Change the way we mark tuples as frozen.

Instead of changing the tuple xmin to FrozenTransactionId, the combination
of HEAP_XMIN_COMMITTED and HEAP_XMIN_INVALID, which were previously never
set together, is now defined as HEAP_XMIN_FROZEN.  A variety of previous
proposals to freeze tuples opportunistically before vacuum_freeze_min_age
is reached have foundered on the objection that replacing xmin by
FrozenTransactionId might hinder debugging efforts when things in this
area go awry; this patch is intended to solve that problem by keeping
the XID around (but largely ignoring the value to which it is set).

Third-party code that checks for HEAP_XMIN_INVALID on tuples where
HEAP_XMIN_COMMITTED might be set will be broken by this change.  To fix,
use the new accessor macros in htup_details.h rather than consulting the
bits directly.  HeapTupleHeaderGetXmin has been modified to return
FrozenTransactionId when the infomask bits indicate that the tuple is
frozen; use HeapTupleHeaderGetRawXmin when you already know that the
tuple isn't marked commited or frozen, or want the raw value anyway.
We currently do this in routines that display the xmin for user consumption,
in tqual.c where it's known to be safe and important for the avoidance of
extra cycles, and in the function-caching code for various procedural
languages, which shouldn't invalidate the cache just because the tuple
gets frozen.

Robert Haas and Andres Freund

Author: robertmhaas

contrib/pageinspect/heapfuncs.c

@@ -162,7 +162,7 @@ heap_page_items(PG_FUNCTION_ARGS)
 
 			tuphdr = (HeapTupleHeader) PageGetItem(page, id);
 
-			values[4] = UInt32GetDatum(HeapTupleHeaderGetXmin(tuphdr));
+			values[4] = UInt32GetDatum(HeapTupleHeaderGetRawXmin(tuphdr));
 			values[5] = UInt32GetDatum(HeapTupleHeaderGetRawXmax(tuphdr));
 			values[6] = UInt32GetDatum(HeapTupleHeaderGetRawCommandId(tuphdr)); /* shared with xvac */
 			values[7] = PointerGetDatum(&tuphdr->t_ctid);

src/backend/access/common/heaptuple.c

@@ -539,7 +539,7 @@ heap_getsysattr(HeapTuple tup, int attnum, TupleDesc tupleDesc, bool *isnull)
 			result = ObjectIdGetDatum(HeapTupleGetOid(tup));
 			break;
 		case MinTransactionIdAttributeNumber:
-			result = TransactionIdGetDatum(HeapTupleHeaderGetXmin(tup->t_data));
+			result = TransactionIdGetDatum(HeapTupleHeaderGetRawXmin(tup->t_data));
 			break;
 		case MaxTransactionIdAttributeNumber:
 			result = TransactionIdGetDatum(HeapTupleHeaderGetRawXmax(tup->t_data));

src/backend/access/heap/heapam.c

@@ -2257,13 +2257,10 @@ heap_prepare_insert(Relation relation, HeapTuple tup, TransactionId xid,
 	tup->t_data->t_infomask &= ~(HEAP_XACT_MASK);
 	tup->t_data->t_infomask2 &= ~(HEAP2_XACT_MASK);
 	tup->t_data->t_infomask |= HEAP_XMAX_INVALID;
+	HeapTupleHeaderSetXmin(tup->t_data, xid);
 	if (options & HEAP_INSERT_FROZEN)
-	{
-		tup->t_data->t_infomask |= HEAP_XMIN_COMMITTED;
-		HeapTupleHeaderSetXmin(tup->t_data, FrozenTransactionId);
-	}
-	else
-		HeapTupleHeaderSetXmin(tup->t_data, xid);
+		HeapTupleHeaderSetXminFrozen(tup->t_data);
+
 	HeapTupleHeaderSetCmin(tup->t_data, cid);
 	HeapTupleHeaderSetXmax(tup->t_data, 0);		/* for cleanliness */
 	tup->t_tableOid = RelationGetRelid(relation);
@@ -5732,13 +5729,7 @@ heap_prepare_freeze_tuple(HeapTupleHeader tuple, TransactionId cutoff_xid,
 	if (TransactionIdIsNormal(xid) &&
 		TransactionIdPrecedes(xid, cutoff_xid))
 	{
-		frz->frzflags |= XLH_FREEZE_XMIN;
-
-		/*
-		 * Might as well fix the hint bits too; usually XMIN_COMMITTED will
-		 * already be set here, but there's a small chance not.
-		 */
-		frz->t_infomask |= HEAP_XMIN_COMMITTED;
+		frz->t_infomask |= HEAP_XMIN_FROZEN;
 		changed = true;
 	}
 
