Line data Source code
1 : /* -------------------------------------------------------------------------
2 : *
3 : * decode.c
4 : * This module decodes WAL records read using xlogreader.h's APIs for the
5 : * purpose of logical decoding by passing information to the
6 : * reorderbuffer module (containing the actual changes) and to the
7 : * snapbuild module to build a fitting catalog snapshot (to be able to
8 : * properly decode the changes in the reorderbuffer).
9 : *
10 : * NOTE:
11 : * This basically tries to handle all low level xlog stuff for
12 : * reorderbuffer.c and snapbuild.c. There's some minor leakage where a
13 : * specific record's struct is used to pass data along, but those just
14 : * happen to contain the right amount of data in a convenient
15 : * format. There isn't and shouldn't be much intelligence about the
16 : * contents of records in here except turning them into a more usable
17 : * format.
18 : *
19 : * Portions Copyright (c) 1996-2025, PostgreSQL Global Development Group
20 : * Portions Copyright (c) 1994, Regents of the University of California
21 : *
22 : * IDENTIFICATION
23 : * src/backend/replication/logical/decode.c
24 : *
25 : * -------------------------------------------------------------------------
26 : */
27 : #include "postgres.h"
28 :
29 : #include "access/heapam_xlog.h"
30 : #include "access/transam.h"
31 : #include "access/xact.h"
32 : #include "access/xlog_internal.h"
33 : #include "access/xlogreader.h"
34 : #include "access/xlogrecord.h"
35 : #include "catalog/pg_control.h"
36 : #include "replication/decode.h"
37 : #include "replication/logical.h"
38 : #include "replication/message.h"
39 : #include "replication/reorderbuffer.h"
40 : #include "replication/snapbuild.h"
41 : #include "storage/standbydefs.h"
42 :
43 : /* individual record(group)'s handlers */
44 : static void DecodeInsert(LogicalDecodingContext *ctx, XLogRecordBuffer *buf);
45 : static void DecodeUpdate(LogicalDecodingContext *ctx, XLogRecordBuffer *buf);
46 : static void DecodeDelete(LogicalDecodingContext *ctx, XLogRecordBuffer *buf);
47 : static void DecodeTruncate(LogicalDecodingContext *ctx, XLogRecordBuffer *buf);
48 : static void DecodeMultiInsert(LogicalDecodingContext *ctx, XLogRecordBuffer *buf);
49 : static void DecodeSpecConfirm(LogicalDecodingContext *ctx, XLogRecordBuffer *buf);
50 :
51 : static void DecodeCommit(LogicalDecodingContext *ctx, XLogRecordBuffer *buf,
52 : xl_xact_parsed_commit *parsed, TransactionId xid,
53 : bool two_phase);
54 : static void DecodeAbort(LogicalDecodingContext *ctx, XLogRecordBuffer *buf,
55 : xl_xact_parsed_abort *parsed, TransactionId xid,
56 : bool two_phase);
57 : static void DecodePrepare(LogicalDecodingContext *ctx, XLogRecordBuffer *buf,
58 : xl_xact_parsed_prepare *parsed);
59 :
60 :
61 : /* common function to decode tuples */
62 : static void DecodeXLogTuple(char *data, Size len, HeapTuple tuple);
63 :
64 : /* helper functions for decoding transactions */
65 : static inline bool FilterPrepare(LogicalDecodingContext *ctx,
66 : TransactionId xid, const char *gid);
67 : static bool DecodeTXNNeedSkip(LogicalDecodingContext *ctx,
68 : XLogRecordBuffer *buf, Oid txn_dbid,
69 : RepOriginId origin_id);
70 :
71 : /*
72 : * Take every XLogReadRecord()ed record and perform the actions required to
73 : * decode it using the output plugin already setup in the logical decoding
74 : * context.
75 : *
76 : * NB: Note that every record's xid needs to be processed by reorderbuffer
77 : * (xids contained in the content of records are not relevant for this rule).
78 : * That means that for records which'd otherwise not go through the
79 : * reorderbuffer ReorderBufferProcessXid() has to be called. We don't want to
80 : * call ReorderBufferProcessXid for each record type by default, because
81 : * e.g. empty xacts can be handled more efficiently if there's no previous
82 : * state for them.
83 : *
84 : * We also support the ability to fast forward thru records, skipping some
85 : * record types completely - see individual record types for details.
86 : */
87 : void
88 4989586 : LogicalDecodingProcessRecord(LogicalDecodingContext *ctx, XLogReaderState *record)
89 : {
90 : XLogRecordBuffer buf;
91 : TransactionId txid;
92 : RmgrData rmgr;
93 :
94 4989586 : buf.origptr = ctx->reader->ReadRecPtr;
95 4989586 : buf.endptr = ctx->reader->EndRecPtr;
96 4989586 : buf.record = record;
97 :
98 4989586 : txid = XLogRecGetTopXid(record);
99 :
100 : /*
101 : * If the top-level xid is valid, we need to assign the subxact to the
102 : * top-level xact. We need to do this for all records, hence we do it
103 : * before the switch.
104 : */
105 4989586 : if (TransactionIdIsValid(txid))
106 : {
107 1366 : ReorderBufferAssignChild(ctx->reorder,
108 : txid,
109 1366 : XLogRecGetXid(record),
110 : buf.origptr);
111 : }
112 :
113 4989586 : rmgr = GetRmgr(XLogRecGetRmid(record));
114 :
115 4989586 : if (rmgr.rm_decode != NULL)
116 3885618 : rmgr.rm_decode(ctx, &buf);
117 : else
118 : {
119 : /* just deal with xid, and done */
120 1103968 : ReorderBufferProcessXid(ctx->reorder, XLogRecGetXid(record),
121 : buf.origptr);
122 : }
123 4989568 : }
124 :
125 : /*
126 : * Handle rmgr XLOG_ID records for LogicalDecodingProcessRecord().
127 : */
128 : void
129 11754 : xlog_decode(LogicalDecodingContext *ctx, XLogRecordBuffer *buf)
130 : {
131 11754 : SnapBuild *builder = ctx->snapshot_builder;
132 11754 : uint8 info = XLogRecGetInfo(buf->record) & ~XLR_INFO_MASK;
133 :
134 11754 : ReorderBufferProcessXid(ctx->reorder, XLogRecGetXid(buf->record),
135 : buf->origptr);
136 :
137 11754 : switch (info)
138 : {
139 : /* this is also used in END_OF_RECOVERY checkpoints */
140 78 : case XLOG_CHECKPOINT_SHUTDOWN:
141 : case XLOG_END_OF_RECOVERY:
142 78 : SnapBuildSerializationPoint(builder, buf->origptr);
143 :
144 78 : break;
145 128 : case XLOG_CHECKPOINT_ONLINE:
146 :
147 : /*
148 : * a RUNNING_XACTS record will have been logged near to this, we
149 : * can restart from there.
150 : */
151 128 : break;
152 2 : case XLOG_PARAMETER_CHANGE:
153 : {
154 2 : xl_parameter_change *xlrec =
155 2 : (xl_parameter_change *) XLogRecGetData(buf->record);
156 :
157 : /*
158 : * If wal_level on the primary is reduced to less than
159 : * logical, we want to prevent existing logical slots from
160 : * being used. Existing logical slots on the standby get
161 : * invalidated when this WAL record is replayed; and further,
162 : * slot creation fails when wal_level is not sufficient; but
163 : * all these operations are not synchronized, so a logical
164 : * slot may creep in while the wal_level is being reduced.
