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
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667/*-------------------------------------------------------------------------
*
* cdbappendonlystoragewrite.c
*
* Portions Copyright (c) 2007-2009, Greenplum inc
* Portions Copyright (c) 2012-Present VMware, Inc. or its affiliates.
*
*
* IDENTIFICATION
* src/backend/cdb/cdbappendonlystoragewrite.c
*
*-------------------------------------------------------------------------
*/
#include "postgres.h"
#ifndef WIN32
#include <sys/fcntl.h>
#else
#include <io.h>
#endif
#include <sys/file.h>
#include <unistd.h>
#include "catalog/catalog.h"
#include "catalog/heap.h"
#include "catalog/pg_compression.h"
#include "cdb/cdbappendonlystorageread.h"
#include "cdb/cdbappendonlystorage.h"
#include "cdb/cdbappendonlystoragelayer.h"
#include "cdb/cdbappendonlystorageformat.h"
#include "cdb/cdbappendonlystoragewrite.h"
#include "cdb/cdbappendonlyxlog.h"
#include "common/relpath.h"
#include "pgstat.h"
#include "storage/gp_compress.h"
#include "utils/faultinjector.h"
#include "utils/guc.h"
/*----------------------------------------------------------------
* Initialization
*----------------------------------------------------------------
*/
/*
* Initialize AppendOnlyStorageWrite.
*
* The AppendOnlyStorageWrite data structure is initialized once for an
* append "session" and can be used to add Append-Only Storage Blocks to 1
* or more segment files.
*
* The current file to write to is opened with the
* AppendOnlyStorageWrite_OpenFile routine.
*
* storageWrite - data structure to initialize
* memoryContext - memory context to use for buffers and other memory
* needs. When NULL, the current memory context is used.
* maxBufferLen - maximum Append-Only Storage Block length including all
* storage headers.
* relationName - name of the relation to use in log and error messages.
* title - A phrase that better describes the purpose of this open.
* The caller manages the storage for this.
* storageAttributes - Append-Only Storage Attributes from relation creation.
*/
void
AppendOnlyStorageWrite_Init(AppendOnlyStorageWrite *storageWrite,
MemoryContext memoryContext,
int32 maxBufferLen,
char *relationName,
Oid reloid,
char *title,
AppendOnlyStorageAttributes *storageAttributes,
bool needsWAL,
const struct f_smgr_ao *smgrAO)
{
uint8 *memory;
int32 memoryLen;
MemoryContext oldMemoryContext;
Assert(storageWrite != NULL);
/* UNDONE: Range check maxBufferLen */
Assert(relationName != NULL);
Assert(storageAttributes != NULL);
Assert(smgrAO != NULL);
/* UNDONE: Range check fields in storageAttributes */
MemSet(storageWrite, 0, sizeof(AppendOnlyStorageWrite));
storageWrite->smgrAO = smgrAO;
storageWrite->maxBufferLen = maxBufferLen;
if (memoryContext == NULL)
storageWrite->memoryContext = CurrentMemoryContext;
else
storageWrite->memoryContext = memoryContext;
oldMemoryContext = MemoryContextSwitchTo(storageWrite->memoryContext);
memcpy(
&storageWrite->storageAttributes,
storageAttributes,
sizeof(AppendOnlyStorageAttributes));
/*
* Determine the fixed header length based on the checksum flag.
*/
storageWrite->regularHeaderLen = AoHeader_RegularSize;
if (storageWrite->storageAttributes.checksum)
storageWrite->regularHeaderLen += 2 * sizeof(pg_crc32);
storageWrite->relationName = pstrdup(relationName);
storageWrite->reloid = reloid;
storageWrite->title = title;
/*
* Set up extra buffers for compression.
*/
storageWrite->compressionOverrunLen = storageAttributes->overflowSize;
elog(DEBUG2, "Requested compression overflow bytes = %d.", storageAttributes->overflowSize);
if (storageWrite->storageAttributes.compress)
{
storageWrite->uncompressedBuffer = (uint8 *) palloc0(storageWrite->maxBufferLen * sizeof(uint8));
}
else
{
Assert(storageWrite->uncompressedBuffer == NULL);
}
/*
* Now that we have determined the compression overrun, we can initialize
* BufferedAppend with the correct maxBufferLen + compressionOverrunLen.
*/
storageWrite->maxBufferWithCompressionOverrrunLen =
storageWrite->maxBufferLen + storageWrite->compressionOverrunLen;
storageWrite->maxLargeWriteLen = 2 * storageWrite->maxBufferLen;
Assert(storageWrite->maxBufferWithCompressionOverrrunLen <= storageWrite->maxLargeWriteLen);
memoryLen = BufferedAppendMemoryLen
(
storageWrite->maxBufferWithCompressionOverrrunLen, /* maxBufferLen */
storageWrite->maxLargeWriteLen);
memory = (uint8 *) palloc(memoryLen);
BufferedAppendInit(&storageWrite->bufferedAppend,
memory,
memoryLen,
storageWrite->maxBufferWithCompressionOverrrunLen,
storageWrite->maxLargeWriteLen,
relationName,
smgrAO);
elogif(Debug_appendonly_print_insert || Debug_appendonly_print_append_block, LOG,
"Append-Only Storage Write initialize for table '%s' (compression = %s, compression level %d, maximum buffer length %d, large write length %d)",
storageWrite->relationName,
(storageWrite->storageAttributes.compress ? "true" : "false"),
storageWrite->storageAttributes.compressLevel,
storageWrite->maxBufferWithCompressionOverrrunLen,
storageWrite->maxLargeWriteLen);
/*
* When doing VerifyBlock, allocate the extra buffers.
*/
if (gp_appendonly_verify_write_block && storageWrite->storageAttributes.compress)
{
storageWrite->verifyWriteBuffer = (uint8 *) palloc(storageWrite->maxBufferLen * sizeof(uint8));
}
else
{
Assert(storageWrite->verifyWriteBuffer == NULL);
}
storageWrite->file = -1;
storageWrite->formatVersion = -1;
storageWrite->needsWAL = needsWAL;
MemoryContextSwitchTo(oldMemoryContext);
storageWrite->isActive = true;
}
/*
* Finish using the AppendOnlyStorageWrite session created with ~Init.