@@ -5882,9 +5873,6 @@ heap_prepare_freeze_tuple(HeapTupleHeader tuple, TransactionId cutoff_xid,
 void
 heap_execute_freeze_tuple(HeapTupleHeader tuple, xl_heap_freeze_tuple *frz)
 {
-	if (frz->frzflags & XLH_FREEZE_XMIN)
-		HeapTupleHeaderSetXmin(tuple, FrozenTransactionId);
-
 	HeapTupleHeaderSetXmax(tuple, frz->xmax);
 
 	if (frz->frzflags & XLH_FREEZE_XVAC)
@@ -6361,10 +6349,8 @@ HeapTupleHeaderAdvanceLatestRemovedXid(HeapTupleHeader tuple,
 	 * This needs to work on both master and standby, where it is used to
 	 * assess btree delete records.
 	 */
-	if ((tuple->t_infomask & HEAP_XMIN_COMMITTED) ||
-		(!(tuple->t_infomask & HEAP_XMIN_COMMITTED) &&
-		 !(tuple->t_infomask & HEAP_XMIN_INVALID) &&
-		 TransactionIdDidCommit(xmin)))
+	if (HeapTupleHeaderXminCommitted(tuple) ||
+		(!HeapTupleHeaderXminInvalid(tuple) && TransactionIdDidCommit(xmin)))
 	{
 		if (xmax != xmin &&
 			TransactionIdFollows(xmax, *latestRemovedXid))
@@ -6882,7 +6868,7 @@ log_heap_new_cid(Relation relation, HeapTuple tup)
 	if (hdr->t_infomask & HEAP_COMBOCID)
 	{
 		Assert(!(hdr->t_infomask & HEAP_XMAX_INVALID));
-		Assert(!(hdr->t_infomask & HEAP_XMIN_INVALID));
+		Assert(!HeapTupleHeaderXminInvalid(hdr));
 		xlrec.cmin = HeapTupleHeaderGetCmin(hdr);
 		xlrec.cmax = HeapTupleHeaderGetCmax(hdr);
 		xlrec.combocid = HeapTupleHeaderGetRawCommandId(hdr);

src/backend/commands/sequence.c

@@ -361,10 +361,10 @@ fill_seq_with_data(Relation rel, HeapTuple tuple)
 		item = PageGetItem((Page) page, itemId);
 
 		HeapTupleHeaderSetXmin((HeapTupleHeader) item, FrozenTransactionId);
-		((HeapTupleHeader) item)->t_infomask |= HEAP_XMIN_COMMITTED;
+		HeapTupleHeaderSetXminFrozen((HeapTupleHeader) item);
 
 		HeapTupleHeaderSetXmin(tuple->t_data, FrozenTransactionId);
-		tuple->t_data->t_infomask |= HEAP_XMIN_COMMITTED;
+		HeapTupleHeaderSetXminFrozen(tuple->t_data);
 	}
 
 	MarkBufferDirty(buf);

src/backend/commands/vacuumlazy.c

@@ -823,14 +823,14 @@ lazy_scan_heap(Relation onerel, LVRelStats *vacrelstats,
 					 * NB: Like with per-tuple hint bits, we can't set the
 					 * PD_ALL_VISIBLE flag if the inserter committed
 					 * asynchronously. See SetHintBits for more info. Check
-					 * that the HEAP_XMIN_COMMITTED hint bit is set because of
-					 * that.
+					 * that the tuple is hinted xmin-committed because
+					 * of that.
 					 */
 					if (all_visible)
 					{
 						TransactionId xmin;
 