165 : * Hence this extra check.
166 : */
167 2 : if (xlrec->wal_level < WAL_LEVEL_LOGICAL)
168 : {
169 : /*
170 : * This can occur only on a standby, as a primary would
171 : * not allow to restart after changing wal_level < logical
172 : * if there is pre-existing logical slot.
173 : */
174 : Assert(RecoveryInProgress());
175 0 : ereport(ERROR,
176 : (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
177 : errmsg("logical decoding on standby requires \"wal_level\" >= \"logical\" on the primary")));
178 : }
179 2 : break;
180 : }
181 11546 : case XLOG_NOOP:
182 : case XLOG_NEXTOID:
183 : case XLOG_SWITCH:
184 : case XLOG_BACKUP_END:
185 : case XLOG_RESTORE_POINT:
186 : case XLOG_FPW_CHANGE:
187 : case XLOG_FPI_FOR_HINT:
188 : case XLOG_FPI:
189 : case XLOG_OVERWRITE_CONTRECORD:
190 : case XLOG_CHECKPOINT_REDO:
191 11546 : break;
192 0 : default:
193 0 : elog(ERROR, "unexpected RM_XLOG_ID record type: %u", info);
194 : }
195 11754 : }
196 :
197 : /*
198 : * Handle rmgr XACT_ID records for LogicalDecodingProcessRecord().
199 : */
200 : void
201 16342 : xact_decode(LogicalDecodingContext *ctx, XLogRecordBuffer *buf)
202 : {
203 16342 : SnapBuild *builder = ctx->snapshot_builder;
204 16342 : ReorderBuffer *reorder = ctx->reorder;
205 16342 : XLogReaderState *r = buf->record;
206 16342 : uint8 info = XLogRecGetInfo(r) & XLOG_XACT_OPMASK;
207 :
208 : /*
209 : * If the snapshot isn't yet fully built, we cannot decode anything, so
210 : * bail out.
211 : */
212 16342 : if (SnapBuildCurrentState(builder) < SNAPBUILD_FULL_SNAPSHOT)
213 40 : return;
214 :
215 16302 : switch (info)
216 : {
217 5674 : case XLOG_XACT_COMMIT:
218 : case XLOG_XACT_COMMIT_PREPARED:
219 : {
220 : xl_xact_commit *xlrec;
221 : xl_xact_parsed_commit parsed;
222 : TransactionId xid;
223 5674 : bool two_phase = false;
224 :
225 5674 : xlrec = (xl_xact_commit *) XLogRecGetData(r);
226 5674 : ParseCommitRecord(XLogRecGetInfo(buf->record), xlrec, &parsed);
227 :
228 5674 : if (!TransactionIdIsValid(parsed.twophase_xid))
229 5480 : xid = XLogRecGetXid(r);
230 : else
231 194 : xid = parsed.twophase_xid;
232 :
233 : /*
234 : * We would like to process the transaction in a two-phase
235 : * manner iff output plugin supports two-phase commits and
236 : * doesn't filter the transaction at prepare time.
237 : */
238 5674 : if (info == XLOG_XACT_COMMIT_PREPARED)
239 194 : two_phase = !(FilterPrepare(ctx, xid,
240 194 : parsed.twophase_gid));
241 :
242 5674 : DecodeCommit(ctx, buf, &parsed, xid, two_phase);
243 5660 : break;
244 : }
245 224 : case XLOG_XACT_ABORT:
246 : case XLOG_XACT_ABORT_PREPARED:
247 : {
248 : xl_xact_abort *xlrec;
249 : xl_xact_parsed_abort parsed;
250 : TransactionId xid;
251 224 : bool two_phase = false;
252 :
253 224 : xlrec = (xl_xact_abort *) XLogRecGetData(r);
254 224 : ParseAbortRecord(XLogRecGetInfo(buf->record), xlrec, &parsed);
255 :
256 224 : if (!TransactionIdIsValid(parsed.twophase_xid))
257 152 : xid = XLogRecGetXid(r);
258 : else
259 72 : xid = parsed.twophase_xid;
260 :
261 : /*
262 : * We would like to process the transaction in a two-phase
263 : * manner iff output plugin supports two-phase commits and
264 : * doesn't filter the transaction at prepare time.
265 : */
266 224 : if (info == XLOG_XACT_ABORT_PREPARED)
267 72 : two_phase = !(FilterPrepare(ctx, xid,
268 72 : parsed.twophase_gid));
269 :
270 224 : DecodeAbort(ctx, buf, &parsed, xid, two_phase);
271 224 : break;
272 : }
273 264 : case XLOG_XACT_ASSIGNMENT:
274 :
275 : /*
276 : * We assign subxact to the toplevel xact while processing each
277 : * record if required. So, we don't need to do anything here. See
278 : * LogicalDecodingProcessRecord.
279 : */
280 264 : break;
281 9832 : case XLOG_XACT_INVALIDATIONS:
282 : {
283 : TransactionId xid;
284 : xl_xact_invals *invals;
285 :
286 9832 : xid = XLogRecGetXid(r);
287 9832 : invals = (xl_xact_invals *) XLogRecGetData(r);
288 :
289 : /*
290 : * Execute the invalidations for xid-less transactions,
291 : * otherwise, accumulate them so that they can be processed at
292 : * the commit time.
293 : */
294 9832 : if (TransactionIdIsValid(xid))
295 : {
296 9828 : if (!ctx->fast_forward)
297 9762 : ReorderBufferAddInvalidations(reorder, xid,
298 : buf->origptr,
299 9762 : invals->nmsgs,
300 9762 : invals->msgs);
301 9828 : ReorderBufferXidSetCatalogChanges(ctx->reorder, xid,
302 : buf->origptr);
303 : }
304 4 : else if (!ctx->fast_forward)
305 4 : ReorderBufferImmediateInvalidation(ctx->reorder,
306 4 : invals->nmsgs,
307 4 : invals->msgs);
308 :
309 9832 : break;
310 : }
311 308 : case XLOG_XACT_PREPARE:
312 : {
313 : xl_xact_parsed_prepare parsed;
314 : xl_xact_prepare *xlrec;
315 :
316 : /* ok, parse it */
317 308 : xlrec = (xl_xact_prepare *) XLogRecGetData(r);
318 308 : ParsePrepareRecord(XLogRecGetInfo(buf->record),
319 : xlrec, &parsed);
320 :
321 : /*
322 : * We would like to process the transaction in a two-phase
323 : * manner iff output plugin supports two-phase commits and
324 : * doesn't filter the transaction at prepare time.
325 : */
326 308 : if (FilterPrepare(ctx, parsed.twophase_xid,
327 : parsed.twophase_gid))
328 : {
329 32 : ReorderBufferProcessXid(reorder, parsed.twophase_xid,
330 : buf->origptr);
331 32 : break;
332 : }
333 :
334 : /*
335 : * Note that if the prepared transaction has locked [user]
336 : * catalog tables exclusively then decoding prepare can block
337 : * till the main transaction is committed because it needs to
338 : * lock the catalog tables.
339 : *
340 : * XXX Now, this can even lead to a deadlock if the prepare
341 : * transaction is waiting to get it logically replicated for
342 : * distributed 2PC. This can be avoided by disallowing
343 : * preparing transactions that have locked [user] catalog
344 : * tables exclusively but as of now, we ask users not to do
345 : * such an operation.
