Exemple #1
0
/*
 * Get the latestRemovedXid from the heap pages pointed at by the index
 * tuples being deleted. This puts the work for calculating latestRemovedXid
 * into the recovery path rather than the primary path.
 *
 * It's possible that this generates a fair amount of I/O, since an index
 * block may have hundreds of tuples being deleted. Repeat accesses to the
 * same heap blocks are common, though are not yet optimised.
 *
 * XXX optimise later with something like XLogPrefetchBuffer()
 */
static TransactionId
btree_xlog_delete_get_latestRemovedXid(XLogReaderState *record)
{
	xl_btree_delete *xlrec = (xl_btree_delete *) XLogRecGetData(record);
	OffsetNumber *unused;
	Buffer		ibuffer,
				hbuffer;
	Page		ipage,
				hpage;
	RelFileNode rnode;
	BlockNumber blkno;
	ItemId		iitemid,
				hitemid;
	IndexTuple	itup;
	HeapTupleHeader htuphdr;
	BlockNumber hblkno;
	OffsetNumber hoffnum;
	TransactionId latestRemovedXid = InvalidTransactionId;
	int			i;

	/*
	 * If there's nothing running on the standby we don't need to derive a
	 * full latestRemovedXid value, so use a fast path out of here.  This
	 * returns InvalidTransactionId, and so will conflict with all HS
	 * transactions; but since we just worked out that that's zero people,
	 * it's OK.
	 *
	 * XXX There is a race condition here, which is that a new backend might
	 * start just after we look.  If so, it cannot need to conflict, but this
	 * coding will result in throwing a conflict anyway.
	 */
	if (CountDBBackends(InvalidOid) == 0)
		return latestRemovedXid;

	/*
	 * In what follows, we have to examine the previous state of the index
	 * page, as well as the heap page(s) it points to.  This is only valid if
	 * WAL replay has reached a consistent database state; which means that
	 * the preceding check is not just an optimization, but is *necessary*. We
	 * won't have let in any user sessions before we reach consistency.
	 */
	if (!reachedConsistency)
		elog(PANIC, "btree_xlog_delete_get_latestRemovedXid: cannot operate with inconsistent data");

	/*
	 * Get index page.  If the DB is consistent, this should not fail, nor
	 * should any of the heap page fetches below.  If one does, we return
	 * InvalidTransactionId to cancel all HS transactions.  That's probably
	 * overkill, but it's safe, and certainly better than panicking here.
	 */
	XLogRecGetBlockTag(record, 0, &rnode, NULL, &blkno);
	ibuffer = XLogReadBufferExtended(rnode, MAIN_FORKNUM, blkno, RBM_NORMAL);
	if (!BufferIsValid(ibuffer))
		return InvalidTransactionId;
	LockBuffer(ibuffer, BT_READ);
	ipage = (Page) BufferGetPage(ibuffer);

	/*
	 * Loop through the deleted index items to obtain the TransactionId from
	 * the heap items they point to.
	 */
	unused = (OffsetNumber *) ((char *) xlrec + SizeOfBtreeDelete);

	for (i = 0; i < xlrec->nitems; i++)
	{
		/*
		 * Identify the index tuple about to be deleted
		 */
		iitemid = PageGetItemId(ipage, unused[i]);
		itup = (IndexTuple) PageGetItem(ipage, iitemid);

		/*
		 * Locate the heap page that the index tuple points at
		 */
		hblkno = ItemPointerGetBlockNumber(&(itup->t_tid));
		hbuffer = XLogReadBufferExtended(xlrec->hnode, MAIN_FORKNUM, hblkno, RBM_NORMAL);
		if (!BufferIsValid(hbuffer))
		{
			UnlockReleaseBuffer(ibuffer);
			return InvalidTransactionId;
		}
		LockBuffer(hbuffer, BUFFER_LOCK_SHARE);
		hpage = (Page) BufferGetPage(hbuffer);

		/*
		 * Look up the heap tuple header that the index tuple points at by
		 * using the heap node supplied with the xlrec. We can't use
		 * heap_fetch, since it uses ReadBuffer rather than XLogReadBuffer.
		 * Note that we are not looking at tuple data here, just headers.
		 */
		hoffnum = ItemPointerGetOffsetNumber(&(itup->t_tid));
		hitemid = PageGetItemId(hpage, hoffnum);