-						if (!(tuple.t_data->t_infomask & HEAP_XMIN_COMMITTED))
+						if (!HeapTupleHeaderXminCommitted(tuple.t_data))
 						{
 							all_visible = false;
 							break;
@@ -1774,7 +1774,7 @@ heap_page_is_all_visible(Relation rel, Buffer buf, TransactionId *visibility_cut
 					TransactionId xmin;
 
 					/* Check comments in lazy_scan_heap. */
-					if (!(tuple.t_data->t_infomask & HEAP_XMIN_COMMITTED))
+					if (!HeapTupleHeaderXminCommitted(tuple.t_data))
 					{
 						all_visible = false;
 						break;

src/backend/storage/buffer/README

@@ -65,6 +65,9 @@ manage to be a conflict it would merely mean that one bit-update would
 be lost and need to be done again later.  These four bits are only hints
 (they cache the results of transaction status lookups in pg_clog), so no
 great harm is done if they get reset to zero by conflicting updates.
+Note, however, that a tuple is frozen by setting both HEAP_XMIN_INVALID
+and HEAP_XMIN_COMMITTED; this is a critical update and accordingly requires
+an exclusive buffer lock (and it must also be WAL-logged).
 
 5. To physically remove a tuple or compact free space on a page, one
 must hold a pin and an exclusive lock, *and* observe while holding the

src/backend/utils/fmgr/fmgr.c

@@ -514,7 +514,7 @@ lookup_C_func(HeapTuple procedureTuple)
 					NULL);
 	if (entry == NULL)
 		return NULL;			/* no such entry */
-	if (entry->fn_xmin == HeapTupleHeaderGetXmin(procedureTuple->t_data) &&
+	if (entry->fn_xmin == HeapTupleHeaderGetRawXmin(procedureTuple->t_data) &&
 		ItemPointerEquals(&entry->fn_tid, &procedureTuple->t_self))
 		return entry;			/* OK */
 	return NULL;				/* entry is out of date */
@@ -552,7 +552,7 @@ record_C_func(HeapTuple procedureTuple,
 					HASH_ENTER,
 					&found);
 	/* OID is already filled in */
-	entry->fn_xmin = HeapTupleHeaderGetXmin(procedureTuple->t_data);
+	entry->fn_xmin = HeapTupleHeaderGetRawXmin(procedureTuple->t_data);
 	entry->fn_tid = procedureTuple->t_self;
 	entry->user_fn = user_fn;
 	entry->inforec = inforec;

src/backend/utils/time/combocid.c

@@ -148,11 +148,11 @@ HeapTupleHeaderAdjustCmax(HeapTupleHeader tup,
 	/*
 	 * If we're marking a tuple deleted that was inserted by (any
 	 * subtransaction of) our transaction, we need to use a combo command id.
-	 * Test for HEAP_XMIN_COMMITTED first, because it's cheaper than a
+	 * Test for HeapTupleHeaderXminCommitted() first, because it's cheaper than a
 	 * TransactionIdIsCurrentTransactionId call.
 	 */
-	if (!(tup->t_infomask & HEAP_XMIN_COMMITTED) &&
-		TransactionIdIsCurrentTransactionId(HeapTupleHeaderGetXmin(tup)))
+	if (!HeapTupleHeaderXminCommitted(tup) &&
+		TransactionIdIsCurrentTransactionId(HeapTupleHeaderGetRawXmin(tup)))
 	{
 		CommandId	cmin = HeapTupleHeaderGetCmin(tup);
 

src/backend/utils/time/tqual.c

@@ -166,9 +166,9 @@ HeapTupleSatisfiesSelf(HeapTuple htup, Snapshot snapshot, Buffer buffer)
 	Assert(ItemPointerIsValid(&htup->t_self));
 	Assert(htup->t_tableOid != InvalidOid);
 
