pattern[0] != '/' && core_pattern[0] != '|') {
2047  printk(KERN_WARNING "Unsafe core_pattern used with "\
2048  "suid_dumpable=2. Pipe handler or fully qualified "\
2049  "core dump path required.\n");
2050  }
2051 #endif
2052 }
2053 
2054 static int proc_dointvec_minmax_coredump(struct ctl_table *table, int write,
2055  void __user *buffer, size_t *lenp, loff_t *ppos)
2056 {
2057  int error = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
2058  if (!error)
2059  validate_coredump_safety();
2060  return error;
2061 }
2062 
2063 #ifdef CONFIG_COREDUMP
2064 static int proc_dostring_coredump(struct ctl_table *table, int write,
2065  void __user *buffer, size_t *lenp, loff_t *ppos)
2066 {
2067  int error = proc_dostring(table, write, buffer, lenp, ppos);
2068  if (!error)
2069  validate_coredump_safety();
2070  return error;
2071 }
2072 #endif
2073 
2074 static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int write,
2075  void __user *buffer,
2076  size_t *lenp, loff_t *ppos,
2077  unsigned long convmul,
2078  unsigned long convdiv)
2079 {
2080  unsigned long *i, *min, *max;
2081  int vleft, first = 1, err = 0;
2082  unsigned long page = 0;
2083  size_t left;
2084  char *kbuf;
2085 
2086  if (!data || !table->maxlen || !*lenp || (*ppos && !write)) {
2087  *lenp = 0;
2088  return 0;
2089  }
2090 
2091  i = (unsigned long *) data;
2092  min = (unsigned long *) table->extra1;
2093  max = (unsigned long *) table->extra2;
2094  vleft = table->maxlen / sizeof(unsigned long);
2095  left = *lenp;
2096 
2097  if (write) {
2098  if (left > PAGE_SIZE - 1)
2099  left = PAGE_SIZE - 1;
2101  kbuf = (char *) page;
2102  if (!kbuf)
2103  return -ENOMEM;
2104  if (copy_from_user(kbuf, buffer, left)) {
2105  err = -EFAULT;
2106  goto free;
2107  }
2108  kbuf[left] = 0;
2109  }
2110 
2111  for (; left && vleft--; i++, first = 0) {
2112  unsigned long val;
2113 
2114  if (write) {
2115  bool neg;
2116 
2117  left -= proc_skip_spaces(&kbuf);
2118 
2119  err = proc_get_long(&kbuf, &left, &val, &neg,
2120  proc_wspace_sep,
2121  sizeof(proc_wspace_sep), NULL);
2122  if (err)
2123  break;
2124  if (neg)
2125  continue;
2126  if ((min && val < *min) || (max && val > *max))
2127  continue;
2128  *i = val;
2129  } else {
2130  val = convdiv * (*i) / convmul;
2131  if (!first)
2132  err = proc_put_char(&buffer, &left, '\t');
2133  err = proc_put_long(&buffer, &left, val, false);
2134  if (err)
2135  break;
2136  }
2137  }
2138 
2139  if (!write && !first && left && !err)
2140  err = proc_put_char(&buffer, &left, '\n');
2141  if (write && !err)
2142  left -= proc_skip_spaces(&kbuf);
2143 free:
2144  if (write) {
2145  free_page(page);
2146  if (first)
2147  return err ? : -EINVAL;
2148  }
2149  *lenp -= left;
2150  *ppos += *lenp;
2151  return err;
2152 }
2153 
2154 static int do_proc_doulongvec_minmax(struct ctl_table *table, int write,
2155  void __user *buffer,
2156  size_t *lenp, loff_t *ppos,
2157  unsigned long convmul,
2158  unsigned long convdiv)
2159 {
2160  return __do_proc_doulongvec_minmax(table->data, table, write,
2161  buffer, lenp, ppos, convmul, convdiv);
2162 }
2163 
2180 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2181  void __user *buffer, size_t *lenp, loff_t *ppos)
2182 {
2183  return do_proc_doulongvec_minmax(table, write, buffer, lenp, ppos, 1l, 1l);
2184 }
2185 
2203 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2204  void __user *buffer,
2205  size_t *lenp, loff_t *ppos)
2206 {
2207  return do_proc_doulongvec_minmax(table, write, buffer,
2208  lenp, ppos, HZ, 1000l);
2209 }
2210 
2211 
2212 static int do_proc_dointvec_jiffies_conv(bool *negp, unsigned long *lvalp,
2213  int *valp,