		/*
		 * Follow any redirections until we find something useful.
		 */
		while (ItemIdIsRedirected(hitemid))
		{
			hoffnum = ItemIdGetRedirect(hitemid);
			hitemid = PageGetItemId(hpage, hoffnum);
			CHECK_FOR_INTERRUPTS();
		}

		/*
		 * If the heap item has storage, then read the header and use that to
		 * set latestRemovedXid.
		 *
		 * Some LP_DEAD items may not be accessible, so we ignore them.
		 */
		if (ItemIdHasStorage(hitemid))
		{
			htuphdr = (HeapTupleHeader) PageGetItem(hpage, hitemid);

			HeapTupleHeaderAdvanceLatestRemovedXid(htuphdr, &latestRemovedXid);
		}
		else if (ItemIdIsDead(hitemid))
		{
			/*
			 * Conjecture: if hitemid is dead then it had xids before the xids
			 * marked on LP_NORMAL items. So we just ignore this item and move
			 * onto the next, for the purposes of calculating
			 * latestRemovedxids.
			 */
		}
		else
			Assert(!ItemIdIsUsed(hitemid));

		UnlockReleaseBuffer(hbuffer);
	}

	UnlockReleaseBuffer(ibuffer);

	/*
	 * If all heap tuples were LP_DEAD then we will be returning
	 * InvalidTransactionId here, which avoids conflicts. This matches
	 * existing logic which assumes that LP_DEAD tuples must already be older
	 * than the latestRemovedXid on the cleanup record that set them as
	 * LP_DEAD, hence must already have generated a conflict.
	 */
	return latestRemovedXid;
}
Exemple #2
0
/*
 * Prune specified item pointer or a HOT chain originating at that item.
 *
 * If the item is an index-referenced tuple (i.e. not a heap-only tuple),
 * the HOT chain is pruned by removing all DEAD tuples at the start of the HOT
 * chain.  We also prune any RECENTLY_DEAD tuples preceding a DEAD tuple.
 * This is OK because a RECENTLY_DEAD tuple preceding a DEAD tuple is really
 * DEAD, the OldestXmin test is just too coarse to detect it.
 *
 * The root line pointer is redirected to the tuple immediately after the
 * latest DEAD tuple.  If all tuples in the chain are DEAD, the root line
 * pointer is marked LP_DEAD.  (This includes the case of a DEAD simple
 * tuple, which we treat as a chain of length 1.)
 *
 * OldestXmin is the cutoff XID used to identify dead tuples.
 *
 * We don't actually change the page here, except perhaps for hint-bit updates
 * caused by HeapTupleSatisfiesVacuum.	We just add entries to the arrays in
 * prstate showing the changes to be made.	Items to be redirected are added
 * to the redirected[] array (two entries per redirection); items to be set to
 * LP_DEAD state are added to nowdead[]; and items to be set to LP_UNUSED
 * state are added to nowunused[].
 *
 * If redirect_move is true, we intend to get rid of redirecting line pointers,
 * not just make redirection entries.
 *
 * Returns the number of tuples (to be) deleted from the page.
 */
static int
heap_prune_chain(Relation relation, Buffer buffer, OffsetNumber rootoffnum,
				 TransactionId OldestXmin,
				 PruneState *prstate,
				 bool redirect_move)
{
	int			ndeleted = 0;
	Page		dp = (Page) BufferGetPage(buffer);
	TransactionId priorXmax = InvalidTransactionId;
	ItemId		rootlp;
	HeapTupleHeader htup;
	OffsetNumber latestdead = InvalidOffsetNumber,
				redirect_target = InvalidOffsetNumber,
				maxoff = PageGetMaxOffsetNumber(dp),
				offnum;
	OffsetNumber chainitems[MaxHeapTuplesPerPage];
	int			nchain = 0,
				i;

	rootlp = PageGetItemId(dp, rootoffnum);

