This repository has been archived by the owner on Mar 4, 2018. It is now read-only.
-
Notifications
You must be signed in to change notification settings - Fork 46
/
Copy pathmdb.c
9364 lines (8513 loc) · 254 KB
/
mdb.c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/** @file mdb.c
* @brief Lightning memory-mapped database library
*
* A Btree-based database management library modeled loosely on the
* BerkeleyDB API, but much simplified.
*/
/*
* Copyright 2011-2014 Howard Chu, Symas Corp.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted only as authorized by the OpenLDAP
* Public License.
*
* A copy of this license is available in the file LICENSE in the
* top-level directory of the distribution or, alternatively, at
* <http://www.OpenLDAP.org/license.html>.
*
* This code is derived from btree.c written by Martin Hedenfalk.
*
* Copyright (c) 2009, 2010 Martin Hedenfalk <[email protected]>
*
* Permission to use, copy, modify, and distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
* copyright notice and this permission notice appear in all copies.
*
* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
#ifndef _GNU_SOURCE
#define _GNU_SOURCE 1
#endif
#ifdef _WIN32
#include <malloc.h>
#include <windows.h>
/** getpid() returns int; MinGW defines pid_t but MinGW64 typedefs it
* as int64 which is wrong. MSVC doesn't define it at all, so just
* don't use it.
*/
#define MDB_PID_T int
#define MDB_THR_T DWORD
#include <sys/types.h>
#include <sys/stat.h>
#ifdef __GNUC__
# include <sys/param.h>
#else
# define LITTLE_ENDIAN 1234
# define BIG_ENDIAN 4321
# define BYTE_ORDER LITTLE_ENDIAN
# ifndef SSIZE_MAX
# define SSIZE_MAX INT_MAX
# endif
#endif
#else
#include <sys/types.h>
#include <sys/stat.h>
#define MDB_PID_T pid_t
#define MDB_THR_T pthread_t
#include <sys/param.h>
#include <sys/uio.h>
#include <sys/mman.h>
#ifdef HAVE_SYS_FILE_H
#include <sys/file.h>
#endif
#include <fcntl.h>
#endif
#if defined(__mips) && defined(__linux)
/* MIPS has cache coherency issues, requires explicit cache control */
#include <asm/cachectl.h>
extern int cacheflush(char *addr, int nbytes, int cache);
#define CACHEFLUSH(addr, bytes, cache) cacheflush(addr, bytes, cache)
#else
#define CACHEFLUSH(addr, bytes, cache)
#endif
#include <errno.h>
#include <limits.h>
#include <stddef.h>
#include <inttypes.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <time.h>
#include <unistd.h>
#if defined(__sun)
/* Most platforms have posix_memalign, older may only have memalign */
#define HAVE_MEMALIGN 1
#include <malloc.h>
#endif
#if !(defined(BYTE_ORDER) || defined(__BYTE_ORDER))
#include <netinet/in.h>
#include <resolv.h> /* defines BYTE_ORDER on HPUX and Solaris */
#endif
#if defined(__APPLE__) || defined (BSD)
# define MDB_USE_POSIX_SEM 1
# define MDB_FDATASYNC fsync
#elif defined(ANDROID)
# define MDB_FDATASYNC fsync
#endif
#ifndef _WIN32
#include <pthread.h>
#ifdef MDB_USE_POSIX_SEM
# define MDB_USE_HASH 1
#include <semaphore.h>
#endif
#endif
#ifdef USE_VALGRIND
#include <valgrind/memcheck.h>
#define VGMEMP_CREATE(h,r,z) VALGRIND_CREATE_MEMPOOL(h,r,z)
#define VGMEMP_ALLOC(h,a,s) VALGRIND_MEMPOOL_ALLOC(h,a,s)
#define VGMEMP_FREE(h,a) VALGRIND_MEMPOOL_FREE(h,a)
#define VGMEMP_DESTROY(h) VALGRIND_DESTROY_MEMPOOL(h)
#define VGMEMP_DEFINED(a,s) VALGRIND_MAKE_MEM_DEFINED(a,s)
#else
#define VGMEMP_CREATE(h,r,z)
#define VGMEMP_ALLOC(h,a,s)
#define VGMEMP_FREE(h,a)
#define VGMEMP_DESTROY(h)
#define VGMEMP_DEFINED(a,s)
#endif
#ifndef BYTE_ORDER
# if (defined(_LITTLE_ENDIAN) || defined(_BIG_ENDIAN)) && !(defined(_LITTLE_ENDIAN) && defined(_BIG_ENDIAN))
/* Solaris just defines one or the other */
# define LITTLE_ENDIAN 1234
# define BIG_ENDIAN 4321
# ifdef _LITTLE_ENDIAN
# define BYTE_ORDER LITTLE_ENDIAN
# else
# define BYTE_ORDER BIG_ENDIAN
# endif
# else
# define BYTE_ORDER __BYTE_ORDER
# endif
#endif
#ifndef LITTLE_ENDIAN
#define LITTLE_ENDIAN __LITTLE_ENDIAN
#endif
#ifndef BIG_ENDIAN
#define BIG_ENDIAN __BIG_ENDIAN
#endif
#if defined(__i386) || defined(__x86_64) || defined(_M_IX86)
#define MISALIGNED_OK 1
#endif
#include "lmdb.h"
#include "midl.h"
#if (BYTE_ORDER == LITTLE_ENDIAN) == (BYTE_ORDER == BIG_ENDIAN)
# error "Unknown or unsupported endianness (BYTE_ORDER)"
#elif (-6 & 5) || CHAR_BIT != 8 || UINT_MAX < 0xffffffff || ULONG_MAX % 0xFFFF
# error "Two's complement, reasonably sized integer types, please"
#endif
#ifdef __GNUC__
/** Put infrequently used env functions in separate section */
# ifdef __APPLE__
# define ESECT __attribute__ ((section("__TEXT,text_env")))
# else
# define ESECT __attribute__ ((section("text_env")))
# endif
#else
#define ESECT
#endif
/** @defgroup internal LMDB Internals
* @{
*/
/** @defgroup compat Compatibility Macros
* A bunch of macros to minimize the amount of platform-specific ifdefs
* needed throughout the rest of the code. When the features this library
* needs are similar enough to POSIX to be hidden in a one-or-two line
* replacement, this macro approach is used.