-	if (!(tuple->t_infomask & HEAP_XMIN_COMMITTED))
+	if (!HeapTupleHeaderXminCommitted(tuple))
 	{
-		if (tuple->t_infomask & HEAP_XMIN_INVALID)
+		if (HeapTupleHeaderXminInvalid(tuple))
 			return false;
 
 		/* Used by pre-9.0 binary upgrades */
@@ -210,7 +210,7 @@ HeapTupleSatisfiesSelf(HeapTuple htup, Snapshot snapshot, Buffer buffer)
 				}
 			}
 		}
-		else if (TransactionIdIsCurrentTransactionId(HeapTupleHeaderGetXmin(tuple)))
+		else if (TransactionIdIsCurrentTransactionId(HeapTupleHeaderGetRawXmin(tuple)))
 		{
 			if (tuple->t_infomask & HEAP_XMAX_INVALID)	/* xid invalid */
 				return true;
@@ -244,11 +244,11 @@ HeapTupleSatisfiesSelf(HeapTuple htup, Snapshot snapshot, Buffer buffer)
 
 			return false;
 		}
-		else if (TransactionIdIsInProgress(HeapTupleHeaderGetXmin(tuple)))
+		else if (TransactionIdIsInProgress(HeapTupleHeaderGetRawXmin(tuple)))
 			return false;
-		else if (TransactionIdDidCommit(HeapTupleHeaderGetXmin(tuple)))
+		else if (TransactionIdDidCommit(HeapTupleHeaderGetRawXmin(tuple)))
 			SetHintBits(tuple, buffer, HEAP_XMIN_COMMITTED,
-						HeapTupleHeaderGetXmin(tuple));
+						HeapTupleHeaderGetRawXmin(tuple));
 		else
 		{
 			/* it must have aborted or crashed */
@@ -356,9 +356,9 @@ HeapTupleSatisfiesToast(HeapTuple htup, Snapshot snapshot,
 	Assert(ItemPointerIsValid(&htup->t_self));
 	Assert(htup->t_tableOid != InvalidOid);
 
-	if (!(tuple->t_infomask & HEAP_XMIN_COMMITTED))
+	if (!HeapTupleHeaderXminCommitted(tuple))
 	{
-		if (tuple->t_infomask & HEAP_XMIN_INVALID)
+		if (HeapTupleHeaderXminInvalid(tuple))
 			return false;
 
 		/* Used by pre-9.0 binary upgrades */
@@ -441,9 +441,9 @@ HeapTupleSatisfiesUpdate(HeapTuple htup, CommandId curcid,
 	Assert(ItemPointerIsValid(&htup->t_self));
 	Assert(htup->t_tableOid != InvalidOid);
 
-	if (!(tuple->t_infomask & HEAP_XMIN_COMMITTED))
+	if (!HeapTupleHeaderXminCommitted(tuple))
 	{
-		if (tuple->t_infomask & HEAP_XMIN_INVALID)
+		if (HeapTupleHeaderXminInvalid(tuple))
 			return HeapTupleInvisible;
 
 		/* Used by pre-9.0 binary upgrades */
@@ -485,7 +485,7 @@ HeapTupleSatisfiesUpdate(HeapTuple htup, CommandId curcid,
 				}
 			}
 		}
-		else if (TransactionIdIsCurrentTransactionId(HeapTupleHeaderGetXmin(tuple)))
+		else if (TransactionIdIsCurrentTransactionId(HeapTupleHeaderGetRawXmin(tuple)))
 		{
 			if (HeapTupleHeaderGetCmin(tuple) >= curcid)
 				return HeapTupleInvisible;		/* inserted after scan started */
@@ -564,11 +564,11 @@ HeapTupleSatisfiesUpdate(HeapTuple htup, CommandId curcid,
 			else
 				return HeapTupleInvisible;		/* updated before scan started */
 		}
-		else if (TransactionIdIsInProgress(HeapTupleHeaderGetXmin(tuple)))
+		else if (TransactionIdIsInProgress(HeapTupleHeaderGetRawXmin(tuple)))
 			return HeapTupleInvisible;
-		else if (TransactionIdDidCommit(HeapTupleHeaderGetXmin(tuple)))
+		else if (TransactionIdDidCommit(HeapTupleHeaderGetRawXmin(tuple)))
 			SetHintBits(tuple, buffer, HEAP_XMIN_COMMITTED,
-						HeapTupleHeaderGetXmin(tuple));
+						HeapTupleHeaderGetRawXmin(tuple));
 		else
 		{
 			/* it must have aborted or crashed */
@@ -715,9 +715,9 @@ HeapTupleSatisfiesDirty(HeapTuple htup, Snapshot snapshot,
 
 	snapshot->xmin = snapshot->xmax = InvalidTransactionId;
 
