Revamp the WAL record format.

Each WAL record now carries information about the modified relation and
block(s) in a standardized format. That makes it easier to write tools that
need that information, like pg_rewind, prefetching the blocks to speed up
recovery, etc.

There's a whole new API for building WAL records, replacing the XLogRecData
chains used previously. The new API consists of XLogRegister* functions,
which are called for each buffer and chunk of data that is added to the
record. The new API also gives more control over when a full-page image is
written, by passing flags to the XLogRegisterBuffer function.

This also simplifies the XLogReadBufferForRedo() calls. The function can dig
the relation and block number from the WAL record, so they no longer need to
be passed as arguments.

For the convenience of redo routines, XLogReader now disects each WAL record
after reading it, copying the main data part and the per-block data into
MAXALIGNed buffers. The data chunks are not aligned within the WAL record,
but the redo routines can assume that the pointers returned by XLogRecGet*
functions are. Redo routines are now passed the XLogReaderState, which
contains the record in the already-disected format, instead of the plain
XLogRecord.

The new record format also makes the fixed size XLogRecord header smaller,
by removing the xl_len field. The length of the "main data" portion is now
stored at the end of the WAL record, and there's a separate header after
XLogRecord for it. The alignment padding at the end of XLogRecord is also
removed. This compansates for the fact that the new format would otherwise
be more bulky than the old format.

Reviewed by Andres Freund, Amit Kapila, Michael Paquier, Alvaro Herrera,
Fujii Masao.

1 files changed, 45 insertions, 84 deletions

diff --git a/src/backend/access/transam/xact.c b/src/backend/access/transam/xact.c
index 6f92bad07ca..763e9deb6f5 100644
--- a/src/backend/access/transam/xact.c
+++ b/src/backend/access/transam/xact.c
@@ -571,7 +571,6 @@ AssignTransactionId(TransactionState s)
 		if (nUnreportedXids >= PGPROC_MAX_CACHED_SUBXIDS ||
 			log_unknown_top)
 		{
-			XLogRecData rdata[2];
 			xl_xact_assignment xlrec;
 
 			/*
@@ -582,17 +581,12 @@ AssignTransactionId(TransactionState s)
 			Assert(TransactionIdIsValid(xlrec.xtop));
 			xlrec.nsubxacts = nUnreportedXids;
 
-			rdata[0].data = (char *) &xlrec;
-			rdata[0].len = MinSizeOfXactAssignment;
-			rdata[0].buffer = InvalidBuffer;
-			rdata[0].next = &rdata[1];
+			XLogBeginInsert();
+			XLogRegisterData((char *) &xlrec, MinSizeOfXactAssignment);
+			XLogRegisterData((char *) unreportedXids,
+							 nUnreportedXids * sizeof(TransactionId));
 
-			rdata[1].data = (char *) unreportedXids;
-			rdata[1].len = nUnreportedXids * sizeof(TransactionId);
-			rdata[1].buffer = InvalidBuffer;
-			rdata[1].next = NULL;
-
-			(void) XLogInsert(RM_XACT_ID, XLOG_XACT_ASSIGNMENT, rdata);
+			(void) XLogInsert(RM_XACT_ID, XLOG_XACT_ASSIGNMENT);
 
 			nUnreportedXids = 0;
 			/* mark top, not current xact as having been logged */
@@ -1087,8 +1081,6 @@ RecordTransactionCommit(void)
 		if (nrels > 0 || nmsgs > 0 || RelcacheInitFileInval || forceSyncCommit ||
 			XLogLogicalInfoActive())
 		{
-			XLogRecData rdata[4];
-			int			lastrdata = 0;
 			xl_xact_commit xlrec;
 