	/*
	 * If it's a heap-only tuple, then it is not the start of a HOT chain.
	 */
	if (ItemIdIsNormal(rootlp))
	{
		htup = (HeapTupleHeader) PageGetItem(dp, rootlp);
		if (HeapTupleHeaderIsHeapOnly(htup))
		{
			/*
			 * If the tuple is DEAD and doesn't chain to anything else, mark
			 * it unused immediately.  (If it does chain, we can only remove
			 * it as part of pruning its chain.)
			 *
			 * We need this primarily to handle aborted HOT updates, that is,
			 * XMIN_INVALID heap-only tuples.  Those might not be linked to by
			 * any chain, since the parent tuple might be re-updated before
			 * any pruning occurs.	So we have to be able to reap them
			 * separately from chain-pruning.  (Note that
			 * HeapTupleHeaderIsHotUpdated will never return true for an
			 * XMIN_INVALID tuple, so this code will work even when there were
			 * sequential updates within the aborted transaction.)
			 *
			 * Note that we might first arrive at a dead heap-only tuple
			 * either here or while following a chain below.  Whichever path
			 * gets there first will mark the tuple unused.
			 */
			if (HeapTupleSatisfiesVacuum(relation, htup, OldestXmin, buffer)
				== HEAPTUPLE_DEAD && !HeapTupleHeaderIsHotUpdated(htup))
			{
				heap_prune_record_unused(prstate, rootoffnum);
				ndeleted++;
			}

			/* Nothing more to do */
			return ndeleted;
		}
	}

	/* Start from the root tuple */
	offnum = rootoffnum;

	/* while not end of the chain */
	for (;;)
	{
		ItemId		lp;
		bool		tupdead,
					recent_dead;

		/* Some sanity checks */
		if (offnum < FirstOffsetNumber || offnum > maxoff)
			break;

		/* If item is already processed, stop --- it must not be same chain */
		if (prstate->marked[offnum])
			break;

		lp = PageGetItemId(dp, offnum);

		/* Unused item obviously isn't part of the chain */
		if (!ItemIdIsUsed(lp))
			break;

		/*
		 * If we are looking at the redirected root line pointer, jump to the
		 * first normal tuple in the chain.  If we find a redirect somewhere
		 * else, stop --- it must not be same chain.
		 */
		if (ItemIdIsRedirected(lp))
		{
			if (nchain > 0)
				break;			/* not at start of chain */
			chainitems[nchain++] = offnum;
			offnum = ItemIdGetRedirect(rootlp);
			continue;
		}

		/*
		 * Likewise, a dead item pointer can't be part of the chain. (We
		 * already eliminated the case of dead root tuple outside this
		 * function.)
		 */
		if (ItemIdIsDead(lp))
			break;

		Assert(ItemIdIsNormal(lp));
		htup = (HeapTupleHeader) PageGetItem(dp, lp);

		/*
		 * Check the tuple XMIN against prior XMAX, if any
		 */
		if (TransactionIdIsValid(priorXmax) &&
			!TransactionIdEquals(HeapTupleHeaderGetXmin(htup), priorXmax))
			break;

		/*
		 * OK, this tuple is indeed a member of the chain.
		 */
		chainitems[nchain++] = offnum;

		/*
		 * Check tuple's visibility status.
		 */
		tupdead = recent_dead = false;

		switch (HeapTupleSatisfiesVacuum(relation, htup, OldestXmin, buffer))
		{
			case HEAPTUPLE_DEAD:
				tupdead = true;
				break;

			case HEAPTUPLE_RECENTLY_DEAD:
				recent_dead = true;

				/*
				 * This tuple may soon become DEAD.  Update the hint field so
				 * that the page is reconsidered for pruning in future.
				 */
				heap_prune_record_prunable(prstate,
										   HeapTupleHeaderGetXmax(htup));
				break;

			case HEAPTUPLE_DELETE_IN_PROGRESS:

				/*
				 * This tuple may soon become DEAD.  Update the hint field so
				 * that the page is reconsidered for pruning in future.
				 */
				heap_prune_record_prunable(prstate,
										   HeapTupleHeaderGetXmax(htup));
				break;

			case HEAPTUPLE_LIVE:
			case HEAPTUPLE_INSERT_IN_PROGRESS:

				/*
				 * If we wanted to optimize for aborts, we might consider
				 * marking the page prunable when we see INSERT_IN_PROGRESS.
				 * But we don't.  See related decisions about when to mark the
				 * page prunable in heapam.c.
				 */
				break;

			default:
				elog(ERROR, "unexpected HeapTupleSatisfiesVacuum result");
				break;
		}