*/
void
AppendOnlyStorageWrite_FinishSession(AppendOnlyStorageWrite *storageWrite)
{
MemoryContext oldMemoryContext;
if (!storageWrite->isActive)
return;
oldMemoryContext = MemoryContextSwitchTo(storageWrite->memoryContext);
/*
* UNDONE: This expects the MemoryContent to be what was used for the
* 'memory' in ~Init
*/
BufferedAppendFinish(&storageWrite->bufferedAppend);
if (storageWrite->relationName != NULL)
{
pfree(storageWrite->relationName);
storageWrite->relationName = NULL;
}
if (storageWrite->uncompressedBuffer != NULL)
{
pfree(storageWrite->uncompressedBuffer);
storageWrite->uncompressedBuffer = NULL;
}
if (storageWrite->verifyWriteBuffer != NULL)
{
pfree(storageWrite->verifyWriteBuffer);
storageWrite->verifyWriteBuffer = NULL;
}
if (storageWrite->segmentFileName != NULL)
{
pfree(storageWrite->segmentFileName);
storageWrite->segmentFileName = NULL;
}
if (storageWrite->compression_functions != NULL)
{
callCompressionDestructor(storageWrite->compression_functions[COMPRESSION_DESTRUCTOR], storageWrite->compressionState);
if (storageWrite->compressionState != NULL)
{
pfree(storageWrite->compressionState);
storageWrite->compressionState = NULL;
}
if (storageWrite->verifyWriteCompressionState != NULL)
{
callCompressionDestructor(storageWrite->compression_functions[COMPRESSION_DESTRUCTOR], storageWrite->verifyWriteCompressionState);
storageWrite->verifyWriteCompressionState = NULL;
}
}
/* Deallocation is done. Go back to caller memory-context. */
MemoryContextSwitchTo(oldMemoryContext);
storageWrite->isActive = false;
}
/*----------------------------------------------------------------
* Open and FlushAndClose
*----------------------------------------------------------------
*/
/*
* Creates an on-demand Append-Only segment file under transaction.
*/
void
AppendOnlyStorageWrite_TransactionCreateFile(AppendOnlyStorageWrite *storageWrite,
RelFileNodeBackend *relFileNode,
int32 segmentFileNum)
{
Assert(segmentFileNum > 0);
/* The file might already exist. that's OK */
// WALREP_FIXME: Pass isRedo == true, so that you don't get an error if it
// exists already. That's currently OK, but in the future, other things
// might depend on the isRedo flag, like whether to WAL-log the creation.
smgrcreate_ao(storageWrite->smgrAO, *relFileNode, segmentFileNum, true);
/*
* Create a WAL record, so that the segfile is also created after crash or
* in possible standby server. Not strictly necessarily, because a 0-length
* segfile and a non-existent segfile are treated the same. But the
* gp_replica_check tool, to compare primary and mirror, will complain if
* a file exists in master but not in mirror, even if it's empty.
*/
if (storageWrite->needsWAL)
xlog_ao_insert(relFileNode->node, segmentFileNum, 0, NULL, 0);
}
/*
* Opens the next segment file to write. The file must already exist.
*
* This routine is responsible for seeking to the proper write location given
* the logical EOF.
*
* filePathName - name of the segment file to open.
* version - AO table format version the file is in.
* logicalEof - last committed write transaction's EOF value to use as
* the end of the segment file.
*/
void
AppendOnlyStorageWrite_OpenFile(AppendOnlyStorageWrite *storageWrite,
char *filePathName,
int version,
int64 logicalEof,
int64 fileLen_uncompressed,
RelFileNodeBackend *relFileNode,
int32 segmentFileNum)
{
File file;
MemoryContext oldMemoryContext;
Assert(storageWrite != NULL);
Assert(storageWrite->isActive);
Assert(filePathName != NULL);
/*
* Assume that we only write in the current latest format. (it's redundant
* to pass the version number as argument, currently)
*/
if (version != AOSegfileFormatVersion_GetLatest())
ereport(ERROR,
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
errmsg("cannot write append-only table version %d", version)));
/*
* Open or create the file for write.
*/
char *path = aorelpath(*relFileNode, segmentFileNum);
errno = 0;
int fileFlags = O_RDWR | PG_BINARY;
file = storageWrite->smgrAO->smgr_AORelOpenSegFile(storageWrite->reloid, path, fileFlags);
if (file < 0)
ereport(ERROR,
(errcode_for_file_access(),
errmsg("Append-only Storage Write could not open segment file %s \"%s\" for relation \"%s\": %m",
path, filePathName,
storageWrite->relationName)));
/*
* Seek to the logical EOF write position.
*/
storageWrite->file = file;
storageWrite->formatVersion = version;
storageWrite->startEof = logicalEof;
storageWrite->relFileNode = *relFileNode;
storageWrite->segmentFileNum = segmentFileNum;
/*
* When writing multiple segment files, we throw away the old segment file
* name strings.
*/
oldMemoryContext = MemoryContextSwitchTo(storageWrite->memoryContext);
if (storageWrite->segmentFileName != NULL)
pfree(storageWrite->segmentFileName);
storageWrite->segmentFileName = pstrdup(filePathName);
/* Allocation is done. Go back to caller memory-context. */
MemoryContextSwitchTo(oldMemoryContext);
/*
* Tell the BufferedAppend module about the file we just opened.
*/
BufferedAppendSetFile(&storageWrite->bufferedAppend,
storageWrite->file,
storageWrite->relFileNode,
storageWrite->segmentFileNum,
storageWrite->segmentFileName,
logicalEof,
fileLen_uncompressed);
}
/*
* Flush and close the current segment file.
*
* No error if the current is already closed.
*
* The new EOF of the segment file is returend in *newLogicalEof.
*/
void
AppendOnlyStorageWrite_FlushAndCloseFile(
AppendOnlyStorageWrite *storageWrite,
int64 *newLogicalEof,
int64 *fileLen_uncompressed)
{
Assert(storageWrite != NULL);
Assert(storageWrite->isActive);
if (storageWrite->file == -1)
{
*newLogicalEof = 0;
*fileLen_uncompressed = 0;
return;
}
/*
* Have the BufferedAppend module let go, but this does not close the
* file.
*/
BufferedAppendCompleteFile(&storageWrite->bufferedAppend,
newLogicalEof,
fileLen_uncompressed,
storageWrite->needsWAL);
/*
* We must take care of fsynching to disk ourselves since a fsync request
* is not enqueued for an AO segment file that is written to disk on
* primary. Temp tables are not crash safe, no need to fsync them.