-	if (!(tuple->t_infomask & HEAP_XMIN_COMMITTED))
+	if (!HeapTupleHeaderXminCommitted(tuple))
 	{
-		if (tuple->t_infomask & HEAP_XMIN_INVALID)
+		if (HeapTupleHeaderXminInvalid(tuple))
 			return false;
 
 		/* Used by pre-9.0 binary upgrades */
@@ -759,7 +759,7 @@ HeapTupleSatisfiesDirty(HeapTuple htup, Snapshot snapshot,
 				}
 			}
 		}
-		else if (TransactionIdIsCurrentTransactionId(HeapTupleHeaderGetXmin(tuple)))
+		else if (TransactionIdIsCurrentTransactionId(HeapTupleHeaderGetRawXmin(tuple)))
 		{
 			if (tuple->t_infomask & HEAP_XMAX_INVALID)	/* xid invalid */
 				return true;
@@ -793,15 +793,15 @@ HeapTupleSatisfiesDirty(HeapTuple htup, Snapshot snapshot,
 
 			return false;
 		}
-		else if (TransactionIdIsInProgress(HeapTupleHeaderGetXmin(tuple)))
+		else if (TransactionIdIsInProgress(HeapTupleHeaderGetRawXmin(tuple)))
 		{
-			snapshot->xmin = HeapTupleHeaderGetXmin(tuple);
+			snapshot->xmin = HeapTupleHeaderGetRawXmin(tuple);
 			/* XXX shouldn't we fall through to look at xmax? */
 			return true;		/* in insertion by other */
 		}
-		else if (TransactionIdDidCommit(HeapTupleHeaderGetXmin(tuple)))
+		else if (TransactionIdDidCommit(HeapTupleHeaderGetRawXmin(tuple)))
 			SetHintBits(tuple, buffer, HEAP_XMIN_COMMITTED,
-						HeapTupleHeaderGetXmin(tuple));
+						HeapTupleHeaderGetRawXmin(tuple));
 		else
 		{
 			/* it must have aborted or crashed */
@@ -909,9 +909,9 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot,
 	Assert(ItemPointerIsValid(&htup->t_self));
 	Assert(htup->t_tableOid != InvalidOid);
 
-	if (!(tuple->t_infomask & HEAP_XMIN_COMMITTED))
+	if (!HeapTupleHeaderXminCommitted(tuple))
 	{
-		if (tuple->t_infomask & HEAP_XMIN_INVALID)
+		if (HeapTupleHeaderXminInvalid(tuple))
 			return false;
 
 		/* Used by pre-9.0 binary upgrades */
@@ -953,7 +953,7 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot,
 				}
 			}
 		}
-		else if (TransactionIdIsCurrentTransactionId(HeapTupleHeaderGetXmin(tuple)))
+		else if (TransactionIdIsCurrentTransactionId(HeapTupleHeaderGetRawXmin(tuple)))
 		{
 			if (HeapTupleHeaderGetCmin(tuple) >= snapshot->curcid)
 				return false;	/* inserted after scan started */
@@ -995,11 +995,11 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot,
 			else
 				return false;	/* deleted before scan started */
 		}
-		else if (TransactionIdIsInProgress(HeapTupleHeaderGetXmin(tuple)))
+		else if (TransactionIdIsInProgress(HeapTupleHeaderGetRawXmin(tuple)))
 			return false;
-		else if (TransactionIdDidCommit(HeapTupleHeaderGetXmin(tuple)))
+		else if (TransactionIdDidCommit(HeapTupleHeaderGetRawXmin(tuple)))
 			SetHintBits(tuple, buffer, HEAP_XMIN_COMMITTED,
-						HeapTupleHeaderGetXmin(tuple));
+						HeapTupleHeaderGetRawXmin(tuple));
 		else
 		{
 			/* it must have aborted or crashed */
@@ -1013,7 +1013,8 @@ HeapTupleSatisfiesMVCC(HeapTuple htup, Snapshot snapshot,
 	 * By here, the inserting transaction has committed - have to check
 	 * when...
 	 */
-	if (XidInMVCCSnapshot(HeapTupleHeaderGetXmin(tuple), snapshot))
+	if (!HeapTupleHeaderXminFrozen(tuple)
+		&& XidInMVCCSnapshot(HeapTupleHeaderGetRawXmin(tuple), snapshot))
 		return false;			/* treat as still in progress */
 