		/*
		 * Remember the last DEAD tuple seen.  We will advance past
		 * RECENTLY_DEAD tuples just in case there's a DEAD one after them;
		 * but we can't advance past anything else.  (XXX is it really worth
		 * continuing to scan beyond RECENTLY_DEAD?  The case where we will
		 * find another DEAD tuple is a fairly unusual corner case.)
		 */
		if (tupdead)
			latestdead = offnum;
		else if (!recent_dead)
			break;

		/*
		 * If the tuple is not HOT-updated, then we are at the end of this
		 * HOT-update chain.
		 */
		if (!HeapTupleHeaderIsHotUpdated(htup))
			break;

		/*
		 * Advance to next chain member.
		 */
		Assert(ItemPointerGetBlockNumber(&htup->t_ctid) ==
			   BufferGetBlockNumber(buffer));
		offnum = ItemPointerGetOffsetNumber(&htup->t_ctid);
		priorXmax = HeapTupleHeaderGetXmax(htup);
	}

	/*
	 * If we found a DEAD tuple in the chain, adjust the HOT chain so that all
	 * the DEAD tuples at the start of the chain are removed and the root line
	 * pointer is appropriately redirected.
	 */
	if (OffsetNumberIsValid(latestdead))
	{
		/*
		 * Mark as unused each intermediate item that we are able to remove
		 * from the chain.
		 *
		 * When the previous item is the last dead tuple seen, we are at the
		 * right candidate for redirection.
		 */
		for (i = 1; (i < nchain) && (chainitems[i - 1] != latestdead); i++)
		{
			heap_prune_record_unused(prstate, chainitems[i]);
			ndeleted++;
		}

		/*
		 * If the root entry had been a normal tuple, we are deleting it, so
		 * count it in the result.	But changing a redirect (even to DEAD
		 * state) doesn't count.
		 */
		if (ItemIdIsNormal(rootlp))
			ndeleted++;

		/*
		 * If the DEAD tuple is at the end of the chain, the entire chain is
		 * dead and the root line pointer can be marked dead.  Otherwise just
		 * redirect the root to the correct chain member.
		 */
		if (i >= nchain)
			heap_prune_record_dead(prstate, rootoffnum);
		else
		{
			heap_prune_record_redirect(prstate, rootoffnum, chainitems[i]);
			/* If the redirection will be a move, need more processing */
			if (redirect_move)
				redirect_target = chainitems[i];
		}
	}
	else if (nchain < 2 && ItemIdIsRedirected(rootlp))
	{
		/*
		 * We found a redirect item that doesn't point to a valid follow-on
		 * item.  This can happen if the loop in heap_page_prune caused us to
		 * visit the dead successor of a redirect item before visiting the
		 * redirect item.  We can clean up by setting the redirect item to
		 * DEAD state.
		 */
		heap_prune_record_dead(prstate, rootoffnum);
	}
	else if (redirect_move && ItemIdIsRedirected(rootlp))
	{
		/*
		 * If we desire to eliminate LP_REDIRECT items by moving tuples, make
		 * a redirection entry for each redirected root item; this will cause
		 * heap_page_prune_execute to actually do the move. (We get here only
		 * when there are no DEAD tuples in the chain; otherwise the
		 * redirection entry was made above.)
		 */
		heap_prune_record_redirect(prstate, rootoffnum, chainitems[1]);
		redirect_target = chainitems[1];
	}

	/*
	 * If we are going to implement a redirect by moving tuples, we have to
	 * issue a cache invalidation against the redirection target tuple,
	 * because its CTID will be effectively changed by the move.  Note that
	 * CacheInvalidateHeapTuple only queues the request, it doesn't send it;
	 * if we fail before reaching EndNonTransactionalInvalidation, nothing
	 * happens and no harm is done.
	 */
	if (OffsetNumberIsValid(redirect_target))
	{
		ItemId		firstlp = PageGetItemId(dp, redirect_target);
		HeapTupleData firsttup;