346 : */
347 276 : DecodePrepare(ctx, buf, &parsed);
348 276 : break;
349 : }
350 0 : default:
351 0 : elog(ERROR, "unexpected RM_XACT_ID record type: %u", info);
352 : }
353 : }
354 :
355 : /*
356 : * Handle rmgr STANDBY_ID records for LogicalDecodingProcessRecord().
357 : */
358 : void
359 7270 : standby_decode(LogicalDecodingContext *ctx, XLogRecordBuffer *buf)
360 : {
361 7270 : SnapBuild *builder = ctx->snapshot_builder;
362 7270 : XLogReaderState *r = buf->record;
363 7270 : uint8 info = XLogRecGetInfo(r) & ~XLR_INFO_MASK;
364 :
365 7270 : ReorderBufferProcessXid(ctx->reorder, XLogRecGetXid(r), buf->origptr);
366 :
367 7270 : switch (info)
368 : {
369 2682 : case XLOG_RUNNING_XACTS:
370 : {
371 2682 : xl_running_xacts *running = (xl_running_xacts *) XLogRecGetData(r);
372 :
373 2682 : SnapBuildProcessRunningXacts(builder, buf->origptr, running);
374 :
375 : /*
376 : * Abort all transactions that we keep track of, that are
377 : * older than the record's oldestRunningXid. This is the most
378 : * convenient spot for doing so since, in contrast to shutdown
379 : * or end-of-recovery checkpoints, we have information about
380 : * all running transactions which includes prepared ones,
381 : * while shutdown checkpoints just know that no non-prepared
382 : * transactions are in progress.
383 : */
384 2678 : ReorderBufferAbortOld(ctx->reorder, running->oldestRunningXid);
385 : }
386 2678 : break;
387 4584 : case XLOG_STANDBY_LOCK:
388 4584 : break;
389 4 : case XLOG_INVALIDATIONS:
390 :
391 : /*
392 : * We are processing the invalidations at the command level via
393 : * XLOG_XACT_INVALIDATIONS. So we don't need to do anything here.
394 : */
395 4 : break;
396 0 : default:
397 0 : elog(ERROR, "unexpected RM_STANDBY_ID record type: %u", info);
398 : }
399 7266 : }
400 :
401 : /*
402 : * Handle rmgr HEAP2_ID records for LogicalDecodingProcessRecord().
403 : */
404 : void
405 64378 : heap2_decode(LogicalDecodingContext *ctx, XLogRecordBuffer *buf)
406 : {
407 64378 : uint8 info = XLogRecGetInfo(buf->record) & XLOG_HEAP_OPMASK;
408 64378 : TransactionId xid = XLogRecGetXid(buf->record);
409 64378 : SnapBuild *builder = ctx->snapshot_builder;
410 :
411 64378 : ReorderBufferProcessXid(ctx->reorder, xid, buf->origptr);
412 :
413 : /*
414 : * If we don't have snapshot or we are just fast-forwarding, there is no
415 : * point in decoding data changes. However, it's crucial to build the base
416 : * snapshot during fast-forward mode (as is done in
417 : * SnapBuildProcessChange()) because we require the snapshot's xmin when
418 : * determining the candidate catalog_xmin for the replication slot. See
419 : * SnapBuildProcessRunningXacts().
420 : */
421 64378 : if (SnapBuildCurrentState(builder) < SNAPBUILD_FULL_SNAPSHOT)
422 22 : return;
423 :
424 64356 : switch (info)
425 : {
426 11580 : case XLOG_HEAP2_MULTI_INSERT:
427 11580 : if (SnapBuildProcessChange(builder, xid, buf->origptr) &&
428 11580 : !ctx->fast_forward)
429 11450 : DecodeMultiInsert(ctx, buf);
430 11580 : break;
431 47826 : case XLOG_HEAP2_NEW_CID:
432 47826 : if (!ctx->fast_forward)
433 : {
434 : xl_heap_new_cid *xlrec;
435 :
436 47464 : xlrec = (xl_heap_new_cid *) XLogRecGetData(buf->record);
437 47464 : SnapBuildProcessNewCid(builder, xid, buf->origptr, xlrec);
438 :
439 47464 : break;
440 : }
441 : case XLOG_HEAP2_REWRITE:
442 :
443 : /*
444 : * Although these records only exist to serve the needs of logical
445 : * decoding, all the work happens as part of crash or archive
446 : * recovery, so we don't need to do anything here.
447 : */
448 542 : break;
449 :
450 : /*
451 : * Everything else here is just low level physical stuff we're not
452 : * interested in.
453 : */
454 4770 : case XLOG_HEAP2_PRUNE_ON_ACCESS:
455 : case XLOG_HEAP2_PRUNE_VACUUM_SCAN:
456 : case XLOG_HEAP2_PRUNE_VACUUM_CLEANUP:
457 : case XLOG_HEAP2_VISIBLE:
458 : case XLOG_HEAP2_LOCK_UPDATED:
459 4770 : break;
460 0 : default:
461 0 : elog(ERROR, "unexpected RM_HEAP2_ID record type: %u", info);
462 : }
463 : }
464 :
465 : /*
466 : * Handle rmgr HEAP_ID records for LogicalDecodingProcessRecord().
467 : */
468 : void
469 3785760 : heap_decode(LogicalDecodingContext *ctx, XLogRecordBuffer *buf)
470 : {
471 3785760 : uint8 info = XLogRecGetInfo(buf->record) & XLOG_HEAP_OPMASK;
472 3785760 : TransactionId xid = XLogRecGetXid(buf->record);
473 3785760 : SnapBuild *builder = ctx->snapshot_builder;
474 :
475 3785760 : ReorderBufferProcessXid(ctx->reorder, xid, buf->origptr);
476 :
477 : /*
478 : * If we don't have snapshot or we are just fast-forwarding, there is no
479 : * point in decoding data changes. However, it's crucial to build the base
480 : * snapshot during fast-forward mode (as is done in
481 : * SnapBuildProcessChange()) because we require the snapshot's xmin when
482 : * determining the candidate catalog_xmin for the replication slot. See
483 : * SnapBuildProcessRunningXacts().
484 : */
485 3785760 : if (SnapBuildCurrentState(builder) < SNAPBUILD_FULL_SNAPSHOT)
486 7340 : return;
487 :
488 3778420 : switch (info)
489 : {
490 2410048 : case XLOG_HEAP_INSERT:
491 2410048 : if (SnapBuildProcessChange(builder, xid, buf->origptr) &&
492 2410048 : !ctx->fast_forward)
493 2409822 : DecodeInsert(ctx, buf);
494 2410048 : break;
495 :
496 : /*
497 : * Treat HOT update as normal updates. There is no useful
498 : * information in the fact that we could make it a HOT update
499 : * locally and the WAL layout is compatible.