*/
if (!RelFileNodeBackendIsTemp(storageWrite->relFileNode) &&
storageWrite->smgrAO->smgr_FileSync(storageWrite->file, WAIT_EVENT_DATA_FILE_SYNC) != 0)
ereport(ERROR,
(errcode_for_file_access(),
errmsg("Could not flush (fsync) Append-Only segment file '%s' to disk for relation '%s': %m",
storageWrite->segmentFileName,
storageWrite->relationName)));
storageWrite->smgrAO->smgr_FileClose(storageWrite->file);
storageWrite->file = -1;
storageWrite->formatVersion = -1;
MemSet(&storageWrite->relFileNode, 0, sizeof(RelFileNode));
storageWrite->segmentFileNum = 0;
}
/*
* Flush and close the current segment file under a transaction.
*
* Handles mirror loss end transaction work.
*
* No error if the current is already closed.
*
* The new EOF of the segment file is returned in *newLogicalEof.
*/
void
AppendOnlyStorageWrite_TransactionFlushAndCloseFile(AppendOnlyStorageWrite *storageWrite,
int64 *newLogicalEof,
int64 *fileLen_uncompressed)
{
Assert(storageWrite != NULL);
Assert(storageWrite->isActive);
if (storageWrite->file == -1)
{
*newLogicalEof = 0;
*fileLen_uncompressed = 0;
return;
}
AppendOnlyStorageWrite_FlushAndCloseFile(storageWrite,
newLogicalEof,
fileLen_uncompressed);
}
/*----------------------------------------------------------------
* Usable Block Length
*----------------------------------------------------------------
*/
/*
* When writing "short" content intended to stay within the maxBufferLen (also
* known as blocksize), some of the buffer will be used for the Append-Only
* Block Header. This function returns that overhead length.
*
* Isn't the length of the Append-Only Storage Block constant? NO.
*
* Currently, there are two things that can make it longer. When checksums
* are configured, we add checksum data to the header. And there is optional
* header data (e.g. firstRowNum).
*
* We call the header portion with the optional checksums the fixed header
* because we need to be able to read and evaluate the checksums before we can
* interpret flags in the fixed header that indicate there is more header
* information.
*
* The complete header length is the fixed header plus optional information.
*/
/*
* Returns the Append-Only Storage Block complete header length in bytes.
*
* Call this routine after specifying all optional header information for the current block
* begin written.
*/
int32
AppendOnlyStorageWrite_CompleteHeaderLen(AppendOnlyStorageWrite *storageWrite,
AoHeaderKind aoHeaderKind)
{
int32 completeHeaderLen;
Assert(storageWrite != NULL);
Assert(storageWrite->isActive);
completeHeaderLen = storageWrite->regularHeaderLen;
if (AoHeader_IsLong(aoHeaderKind))
{
completeHeaderLen += (AoHeader_LongSize - AoHeader_RegularSize);
}
if (storageWrite->isFirstRowNumSet)
completeHeaderLen += sizeof(int64);
return completeHeaderLen;
}
/*
* Returns the Append-Only Storage large content metadata header length in
* bytes.
*
* Call this routine after specifying all optional header information for the
* current block begin written.
*/
static int32
AppendOnlyStorageWrite_LargeContentHeaderLen(AppendOnlyStorageWrite *storageWrite)
{
int32 completeHeaderLen;
Assert(storageWrite != NULL);
Assert(storageWrite->isActive);
completeHeaderLen = storageWrite->regularHeaderLen;
if (storageWrite->isFirstRowNumSet)
completeHeaderLen += sizeof(int64);
/* UNDONE: Right alignment? */
return completeHeaderLen;
}
char *
AppendOnlyStorageWrite_ContextStr(AppendOnlyStorageWrite *storageWrite)
{
int64 headerOffsetInFile =
BufferedAppendCurrentBufferPosition(&storageWrite->bufferedAppend);
return psprintf("%s. Append-Only segment file '%s', header offset in file " INT64_FORMAT,
storageWrite->title,
storageWrite->segmentFileName,
headerOffsetInFile);
}
static char *
AppendOnlyStorageWrite_BlockHeaderStr(AppendOnlyStorageWrite *storageWrite,
uint8 *header)
{
return AppendOnlyStorageFormat_BlockHeaderStr(header,
storageWrite->storageAttributes.checksum,
storageWrite->formatVersion);
}
static void
AppendOnlyStorageWrite_LogBlockHeader(AppendOnlyStorageWrite *storageWrite,
int64 headerOffsetInFile,
uint8 *header)
{
char *blockHeaderStr;
blockHeaderStr = AppendOnlyStorageWrite_BlockHeaderStr(storageWrite,
header);
ereport(LOG,
(errmsg("%s. Append-Only segment file '%s', header offset in file " INT64_FORMAT ". %s",
storageWrite->title,
storageWrite->segmentFileName,
headerOffsetInFile,
blockHeaderStr)));
pfree(blockHeaderStr);
}
/*----------------------------------------------------------------
* errcontext and errdetail
*----------------------------------------------------------------
*/
/*
* errcontext_appendonly_write_storage_block
*
* Add an errcontext() line showing the table, segment file, offset in file,
* block count of the storage block being read.
*/
static int
errcontext_appendonly_write_storage_block(AppendOnlyStorageWrite *storageWrite)
{
int64 headerOffsetInFile;
headerOffsetInFile = BufferedAppendCurrentBufferPosition(&storageWrite->bufferedAppend);
errcontext("Append-Only table '%s', segment file '%s', block header offset in file = " INT64_FORMAT ", bufferCount " INT64_FORMAT ")",
storageWrite->relationName,
storageWrite->segmentFileName,
headerOffsetInFile,
storageWrite->bufferCount);
return 0;
}
/*
* errdetail_appendonly_write_storage_block_header
*
* Add an errdetail() line showing the Append-Only Storage header being written.