 			/*
@@ -1107,63 +1099,38 @@ RecordTransactionCommit(void)
 			xlrec.nrels = nrels;
 			xlrec.nsubxacts = nchildren;
 			xlrec.nmsgs = nmsgs;
-			rdata[0].data = (char *) (&xlrec);
-			rdata[0].len = MinSizeOfXactCommit;
-			rdata[0].buffer = InvalidBuffer;
+
+			XLogBeginInsert();
+			XLogRegisterData((char *) (&xlrec), MinSizeOfXactCommit);
 			/* dump rels to delete */
 			if (nrels > 0)
-			{
-				rdata[0].next = &(rdata[1]);
-				rdata[1].data = (char *) rels;
-				rdata[1].len = nrels * sizeof(RelFileNode);
-				rdata[1].buffer = InvalidBuffer;
-				lastrdata = 1;
-			}
+				XLogRegisterData((char *) rels,
+								 nrels * sizeof(RelFileNode));
 			/* dump committed child Xids */
 			if (nchildren > 0)
-			{
-				rdata[lastrdata].next = &(rdata[2]);
-				rdata[2].data = (char *) children;
-				rdata[2].len = nchildren * sizeof(TransactionId);
-				rdata[2].buffer = InvalidBuffer;
-				lastrdata = 2;
-			}
+				XLogRegisterData((char *) children,
+								 nchildren * sizeof(TransactionId));
 			/* dump shared cache invalidation messages */
 			if (nmsgs > 0)
-			{
-				rdata[lastrdata].next = &(rdata[3]);
-				rdata[3].data = (char *) invalMessages;
-				rdata[3].len = nmsgs * sizeof(SharedInvalidationMessage);
-				rdata[3].buffer = InvalidBuffer;
-				lastrdata = 3;
-			}
-			rdata[lastrdata].next = NULL;
-
-			(void) XLogInsert(RM_XACT_ID, XLOG_XACT_COMMIT, rdata);
+				XLogRegisterData((char *) invalMessages,
+								 nmsgs * sizeof(SharedInvalidationMessage));
+			(void) XLogInsert(RM_XACT_ID, XLOG_XACT_COMMIT);
 		}
 		else
 		{
-			XLogRecData rdata[2];
-			int			lastrdata = 0;
 			xl_xact_commit_compact xlrec;
 
 			xlrec.xact_time = xactStopTimestamp;
 			xlrec.nsubxacts = nchildren;
-			rdata[0].data = (char *) (&xlrec);
-			rdata[0].len = MinSizeOfXactCommitCompact;
-			rdata[0].buffer = InvalidBuffer;
+
+			XLogBeginInsert();
+			XLogRegisterData((char *) (&xlrec), MinSizeOfXactCommitCompact);
 			/* dump committed child Xids */
 			if (nchildren > 0)
-			{
-				rdata[0].next = &(rdata[1]);
-				rdata[1].data = (char *) children;
-				rdata[1].len = nchildren * sizeof(TransactionId);
-				rdata[1].buffer = InvalidBuffer;
-				lastrdata = 1;
-			}
-			rdata[lastrdata].next = NULL;
+				XLogRegisterData((char *) children,
+								 nchildren * sizeof(TransactionId));
 
-			(void) XLogInsert(RM_XACT_ID, XLOG_XACT_COMMIT_COMPACT, rdata);
+			(void) XLogInsert(RM_XACT_ID, XLOG_XACT_COMMIT_COMPACT);
 		}
 	}
 
@@ -1436,8 +1403,6 @@ RecordTransactionAbort(bool isSubXact)
 	RelFileNode *rels;
 	int			nchildren;
 	TransactionId *children;
-	XLogRecData rdata[3];
-	int			lastrdata = 0;
 	xl_xact_abort xlrec;
 
 	/*
@@ -1486,30 +1451,20 @@ RecordTransactionAbort(bool isSubXact)
 	}
 	xlrec.nrels = nrels;
 	xlrec.nsubxacts = nchildren;
-	rdata[0].data = (char *) (&xlrec);
-	rdata[0].len = MinSizeOfXactAbort;
-	rdata[0].buffer = InvalidBuffer;
+
+	XLogBeginInsert();
+	XLogRegisterData((char *) (&xlrec), MinSizeOfXactAbort);
+
 	/* dump rels to delete */
 	if (nrels > 0)
-	{
-		rdata[0].next = &(rdata[1]);
-		rdata[1].data = (char *) rels;
-		rdata[1].len = nrels * sizeof(RelFileNode);
-		rdata[1].buffer = InvalidBuffer;
-		lastrdata = 1;
-	}
+		XLogRegisterData((char *) rels, nrels * sizeof(RelFileNode));
+
 	/* dump committed child Xids */
 	if (nchildren > 0)
-	{
-		rdata[lastrdata].next = &(rdata[2]);
-		rdata[2].data = (char *) children;
-		rdata[2].len = nchildren * sizeof(TransactionId);
-		rdata[2].buffer = InvalidBuffer;
-		lastrdata = 2;
-	}
-	rdata[lastrdata].next = NULL;
+		XLogRegisterData((char *) children,
+						 nchildren * sizeof(TransactionId));
 