* @{
*/
/** Features under development */
#ifndef MDB_DEVEL
#define MDB_DEVEL 0
#endif
/** Wrapper around __func__, which is a C99 feature */
#if __STDC_VERSION__ >= 199901L
# define mdb_func_ __func__
#elif __GNUC__ >= 2 || _MSC_VER >= 1300
# define mdb_func_ __FUNCTION__
#else
/* If a debug message says <mdb_unknown>(), update the #if statements above */
# define mdb_func_ "<mdb_unknown>"
#endif
#ifdef _WIN32
#define MDB_USE_HASH 1
#define MDB_PIDLOCK 0
#define THREAD_RET DWORD
#define pthread_t HANDLE
#define pthread_mutex_t HANDLE
#define pthread_cond_t HANDLE
#define pthread_key_t DWORD
#define pthread_self() GetCurrentThreadId()
#define pthread_key_create(x,y) \
((*(x) = TlsAlloc()) == TLS_OUT_OF_INDEXES ? ErrCode() : 0)
#define pthread_key_delete(x) TlsFree(x)
#define pthread_getspecific(x) TlsGetValue(x)
#define pthread_setspecific(x,y) (TlsSetValue(x,y) ? 0 : ErrCode())
#define pthread_mutex_unlock(x) ReleaseMutex(*x)
#define pthread_mutex_lock(x) WaitForSingleObject(*x, INFINITE)
#define pthread_cond_signal(x) SetEvent(*x)
#define pthread_cond_wait(cond,mutex) do{SignalObjectAndWait(*mutex, *cond, INFINITE, FALSE); WaitForSingleObject(*mutex, INFINITE);}while(0)
#define THREAD_CREATE(thr,start,arg) thr=CreateThread(NULL,0,start,arg,0,NULL)
#define THREAD_FINISH(thr) WaitForSingleObject(thr, INFINITE)
#define LOCK_MUTEX_R(env) pthread_mutex_lock(&(env)->me_rmutex)
#define UNLOCK_MUTEX_R(env) pthread_mutex_unlock(&(env)->me_rmutex)
#define LOCK_MUTEX_W(env) pthread_mutex_lock(&(env)->me_wmutex)
#define UNLOCK_MUTEX_W(env) pthread_mutex_unlock(&(env)->me_wmutex)
#define getpid() GetCurrentProcessId()
#define MDB_FDATASYNC(fd) (!FlushFileBuffers(fd))
#define MDB_MSYNC(addr,len,flags) (!FlushViewOfFile(addr,len))
#define ErrCode() GetLastError()
#define GET_PAGESIZE(x) {SYSTEM_INFO si; GetSystemInfo(&si); (x) = si.dwPageSize;}
#define close(fd) (CloseHandle(fd) ? 0 : -1)
#define munmap(ptr,len) UnmapViewOfFile(ptr)
#ifdef PROCESS_QUERY_LIMITED_INFORMATION
#define MDB_PROCESS_QUERY_LIMITED_INFORMATION PROCESS_QUERY_LIMITED_INFORMATION
#else
#define MDB_PROCESS_QUERY_LIMITED_INFORMATION 0x1000
#endif
#define Z "I"
#else
#define THREAD_RET void *
#define THREAD_CREATE(thr,start,arg) pthread_create(&thr,NULL,start,arg)
#define THREAD_FINISH(thr) pthread_join(thr,NULL)
#define Z "z" /**< printf format modifier for size_t */
/** For MDB_LOCK_FORMAT: True if readers take a pid lock in the lockfile */
#define MDB_PIDLOCK 1
#ifdef MDB_USE_POSIX_SEM
#define LOCK_MUTEX_R(env) mdb_sem_wait((env)->me_rmutex)
#define UNLOCK_MUTEX_R(env) sem_post((env)->me_rmutex)
#define LOCK_MUTEX_W(env) mdb_sem_wait((env)->me_wmutex)
#define UNLOCK_MUTEX_W(env) sem_post((env)->me_wmutex)
static int
mdb_sem_wait(sem_t *sem)
{
int rc;
while ((rc = sem_wait(sem)) && (rc = errno) == EINTR) ;
return rc;
}
#else
/** Lock the reader mutex.