*/
static void
errdetail_appendonly_write_storage_block_header(AppendOnlyStorageWrite *storageWrite)
{
uint8 *header;
bool checksum;
int version;
header = BufferedAppendGetCurrentBuffer(&storageWrite->bufferedAppend);
checksum = storageWrite->storageAttributes.checksum;
version = storageWrite->formatVersion;
errdetail_appendonly_storage_smallcontent_header(header, checksum,
version);
}
/*----------------------------------------------------------------
* Writing Small Content Efficiently that is not being Bulk Compressed
*----------------------------------------------------------------
*/
/*
* This section describes for writing content that is less than or equal to
* the blocksize (e.g. 32k) bytes that is not being bulk compressed by the
* Append-Only Storage Layer.
*
* Actually, the content is limited to blocksize minus the Append-Only header
* size (see the AppendOnlyStorageWrite_HeaderLen routine).
*/
/*
* Get a pointer to next maximum length buffer space for
* appending small content.
*
* You must decide whether you are supplying the optional first row number
* BEFORE calling this routine!
*
* NOTE: The maximum length buffer space =
* maxBufferLen +
* AppendOnlyStorageWrite_CompleteHeaderLen(...)
*
* When compression is not being used, this interface provides a pointer
* directly into the write buffer for efficient data generation. Otherwise,
* a pointer to a temporary buffer to collect the uncompressed contents will
* be provided.
*
* Returns NULL when the current file does not have enough room for another
* buffer.
*/
uint8 *
AppendOnlyStorageWrite_GetBuffer(AppendOnlyStorageWrite *storageWrite,
int aoHeaderKind)
{
Assert(storageWrite != NULL);
Assert(storageWrite->isActive);
Assert(aoHeaderKind == AoHeaderKind_SmallContent ||
aoHeaderKind == AoHeaderKind_NonBulkDenseContent ||
aoHeaderKind == AoHeaderKind_BulkDenseContent);
storageWrite->getBufferAoHeaderKind = aoHeaderKind;
/*
* Both headers (Small and NonBulkDense) have the same length. BulkDense
* is a long header.
*/
storageWrite->currentCompleteHeaderLen =
AppendOnlyStorageWrite_CompleteHeaderLen(
storageWrite,
aoHeaderKind);
/*
* If compression configured, the supply the temporary buffer instead.
*/
if (storageWrite->storageAttributes.compress)
{
storageWrite->currentBuffer = NULL;
return storageWrite->uncompressedBuffer;
}
else
{
storageWrite->currentBuffer =
BufferedAppendGetMaxBuffer(&storageWrite->bufferedAppend);
return &storageWrite->currentBuffer[storageWrite->currentCompleteHeaderLen];
}
}
/*
* Test if a buffer is currently allocated.
*/
bool
AppendOnlyStorageWrite_IsBufferAllocated(AppendOnlyStorageWrite *storageWrite)
{
Assert(storageWrite != NULL);
Assert(storageWrite->isActive);
return (storageWrite->currentCompleteHeaderLen > 0);
}
/*
* Return the beginning of the last write position of the write buffer.
*/
int64
AppendOnlyStorageWrite_LogicalBlockStartOffset(AppendOnlyStorageWrite *storageWrite)
{
return storageWrite->logicalBlockStartOffset;
}
static void
AppendOnlyStorageWrite_VerifyWriteBlock(AppendOnlyStorageWrite *storageWrite,
int64 headerOffsetInFile,
int32 bufferLen,
uint8 *expectedContent,
int32 expectedUncompressedLen,
int expectedExecutorBlockKind,
int expectedRowCount,
int32 expectedCompressedLen)
{
uint8 *header;
AOHeaderCheckError checkError;
AoHeaderKind headerKind;
int32 actualHeaderLen;
int32 offset;
int32 uncompressedLen = 0;
bool isCompressed = false;
int32 overallBlockLen;
int32 compressedLen = 0;
int executorBlockKind = 0;
bool hasFirstRowNum;
int64 firstRowNum;
int rowCount = 0;
pg_crc32 storedChecksum;
pg_crc32 computedChecksum;
if (storageWrite->storageAttributes.compress && storageWrite->verifyWriteBuffer == NULL)
return; /* GUC must have been turned on
* mid-transaction. */
if (gp_appendonly_verify_write_block == false)
elog(WARNING, "The GUC gp_appendonly_verify_write_block is false. Compressed write not checked.");
header = BufferedAppendGetCurrentBuffer(&storageWrite->bufferedAppend);
/*
* Code is similar to that in getNextStorageBlockInFile routine.
*/
/*----------
* Proceed very carefully:
* [ 1. Verify header checksum ]
* 2. Examine (basic) header.
* 3. Examine specific header.
* [ 4. Verify the block checksum ]
*----------
*/
if (storageWrite->storageAttributes.checksum)
{
if (!AppendOnlyStorageFormat_VerifyHeaderChecksum(
header,
&storedChecksum,
&computedChecksum))
ereport(ERROR,
(errmsg("Verify block during write found header checksum does not match. Expected 0x%08X and found 0x%08X",
storedChecksum,
computedChecksum),
errdetail_appendonly_write_storage_block_header(storageWrite),
errcontext_appendonly_write_storage_block(storageWrite)));
}
/*
* Check the (basic) header information.
*/
checkError = AppendOnlyStorageFormat_GetHeaderInfo(
header,
storageWrite->storageAttributes.checksum,
&headerKind,
&actualHeaderLen);
if (checkError != AOHeaderCheckOk)
ereport(ERROR,
(errmsg("Verify block during write found bad append-only storage header. Header check error %d, detail '%s'",
(int) checkError,
AppendOnlyStorageFormat_GetHeaderCheckErrorStr()),
errdetail_appendonly_write_storage_block_header(storageWrite),
errcontext_appendonly_write_storage_block(storageWrite)));
switch (headerKind)
{
case AoHeaderKind_SmallContent:
/*
* Check the Block header information.
*/
checkError = AppendOnlyStorageFormat_GetSmallContentHeaderInfo(header,
actualHeaderLen,
storageWrite->storageAttributes.checksum,
bufferLen,
&overallBlockLen,
&offset, //Offset to data.
& uncompressedLen,
&executorBlockKind,
&hasFirstRowNum,
storageWrite->formatVersion,
&firstRowNum,
&rowCount,
&isCompressed,
&compressedLen);
break;
case AoHeaderKind_LargeContent:
/*
* Check the LargeContent header information.