		Assert(ItemIdIsNormal(firstlp));
		/* Set up firsttup to reference the tuple at its existing CTID */
		firsttup.t_data = (HeapTupleHeader) PageGetItem(dp, firstlp);
		firsttup.t_len = ItemIdGetLength(firstlp);
		ItemPointerSet(&firsttup.t_self,
					   BufferGetBlockNumber(buffer),
					   redirect_target);
		CacheInvalidateHeapTuple(relation, &firsttup);
	}

	return ndeleted;
}
Exemple #3
0
/*
 * For all items in this page, find their respective root line pointers.
 * If item k is part of a HOT-chain with root at item j, then we set
 * root_offsets[k - 1] = j.
 *
 * The passed-in root_offsets array must have MaxHeapTuplesPerPage entries.
 * We zero out all unused entries.
 *
 * The function must be called with at least share lock on the buffer, to
 * prevent concurrent prune operations.
 *
 * Note: The information collected here is valid only as long as the caller
 * holds a pin on the buffer. Once pin is released, a tuple might be pruned
 * and reused by a completely unrelated tuple.
 */
void
heap_get_root_tuples(Page page, OffsetNumber *root_offsets)
{
	OffsetNumber offnum,
				maxoff;

	MemSet(root_offsets, 0, MaxHeapTuplesPerPage * sizeof(OffsetNumber));

	maxoff = PageGetMaxOffsetNumber(page);
	for (offnum = FirstOffsetNumber; offnum <= maxoff; offnum = OffsetNumberNext(offnum))
	{
		ItemId		lp = PageGetItemId(page, offnum);
		HeapTupleHeader htup;
		OffsetNumber nextoffnum;
		TransactionId priorXmax;

		/* skip unused and dead items */
		if (!ItemIdIsUsed(lp) || ItemIdIsDead(lp))
			continue;

		if (ItemIdIsNormal(lp))
		{
			htup = (HeapTupleHeader) PageGetItem(page, lp);

			/*
			 * Check if this tuple is part of a HOT-chain rooted at some other
			 * tuple. If so, skip it for now; we'll process it when we find
			 * its root.
			 */
			if (HeapTupleHeaderIsHeapOnly(htup))
				continue;

			/*
			 * This is either a plain tuple or the root of a HOT-chain.
			 * Remember it in the mapping.
			 */
			root_offsets[offnum - 1] = offnum;

			/* If it's not the start of a HOT-chain, we're done with it */
			if (!HeapTupleHeaderIsHotUpdated(htup))
				continue;

			/* Set up to scan the HOT-chain */
			nextoffnum = ItemPointerGetOffsetNumber(&htup->t_ctid);
			priorXmax = HeapTupleHeaderGetXmax(htup);
		}
		else
		{
			/* Must be a redirect item. We do not set its root_offsets entry */
			Assert(ItemIdIsRedirected(lp));
			/* Set up to scan the HOT-chain */
			nextoffnum = ItemIdGetRedirect(lp);
			priorXmax = InvalidTransactionId;
		}

		/*
		 * Now follow the HOT-chain and collect other tuples in the chain.
		 *
		 * Note: Even though this is a nested loop, the complexity of the
		 * function is O(N) because a tuple in the page should be visited not
		 * more than twice, once in the outer loop and once in HOT-chain
		 * chases.
		 */
		for (;;)
		{
			lp = PageGetItemId(page, nextoffnum);

			/* Check for broken chains */
			if (!ItemIdIsNormal(lp))
				break;

			htup = (HeapTupleHeader) PageGetItem(page, lp);

			if (TransactionIdIsValid(priorXmax) &&
				!TransactionIdEquals(priorXmax, HeapTupleHeaderGetXmin(htup)))
				break;

			/* Remember the root line pointer for this item */
			root_offsets[nextoffnum - 1] = offnum;

			/* Advance to next chain member, if any */
			if (!HeapTupleHeaderIsHotUpdated(htup))
				break;

			nextoffnum = ItemPointerGetOffsetNumber(&htup->t_ctid);
			priorXmax = HeapTupleHeaderGetXmax(htup);
		}
	}
}
Exemple #4
0
/* ----------------
 *		index_getnext - get the next heap tuple from a scan
 *
 * The result is the next heap tuple satisfying the scan keys and the
 * snapshot, or NULL if no more matching tuples exist.	On success,
 * the buffer containing the heap tuple is pinned (the pin will be dropped
 * at the next index_getnext or index_endscan).
 *
 * Note: caller must check scan->xs_recheck, and perform rechecking of the
 * scan keys if required.  We do not do that here because we don't have
 * enough information to do it efficiently in the general case.
 * ----------------
 */
HeapTuple
index_getnext(IndexScanDesc scan, ScanDirection direction)
{
	HeapTuple	heapTuple = &scan->xs_ctup;
	ItemPointer tid = &heapTuple->t_self;
	FmgrInfo   *procedure;