*/
#define LOCK_MUTEX_R(env) pthread_mutex_lock(&(env)->me_txns->mti_mutex)
/** Unlock the reader mutex.
*/
#define UNLOCK_MUTEX_R(env) pthread_mutex_unlock(&(env)->me_txns->mti_mutex)
/** Lock the writer mutex.
* Only a single write transaction is allowed at a time. Other writers
* will block waiting for this mutex.
*/
#define LOCK_MUTEX_W(env) pthread_mutex_lock(&(env)->me_txns->mti_wmutex)
/** Unlock the writer mutex.
*/
#define UNLOCK_MUTEX_W(env) pthread_mutex_unlock(&(env)->me_txns->mti_wmutex)
#endif /* MDB_USE_POSIX_SEM */
/** Get the error code for the last failed system function.
*/
#define ErrCode() errno
/** An abstraction for a file handle.
* On POSIX systems file handles are small integers. On Windows
* they're opaque pointers.
*/
#define HANDLE int
/** A value for an invalid file handle.
* Mainly used to initialize file variables and signify that they are
* unused.
*/
#define INVALID_HANDLE_VALUE (-1)
/** Get the size of a memory page for the system.
* This is the basic size that the platform's memory manager uses, and is
* fundamental to the use of memory-mapped files.
*/
#define GET_PAGESIZE(x) ((x) = sysconf(_SC_PAGE_SIZE))
#endif
#if defined(_WIN32) || defined(MDB_USE_POSIX_SEM)
#define MNAME_LEN 32
#else
#define MNAME_LEN (sizeof(pthread_mutex_t))
#endif
/** @} */
#ifndef _WIN32
/** A flag for opening a file and requesting synchronous data writes.
* This is only used when writing a meta page. It's not strictly needed;
* we could just do a normal write and then immediately perform a flush.
* But if this flag is available it saves us an extra system call.
*
* @note If O_DSYNC is undefined but exists in /usr/include,
* preferably set some compiler flag to get the definition.
* Otherwise compile with the less efficient -DMDB_DSYNC=O_SYNC.
*/
#ifndef MDB_DSYNC
# define MDB_DSYNC O_DSYNC
#endif
#endif
/** Function for flushing the data of a file. Define this to fsync
* if fdatasync() is not supported.
*/
#ifndef MDB_FDATASYNC
# define MDB_FDATASYNC fdatasync
#endif
#ifndef MDB_MSYNC
# define MDB_MSYNC(addr,len,flags) msync(addr,len,flags)
#endif
#ifndef MS_SYNC
#define MS_SYNC 1
#endif
#ifndef MS_ASYNC
#define MS_ASYNC 0
#endif
/** A page number in the database.
* Note that 64 bit page numbers are overkill, since pages themselves
* already represent 12-13 bits of addressable memory, and the OS will
* always limit applications to a maximum of 63 bits of address space.
*
* @note In the #MDB_node structure, we only store 48 bits of this value,
* which thus limits us to only 60 bits of addressable data.
*/
typedef MDB_ID pgno_t;
/** A transaction ID.
* See struct MDB_txn.mt_txnid for details.
*/
typedef MDB_ID txnid_t;
/** @defgroup debug Debug Macros
* @{
*/
#ifndef MDB_DEBUG
/** Enable debug output. Needs variable argument macros (a C99 feature).
* Set this to 1 for copious tracing. Set to 2 to add dumps of all IDLs
* read from and written to the database (used for free space management).
*/
#define MDB_DEBUG 0
#endif
#if MDB_DEBUG
static int mdb_debug;
static txnid_t mdb_debug_start;
/** Print a debug message with printf formatting.
* Requires double parenthesis around 2 or more args.
*/
# define DPRINTF(args) ((void) ((mdb_debug) && DPRINTF0 args))
# define DPRINTF0(fmt, ...) \
fprintf(stderr, "%s:%d " fmt "\n", mdb_func_, __LINE__, __VA_ARGS__)
#else
# define DPRINTF(args) ((void) 0)
#endif
/** Print a debug string.
* The string is printed literally, with no format processing.
*/
#define DPUTS(arg) DPRINTF(("%s", arg))
/** Debuging output value of a cursor DBI: Negative in a sub-cursor. */
#define DDBI(mc) \
(((mc)->mc_flags & C_SUB) ? -(int)(mc)->mc_dbi : (int)(mc)->mc_dbi)
/** @} */
/** @brief The maximum size of a database page.
*
* It is 32k or 64k, since value-PAGEBASE must fit in
* #MDB_page.%mp_upper.