2214  int write, void *data)
2215 {
2216  if (write) {
2217  if (*lvalp > LONG_MAX / HZ)
2218  return 1;
2219  *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2220  } else {
2221  int val = *valp;
2222  unsigned long lval;
2223  if (val < 0) {
2224  *negp = true;
2225  lval = (unsigned long)-val;
2226  } else {
2227  *negp = false;
2228  lval = (unsigned long)val;
2229  }
2230  *lvalp = lval / HZ;
2231  }
2232  return 0;
2233 }
2234 
2235 static int do_proc_dointvec_userhz_jiffies_conv(bool *negp, unsigned long *lvalp,
2236  int *valp,
2237  int write, void *data)
2238 {
2239  if (write) {
2240  if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2241  return 1;
2242  *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2243  } else {
2244  int val = *valp;
2245  unsigned long lval;
2246  if (val < 0) {
2247  *negp = true;
2248  lval = (unsigned long)-val;
2249  } else {
2250  *negp = false;
2251  lval = (unsigned long)val;
2252  }
2253  *lvalp = jiffies_to_clock_t(lval);
2254  }
2255  return 0;
2256 }
2257 
2258 static int do_proc_dointvec_ms_jiffies_conv(bool *negp, unsigned long *lvalp,
2259  int *valp,
2260  int write, void *data)
2261 {
2262  if (write) {
2263  *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2264  } else {
2265  int val = *valp;
2266  unsigned long lval;
2267  if (val < 0) {
2268  *negp = true;
2269  lval = (unsigned long)-val;
2270  } else {
2271  *negp = false;
2272  lval = (unsigned long)val;
2273  }
2274  *lvalp = jiffies_to_msecs(lval);
2275  }
2276  return 0;
2277 }
2278 
2294 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2295  void __user *buffer, size_t *lenp, loff_t *ppos)
2296 {
2297  return do_proc_dointvec(table,write,buffer,lenp,ppos,
2298  do_proc_dointvec_jiffies_conv,NULL);
2299 }
2300 
2316 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2317  void __user *buffer, size_t *lenp, loff_t *ppos)
2318 {
2319  return do_proc_dointvec(table,write,buffer,lenp,ppos,
2320  do_proc_dointvec_userhz_jiffies_conv,NULL);
2321 }
2322 
2339 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2340  void __user *buffer, size_t *lenp, loff_t *ppos)
2341 {
2342  return do_proc_dointvec(table, write, buffer, lenp, ppos,
2343  do_proc_dointvec_ms_jiffies_conv, NULL);
2344 }
2345 
2346 static int proc_do_cad_pid(struct ctl_table *table, int write,
2347  void __user *buffer, size_t *lenp, loff_t *ppos)
2348 {
2349  struct pid *new_pid;
2350  pid_t tmp;
2351  int r;
2352 
2353  tmp = pid_vnr(cad_pid);
2354 
2355  r = __do_proc_dointvec(&tmp, table, write, buffer,
2356  lenp, ppos, NULL, NULL);
2357  if (r || !write)
2358  return r;
2359 
2360  new_pid = find_get_pid(tmp);
2361  if (!new_pid)
2362  return -ESRCH;
2363 
2364  put_pid(xchg(&cad_pid, new_pid));
2365  return 0;
2366 }
2367 
2385 int proc_do_large_bitmap(struct ctl_table *table, int write,
2386  void __user *buffer, size_t *lenp, loff_t *ppos)
2387 {
2388  int err = 0;
2389  bool first = 1;
2390  size_t left = *lenp;
2391  unsigned long bitmap_len = table->maxlen;
2392  unsigned long *bitmap = (unsigned long *) table->data;
2393  unsigned long *tmp_bitmap = NULL;
2394  char tr_a[] = { '-', ',', '\n' }, tr_b[] = { ',', '\n', 0 }, c;
2395 
2396  if (!bitmap_len || !left || (*ppos && !write)) {
2397  *lenp = 0;
2398  return 0;
2399  }
2400 
2401  if (write) {
2402  unsigned long page = 0;
2403  char *kbuf;
2404 
2405  if (left > PAGE_SIZE - 1)
2406  left = PAGE_SIZE - 1;
2407 
2409  kbuf = (char *) page;
2410  if (!kbuf)
2411  return -ENOMEM;
2412  if (copy_from_user(kbuf, buffer, left)) {
2413  free_page(page);
2414  return -EFAULT;
2415  }
2416  kbuf[left] = 0;
2417 