	SCAN_CHECKS;
	GET_SCAN_PROCEDURE(amgettuple);

	Assert(TransactionIdIsValid(RecentGlobalXmin));

	/*
	 * We always reset xs_hot_dead; if we are here then either we are just
	 * starting the scan, or we previously returned a visible tuple, and in
	 * either case it's inappropriate to kill the prior index entry.
	 */
	scan->xs_hot_dead = false;

	for (;;)
	{
		OffsetNumber offnum;
		bool		at_chain_start;
		Page		dp;

		if (scan->xs_next_hot != InvalidOffsetNumber)
		{
			/*
			 * We are resuming scan of a HOT chain after having returned an
			 * earlier member.	Must still hold pin on current heap page.
			 */
			Assert(BufferIsValid(scan->xs_cbuf));
			Assert(ItemPointerGetBlockNumber(tid) ==
				   BufferGetBlockNumber(scan->xs_cbuf));
			Assert(TransactionIdIsValid(scan->xs_prev_xmax));
			offnum = scan->xs_next_hot;
			at_chain_start = false;
			scan->xs_next_hot = InvalidOffsetNumber;
		}
		else
		{
			bool		found;
			Buffer		prev_buf;

			/*
			 * If we scanned a whole HOT chain and found only dead tuples,
			 * tell index AM to kill its entry for that TID. We do not do this
			 * when in recovery because it may violate MVCC to do so. see
			 * comments in RelationGetIndexScan().
			 */
			if (!scan->xactStartedInRecovery)
				scan->kill_prior_tuple = scan->xs_hot_dead;

			/*
			 * The AM's gettuple proc finds the next index entry matching the
			 * scan keys, and puts the TID in xs_ctup.t_self (ie, *tid). It
			 * should also set scan->xs_recheck, though we pay no attention to
			 * that here.
			 */
			found = DatumGetBool(FunctionCall2(procedure,
											   PointerGetDatum(scan),
											   Int32GetDatum(direction)));

			/* Reset kill flag immediately for safety */
			scan->kill_prior_tuple = false;

			/* If we're out of index entries, break out of outer loop */
			if (!found)
				break;

			pgstat_count_index_tuples(scan->indexRelation, 1);

			/* Switch to correct buffer if we don't have it already */
			prev_buf = scan->xs_cbuf;
			scan->xs_cbuf = ReleaseAndReadBuffer(scan->xs_cbuf,
												 scan->heapRelation,
											 ItemPointerGetBlockNumber(tid));

			/*
			 * Prune page, but only if we weren't already on this page
			 */
			if (prev_buf != scan->xs_cbuf)
				heap_page_prune_opt(scan->heapRelation, scan->xs_cbuf,
									RecentGlobalXmin);

			/* Prepare to scan HOT chain starting at index-referenced offnum */
			offnum = ItemPointerGetOffsetNumber(tid);
			at_chain_start = true;

			/* We don't know what the first tuple's xmin should be */
			scan->xs_prev_xmax = InvalidTransactionId;

			/* Initialize flag to detect if all entries are dead */
			scan->xs_hot_dead = true;
		}

		/* Obtain share-lock on the buffer so we can examine visibility */
		LockBuffer(scan->xs_cbuf, BUFFER_LOCK_SHARE);

		dp = (Page) BufferGetPage(scan->xs_cbuf);

		/* Scan through possible multiple members of HOT-chain */
		for (;;)
		{
			ItemId		lp;
			ItemPointer ctid;
			bool		valid;

			/* check for bogus TID */
			if (offnum < FirstOffsetNumber ||
				offnum > PageGetMaxOffsetNumber(dp))
				break;

			lp = PageGetItemId(dp, offnum);

			/* check for unused, dead, or redirected items */
			if (!ItemIdIsNormal(lp))
			{
				/* We should only see a redirect at start of chain */
				if (ItemIdIsRedirected(lp) && at_chain_start)
				{
					/* Follow the redirect */
					offnum = ItemIdGetRedirect(lp);
					at_chain_start = false;
					continue;
				}
				/* else must be end of chain */
				break;
			}