*
* LMDB will use database pages < OS pages if needed.
* That causes more I/O in write transactions: The OS must
* know (read) the whole page before writing a partial page.
*
* Note that we don't currently support Huge pages. On Linux,
* regular data files cannot use Huge pages, and in general
* Huge pages aren't actually pageable. We rely on the OS
* demand-pager to read our data and page it out when memory
* pressure from other processes is high. So until OSs have
* actual paging support for Huge pages, they're not viable.
*/
#define MAX_PAGESIZE (PAGEBASE ? 0x10000 : 0x8000)
/** The minimum number of keys required in a database page.
* Setting this to a larger value will place a smaller bound on the
* maximum size of a data item. Data items larger than this size will
* be pushed into overflow pages instead of being stored directly in
* the B-tree node. This value used to default to 4. With a page size
* of 4096 bytes that meant that any item larger than 1024 bytes would
* go into an overflow page. That also meant that on average 2-3KB of
* each overflow page was wasted space. The value cannot be lower than
* 2 because then there would no longer be a tree structure. With this
* value, items larger than 2KB will go into overflow pages, and on
* average only 1KB will be wasted.
*/
#define MDB_MINKEYS 2
/** A stamp that identifies a file as an LMDB file.
* There's nothing special about this value other than that it is easily
* recognizable, and it will reflect any byte order mismatches.
*/
#define MDB_MAGIC 0xBEEFC0DE
/** The version number for a database's datafile format. */
#define MDB_DATA_VERSION ((MDB_DEVEL) ? 999 : 1)
/** The version number for a database's lockfile format. */
#define MDB_LOCK_VERSION 1
/** @brief The max size of a key we can write, or 0 for dynamic max.
*
* Define this as 0 to compute the max from the page size. 511
* is default for backwards compat: liblmdb <= 0.9.10 can break
* when modifying a DB with keys/dupsort data bigger than its max.
* #MDB_DEVEL sets the default to 0.
*
* Data items in an #MDB_DUPSORT database are also limited to
* this size, since they're actually keys of a sub-DB. Keys and
* #MDB_DUPSORT data items must fit on a node in a regular page.
*/
#ifndef MDB_MAXKEYSIZE
#define MDB_MAXKEYSIZE ((MDB_DEVEL) ? 0 : 511)
#endif
/** The maximum size of a key we can write to the environment. */
#if MDB_MAXKEYSIZE
#define ENV_MAXKEY(env) (MDB_MAXKEYSIZE)
#else
#define ENV_MAXKEY(env) ((env)->me_maxkey)
#endif
/** @brief The maximum size of a data item.
*
* We only store a 32 bit value for node sizes.
*/
#define MAXDATASIZE 0xffffffffUL
#if MDB_DEBUG
/** Key size which fits in a #DKBUF.
* @ingroup debug
*/
#define DKBUF_MAXKEYSIZE ((MDB_MAXKEYSIZE) > 0 ? (MDB_MAXKEYSIZE) : 511)
/** A key buffer.
* @ingroup debug
* This is used for printing a hex dump of a key's contents.
*/
#define DKBUF char kbuf[DKBUF_MAXKEYSIZE*2+1]
/** Display a key in hex.
* @ingroup debug
* Invoke a function to display a key in hex.
*/
#define DKEY(x) mdb_dkey(x, kbuf)
#else
#define DKBUF
#define DKEY(x) 0
#endif
/** An invalid page number.
* Mainly used to denote an empty tree.
*/
#define P_INVALID (~(pgno_t)0)
/** Test if the flags \b f are set in a flag word \b w. */
#define F_ISSET(w, f) (((w) & (f)) == (f))
/** Round \b n up to an even number. */
#define EVEN(n) (((n) + 1U) & -2) /* sign-extending -2 to match n+1U */
/** Used for offsets within a single page.
* Since memory pages are typically 4 or 8KB in size, 12-13 bits,
* this is plenty.
*/
typedef uint16_t indx_t;
/** Default size of memory map.
* This is certainly too small for any actual applications. Apps should always set
* the size explicitly using #mdb_env_set_mapsize().
*/
#define DEFAULT_MAPSIZE 1048576
/** @defgroup readers Reader Lock Table
* Readers don't acquire any locks for their data access. Instead, they
* simply record their transaction ID in the reader table. The reader
* mutex is needed just to find an empty slot in the reader table. The
* slot's address is saved in thread-specific data so that subsequent read
* transactions started by the same thread need no further locking to proceed.
*
* If #MDB_NOTLS is set, the slot address is not saved in thread-specific data.
*
* No reader table is used if the database is on a read-only filesystem, or
* if #MDB_NOLOCK is set.
*
* Since the database uses multi-version concurrency control, readers don't
* actually need any locking. This table is used to keep track of which
* readers are using data from which old transactions, so that we'll know
* when a particular old transaction is no longer in use. Old transactions
* that have discarded any data pages can then have those pages reclaimed
* for use by a later write transaction.