500 : */
501 414420 : case XLOG_HEAP_HOT_UPDATE:
502 : case XLOG_HEAP_UPDATE:
503 414420 : if (SnapBuildProcessChange(builder, xid, buf->origptr) &&
504 414420 : !ctx->fast_forward)
505 414406 : DecodeUpdate(ctx, buf);
506 414420 : break;
507 :
508 535194 : case XLOG_HEAP_DELETE:
509 535194 : if (SnapBuildProcessChange(builder, xid, buf->origptr) &&
510 535194 : !ctx->fast_forward)
511 535188 : DecodeDelete(ctx, buf);
512 535194 : break;
513 :
514 100 : case XLOG_HEAP_TRUNCATE:
515 100 : if (SnapBuildProcessChange(builder, xid, buf->origptr) &&
516 100 : !ctx->fast_forward)
517 98 : DecodeTruncate(ctx, buf);
518 100 : break;
519 :
520 2000 : case XLOG_HEAP_INPLACE:
521 :
522 : /*
523 : * Inplace updates are only ever performed on catalog tuples and
524 : * can, per definition, not change tuple visibility. Inplace
525 : * updates don't affect storage or interpretation of table rows,
526 : * so they don't affect logicalrep_write_tuple() outcomes. Hence,
527 : * we don't process invalidations from the original operation. If
528 : * inplace updates did affect those things, invalidations wouldn't
529 : * make it work, since there are no snapshot-specific versions of
530 : * inplace-updated values. Since we also don't decode catalog
531 : * tuples, we're not interested in the record's contents.
532 : *
533 : * WAL contains likely-unnecessary commit-time invals from the
534 : * CacheInvalidateHeapTuple() call in
535 : * heap_inplace_update_and_unlock(). Excess invalidation is safe.
536 : */
537 2000 : break;
538 :
539 35832 : case XLOG_HEAP_CONFIRM:
540 35832 : if (SnapBuildProcessChange(builder, xid, buf->origptr) &&
541 35832 : !ctx->fast_forward)
542 35832 : DecodeSpecConfirm(ctx, buf);
543 35832 : break;
544 :
545 380826 : case XLOG_HEAP_LOCK:
546 : /* we don't care about row level locks for now */
547 380826 : break;
548 :
549 0 : default:
550 0 : elog(ERROR, "unexpected RM_HEAP_ID record type: %u", info);
551 : break;
552 : }
553 : }
554 :
555 : /*
556 : * Ask output plugin whether we want to skip this PREPARE and send
557 : * this transaction as a regular commit later.
558 : */
559 : static inline bool
560 574 : FilterPrepare(LogicalDecodingContext *ctx, TransactionId xid,
561 : const char *gid)
562 : {
563 : /*
564 : * Skip if decoding of two-phase transactions at PREPARE time is not
565 : * enabled. In that case, all two-phase transactions are considered
566 : * filtered out and will be applied as regular transactions at COMMIT
567 : * PREPARED.
568 : */
569 574 : if (!ctx->twophase)
570 32 : return true;
571 :
572 : /*
573 : * The filter_prepare callback is optional. When not supplied, all
574 : * prepared transactions should go through.
575 : */
576 542 : if (ctx->callbacks.filter_prepare_cb == NULL)
577 246 : return false;
578 :
579 296 : return filter_prepare_cb_wrapper(ctx, xid, gid);
580 : }
581 :
582 : static inline bool
583 3390936 : FilterByOrigin(LogicalDecodingContext *ctx, RepOriginId origin_id)
584 : {
585 3390936 : if (ctx->callbacks.filter_by_origin_cb == NULL)
586 46 : return false;
587 :
588 3390890 : return filter_by_origin_cb_wrapper(ctx, origin_id);
589 : }
590 :
591 : /*
592 : * Handle rmgr LOGICALMSG_ID records for LogicalDecodingProcessRecord().
593 : */
594 : void
595 114 : logicalmsg_decode(LogicalDecodingContext *ctx, XLogRecordBuffer *buf)
596 : {
597 114 : SnapBuild *builder = ctx->snapshot_builder;
598 114 : XLogReaderState *r = buf->record;
599 114 : TransactionId xid = XLogRecGetXid(r);
600 114 : uint8 info = XLogRecGetInfo(r) & ~XLR_INFO_MASK;
601 114 : RepOriginId origin_id = XLogRecGetOrigin(r);
602 114 : Snapshot snapshot = NULL;
603 : xl_logical_message *message;
604 :
605 114 : if (info != XLOG_LOGICAL_MESSAGE)
606 0 : elog(ERROR, "unexpected RM_LOGICALMSG_ID record type: %u", info);
607 :
608 114 : ReorderBufferProcessXid(ctx->reorder, XLogRecGetXid(r), buf->origptr);
609 :
610 : /* If we don't have snapshot, there is no point in decoding messages */
611 114 : if (SnapBuildCurrentState(builder) < SNAPBUILD_FULL_SNAPSHOT)
612 0 : return;
613 :
614 114 : message = (xl_logical_message *) XLogRecGetData(r);
615 :
616 224 : if (message->dbId != ctx->slot->data.database ||
617 110 : FilterByOrigin(ctx, origin_id))
618 8 : return;
619 :
620 106 : if (message->transactional &&
621 78 : !SnapBuildProcessChange(builder, xid, buf->origptr))
622 0 : return;
623 134 : else if (!message->transactional &&
624 56 : (SnapBuildCurrentState(builder) != SNAPBUILD_CONSISTENT ||
625 28 : SnapBuildXactNeedsSkip(builder, buf->origptr)))
626 8 : return;
627 :
628 : /*
629 : * We also skip decoding in fast_forward mode. This check must be last
630 : * because we don't want to set the processing_required flag unless we
631 : * have a decodable message.
632 : */
633 98 : if (ctx->fast_forward)
634 : {
635 : /*
636 : * We need to set processing_required flag to notify the message's
637 : * existence to the caller. Usually, the flag is set when either the
638 : * COMMIT or ABORT records are decoded, but this must be turned on
639 : * here because the non-transactional logical message is decoded
640 : * without waiting for these records.
641 : */
642 4 : if (!message->transactional)
643 4 : ctx->processing_required = true;
644 :
645 4 : return;
646 : }
647 :
648 : /*
649 : * If this is a non-transactional change, get the snapshot we're expected
650 : * to use. We only get here when the snapshot is consistent, and the
651 : * change is not meant to be skipped.
652 : *
653 : * For transactional changes we don't need a snapshot, we'll use the
654 : * regular snapshot maintained by ReorderBuffer. We just leave it NULL.
655 : */
656 94 : if (!message->transactional)
657 16 : snapshot = SnapBuildGetOrBuildSnapshot(builder);
658 :
659 94 : ReorderBufferQueueMessage(ctx->reorder, xid, snapshot, buf->endptr,
660 94 : message->transactional,
661 94 : message->message, /* first part of message is
662 : * prefix */
663 : message->message_size,
664 94 : message->message + message->prefix_size);
665 : }
666 :
667 : /*
668 : * Consolidated commit record handling between the different form of commit
669 : * records.
670 : *
671 : * 'two_phase' indicates that caller wants to process the transaction in two
672 : * phases, first process prepare if not already done and then process
673 : * commit_prepared.
674 : */
675 : static void
676 5674 : DecodeCommit(LogicalDecodingContext *ctx, XLogRecordBuffer *buf,
677 : xl_xact_parsed_commit *parsed, TransactionId xid,
678 : bool two_phase)
679 : {
680 5674 : XLogRecPtr origin_lsn = InvalidXLogRecPtr;
681 5674 : TimestampTz commit_time = parsed->xact_time;
682 5674 : RepOriginId origin_id = XLogRecGetOrigin(buf->record);
683 : int i;
684 :
685 5674 : if (parsed->xinfo & XACT_XINFO_HAS_ORIGIN)
686 : {
687 138 : origin_lsn = parsed->origin_lsn;
688 138 : commit_time = parsed->origin_timestamp;
689 : }
690 :
691 5674 : SnapBuildCommitTxn(ctx->snapshot_builder, buf->origptr, xid,
692 : parsed->nsubxacts, parsed->subxacts,
693 : parsed->xinfo);
694 :
695 : /* ----
696 : * Check whether we are interested in this specific transaction, and tell
697 : * the reorderbuffer to forget the content of the (sub-)transactions
698 : * if not.