*/
checkError = AppendOnlyStorageFormat_GetLargeContentHeaderInfo(header,
actualHeaderLen,
storageWrite->storageAttributes.checksum,
&uncompressedLen,
&executorBlockKind,
&hasFirstRowNum,
&firstRowNum,
&rowCount);
break;
case AoHeaderKind_NonBulkDenseContent:
checkError = AppendOnlyStorageFormat_GetNonBulkDenseContentHeaderInfo(header,
actualHeaderLen,
storageWrite->storageAttributes.checksum,
bufferLen,
&overallBlockLen,
&offset, //Offset to data.
&uncompressedLen,
&executorBlockKind,
&hasFirstRowNum,
storageWrite->formatVersion,
&firstRowNum,
&rowCount);
break;
case AoHeaderKind_BulkDenseContent:
checkError = AppendOnlyStorageFormat_GetBulkDenseContentHeaderInfo(header,
actualHeaderLen,
storageWrite->storageAttributes.checksum,
bufferLen,
&overallBlockLen,
&offset, //Offset to data.
&uncompressedLen,
&executorBlockKind,
&hasFirstRowNum,
storageWrite->formatVersion,
&firstRowNum,
&rowCount,
&isCompressed,
&compressedLen);
break;
default:
elog(ERROR, "Unexpected Append-Only header kind %d",
headerKind);
break;
}
if (checkError != AOHeaderCheckOk)
ereport(ERROR,
(errmsg("Verify block during write found bad append-only storage block header. Block type: %d "
"Header check error %d, detail '%s'",
headerKind,
(int) checkError,
AppendOnlyStorageFormat_GetHeaderCheckErrorStr()),
errdetail_appendonly_write_storage_block_header(storageWrite),
errcontext_appendonly_write_storage_block(storageWrite)));
if (uncompressedLen != expectedUncompressedLen)
ereport(ERROR,
(errmsg("Verify block during write found append-only storage block header. "
"DataLen %d does not equal expected length %d, ",
uncompressedLen,
expectedUncompressedLen),
errdetail_appendonly_write_storage_block_header(storageWrite),
errcontext_appendonly_write_storage_block(storageWrite)));
if (compressedLen != expectedCompressedLen)
ereport(ERROR,
(errmsg("Verify block during write found append-only storage block header. "
"CompressedLen %d does not equal expected length %d",
compressedLen,
expectedCompressedLen),
errdetail_appendonly_write_storage_block_header(storageWrite),
errcontext_appendonly_write_storage_block(storageWrite)));
/*
* Now verify the executor portion of the block.
*/
if (executorBlockKind != expectedExecutorBlockKind)
ereport(ERROR,
(errmsg("Verify block during write found append-only storage block header. "
"ExecutorBlockKind %d does not equal expected value %d.",
executorBlockKind,
expectedExecutorBlockKind),
errdetail_appendonly_write_storage_block_header(storageWrite),
errcontext_appendonly_write_storage_block(storageWrite)));
/* UNDONE: Check hasFirstRowNum */
if (rowCount != expectedRowCount)
ereport(ERROR,
(errmsg("Verify block during write found append-only storage block header. "
"RowCount %d does not equal expected value %d",
rowCount,
expectedRowCount),
errdetail_appendonly_write_storage_block_header(storageWrite),
errcontext_appendonly_write_storage_block(storageWrite)));
if (Debug_appendonly_print_verify_write_block)
{
AppendOnlyStorageWrite_LogBlockHeader(storageWrite,
headerOffsetInFile,
header);
}
if (isCompressed)
{
int test;
PGFunction decompressor;
PGFunction *cfns = storageWrite->compression_functions;
Assert(gp_appendonly_verify_write_block == true);
Assert(storageWrite->verifyWriteCompressionState != NULL);
if (cfns == NULL)
ereport(ERROR,
(errcode(ERRCODE_INTERNAL_ERROR),
errmsg("decompression information missing")));
decompressor = cfns[COMPRESSION_DECOMPRESS];
gp_decompress(&header[offset], //Compressed data in block.
compressedLen,
storageWrite->verifyWriteBuffer, //Temporary buffer to hold uncompressed data.
uncompressedLen,
decompressor,
storageWrite->verifyWriteCompressionState,
storageWrite->bufferCount);
/*
* Compare.
*/
test = memcmp(expectedContent,
storageWrite->verifyWriteBuffer,
uncompressedLen);
if (test != 0)
ereport(ERROR,
(errmsg("Verify block during write found decompress did not produce the exact same bits passed to compress! "
"Memcmp result %d",
test),
errdetail_appendonly_write_storage_block_header(storageWrite),
errcontext_appendonly_write_storage_block(storageWrite)));
}
else
{
/* UNDONE: Do comparison here */
}
#ifdef NeedCallBack
if (executorBlockKind == AoExecutorBlockKind_VarBlock)
{
VarBlockCheckError varBlockCheckError;
VarBlockReader varBlockReader;
int readerItemCount;
/*
* Now use the VarBlock module to extract the items out.
*/
VarBlockReaderInit(&varBlockReader,
data,
uncompressedLen,
true,
&storageWrite->relFileNode.node);
varBlockCheckError = VarBlockIsValid(data, uncompressedLen);
if (varBlockCheckError != VarBlockCheckOk)
ereport(ERROR,
(errmsg("Verify block during write found VarBlock is not valid "
"Valid block check error %d, detail '%s'",
varBlockCheckError,
VarBlockGetCheckErrorStr()),
errdetail_appendonly_write_storage_block_header(storageWrite),
errcontext_appendonly_write_storage_block(storageWrite)));
readerItemCount = VarBlockReaderItemCount(&varBlockReader);
if (rowCount != readerItemCount)
{
ereport(ERROR,
(errmsg("Verify block during write found row count %d in append-only storage header does not match VarBlock item count %d",
rowCount,
readerItemCount),
errdetail_appendonly_write_storage_block_header(storageWrite),
errcontext_appendonly_write_storage_block(storageWrite)));
}
}
#endif
}
static void
AppendOnlyStorageWrite_CompressAppend(AppendOnlyStorageWrite *storageWrite,
uint8 *sourceData,
int32 sourceLen,
int executorBlockKind,
int itemCount,
int32 *compressedLen,
int32 *bufferLen)
{
uint8 *header;
uint8 *dataBuffer;
int32 dataBufferWithOverrrunLen;
PGFunction *cfns = storageWrite->compression_functions;
PGFunction compressor;
if (cfns == NULL)
compressor = NULL;
else
compressor = cfns[COMPRESSION_COMPRESS];
/* UNDONE: This can be a duplicate call... */
storageWrite->currentCompleteHeaderLen =
AppendOnlyStorageWrite_CompleteHeaderLen(
storageWrite,
storageWrite->getBufferAoHeaderKind);
header = BufferedAppendGetMaxBuffer(&storageWrite->bufferedAppend);
if (header == NULL)
ereport(ERROR,
(errcode(ERRCODE_INTERNAL_ERROR),
errmsg("We do not expect files to be have a maximum length"),
errcontext_appendonly_write_storage_block(storageWrite)));
dataBuffer = &header[storageWrite->currentCompleteHeaderLen];
dataBufferWithOverrrunLen =
storageWrite->maxBufferWithCompressionOverrrunLen
- storageWrite->currentCompleteHeaderLen;
/*
* Compress into the BufferedAppend buffer after the large header (and
* optional checksum, etc.