*
* The lock table is constructed such that reader slots are aligned with the
* processor's cache line size. Any slot is only ever used by one thread.
* This alignment guarantees that there will be no contention or cache
* thrashing as threads update their own slot info, and also eliminates
* any need for locking when accessing a slot.
*
* A writer thread will scan every slot in the table to determine the oldest
* outstanding reader transaction. Any freed pages older than this will be
* reclaimed by the writer. The writer doesn't use any locks when scanning
* this table. This means that there's no guarantee that the writer will
* see the most up-to-date reader info, but that's not required for correct
* operation - all we need is to know the upper bound on the oldest reader,
* we don't care at all about the newest reader. So the only consequence of
* reading stale information here is that old pages might hang around a
* while longer before being reclaimed. That's actually good anyway, because
* the longer we delay reclaiming old pages, the more likely it is that a
* string of contiguous pages can be found after coalescing old pages from
* many old transactions together.
* @{
*/
/** Number of slots in the reader table.
* This value was chosen somewhat arbitrarily. 126 readers plus a
* couple mutexes fit exactly into 8KB on my development machine.
* Applications should set the table size using #mdb_env_set_maxreaders().
*/
#define DEFAULT_READERS 126
/** The size of a CPU cache line in bytes. We want our lock structures
* aligned to this size to avoid false cache line sharing in the
* lock table.
* This value works for most CPUs. For Itanium this should be 128.
*/
#ifndef CACHELINE
#define CACHELINE 64
#endif
/** The information we store in a single slot of the reader table.
* In addition to a transaction ID, we also record the process and
* thread ID that owns a slot, so that we can detect stale information,
* e.g. threads or processes that went away without cleaning up.
* @note We currently don't check for stale records. We simply re-init
* the table when we know that we're the only process opening the
* lock file.
*/
typedef struct MDB_rxbody {
/** Current Transaction ID when this transaction began, or (txnid_t)-1.
* Multiple readers that start at the same time will probably have the
* same ID here. Again, it's not important to exclude them from
* anything; all we need to know is which version of the DB they
* started from so we can avoid overwriting any data used in that
* particular version.
*/
txnid_t mrb_txnid;
/** The process ID of the process owning this reader txn. */
MDB_PID_T mrb_pid;
/** The thread ID of the thread owning this txn. */
MDB_THR_T mrb_tid;
} MDB_rxbody;
/** The actual reader record, with cacheline padding. */
typedef struct MDB_reader {
union {
MDB_rxbody mrx;
/** shorthand for mrb_txnid */
#define mr_txnid mru.mrx.mrb_txnid
#define mr_pid mru.mrx.mrb_pid
#define mr_tid mru.mrx.mrb_tid
/** cache line alignment */
char pad[(sizeof(MDB_rxbody)+CACHELINE-1) & ~(CACHELINE-1)];
} mru;
} MDB_reader;
/** The header for the reader table.
* The table resides in a memory-mapped file. (This is a different file
* than is used for the main database.)
*
* For POSIX the actual mutexes reside in the shared memory of this
* mapped file. On Windows, mutexes are named objects allocated by the
* kernel; we store the mutex names in this mapped file so that other
* processes can grab them. This same approach is also used on
* MacOSX/Darwin (using named semaphores) since MacOSX doesn't support
* process-shared POSIX mutexes. For these cases where a named object
* is used, the object name is derived from a 64 bit FNV hash of the
* environment pathname. As such, naming collisions are extremely
* unlikely. If a collision occurs, the results are unpredictable.
*/
typedef struct MDB_txbody {
/** Stamp identifying this as an LMDB file. It must be set
* to #MDB_MAGIC. */
uint32_t mtb_magic;
/** Format of this lock file. Must be set to #MDB_LOCK_FORMAT. */
uint32_t mtb_format;
#if defined(_WIN32) || defined(MDB_USE_POSIX_SEM)
char mtb_rmname[MNAME_LEN];
#else
/** Mutex protecting access to this table.
* This is the reader lock that #LOCK_MUTEX_R acquires.
*/
pthread_mutex_t mtb_mutex;
#endif
/** The ID of the last transaction committed to the database.
* This is recorded here only for convenience; the value can always
* be determined by reading the main database meta pages.
*/
txnid_t mtb_txnid;
/** The number of slots that have been used in the reader table.
* This always records the maximum count, it is not decremented
* when readers release their slots.