 	if (tuple->t_infomask & HEAP_XMAX_INVALID)	/* xid invalid or aborted */
@@ -1116,9 +1117,9 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin,
 	 * If the inserting transaction aborted, then the tuple was never visible
 	 * to any other transaction, so we can delete it immediately.
 	 */
-	if (!(tuple->t_infomask & HEAP_XMIN_COMMITTED))
+	if (!HeapTupleHeaderXminCommitted(tuple))
 	{
-		if (tuple->t_infomask & HEAP_XMIN_INVALID)
+		if (HeapTupleHeaderXminInvalid(tuple))
 			return HEAPTUPLE_DEAD;
 		/* Used by pre-9.0 binary upgrades */
 		else if (tuple->t_infomask & HEAP_MOVED_OFF)
@@ -1157,7 +1158,7 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin,
 				return HEAPTUPLE_DEAD;
 			}
 		}
-		else if (TransactionIdIsInProgress(HeapTupleHeaderGetXmin(tuple)))
+		else if (TransactionIdIsInProgress(HeapTupleHeaderGetRawXmin(tuple)))
 		{
 			if (tuple->t_infomask & HEAP_XMAX_INVALID)	/* xid invalid */
 				return HEAPTUPLE_INSERT_IN_PROGRESS;
@@ -1168,9 +1169,9 @@ HeapTupleSatisfiesVacuum(HeapTuple htup, TransactionId OldestXmin,
 			/* inserted and then deleted by same xact */
 			return HEAPTUPLE_DELETE_IN_PROGRESS;
 		}
-		else if (TransactionIdDidCommit(HeapTupleHeaderGetXmin(tuple)))
+		else if (TransactionIdDidCommit(HeapTupleHeaderGetRawXmin(tuple)))
 			SetHintBits(tuple, buffer, HEAP_XMIN_COMMITTED,
-						HeapTupleHeaderGetXmin(tuple));
+						HeapTupleHeaderGetRawXmin(tuple));
 		else
 		{
 			/*
@@ -1347,8 +1348,8 @@ HeapTupleIsSurelyDead(HeapTuple htup, TransactionId OldestXmin)
 	 * invalid, then we assume it's still alive (since the presumption is that
 	 * all relevant hint bits were just set moments ago).
 	 */
-	if (!(tuple->t_infomask & HEAP_XMIN_COMMITTED))
-		return (tuple->t_infomask & HEAP_XMIN_INVALID) != 0 ? true : false;
+	if (!HeapTupleHeaderXminCommitted(tuple))
+		return HeapTupleHeaderXminInvalid(tuple) ? true : false;
 
 	/*
 	 * If the inserting transaction committed, but any deleting transaction

src/include/access/heapam_xlog.h

@@ -274,7 +274,7 @@ typedef struct xl_heap_inplace
  * This struct represents a 'freeze plan', which is what we need to know about
  * a single tuple being frozen during vacuum.
  */
-#define		XLH_FREEZE_XMIN		0x01
+/* 0x01 was XLH_FREEZE_XMIN */
 #define		XLH_FREEZE_XVAC		0x02
 #define		XLH_INVALID_XVAC	0x04