-	(void) XLogInsert(RM_XACT_ID, XLOG_XACT_ABORT, rdata);
+	(void) XLogInsert(RM_XACT_ID, XLOG_XACT_ABORT);
 
 	/*
 	 * Report the latest async abort LSN, so that the WAL writer knows to
@@ -2351,6 +2306,9 @@ AbortTransaction(void)
 	AbortBufferIO();
 	UnlockBuffers();
 
+	/* Reset WAL record construction state */
+	XLogResetInsertion();
+
 	/*
 	 * Also clean up any open wait for lock, since the lock manager will choke
 	 * if we try to wait for another lock before doing this.
@@ -4299,6 +4257,9 @@ AbortSubTransaction(void)
 	AbortBufferIO();
 	UnlockBuffers();
 
+	/* Reset WAL record construction state */
+	XLogResetInsertion();
+
 	/*
 	 * Also clean up any open wait for lock, since the lock manager will choke
 	 * if we try to wait for another lock before doing this.
@@ -4938,42 +4899,42 @@ xact_redo_abort(xl_xact_abort *xlrec, TransactionId xid)
 }
 
 void
-xact_redo(XLogRecPtr lsn, XLogRecord *record)
+xact_redo(XLogReaderState *record)
 {
-	uint8		info = record->xl_info & ~XLR_INFO_MASK;
+	uint8		info = XLogRecGetInfo(record) & ~XLR_INFO_MASK;
 
 	/* Backup blocks are not used in xact records */
-	Assert(!(record->xl_info & XLR_BKP_BLOCK_MASK));
+	Assert(!XLogRecHasAnyBlockRefs(record));
 
 	if (info == XLOG_XACT_COMMIT_COMPACT)
 	{
 		xl_xact_commit_compact *xlrec = (xl_xact_commit_compact *) XLogRecGetData(record);
 
-		xact_redo_commit_compact(xlrec, record->xl_xid, lsn);
+		xact_redo_commit_compact(xlrec, XLogRecGetXid(record), record->EndRecPtr);
 	}
 	else if (info == XLOG_XACT_COMMIT)
 	{
 		xl_xact_commit *xlrec = (xl_xact_commit *) XLogRecGetData(record);
 
-		xact_redo_commit(xlrec, record->xl_xid, lsn);
+		xact_redo_commit(xlrec, XLogRecGetXid(record), record->EndRecPtr);
 	}
 	else if (info == XLOG_XACT_ABORT)
 	{
 		xl_xact_abort *xlrec = (xl_xact_abort *) XLogRecGetData(record);
 
-		xact_redo_abort(xlrec, record->xl_xid);
+		xact_redo_abort(xlrec, XLogRecGetXid(record));
 	}
 	else if (info == XLOG_XACT_PREPARE)
 	{
 		/* the record contents are exactly the 2PC file */
-		RecreateTwoPhaseFile(record->xl_xid,
-							 XLogRecGetData(record), record->xl_len);
+		RecreateTwoPhaseFile(XLogRecGetXid(record),
+						  XLogRecGetData(record), XLogRecGetDataLen(record));
 	}
 	else if (info == XLOG_XACT_COMMIT_PREPARED)
 	{
 		xl_xact_commit_prepared *xlrec = (xl_xact_commit_prepared *) XLogRecGetData(record);
 
-		xact_redo_commit(&xlrec->crec, xlrec->xid, lsn);
+		xact_redo_commit(&xlrec->crec, xlrec->xid, record->EndRecPtr);
 		RemoveTwoPhaseFile(xlrec->xid, false);
 	}
 	else if (info == XLOG_XACT_ABORT_PREPARED)