*/
gp_trycompress(sourceData,
sourceLen,
dataBuffer,
dataBufferWithOverrrunLen,
compressedLen,
compressor,
storageWrite->compressionState);
#ifdef FAULT_INJECTOR
/* Simulate that compression is not possible if the fault is set. */
if (FaultInjector_InjectFaultIfSet(
"appendonly_skip_compression",
DDLNotSpecified,
"",
storageWrite->relationName) == FaultInjectorTypeSkip)
{
*compressedLen = sourceLen + 1;
}
#endif
/*
* We always store the data compressed if the compressed length is less
* than the uncompressed length.
*
* TODO: this is a weak assumption. It doesn't account for the fact that
* it's not worth paying the CPU cost of decompression for a potentially
* trivial saving.
*
* The best solution to this seems to be to make the threshold at which we
* compress data user configurable.
*/
int32 dataLen = *compressedLen;
if (*compressedLen >= sourceLen)
{
dataLen = sourceLen;
memcpy(dataBuffer, sourceData, sourceLen);
*compressedLen = 0;
}
int32 dataRoundedUpLen =
AOStorage_RoundUp(dataLen, storageWrite->formatVersion);
AOStorage_ZeroPad(dataBuffer, dataLen, dataRoundedUpLen);
/* Make the header and compute the checksum if necessary. */
switch (storageWrite->getBufferAoHeaderKind)
{
case AoHeaderKind_SmallContent:
AppendOnlyStorageFormat_MakeSmallContentHeader
(header,
storageWrite->storageAttributes.checksum,
storageWrite->isFirstRowNumSet,
storageWrite->formatVersion,
storageWrite->firstRowNum,
executorBlockKind,
itemCount,
sourceLen,
*compressedLen);
break;
case AoHeaderKind_BulkDenseContent:
/*
* This header is used when compresslevel > 1 is specified with
* rle_type compression. The sourceData is already encoded with
* RLE. It is further compressed with bulk compression.
* Corresponding datumstream version is
* DatumStreamVersion_Dense_Enhanced.
*/
AppendOnlyStorageFormat_MakeBulkDenseContentHeader
(header,
storageWrite->storageAttributes.checksum,
storageWrite->isFirstRowNumSet,
storageWrite->formatVersion,
storageWrite->firstRowNum,
executorBlockKind,
itemCount,
sourceLen,
*compressedLen);
break;
default:
elog(ERROR, "unexpected Append-Only header kind %d",
storageWrite->getBufferAoHeaderKind);
break;
}
if (Debug_appendonly_print_storage_headers)
{
AppendOnlyStorageWrite_LogBlockHeader(
storageWrite,
BufferedAppendCurrentBufferPosition(&storageWrite->bufferedAppend),
header);
}
elogif(Debug_appendonly_print_insert, LOG,
"Append-only insert block for table '%s' stored %s "
"(segment file '%s', header offset in file " INT64_FORMAT ", "
"source length = %d, result length %d item count %d, block count "
INT64_FORMAT ")",
storageWrite->relationName,
(*compressedLen > 0) ? "compressed" : "uncompressed",
storageWrite->segmentFileName,
BufferedAppendCurrentBufferPosition(&storageWrite->bufferedAppend),
sourceLen,
dataLen,
itemCount,
storageWrite->bufferCount);
*bufferLen = storageWrite->currentCompleteHeaderLen + dataRoundedUpLen;
}
/*
* Mark the current buffer "small" buffer as finished.
*
* If compression is configured, we will try to compress the contents in
* the temporary uncompressed buffer into the write buffer.
*
* The buffer can be scheduled for writing and reused.
*
* contentLen - byte length of the content generated directly into the
* buffer returned by AppendOnlyStorageWrite_GetBuffer.
* executorBlockKind - A value defined externally by the executor that
* describes in content stored in the Append-Only Storage
* Block.
* rowCount - number of rows stored in the content.
*/
void
AppendOnlyStorageWrite_FinishBuffer(AppendOnlyStorageWrite *storageWrite,
int32 contentLen,
int executorBlockKind,
int rowCount)
{
int64 headerOffsetInFile;
int32 bufferLen;
Assert(storageWrite != NULL);
Assert(storageWrite->isActive);
Assert(storageWrite->currentCompleteHeaderLen > 0);
if (contentLen >
storageWrite->maxBufferLen - storageWrite->currentCompleteHeaderLen)
elog(ERROR,
"Append-only content too large AO storage block (table '%s', "
"content length = %d, maximum buffer length %d, complete header length %d, first row number is set %s)",
storageWrite->relationName,
contentLen,
storageWrite->maxBufferLen,
storageWrite->currentCompleteHeaderLen,
(storageWrite->isFirstRowNumSet ? "true" : "false"));
headerOffsetInFile = BufferedAppendCurrentBufferPosition(&storageWrite->bufferedAppend);
if (!storageWrite->storageAttributes.compress)
{
uint8 *nonCompressedHeader;
uint8 *nonCompressedData;
int32 dataRoundedUpLen;
int32 uncompressedlen;
nonCompressedHeader = storageWrite->currentBuffer;
nonCompressedData = &nonCompressedHeader[storageWrite->currentCompleteHeaderLen];
dataRoundedUpLen = AOStorage_RoundUp(contentLen, storageWrite->formatVersion);
AOStorage_ZeroPad(
nonCompressedData,
contentLen,
dataRoundedUpLen);
switch (storageWrite->getBufferAoHeaderKind)
{
case AoHeaderKind_SmallContent:
/*
* Make the header and compute the checksum if necessary.