2418  tmp_bitmap = kzalloc(BITS_TO_LONGS(bitmap_len) * sizeof(unsigned long),
2419  GFP_KERNEL);
2420  if (!tmp_bitmap) {
2421  free_page(page);
2422  return -ENOMEM;
2423  }
2424  proc_skip_char(&kbuf, &left, '\n');
2425  while (!err && left) {
2426  unsigned long val_a, val_b;
2427  bool neg;
2428 
2429  err = proc_get_long(&kbuf, &left, &val_a, &neg, tr_a,
2430  sizeof(tr_a), &c);
2431  if (err)
2432  break;
2433  if (val_a >= bitmap_len || neg) {
2434  err = -EINVAL;
2435  break;
2436  }
2437 
2438  val_b = val_a;
2439  if (left) {
2440  kbuf++;
2441  left--;
2442  }
2443 
2444  if (c == '-') {
2445  err = proc_get_long(&kbuf, &left, &val_b,
2446  &neg, tr_b, sizeof(tr_b),
2447  &c);
2448  if (err)
2449  break;
2450  if (val_b >= bitmap_len || neg ||
2451  val_a > val_b) {
2452  err = -EINVAL;
2453  break;
2454  }
2455  if (left) {
2456  kbuf++;
2457  left--;
2458  }
2459  }
2460 
2461  bitmap_set(tmp_bitmap, val_a, val_b - val_a + 1);
2462  first = 0;
2463  proc_skip_char(&kbuf, &left, '\n');
2464  }
2465  free_page(page);
2466  } else {
2467  unsigned long bit_a, bit_b = 0;
2468 
2469  while (left) {
2470  bit_a = find_next_bit(bitmap, bitmap_len, bit_b);
2471  if (bit_a >= bitmap_len)
2472  break;
2473  bit_b = find_next_zero_bit(bitmap, bitmap_len,
2474  bit_a + 1) - 1;
2475 
2476  if (!first) {
2477  err = proc_put_char(&buffer, &left, ',');
2478  if (err)
2479  break;
2480  }
2481  err = proc_put_long(&buffer, &left, bit_a, false);
2482  if (err)
2483  break;
2484  if (bit_a != bit_b) {
2485  err = proc_put_char(&buffer, &left, '-');
2486  if (err)
2487  break;
2488  err = proc_put_long(&buffer, &left, bit_b, false);
2489  if (err)
2490  break;
2491  }
2492 
2493  first = 0; bit_b++;
2494  }
2495  if (!err)
2496  err = proc_put_char(&buffer, &left, '\n');
2497  }
2498 
2499  if (!err) {
2500  if (write) {
2501  if (*ppos)
2502  bitmap_or(bitmap, bitmap, tmp_bitmap, bitmap_len);
2503  else
2504  bitmap_copy(bitmap, tmp_bitmap, bitmap_len);
2505  }
2506  kfree(tmp_bitmap);
2507  *lenp -= left;
2508  *ppos += *lenp;
2509  return 0;
2510  } else {
2511  kfree(tmp_bitmap);
2512  return err;
2513  }
2514 }
2515 
2516 #else /* CONFIG_PROC_SYSCTL */
2517 
2518 int proc_dostring(struct ctl_table *table, int write,
2519  void __user *buffer, size_t *lenp, loff_t *ppos)
2520 {
2521  return -ENOSYS;
2522 }
2523 
2524 int proc_dointvec(struct ctl_table *table, int write,
2525  void __user *buffer, size_t *lenp, loff_t *ppos)
2526 {
2527  return -ENOSYS;
2528 }
2529 
2530 int proc_dointvec_minmax(struct ctl_table *table, int write,
2531  void __user *buffer, size_t *lenp, loff_t *ppos)
2532 {
2533  return -ENOSYS;
2534 }
2535 
2536 int proc_dointvec_jiffies(struct ctl_table *table, int write,
2537  void __user *buffer, size_t *lenp, loff_t *ppos)
2538 {
2539  return -ENOSYS;
2540 }
2541 
2542 int proc_dointvec_userhz_jiffies(struct ctl_table *table, int write,
2543  void __user *buffer, size_t *lenp, loff_t *ppos)
2544 {
2545  return -ENOSYS;
2546 }
2547 
2548 int proc_dointvec_ms_jiffies(struct ctl_table *table, int write,
2549  void __user *buffer, size_t *lenp, loff_t *ppos)
2550 {
2551  return -ENOSYS;
2552 }
2553 
2554 int proc_doulongvec_minmax(struct ctl_table *table, int write,
2555  void __user *buffer, size_t *lenp, loff_t *ppos)
2556 {
2557  return -ENOSYS;
2558 }
2559 
2560 int proc_doulongvec_ms_jiffies_minmax(struct ctl_table *table, int write,
2561  void __user *buffer,
2562  size_t *lenp, loff_t *ppos)
2563 {
2564  return -ENOSYS;
2565 }
2566 
2567 
2568 #endif /* CONFIG_PROC_SYSCTL */
2569 
2570 /*
2571  * No sense putting this after each symbol definition, twice,
2572  * exception granted :-)
2573  */