Refactor inCommit flag into generic delayChkpt flag.

Rename PGXACT->inCommit flag into delayChkpt flag,
and generalise comments to allow use in other situations,
such as the forthcoming potential use in checksum patch.
Replace wait loop to look for VXIDs with delayChkpt set.
No user visible changes, not behaviour changes at present.

Simon Riggs, reviewed and rebased by Jeff Davis

src/backend/access/transam/twophase.c

@@ -318,7 +318,7 @@ MarkAsPreparing(TransactionId xid, const char *gid,
 	proc->lxid = (LocalTransactionId) xid;
 	pgxact->xid = xid;
 	pgxact->xmin = InvalidTransactionId;
-	pgxact->inCommit = false;
+	pgxact->delayChkpt = false;
 	pgxact->vacuumFlags = 0;
 	proc->pid = 0;
 	proc->backendId = InvalidBackendId;
@@ -1034,18 +1034,18 @@ EndPrepare(GlobalTransaction gxact)
 	 * odds of a PANIC actually occurring should be very tiny given that we
 	 * were able to write the bogus CRC above.
 	 *
-	 * We have to set inCommit here, too; otherwise a checkpoint starting
+	 * We have to set delayChkpt here, too; otherwise a checkpoint starting
 	 * immediately after the WAL record is inserted could complete without
 	 * fsync'ing our state file.  (This is essentially the same kind of race
 	 * condition as the COMMIT-to-clog-write case that RecordTransactionCommit
-	 * uses inCommit for; see notes there.)
+	 * uses delayChkpt for; see notes there.)
 	 *
 	 * We save the PREPARE record's location in the gxact for later use by
 	 * CheckPointTwoPhase.
 	 */
 	START_CRIT_SECTION();
 
-	MyPgXact->inCommit = true;
+	MyPgXact->delayChkpt = true;
 
 	gxact->prepare_lsn = XLogInsert(RM_XACT_ID, XLOG_XACT_PREPARE,
 									records.head);
@@ -1086,7 +1086,7 @@ EndPrepare(GlobalTransaction gxact)
 	 * checkpoint starting after this will certainly see the gxact as a
 	 * candidate for fsyncing.
 	 */
-	MyPgXact->inCommit = false;
+	MyPgXact->delayChkpt = false;
 
 	END_CRIT_SECTION();
 
@@ -1972,7 +1972,7 @@ RecoverPreparedTransactions(void)
  *	RecordTransactionCommitPrepared
  *
  * This is basically the same as RecordTransactionCommit: in particular,
- * we must set the inCommit flag to avoid a race condition.
+ * we must set the delayChkpt flag to avoid a race condition.
  *
  * We know the transaction made at least one XLOG entry (its PREPARE),
  * so it is never possible to optimize out the commit record.
@@ -1995,7 +1995,7 @@ RecordTransactionCommitPrepared(TransactionId xid,
 	START_CRIT_SECTION();
 
 	/* See notes in RecordTransactionCommit */
-	MyPgXact->inCommit = true;
+	MyPgXact->delayChkpt = true;
 
 	/* Emit the XLOG commit record */
 	xlrec.xid = xid;
@@ -2053,7 +2053,7 @@ RecordTransactionCommitPrepared(TransactionId xid,
 	TransactionIdCommitTree(xid, nchildren, children);
 
 	/* Checkpoint can proceed now */
-	MyPgXact->inCommit = false;
+	MyPgXact->delayChkpt = false;
 
 	END_CRIT_SECTION();
 

src/backend/access/transam/xact.c

@@ -1001,13 +1001,13 @@ RecordTransactionCommit(void)
 		 * RecordTransactionAbort.	That's because loss of a transaction abort
 		 * is noncritical; the presumption would be that it aborted, anyway.
 		 *
-		 * It's safe to change the inCommit flag of our own backend without
+		 * It's safe to change the delayChkpt flag of our own backend without
 		 * holding the ProcArrayLock, since we're the only one modifying it.
-		 * This makes checkpoint's determination of which xacts are inCommit a
+		 * This makes checkpoint's determination of which xacts are delayChkpt a
 		 * bit fuzzy, but it doesn't matter.
 		 */
 		START_CRIT_SECTION();
-		MyPgXact->inCommit = true;
+		MyPgXact->delayChkpt = true;
 
 		SetCurrentTransactionStopTimestamp();
 
@@ -1160,7 +1160,7 @@ RecordTransactionCommit(void)
 	 */
 	if (markXidCommitted)
 	{
-		MyPgXact->inCommit = false;
+		MyPgXact->delayChkpt = false;
 		END_CRIT_SECTION();
 	}
 

src/backend/access/transam/xlog.c

@@ -6884,8 +6884,8 @@ CreateCheckPoint(int flags)
 	XLogRecData rdata;
 	uint32		freespace;
 	XLogSegNo	_logSegNo;
-	TransactionId *inCommitXids;
-	int			nInCommit;
+	VirtualTransactionId *vxids;
+	int	nvxids;
 
 	/*
 	 * An end-of-recovery checkpoint is really a shutdown checkpoint, just
@@ -7056,9 +7056,14 @@ CreateCheckPoint(int flags)
 	TRACE_POSTGRESQL_CHECKPOINT_START(flags);
 
 	/*
-	 * Before flushing data, we must wait for any transactions that are
-	 * currently in their commit critical sections.  If an xact inserted its
-	 * commit record into XLOG just before the REDO point, then a crash
+	 * In some cases there are groups of actions that must all occur on
+	 * one side or the other of a checkpoint record. Before flushing the
+	 * checkpoint record we must explicitly wait for any backend currently
+	 * performing those groups of actions.
+	 *
+	 * One example is end of transaction, so we must wait for any transactions
+	 * that are currently in commit critical sections.  If an xact inserted
+	 * its commit record into XLOG just before the REDO point, then a crash
 	 * restart from the REDO point would not replay that record, which means
 	 * that our flushing had better include the xact's update of pg_clog.  So
 	 * we wait till he's out of his commit critical section before proceeding.
@@ -7073,21 +7078,24 @@ CreateCheckPoint(int flags)
 	 * protected by different locks, but again that seems best on grounds of
 	 * minimizing lock contention.)
 	 *
-	 * A transaction that has not yet set inCommit when we look cannot be at
+	 * A transaction that has not yet set delayChkpt when we look cannot be at
 	 * risk, since he's not inserted his commit record yet; and one that's
 	 * already cleared it is not at risk either, since he's done fixing clog
 	 * and we will correctly flush the update below.  So we cannot miss any
 	 * xacts we need to wait for.
 	 */
-	nInCommit = GetTransactionsInCommit(&inCommitXids);
-	if (nInCommit > 0)
+	vxids = GetVirtualXIDsDelayingChkpt(&nvxids);
+	if (nvxids > 0)
 	{
+		uint	nwaits = 0;
+
 		do
 		{
 			pg_usleep(10000L);	/* wait for 10 msec */
-		} while (HaveTransactionsInCommit(inCommitXids, nInCommit));
+			nwaits++;
+		} while (HaveVirtualXIDsDelayingChkpt(vxids, nvxids));
 	}
-	pfree(inCommitXids);
+	pfree(vxids);
 
 	/*
 	 * Get the other info we need for the checkpoint record.