*/
AppendOnlyStorageFormat_MakeSmallContentHeader
(nonCompressedHeader,
storageWrite->storageAttributes.checksum,
storageWrite->isFirstRowNumSet,
storageWrite->formatVersion,
storageWrite->firstRowNum,
executorBlockKind,
rowCount,
contentLen,
/* compressedLength */ 0);
break;
case AoHeaderKind_NonBulkDenseContent:
/*
* Make the header and compute the checksum if necessary.
*/
AppendOnlyStorageFormat_MakeNonBulkDenseContentHeader
(nonCompressedHeader,
storageWrite->storageAttributes.checksum,
storageWrite->isFirstRowNumSet,
storageWrite->formatVersion,
storageWrite->firstRowNum,
executorBlockKind,
rowCount,
contentLen);
break;
default:
elog(ERROR, "Unexpected Append-Only header kind %d",
storageWrite->getBufferAoHeaderKind);
break;
}
if (Debug_appendonly_print_storage_headers)
{
AppendOnlyStorageWrite_LogBlockHeader(storageWrite,
headerOffsetInFile,
nonCompressedHeader);
}
bufferLen = storageWrite->currentCompleteHeaderLen + dataRoundedUpLen;
uncompressedlen = bufferLen; /* since there's no compression.. */
/*
* Just before finishing the AO Storage buffer with our non-compressed
* content, let's verify it.
*/
if (gp_appendonly_verify_write_block)
AppendOnlyStorageWrite_VerifyWriteBlock(storageWrite,
headerOffsetInFile,
bufferLen,
nonCompressedData,
contentLen,
executorBlockKind,
rowCount,
/* expectedCompressedLen */ 0);
BufferedAppendFinishBuffer(&storageWrite->bufferedAppend,
bufferLen,
uncompressedlen,
storageWrite->needsWAL);
/* Declare it finished. */
storageWrite->currentCompleteHeaderLen = 0;
elogif(Debug_appendonly_print_insert, LOG,
"Append-only insert finished uncompressed block for table '%s' "
"(length = %d, executor block kind %d, item count %d, block count " INT64_FORMAT ")",
storageWrite->relationName,
contentLen,
executorBlockKind,
rowCount,
storageWrite->bufferCount);
}
else
{
int32 compressedLen = 0;
AppendOnlyStorageWrite_CompressAppend(storageWrite,
storageWrite->uncompressedBuffer,
contentLen,
executorBlockKind,
rowCount,
&compressedLen,
&bufferLen);
/*
* Just before finishing the AO Storage buffer with our non-compressed
* content, let's verify it.
*/
if (gp_appendonly_verify_write_block)
AppendOnlyStorageWrite_VerifyWriteBlock(storageWrite,
headerOffsetInFile,
bufferLen,
storageWrite->uncompressedBuffer,
contentLen,
executorBlockKind,
rowCount,
compressedLen);
/*
* Finish the current buffer by specifying the used length.
*/
BufferedAppendFinishBuffer(&storageWrite->bufferedAppend,
bufferLen,
(storageWrite->currentCompleteHeaderLen +
AOStorage_RoundUp(contentLen, storageWrite->formatVersion) /* non-compressed size */ ),
storageWrite->needsWAL);
/* Declare it finished. */
storageWrite->currentCompleteHeaderLen = 0;
}
Assert(storageWrite->currentCompleteHeaderLen == 0);
storageWrite->currentBuffer = NULL;
storageWrite->isFirstRowNumSet = false;
}
/*
* Cancel the last ~GetBuffer call.
*
* This will also turn off the firstRowNum flag.
*/
void
AppendOnlyStorageWrite_CancelLastBuffer(
AppendOnlyStorageWrite *storageWrite)
{
Assert(storageWrite != NULL);
Assert(storageWrite->isActive);
Assert(storageWrite->currentCompleteHeaderLen > 0);
if (storageWrite->currentBuffer != NULL)
{
BufferedAppendCancelLastBuffer(&storageWrite->bufferedAppend);
storageWrite->currentBuffer = NULL;
}
storageWrite->currentCompleteHeaderLen = 0;
/*
* Since we don't know if AppendOnlyStorageWrite_Content will be called
* next or the writer is doing something else, let's turn off the
* firstRowNum flag.
*/
storageWrite->isFirstRowNumSet = false;
}
/*----------------------------------------------------------------
* Writing "Large" Content
*----------------------------------------------------------------
*/
/*
* This section describes for writing long content that can be up to 1 Gb
* long and/or content that will be bulk-compressed when configured.
*/
/*
* Write content up to 1Gb.
*
* Large content will be writen in fragment blocks by the Append-Only
* Storage Layer.
*
* If compression is configured, then the content will be compressed in
* fragments.
*
* Returns NULL when the current file does not have enough room for another
* buffer.
*
* content - Content to store. All contiguous.
* contentLen - byte length of the data to store.
* executorBlockKind - a value defined externally by the executor that
* describes in content stored in the Append-Only Storage
* Block.
* rowCount - number of rows stored in the content.
*/
void
AppendOnlyStorageWrite_Content(AppendOnlyStorageWrite *storageWrite,
uint8 *content,
int32 contentLen,
int executorBlockKind,
int rowCount)
{
int32 completeHeaderLen;
int32 compressedLen;
int32 bufferLen;
uint8 *data;
Assert(storageWrite != NULL);
Assert(storageWrite->isActive);
completeHeaderLen =
AppendOnlyStorageWrite_CompleteHeaderLen(storageWrite,
AoHeaderKind_SmallContent);
if (contentLen <= storageWrite->maxBufferLen - completeHeaderLen)
{
/*
* This is "small" content.
*/
if (!storageWrite->storageAttributes.compress)
{
data = AppendOnlyStorageWrite_GetBuffer(storageWrite,
AoHeaderKind_SmallContent);
memcpy(data, content, contentLen);
storageWrite->logicalBlockStartOffset =
BufferedAppendNextBufferPosition(&(storageWrite->bufferedAppend));
AppendOnlyStorageWrite_FinishBuffer(storageWrite,
contentLen,
executorBlockKind,
rowCount);
Assert(storageWrite->currentCompleteHeaderLen == 0);
}
else
{
/*
* Since ~_GetBuffer now takes in a specification of the header
* kind, we need to set the header kind so general routines like
* ~_CompressAppend will work correctly when writing the small
* "fragments
*/
storageWrite->getBufferAoHeaderKind = AoHeaderKind_SmallContent;
AppendOnlyStorageWrite_CompressAppend(storageWrite,
content,
contentLen,
executorBlockKind,
rowCount,
&compressedLen,
&bufferLen);
/*
* Just before finishing the AO Storage buffer with our
* non-compressed content, let's verify it.