			/*
			 * We must initialize all of *heapTuple (ie, scan->xs_ctup) since
			 * it is returned to the executor on success.
			 */
			heapTuple->t_data = (HeapTupleHeader) PageGetItem(dp, lp);
			heapTuple->t_len = ItemIdGetLength(lp);
			ItemPointerSetOffsetNumber(tid, offnum);
			heapTuple->t_tableOid = RelationGetRelid(scan->heapRelation);
			ctid = &heapTuple->t_data->t_ctid;

			/*
			 * Shouldn't see a HEAP_ONLY tuple at chain start.  (This test
			 * should be unnecessary, since the chain root can't be removed
			 * while we have pin on the index entry, but let's make it
			 * anyway.)
			 */
			if (at_chain_start && HeapTupleIsHeapOnly(heapTuple))
				break;

			/*
			 * The xmin should match the previous xmax value, else chain is
			 * broken.	(Note: this test is not optional because it protects
			 * us against the case where the prior chain member's xmax aborted
			 * since we looked at it.)
			 */
			if (TransactionIdIsValid(scan->xs_prev_xmax) &&
				!TransactionIdEquals(scan->xs_prev_xmax,
								  HeapTupleHeaderGetXmin(heapTuple->t_data)))
				break;

			/* If it's visible per the snapshot, we must return it */
			valid = HeapTupleSatisfiesVisibility(heapTuple, scan->xs_snapshot,
												 scan->xs_cbuf);

			CheckForSerializableConflictOut(valid, scan->heapRelation,
											heapTuple, scan->xs_cbuf);

			if (valid)
			{
				/*
				 * If the snapshot is MVCC, we know that it could accept at
				 * most one member of the HOT chain, so we can skip examining
				 * any more members.  Otherwise, check for continuation of the
				 * HOT-chain, and set state for next time.
				 */
				if (IsMVCCSnapshot(scan->xs_snapshot)
					&& !IsolationIsSerializable())
					scan->xs_next_hot = InvalidOffsetNumber;
				else if (HeapTupleIsHotUpdated(heapTuple))
				{
					Assert(ItemPointerGetBlockNumber(ctid) ==
						   ItemPointerGetBlockNumber(tid));
					scan->xs_next_hot = ItemPointerGetOffsetNumber(ctid);
					scan->xs_prev_xmax = HeapTupleHeaderGetXmax(heapTuple->t_data);
				}
				else
					scan->xs_next_hot = InvalidOffsetNumber;

				PredicateLockTuple(scan->heapRelation, heapTuple);

				LockBuffer(scan->xs_cbuf, BUFFER_LOCK_UNLOCK);

				pgstat_count_heap_fetch(scan->indexRelation);

				return heapTuple;
			}

			/*
			 * If we can't see it, maybe no one else can either.  Check to see
			 * if the tuple is dead to all transactions.  If we find that all
			 * the tuples in the HOT chain are dead, we'll signal the index AM
			 * to not return that TID on future indexscans.
			 */
			if (scan->xs_hot_dead &&
				HeapTupleSatisfiesVacuum(heapTuple->t_data, RecentGlobalXmin,
										 scan->xs_cbuf) != HEAPTUPLE_DEAD)
				scan->xs_hot_dead = false;

			/*
			 * Check to see if HOT chain continues past this tuple; if so
			 * fetch the next offnum (we don't bother storing it into
			 * xs_next_hot, but must store xs_prev_xmax), and loop around.
			 */
			if (HeapTupleIsHotUpdated(heapTuple))
			{
				Assert(ItemPointerGetBlockNumber(ctid) ==
					   ItemPointerGetBlockNumber(tid));
				offnum = ItemPointerGetOffsetNumber(ctid);
				at_chain_start = false;
				scan->xs_prev_xmax = HeapTupleHeaderGetXmax(heapTuple->t_data);
			}
			else
				break;			/* end of chain */
		}						/* loop over a single HOT chain */

		LockBuffer(scan->xs_cbuf, BUFFER_LOCK_UNLOCK);

		/* Loop around to ask index AM for another TID */
		scan->xs_next_hot = InvalidOffsetNumber;
	}

	/* Release any held pin on a heap page */
	if (BufferIsValid(scan->xs_cbuf))
	{
		ReleaseBuffer(scan->xs_cbuf);
		scan->xs_cbuf = InvalidBuffer;
	}

	return NULL;				/* failure exit */
}