*/
unsigned mtb_numreaders;
} MDB_txbody;
/** The actual reader table definition. */
typedef struct MDB_txninfo {
union {
MDB_txbody mtb;
#define mti_magic mt1.mtb.mtb_magic
#define mti_format mt1.mtb.mtb_format
#define mti_mutex mt1.mtb.mtb_mutex
#define mti_rmname mt1.mtb.mtb_rmname
#define mti_txnid mt1.mtb.mtb_txnid
#define mti_numreaders mt1.mtb.mtb_numreaders
char pad[(sizeof(MDB_txbody)+CACHELINE-1) & ~(CACHELINE-1)];
} mt1;
union {
#if defined(_WIN32) || defined(MDB_USE_POSIX_SEM)
char mt2_wmname[MNAME_LEN];
#define mti_wmname mt2.mt2_wmname
#else
pthread_mutex_t mt2_wmutex;
#define mti_wmutex mt2.mt2_wmutex
#endif
char pad[(MNAME_LEN+CACHELINE-1) & ~(CACHELINE-1)];
} mt2;
MDB_reader mti_readers[1];
} MDB_txninfo;
/** Lockfile format signature: version, features and field layout */
#define MDB_LOCK_FORMAT \
((uint32_t) \
((MDB_LOCK_VERSION) \
/* Flags which describe functionality */ \
+ (((MDB_PIDLOCK) != 0) << 16)))
/** @} */
/** Common header for all page types.
* Overflow records occupy a number of contiguous pages with no
* headers on any page after the first.
*/
typedef struct MDB_page {
#define mp_pgno mp_p.p_pgno
#define mp_next mp_p.p_next
union {
pgno_t p_pgno; /**< page number */
struct MDB_page *p_next; /**< for in-memory list of freed pages */
} mp_p;
uint16_t mp_pad;
/** @defgroup mdb_page Page Flags
* @ingroup internal
* Flags for the page headers.
* @{
*/
#define P_BRANCH 0x01 /**< branch page */
#define P_LEAF 0x02 /**< leaf page */
#define P_OVERFLOW 0x04 /**< overflow page */
#define P_META 0x08 /**< meta page */
#define P_DIRTY 0x10 /**< dirty page, also set for #P_SUBP pages */
#define P_LEAF2 0x20 /**< for #MDB_DUPFIXED records */
#define P_SUBP 0x40 /**< for #MDB_DUPSORT sub-pages */
#define P_LOOSE 0x4000 /**< page was dirtied then freed, can be reused */
#define P_KEEP 0x8000 /**< leave this page alone during spill */
/** @} */
uint16_t mp_flags; /**< @ref mdb_page */
#define mp_lower mp_pb.pb.pb_lower
#define mp_upper mp_pb.pb.pb_upper
#define mp_pages mp_pb.pb_pages
union {
struct {
indx_t pb_lower; /**< lower bound of free space */
indx_t pb_upper; /**< upper bound of free space */
} pb;
uint32_t pb_pages; /**< number of overflow pages */
} mp_pb;
indx_t mp_ptrs[1]; /**< dynamic size */
} MDB_page;
/** Size of the page header, excluding dynamic data at the end */
#define PAGEHDRSZ ((unsigned) offsetof(MDB_page, mp_ptrs))
/** Address of first usable data byte in a page, after the header */
#define METADATA(p) ((void *)((char *)(p) + PAGEHDRSZ))
/** ITS#7713, change PAGEBASE to handle 65536 byte pages */
#define PAGEBASE ((MDB_DEVEL) ? PAGEHDRSZ : 0)
/** Number of nodes on a page */
#define NUMKEYS(p) (((p)->mp_lower - (PAGEHDRSZ-PAGEBASE)) >> 1)
/** The amount of space remaining in the page */
#define SIZELEFT(p) (indx_t)((p)->mp_upper - (p)->mp_lower)
/** The percentage of space used in the page, in tenths of a percent. */
#define PAGEFILL(env, p) (1000L * ((env)->me_psize - PAGEHDRSZ - SIZELEFT(p)) / \
((env)->me_psize - PAGEHDRSZ))
/** The minimum page fill factor, in tenths of a percent.
* Pages emptier than this are candidates for merging.
*/
#define FILL_THRESHOLD 250
/** Test if a page is a leaf page */
#define IS_LEAF(p) F_ISSET((p)->mp_flags, P_LEAF)
/** Test if a page is a LEAF2 page */
#define IS_LEAF2(p) F_ISSET((p)->mp_flags, P_LEAF2)
/** Test if a page is a branch page */
#define IS_BRANCH(p) F_ISSET((p)->mp_flags, P_BRANCH)
/** Test if a page is an overflow page */
#define IS_OVERFLOW(p) F_ISSET((p)->mp_flags, P_OVERFLOW)
/** Test if a page is a sub page */
#define IS_SUBP(p) F_ISSET((p)->mp_flags, P_SUBP)
/** The number of overflow pages needed to store the given size. */
#define OVPAGES(size, psize) ((PAGEHDRSZ-1 + (size)) / (psize) + 1)
/** Link in #MDB_txn.%mt_loose_pgs list */
#define NEXT_LOOSE_PAGE(p) (*(MDB_page **)((p) + 2))
/** Header for a single key/data pair within a page.
* Used in pages of type #P_BRANCH and #P_LEAF without #P_LEAF2.
* We guarantee 2-byte alignment for 'MDB_node's.
*/
typedef struct MDB_node {
/** lo and hi are used for data size on leaf nodes and for
* child pgno on branch nodes. On 64 bit platforms, flags
* is also used for pgno. (Branch nodes have no flags).
* They are in host byte order in case that lets some
* accesses be optimized into a 32-bit word access.