699 : *
700 : * We can't just use ReorderBufferAbort() here, because we need to execute
701 : * the transaction's invalidations. This currently won't be needed if
702 : * we're just skipping over the transaction because currently we only do
703 : * so during startup, to get to the first transaction the client needs. As
704 : * we have reset the catalog caches before starting to read WAL, and we
705 : * haven't yet touched any catalogs, there can't be anything to invalidate.
706 : * But if we're "forgetting" this commit because it happened in another
707 : * database, the invalidations might be important, because they could be
708 : * for shared catalogs and we might have loaded data into the relevant
709 : * syscaches.
710 : * ---
711 : */
712 5674 : if (DecodeTXNNeedSkip(ctx, buf, parsed->dbId, origin_id))
713 : {
714 5040 : for (i = 0; i < parsed->nsubxacts; i++)
715 : {
716 1966 : ReorderBufferForget(ctx->reorder, parsed->subxacts[i], buf->origptr);
717 : }
718 3074 : ReorderBufferForget(ctx->reorder, xid, buf->origptr);
719 :
720 3074 : return;
721 : }
722 :
723 : /* tell the reorderbuffer about the surviving subtransactions */
724 3132 : for (i = 0; i < parsed->nsubxacts; i++)
725 : {
726 532 : ReorderBufferCommitChild(ctx->reorder, xid, parsed->subxacts[i],
727 : buf->origptr, buf->endptr);
728 : }
729 :
730 : /*
731 : * Send the final commit record if the transaction data is already
732 : * decoded, otherwise, process the entire transaction.
733 : */
734 2600 : if (two_phase)
735 : {
736 62 : ReorderBufferFinishPrepared(ctx->reorder, xid, buf->origptr, buf->endptr,
737 62 : SnapBuildGetTwoPhaseAt(ctx->snapshot_builder),
738 : commit_time, origin_id, origin_lsn,
739 62 : parsed->twophase_gid, true);
740 : }
741 : else
742 : {
743 2538 : ReorderBufferCommit(ctx->reorder, xid, buf->origptr, buf->endptr,
744 : commit_time, origin_id, origin_lsn);
745 : }
746 :
747 : /*
748 : * Update the decoding stats at transaction prepare/commit/abort.
749 : * Additionally we send the stats when we spill or stream the changes to
750 : * avoid losing them in case the decoding is interrupted. It is not clear
751 : * that sending more or less frequently than this would be better.
752 : */
753 2586 : UpdateDecodingStats(ctx);
754 : }
755 :
756 : /*
757 : * Decode PREPARE record. Similar logic as in DecodeCommit.
758 : *
759 : * Note that we don't skip prepare even if have detected concurrent abort
760 : * because it is quite possible that we had already sent some changes before we
761 : * detect abort in which case we need to abort those changes in the subscriber.
762 : * To abort such changes, we do send the prepare and then the rollback prepared
763 : * which is what happened on the publisher-side as well. Now, we can invent a
764 : * new abort API wherein in such cases we send abort and skip sending prepared
765 : * and rollback prepared but then it is not that straightforward because we
766 : * might have streamed this transaction by that time in which case it is
767 : * handled when the rollback is encountered. It is not impossible to optimize
768 : * the concurrent abort case but it can introduce design complexity w.r.t
769 : * handling different cases so leaving it for now as it doesn't seem worth it.
770 : */
771 : static void
772 276 : DecodePrepare(LogicalDecodingContext *ctx, XLogRecordBuffer *buf,
773 : xl_xact_parsed_prepare *parsed)
774 : {
775 276 : SnapBuild *builder = ctx->snapshot_builder;
776 276 : XLogRecPtr origin_lsn = parsed->origin_lsn;
777 276 : TimestampTz prepare_time = parsed->xact_time;
778 276 : RepOriginId origin_id = XLogRecGetOrigin(buf->record);
779 : int i;
780 276 : TransactionId xid = parsed->twophase_xid;
781 :
782 276 : if (parsed->origin_timestamp != 0)
783 16 : prepare_time = parsed->origin_timestamp;
784 :
785 : /*
786 : * Remember the prepare info for a txn so that it can be used later in
787 : * commit prepared if required. See ReorderBufferFinishPrepared.
788 : */
789 276 : if (!ReorderBufferRememberPrepareInfo(ctx->reorder, xid, buf->origptr,
790 : buf->endptr, prepare_time, origin_id,
791 : origin_lsn))
792 0 : return;
793 :
794 : /* We can't start streaming unless a consistent state is reached. */
795 276 : if (SnapBuildCurrentState(builder) < SNAPBUILD_CONSISTENT)
796 : {
797 6 : ReorderBufferSkipPrepare(ctx->reorder, xid);
798 6 : return;
799 : }
800 :
801 : /*
802 : * Check whether we need to process this transaction. See
803 : * DecodeTXNNeedSkip for the reasons why we sometimes want to skip the
804 : * transaction.
805 : *
806 : * We can't call ReorderBufferForget as we did in DecodeCommit as the txn
807 : * hasn't yet been committed, removing this txn before a commit might
808 : * result in the computation of an incorrect restart_lsn. See
809 : * SnapBuildProcessRunningXacts. But we need to process cache
810 : * invalidations if there are any for the reasons mentioned in
811 : * DecodeCommit.
812 : */
813 270 : if (DecodeTXNNeedSkip(ctx, buf, parsed->dbId, origin_id))
814 : {
815 192 : ReorderBufferSkipPrepare(ctx->reorder, xid);
816 192 : ReorderBufferInvalidate(ctx->reorder, xid, buf->origptr);
817 192 : return;
818 : }
819 :
820 : /* Tell the reorderbuffer about the surviving subtransactions. */
821 80 : for (i = 0; i < parsed->nsubxacts; i++)
822 : {
823 2 : ReorderBufferCommitChild(ctx->reorder, xid, parsed->subxacts[i],
824 : buf->origptr, buf->endptr);
825 : }
826 :
827 : /* replay actions of all transaction + subtransactions in order */
828 78 : ReorderBufferPrepare(ctx->reorder, xid, parsed->twophase_gid);
829 :
830 : /*
831 : * Update the decoding stats at transaction prepare/commit/abort.
832 : * Additionally we send the stats when we spill or stream the changes to
833 : * avoid losing them in case the decoding is interrupted. It is not clear
834 : * that sending more or less frequently than this would be better.
835 : */
836 78 : UpdateDecodingStats(ctx);
837 : }
838 :
839 :
840 : /*
841 : * Get the data from the various forms of abort records and pass it on to
842 : * snapbuild.c and reorderbuffer.c.
843 : *
844 : * 'two_phase' indicates to finish prepared transaction.