*/
if (gp_appendonly_verify_write_block)
AppendOnlyStorageWrite_VerifyWriteBlock(storageWrite,
BufferedAppendCurrentBufferPosition(&storageWrite->bufferedAppend),
bufferLen,
content,
contentLen,
executorBlockKind,
rowCount,
compressedLen);
storageWrite->logicalBlockStartOffset =
BufferedAppendNextBufferPosition(&(storageWrite->bufferedAppend));
/*
* Finish the current buffer by specifying the used length.
*/
BufferedAppendFinishBuffer(&storageWrite->bufferedAppend,
bufferLen,
(storageWrite->currentCompleteHeaderLen +
AOStorage_RoundUp(contentLen, storageWrite->formatVersion) /* non-compressed size */ ),
storageWrite->needsWAL);
/* Declare it finished. */
storageWrite->currentCompleteHeaderLen = 0;
}
}
else
{
int32 largeContentHeaderLen;
uint8 *largeContentHeader;
int32 smallContentHeaderLen;
int32 maxSmallContentLen;
int32 countdownContentLen;
uint8 *contentNext;
int32 smallContentLen;
/*
* Write the "Large" content in fragments.
*/
storageWrite->logicalBlockStartOffset =
BufferedAppendNextBufferPosition(&(storageWrite->bufferedAppend));
/*
* First is a LargeContent header-only block that has the "large"
* content length and "large" row count.
*/
largeContentHeaderLen = AppendOnlyStorageWrite_LargeContentHeaderLen(storageWrite);
largeContentHeader =
BufferedAppendGetBuffer(&storageWrite->bufferedAppend,
largeContentHeaderLen);
AppendOnlyStorageFormat_MakeLargeContentHeader(largeContentHeader,
storageWrite->storageAttributes.checksum,
storageWrite->isFirstRowNumSet,
storageWrite->formatVersion,
storageWrite->firstRowNum,
executorBlockKind,
rowCount,
contentLen);
BufferedAppendFinishBuffer(&storageWrite->bufferedAppend,
largeContentHeaderLen,
largeContentHeaderLen,
storageWrite->needsWAL);
/* Declare it finished. */
storageWrite->currentCompleteHeaderLen = 0;
/*
* Now write the fragments as type Block.
*/
storageWrite->isFirstRowNumSet = false; /* Not written with fragments. */
smallContentHeaderLen =
AppendOnlyStorageWrite_CompleteHeaderLen(storageWrite,
AoHeaderKind_SmallContent);
maxSmallContentLen = storageWrite->maxBufferLen - smallContentHeaderLen;
countdownContentLen = contentLen;
contentNext = content;
while (true)
{
if (countdownContentLen <= maxSmallContentLen)
smallContentLen = countdownContentLen;
else
smallContentLen = maxSmallContentLen;
if (!storageWrite->storageAttributes.compress)
{
data = AppendOnlyStorageWrite_GetBuffer(storageWrite,
AoHeaderKind_SmallContent);
memcpy(data, contentNext, smallContentLen);
AppendOnlyStorageWrite_FinishBuffer(storageWrite,
smallContentLen,
executorBlockKind,
/* rowCount */ 0);
}
else
{
/*
* Since ~_GetBuffer now takes in a specification of the
* header kind, we need to set the header kind so general
* routines like ~_CompressAppend will work correctly when
* writing the small "fragments
*/
storageWrite->getBufferAoHeaderKind = AoHeaderKind_SmallContent;
AppendOnlyStorageWrite_CompressAppend(storageWrite,
contentNext,
smallContentLen,
executorBlockKind,
/* rowCount */ 0,
&compressedLen,
&bufferLen);
/*
* Just before finishing the AO Storage buffer with our
* non-compressed content, let's verify it.
*/
if (gp_appendonly_verify_write_block)
AppendOnlyStorageWrite_VerifyWriteBlock(storageWrite,
BufferedAppendCurrentBufferPosition(&storageWrite->bufferedAppend),
bufferLen,
contentNext,
smallContentLen,
executorBlockKind,
/* rowCount */ 0,
compressedLen);
/*
* Finish the current buffer by specifying the used length.
*/
BufferedAppendFinishBuffer(&storageWrite->bufferedAppend,
bufferLen,
(smallContentHeaderLen +
AOStorage_RoundUp(smallContentLen, storageWrite->formatVersion) /* non-compressed size */ ),
storageWrite->needsWAL);
/* Declare it finished. */
storageWrite->currentCompleteHeaderLen = 0;
}
countdownContentLen -= smallContentLen;
if (countdownContentLen == 0)
break;
contentNext += smallContentLen;
}
}
storageWrite->isFirstRowNumSet = false;
/* Verify we have no buffer allocated. */
Assert(storageWrite->currentCompleteHeaderLen == 0);
}
/*----------------------------------------------------------------
* Optional: Set First Row Number
*----------------------------------------------------------------
*/
/*
* Normally, the first row of an Append-Only Storage Block is implicit. It
* is the last row number of the previous block + 1. However, to support
* BTree indexes that stored TIDs in shared-memory/disk before the
* transaction commits, we may need to not reuse row numbers of aborted
* transactions. So, this routine tells the Append-Only Storage Layer to
* explicitly keep the first row number. This will take up more header
* overhead, so the AppendOnlyStorageWrite_HeaderLen routine should be
* called afterwards to get the new overhead length.
*/
/*
* Set the first row value for the next Append-Only Storage Block to be
* written. Only applies to the next block.
*/
void
AppendOnlyStorageWrite_SetFirstRowNum(AppendOnlyStorageWrite *storageWrite,
int64 firstRowNum)
{
Assert(storageWrite != NULL);
Assert(storageWrite->isActive);
/* UNDONE: Range check firstRowNum */
storageWrite->isFirstRowNumSet = true;
storageWrite->firstRowNum = firstRowNum;
}