*/
#if BYTE_ORDER == LITTLE_ENDIAN
unsigned short mn_lo, mn_hi; /**< part of data size or pgno */
#else
unsigned short mn_hi, mn_lo;
#endif
/** @defgroup mdb_node Node Flags
* @ingroup internal
* Flags for node headers.
* @{
*/
#define F_BIGDATA 0x01 /**< data put on overflow page */
#define F_SUBDATA 0x02 /**< data is a sub-database */
#define F_DUPDATA 0x04 /**< data has duplicates */
/** valid flags for #mdb_node_add() */
#define NODE_ADD_FLAGS (F_DUPDATA|F_SUBDATA|MDB_RESERVE|MDB_APPEND)
/** @} */
unsigned short mn_flags; /**< @ref mdb_node */
unsigned short mn_ksize; /**< key size */
char mn_data[1]; /**< key and data are appended here */
} MDB_node;
/** Size of the node header, excluding dynamic data at the end */
#define NODESIZE offsetof(MDB_node, mn_data)
/** Bit position of top word in page number, for shifting mn_flags */
#define PGNO_TOPWORD ((pgno_t)-1 > 0xffffffffu ? 32 : 0)
/** Size of a node in a branch page with a given key.
* This is just the node header plus the key, there is no data.
*/
#define INDXSIZE(k) (NODESIZE + ((k) == NULL ? 0 : (k)->mv_size))
/** Size of a node in a leaf page with a given key and data.
* This is node header plus key plus data size.
*/
#define LEAFSIZE(k, d) (NODESIZE + (k)->mv_size + (d)->mv_size)
/** Address of node \b i in page \b p */
#define NODEPTR(p, i) ((MDB_node *)((char *)(p) + (p)->mp_ptrs[i] + PAGEBASE))
/** Address of the key for the node */
#define NODEKEY(node) (void *)((node)->mn_data)
/** Address of the data for a node */
#define NODEDATA(node) (void *)((char *)(node)->mn_data + (node)->mn_ksize)
/** Get the page number pointed to by a branch node */
#define NODEPGNO(node) \
((node)->mn_lo | ((pgno_t) (node)->mn_hi << 16) | \
(PGNO_TOPWORD ? ((pgno_t) (node)->mn_flags << PGNO_TOPWORD) : 0))
/** Set the page number in a branch node */
#define SETPGNO(node,pgno) do { \
(node)->mn_lo = (pgno) & 0xffff; (node)->mn_hi = (pgno) >> 16; \
if (PGNO_TOPWORD) (node)->mn_flags = (pgno) >> PGNO_TOPWORD; } while(0)
/** Get the size of the data in a leaf node */
#define NODEDSZ(node) ((node)->mn_lo | ((unsigned)(node)->mn_hi << 16))
/** Set the size of the data for a leaf node */
#define SETDSZ(node,size) do { \
(node)->mn_lo = (size) & 0xffff; (node)->mn_hi = (size) >> 16;} while(0)
/** The size of a key in a node */
#define NODEKSZ(node) ((node)->mn_ksize)
/** Copy a page number from src to dst */
#ifdef MISALIGNED_OK
#define COPY_PGNO(dst,src) dst = src
#else
#if SIZE_MAX > 4294967295UL
#define COPY_PGNO(dst,src) do { \
unsigned short *s, *d; \
s = (unsigned short *)&(src); \
d = (unsigned short *)&(dst); \
*d++ = *s++; \
*d++ = *s++; \
*d++ = *s++; \
*d = *s; \
} while (0)
#else
#define COPY_PGNO(dst,src) do { \
unsigned short *s, *d; \
s = (unsigned short *)&(src); \
d = (unsigned short *)&(dst); \
*d++ = *s++; \
*d = *s; \
} while (0)
#endif
#endif
/** The address of a key in a LEAF2 page.
* LEAF2 pages are used for #MDB_DUPFIXED sorted-duplicate sub-DBs.
* There are no node headers, keys are stored contiguously.
*/
#define LEAF2KEY(p, i, ks) ((char *)(p) + PAGEHDRSZ + ((i)*(ks)))
/** Set the \b node's key into \b keyptr, if requested. */
#define MDB_GET_KEY(node, keyptr) { if ((keyptr) != NULL) { \
(keyptr)->mv_size = NODEKSZ(node); (keyptr)->mv_data = NODEKEY(node); } }
/** Set the \b node's key into \b key. */
#define MDB_GET_KEY2(node, key) { key.mv_size = NODEKSZ(node); key.mv_data = NODEKEY(node); }
/** Information about a single database in the environment. */
typedef struct MDB_db {
uint32_t md_pad; /**< also ksize for LEAF2 pages */
uint16_t md_flags; /**< @ref mdb_dbi_open */
uint16_t md_depth; /**< depth of this tree */
pgno_t md_branch_pages; /**< number of internal pages */
pgno_t md_leaf_pages; /**< number of leaf pages */
pgno_t md_overflow_pages; /**< number of overflow pages */
size_t md_entries; /**< number of data items */
pgno_t md_root; /**< the root page of this tree */
} MDB_db;
/** mdb_dbi_open flags */
#define MDB_VALID 0x8000 /**< DB handle is valid, for me_dbflags */
#define PERSISTENT_FLAGS (0xffff & ~(MDB_VALID))
#define VALID_FLAGS (MDB_REVERSEKEY|MDB_DUPSORT|MDB_INTEGERKEY|MDB_DUPFIXED|\
MDB_INTEGERDUP|MDB_REVERSEDUP|MDB_CREATE)
/** Handle for the DB used to track free pages. */
#define FREE_DBI 0
/** Handle for the default DB. */
#define MAIN_DBI 1
/** Meta page content.