845 : */
846 : static void
847 224 : DecodeAbort(LogicalDecodingContext *ctx, XLogRecordBuffer *buf,
848 : xl_xact_parsed_abort *parsed, TransactionId xid,
849 : bool two_phase)
850 : {
851 : int i;
852 224 : XLogRecPtr origin_lsn = InvalidXLogRecPtr;
853 224 : TimestampTz abort_time = parsed->xact_time;
854 224 : RepOriginId origin_id = XLogRecGetOrigin(buf->record);
855 : bool skip_xact;
856 :
857 224 : if (parsed->xinfo & XACT_XINFO_HAS_ORIGIN)
858 : {
859 8 : origin_lsn = parsed->origin_lsn;
860 8 : abort_time = parsed->origin_timestamp;
861 : }
862 :
863 : /*
864 : * Check whether we need to process this transaction. See
865 : * DecodeTXNNeedSkip for the reasons why we sometimes want to skip the
866 : * transaction.
867 : */
868 224 : skip_xact = DecodeTXNNeedSkip(ctx, buf, parsed->dbId, origin_id);
869 :
870 : /*
871 : * Send the final rollback record for a prepared transaction unless we
872 : * need to skip it. For non-two-phase xacts, simply forget the xact.
873 : */
874 224 : if (two_phase && !skip_xact)
875 : {
876 20 : ReorderBufferFinishPrepared(ctx->reorder, xid, buf->origptr, buf->endptr,
877 : InvalidXLogRecPtr,
878 : abort_time, origin_id, origin_lsn,
879 20 : parsed->twophase_gid, false);
880 : }
881 : else
882 : {
883 216 : for (i = 0; i < parsed->nsubxacts; i++)
884 : {
885 12 : ReorderBufferAbort(ctx->reorder, parsed->subxacts[i],
886 12 : buf->record->EndRecPtr, abort_time);
887 : }
888 :
889 204 : ReorderBufferAbort(ctx->reorder, xid, buf->record->EndRecPtr,
890 : abort_time);
891 : }
892 :
893 : /* update the decoding stats */
894 224 : UpdateDecodingStats(ctx);
895 224 : }
896 :
897 : /*
898 : * Parse XLOG_HEAP_INSERT (not MULTI_INSERT!) records into tuplebufs.
899 : *
900 : * Inserts can contain the new tuple.
901 : */
902 : static void
903 2409822 : DecodeInsert(LogicalDecodingContext *ctx, XLogRecordBuffer *buf)
904 : {
905 : Size datalen;
906 : char *tupledata;
907 : Size tuplelen;
908 2409822 : XLogReaderState *r = buf->record;
909 : xl_heap_insert *xlrec;
910 : ReorderBufferChange *change;
911 : RelFileLocator target_locator;
912 :
913 2409822 : xlrec = (xl_heap_insert *) XLogRecGetData(r);
914 :
915 : /*
916 : * Ignore insert records without new tuples (this does happen when
917 : * raw_heap_insert marks the TOAST record as HEAP_INSERT_NO_LOGICAL).
918 : */
919 2409822 : if (!(xlrec->flags & XLH_INSERT_CONTAINS_NEW_TUPLE))
920 7374 : return;
921 :
922 : /* only interested in our database */
923 2402482 : XLogRecGetBlockTag(r, 0, &target_locator, NULL, NULL);
924 2402482 : if (target_locator.dbOid != ctx->slot->data.database)
925 0 : return;
926 :
927 : /* output plugin doesn't look for this origin, no need to queue */
928 2402482 : if (FilterByOrigin(ctx, XLogRecGetOrigin(r)))
929 34 : return;
930 :
931 2402448 : change = ReorderBufferAllocChange(ctx->reorder);
932 2402448 : if (!(xlrec->flags & XLH_INSERT_IS_SPECULATIVE))
933 2366616 : change->action = REORDER_BUFFER_CHANGE_INSERT;
934 : else
935 35832 : change->action = REORDER_BUFFER_CHANGE_INTERNAL_SPEC_INSERT;
936 2402448 : change->origin_id = XLogRecGetOrigin(r);
937 :
938 2402448 : memcpy(&change->data.tp.rlocator, &target_locator, sizeof(RelFileLocator));
939 :
940 2402448 : tupledata = XLogRecGetBlockData(r, 0, &datalen);
941 2402448 : tuplelen = datalen - SizeOfHeapHeader;
942 :
943 2402448 : change->data.tp.newtuple =
944 2402448 : ReorderBufferAllocTupleBuf(ctx->reorder, tuplelen);
945 :
946 2402448 : DecodeXLogTuple(tupledata, datalen, change->data.tp.newtuple);
947 :
948 2402448 : change->data.tp.clear_toast_afterwards = true;
949 :
950 2402448 : ReorderBufferQueueChange(ctx->reorder, XLogRecGetXid(r), buf->origptr,
951 : change,
952 2402448 : xlrec->flags & XLH_INSERT_ON_TOAST_RELATION);
953 : }
954 :
955 : /*
956 : * Parse XLOG_HEAP_UPDATE and XLOG_HEAP_HOT_UPDATE, which have the same layout
957 : * in the record, from wal into proper tuplebufs.
958 : *
959 : * Updates can possibly contain a new tuple and the old primary key.
960 : */
961 : static void
962 414406 : DecodeUpdate(LogicalDecodingContext *ctx, XLogRecordBuffer *buf)
963 : {
964 414406 : XLogReaderState *r = buf->record;
965 : xl_heap_update *xlrec;
966 : ReorderBufferChange *change;
967 : char *data;
968 : RelFileLocator target_locator;
969 :
970 414406 : xlrec = (xl_heap_update *) XLogRecGetData(r);
971 :
972 : /* only interested in our database */
973 414406 : XLogRecGetBlockTag(r, 0, &target_locator, NULL, NULL);
974 414406 : if (target_locator.dbOid != ctx->slot->data.database)
975 68 : return;
976 :
977 : /* output plugin doesn't look for this origin, no need to queue */
978 414382 : if (FilterByOrigin(ctx, XLogRecGetOrigin(r)))
979 44 : return;
980 :
981 414338 : change = ReorderBufferAllocChange(ctx->reorder);
982 414338 : change->action = REORDER_BUFFER_CHANGE_UPDATE;
983 414338 : change->origin_id = XLogRecGetOrigin(r);
984 414338 : memcpy(&change->data.tp.rlocator, &target_locator, sizeof(RelFileLocator));
985 :
986 414338 : if (xlrec->flags & XLH_UPDATE_CONTAINS_NEW_TUPLE)
987 : {
988 : Size datalen;
989 : Size tuplelen;
990 :
991 410992 : data = XLogRecGetBlockData(r, 0, &datalen);
992 :
993 410992 : tuplelen = datalen - SizeOfHeapHeader;
994 :
995 410992 : change->data.tp.newtuple =
996 410992 : ReorderBufferAllocTupleBuf(ctx->reorder, tuplelen);
997 :
998 410992 : DecodeXLogTuple(data, datalen, change->data.tp.newtuple);
999 : }
1000 :
1001 414338 : if (xlrec->flags & XLH_UPDATE_CONTAINS_OLD)
1002 : {
1003 : Size datalen;
1004 : Size tuplelen;
1005 :
1006 : /* caution, remaining data in record is not aligned */
1007 732 : data = XLogRecGetData(r) + SizeOfHeapUpdate;
1008 732 : datalen = XLogRecGetDataLen(r) - SizeOfHeapUpdate;
1009 732 : tuplelen = datalen - SizeOfHeapHeader;
1010 :
1011 732 : change->data.tp.oldtuple =
1012 732 : ReorderBufferAllocTupleBuf(ctx->reorder, tuplelen);
1013 :
1014 732 : DecodeXLogTuple(data, datalen, change->data.tp.oldtuple);
1015 : }
1016 :
1017 414338 : change->data.tp.clear_toast_afterwards = true;
1018 :
1019 414338 : ReorderBufferQueueChange(ctx->reorder, XLogRecGetXid(r), buf->origptr,
1020 : change, false);
1021 : }
1022 :
1023 : /*
1024 : * Parse XLOG_HEAP_DELETE from wal into proper tuplebufs.
1025 : *
1026 : * Deletes can possibly contain the old primary key.
1027 : */
1028 : static void
1029 535188 : DecodeDelete(LogicalDecodingContext *ctx, XLogRecordBuffer *buf)
1030 : {
1031 535188 : XLogReaderState *r = buf->record;
1032 : xl_heap_delete *xlrec;
1033 : ReorderBufferChange *change;
1034 : RelFileLocator target_locator;
1035 :
1036 535188 : xlrec = (xl_heap_delete *) XLogRecGetData(r);
1037 :
1038 : /* only interested in our database */
1039 535188 : XLogRecGetBlockTag(r, 0, &target_locator, NULL, NULL);
1040 535188 : if (target_locator.dbOid != ctx->slot->data.database)
1041 74 : return;
1042 :
1043 : /* output plugin doesn't look for this origin, no need to queue */
1044 535126 : if (FilterByOrigin(ctx, XLogRecGetOrigin(r)))
1045 12 : return;
1046 :
1047 535114 : change = ReorderBufferAllocChange(ctx->reorder);
1048 :
1049 535114 : if (xlrec->flags & XLH_DELETE_IS_SUPER)
1050 0 : change->action = REORDER_BUFFER_CHANGE_INTERNAL_SPEC_ABORT;
1051 : else
1052 535114 : change->action = REORDER_BUFFER_CHANGE_DELETE;
1053 :
1054 535114 : change->origin_id = XLogRecGetOrigin(r);
1055 :
1056 535114 : memcpy(&change->data.tp.rlocator, &target_locator, sizeof(RelFileLocator));
1057 :
1058 : /* old primary key stored */
1059 535114 : if (xlrec->flags & XLH_DELETE_CONTAINS_OLD)
1060 : {
1061 411392 : Size datalen = XLogRecGetDataLen(r) - SizeOfHeapDelete;
1062 411392 : Size tuplelen = datalen - SizeOfHeapHeader;
1063 :
1064 : Assert(XLogRecGetDataLen(r) > (SizeOfHeapDelete + SizeOfHeapHeader));
1065 :
1066 411392 : change->data.tp.oldtuple =
1067 411392 : ReorderBufferAllocTupleBuf(ctx->reorder, tuplelen);
1068 :
1069 411392 : DecodeXLogTuple((char *) xlrec + SizeOfHeapDelete,
1070 : datalen, change->data.tp.oldtuple);
1071 : }
1072 :
1073 535114 : change->data.tp.clear_toast_afterwards = true;
1074 :
1075 535114 : ReorderBufferQueueChange(ctx->reorder, XLogRecGetXid(r), buf->origptr,
1076 : change, false);
1077 : }
1078 :
1079 : /*
1080 : * Parse XLOG_HEAP_TRUNCATE from wal
1081 : */
1082 : static void
1083 98 : DecodeTruncate(LogicalDecodingContext *ctx, XLogRecordBuffer *buf)
1084 : {
1085 98 : XLogReaderState *r = buf->record;
1086 : xl_heap_truncate *xlrec;
1087 : ReorderBufferChange *change;
1088 :
1089 98 : xlrec = (xl_heap_truncate *) XLogRecGetData(r);
1090 :
1091 : /* only interested in our database */
1092 98 : if (xlrec->dbId != ctx->slot->data.database)
1093 0 : return;
1094 :
1095 : /* output plugin doesn't look for this origin, no need to queue */
1096 98 : if (FilterByOrigin(ctx, XLogRecGetOrigin(r)))
1097 0 : return;
1098 :
1099 98 : change = ReorderBufferAllocChange(ctx->reorder);
1100 98 : change->action = REORDER_BUFFER_CHANGE_TRUNCATE;
1101 98 : change->origin_id = XLogRecGetOrigin(r);
1102 98 : if (xlrec->flags & XLH_TRUNCATE_CASCADE)
1103 2 : change->data.truncate.cascade = true;
1104 98 : if (xlrec->flags & XLH_TRUNCATE_RESTART_SEQS)
1105 4 : change->data.truncate.restart_seqs = true;
1106 98 : change->data.truncate.nrelids = xlrec->nrelids;
1107 196 : change->data.truncate.relids = ReorderBufferAllocRelids(ctx->reorder,
1108 98 : xlrec->nrelids);
1109 98 : memcpy(change->data.truncate.relids, xlrec->relids,
1110 98 : xlrec->nrelids * sizeof(Oid));
1111 98 : ReorderBufferQueueChange(ctx->reorder, XLogRecGetXid(r),
1112 : buf->origptr, change, false);
1113 : }
1114 :
1115 : /*
1116 : * Decode XLOG_HEAP2_MULTI_INSERT record into multiple tuplebufs.
1117 : *
1118 : * Currently MULTI_INSERT will always contain the full tuples.
1119 : */
1120 : static void
1121 11450 : DecodeMultiInsert(LogicalDecodingContext *ctx, XLogRecordBuffer *buf)
1122 : {
1123 11450 : XLogReaderState *r = buf->record;
1124 : xl_heap_multi_insert *xlrec;
1125 : int i;
1126 : char *data;
1127 : char *tupledata;
1128 : Size tuplelen;
1129 : RelFileLocator rlocator;
1130 :
1131 11450 : xlrec = (xl_heap_multi_insert *) XLogRecGetData(r);
1132 :
1133 : /*
1134 : * Ignore insert records without new tuples. This happens when a
1135 : * multi_insert is done on a catalog or on a non-persistent relation.
1136 : */
1137 11450 : if (!(xlrec->flags & XLH_INSERT_CONTAINS_NEW_TUPLE))
1138 11418 : return;
1139 :
1140 : /* only interested in our database */
1141 134 : XLogRecGetBlockTag(r, 0, &rlocator, NULL, NULL);
1142 134 : if (rlocator.dbOid != ctx->slot->data.database)
1143 102 : return;
1144 :
1145 : /* output plugin doesn't look for this origin, no need to queue */
1146 32 : if (FilterByOrigin(ctx, XLogRecGetOrigin(r)))
1147 0 : return;
1148 :
1149 : /*
1150 : * We know that this multi_insert isn't for a catalog, so the block should
1151 : * always have data even if a full-page write of it is taken.