* A meta page is the start point for accessing a database snapshot.
* Pages 0-1 are meta pages. Transaction N writes meta page #(N % 2).
*/
typedef struct MDB_meta {
/** Stamp identifying this as an LMDB file. It must be set
* to #MDB_MAGIC. */
uint32_t mm_magic;
/** Version number of this lock file. Must be set to #MDB_DATA_VERSION. */
uint32_t mm_version;
void *mm_address; /**< address for fixed mapping */
size_t mm_mapsize; /**< size of mmap region */
MDB_db mm_dbs[2]; /**< first is free space, 2nd is main db */
/** The size of pages used in this DB */
#define mm_psize mm_dbs[0].md_pad
/** Any persistent environment flags. @ref mdb_env */
#define mm_flags mm_dbs[0].md_flags
pgno_t mm_last_pg; /**< last used page in file */
txnid_t mm_txnid; /**< txnid that committed this page */
} MDB_meta;
/** Buffer for a stack-allocated meta page.
* The members define size and alignment, and silence type
* aliasing warnings. They are not used directly; that could
* mean incorrectly using several union members in parallel.
*/
typedef union MDB_metabuf {
MDB_page mb_page;
struct {
char mm_pad[PAGEHDRSZ];
MDB_meta mm_meta;
} mb_metabuf;
} MDB_metabuf;
/** Auxiliary DB info.
* The information here is mostly static/read-only. There is
* only a single copy of this record in the environment.
*/
typedef struct MDB_dbx {
MDB_val md_name; /**< name of the database */
MDB_cmp_func *md_cmp; /**< function for comparing keys */
MDB_cmp_func *md_dcmp; /**< function for comparing data items */
MDB_rel_func *md_rel; /**< user relocate function */
void *md_relctx; /**< user-provided context for md_rel */
} MDB_dbx;
/** A database transaction.
* Every operation requires a transaction handle.
*/
struct MDB_txn {
MDB_txn *mt_parent; /**< parent of a nested txn */
MDB_txn *mt_child; /**< nested txn under this txn */
pgno_t mt_next_pgno; /**< next unallocated page */
/** The ID of this transaction. IDs are integers incrementing from 1.
* Only committed write transactions increment the ID. If a transaction
* aborts, the ID may be re-used by the next writer.
*/
txnid_t mt_txnid;
MDB_env *mt_env; /**< the DB environment */
/** The list of pages that became unused during this transaction.
*/
MDB_IDL mt_free_pgs;
/** The list of loose pages that became unused and may be reused
* in this transaction, linked through #NEXT_LOOSE_PAGE(page).
*/
MDB_page *mt_loose_pgs;
/* #Number of loose pages (#mt_loose_pgs) */
int mt_loose_count;
/** The sorted list of dirty pages we temporarily wrote to disk
* because the dirty list was full. page numbers in here are
* shifted left by 1, deleted slots have the LSB set.
*/
MDB_IDL mt_spill_pgs;
union {
/** For write txns: Modified pages. Sorted when not MDB_WRITEMAP. */
MDB_ID2L dirty_list;
/** For read txns: This thread/txn's reader table slot, or NULL. */
MDB_reader *reader;
} mt_u;
/** Array of records for each DB known in the environment. */
MDB_dbx *mt_dbxs;
/** Array of MDB_db records for each known DB */
MDB_db *mt_dbs;
/** Array of sequence numbers for each DB handle */
unsigned int *mt_dbiseqs;
/** @defgroup mt_dbflag Transaction DB Flags
* @ingroup internal
* @{
*/
#define DB_DIRTY 0x01 /**< DB was modified or is DUPSORT data */
#define DB_STALE 0x02 /**< Named-DB record is older than txnID */
#define DB_NEW 0x04 /**< Named-DB handle opened in this txn */
#define DB_VALID 0x08 /**< DB handle is valid, see also #MDB_VALID */
/** @} */
/** In write txns, array of cursors for each DB */
MDB_cursor **mt_cursors;
/** Array of flags for each DB */
unsigned char *mt_dbflags;
/** Number of DB records in use. This number only ever increments;
* we don't decrement it when individual DB handles are closed.
*/
MDB_dbi mt_numdbs;
/** @defgroup mdb_txn Transaction Flags
* @ingroup internal
* @{
*/
#define MDB_TXN_RDONLY 0x01 /**< read-only transaction */