1152 : */
1153 32 : tupledata = XLogRecGetBlockData(r, 0, &tuplelen);
1154 : Assert(tupledata != NULL);
1155 :
1156 32 : data = tupledata;
1157 2150 : for (i = 0; i < xlrec->ntuples; i++)
1158 : {
1159 : ReorderBufferChange *change;
1160 : xl_multi_insert_tuple *xlhdr;
1161 : int datalen;
1162 : HeapTuple tuple;
1163 : HeapTupleHeader header;
1164 :
1165 2118 : change = ReorderBufferAllocChange(ctx->reorder);
1166 2118 : change->action = REORDER_BUFFER_CHANGE_INSERT;
1167 2118 : change->origin_id = XLogRecGetOrigin(r);
1168 :
1169 2118 : memcpy(&change->data.tp.rlocator, &rlocator, sizeof(RelFileLocator));
1170 :
1171 2118 : xlhdr = (xl_multi_insert_tuple *) SHORTALIGN(data);
1172 2118 : data = ((char *) xlhdr) + SizeOfMultiInsertTuple;
1173 2118 : datalen = xlhdr->datalen;
1174 :
1175 2118 : change->data.tp.newtuple =
1176 2118 : ReorderBufferAllocTupleBuf(ctx->reorder, datalen);
1177 :
1178 2118 : tuple = change->data.tp.newtuple;
1179 2118 : header = tuple->t_data;
1180 :
1181 : /* not a disk based tuple */
1182 2118 : ItemPointerSetInvalid(&tuple->t_self);
1183 :
1184 : /*
1185 : * We can only figure this out after reassembling the transactions.
1186 : */
1187 2118 : tuple->t_tableOid = InvalidOid;
1188 :
1189 2118 : tuple->t_len = datalen + SizeofHeapTupleHeader;
1190 :
1191 2118 : memset(header, 0, SizeofHeapTupleHeader);
1192 :
1193 2118 : memcpy((char *) tuple->t_data + SizeofHeapTupleHeader, data, datalen);
1194 2118 : header->t_infomask = xlhdr->t_infomask;
1195 2118 : header->t_infomask2 = xlhdr->t_infomask2;
1196 2118 : header->t_hoff = xlhdr->t_hoff;
1197 :
1198 : /*
1199 : * Reset toast reassembly state only after the last row in the last
1200 : * xl_multi_insert_tuple record emitted by one heap_multi_insert()
1201 : * call.
1202 : */
1203 2118 : if (xlrec->flags & XLH_INSERT_LAST_IN_MULTI &&
1204 398 : (i + 1) == xlrec->ntuples)
1205 22 : change->data.tp.clear_toast_afterwards = true;
1206 : else
1207 2096 : change->data.tp.clear_toast_afterwards = false;
1208 :
1209 2118 : ReorderBufferQueueChange(ctx->reorder, XLogRecGetXid(r),
1210 : buf->origptr, change, false);
1211 :
1212 : /* move to the next xl_multi_insert_tuple entry */
1213 2118 : data += datalen;
1214 : }
1215 : Assert(data == tupledata + tuplelen);
1216 : }
1217 :
1218 : /*
1219 : * Parse XLOG_HEAP_CONFIRM from wal into a confirmation change.
1220 : *
1221 : * This is pretty trivial, all the state essentially already setup by the
1222 : * speculative insertion.
1223 : */
1224 : static void
1225 35832 : DecodeSpecConfirm(LogicalDecodingContext *ctx, XLogRecordBuffer *buf)
1226 : {
1227 35832 : XLogReaderState *r = buf->record;
1228 : ReorderBufferChange *change;
1229 : RelFileLocator target_locator;
1230 :
1231 : /* only interested in our database */
1232 35832 : XLogRecGetBlockTag(r, 0, &target_locator, NULL, NULL);
1233 35832 : if (target_locator.dbOid != ctx->slot->data.database)
1234 0 : return;
1235 :
1236 : /* output plugin doesn't look for this origin, no need to queue */
1237 35832 : if (FilterByOrigin(ctx, XLogRecGetOrigin(r)))
1238 0 : return;
1239 :
1240 35832 : change = ReorderBufferAllocChange(ctx->reorder);
1241 35832 : change->action = REORDER_BUFFER_CHANGE_INTERNAL_SPEC_CONFIRM;
1242 35832 : change->origin_id = XLogRecGetOrigin(r);
1243 :
1244 35832 : memcpy(&change->data.tp.rlocator, &target_locator, sizeof(RelFileLocator));
1245 :
1246 35832 : change->data.tp.clear_toast_afterwards = true;
1247 :
1248 35832 : ReorderBufferQueueChange(ctx->reorder, XLogRecGetXid(r), buf->origptr,
1249 : change, false);
1250 : }
1251 :
1252 :
1253 : /*
1254 : * Read a HeapTuple as WAL logged by heap_insert, heap_update and heap_delete
1255 : * (but not by heap_multi_insert) into a tuplebuf.
1256 : *
1257 : * The size 'len' and the pointer 'data' in the record need to be
1258 : * computed outside as they are record specific.
1259 : */
1260 : static void
1261 3225564 : DecodeXLogTuple(char *data, Size len, HeapTuple tuple)
1262 : {
1263 : xl_heap_header xlhdr;
1264 3225564 : int datalen = len - SizeOfHeapHeader;
1265 : HeapTupleHeader header;
1266 :
1267 : Assert(datalen >= 0);
1268 :
1269 3225564 : tuple->t_len = datalen + SizeofHeapTupleHeader;
1270 3225564 : header = tuple->t_data;
1271 :
1272 : /* not a disk based tuple */
1273 3225564 : ItemPointerSetInvalid(&tuple->t_self);
1274 :
1275 : /* we can only figure this out after reassembling the transactions */
1276 3225564 : tuple->t_tableOid = InvalidOid;
1277 :
1278 : /* data is not stored aligned, copy to aligned storage */
1279 3225564 : memcpy(&xlhdr, data, SizeOfHeapHeader);
1280 :
1281 3225564 : memset(header, 0, SizeofHeapTupleHeader);
1282 :
1283 3225564 : memcpy(((char *) tuple->t_data) + SizeofHeapTupleHeader,
1284 3225564 : data + SizeOfHeapHeader,
1285 : datalen);
1286 :
1287 3225564 : header->t_infomask = xlhdr.t_infomask;
1288 3225564 : header->t_infomask2 = xlhdr.t_infomask2;
1289 3225564 : header->t_hoff = xlhdr.t_hoff;
1290 3225564 : }
1291 :
1292 : /*
1293 : * Check whether we are interested in this specific transaction.
1294 : *
1295 : * There can be several reasons we might not be interested in this
1296 : * transaction:
1297 : * 1) We might not be interested in decoding transactions up to this
1298 : * LSN. This can happen because we previously decoded it and now just
1299 : * are restarting or if we haven't assembled a consistent snapshot yet.
1300 : * 2) The transaction happened in another database.
1301 : * 3) The output plugin is not interested in the origin.
1302 : * 4) We are doing fast-forwarding
1303 : */
1304 : static bool
1305 6168 : DecodeTXNNeedSkip(LogicalDecodingContext *ctx, XLogRecordBuffer *buf,
1306 : Oid txn_dbid, RepOriginId origin_id)
1307 : {
1308 6168 : if (SnapBuildXactNeedsSkip(ctx->snapshot_builder, buf->origptr) ||
1309 5720 : (txn_dbid != InvalidOid && txn_dbid != ctx->slot->data.database) ||
1310 2874 : FilterByOrigin(ctx, origin_id))
1311 3346 : return true;
1312 :
1313 : /*
1314 : * We also skip decoding in fast_forward mode. In passing set the
1315 : * processing_required flag to indicate that if it were not for
1316 : * fast_forward mode, processing would have been required.
1317 : */
1318 2822 : if (ctx->fast_forward)
1319 : {
1320 42 : ctx->processing_required = true;
1321 42 : return true;
1322 : }
1323 :
1324 2780 : return false;
1325 : }
|
