HOT updates. When we update a tuple without changing any of its indexed

columns, and the new version can be stored on the same heap page, we no longer
generate extra index entries for the new version.  Instead, index searches
follow the HOT-chain links to ensure they find the correct tuple version.

In addition, this patch introduces the ability to "prune" dead tuples on a
per-page basis, without having to do a complete VACUUM pass to recover space.
VACUUM is still needed to clean up dead index entries, however.

Pavan Deolasee, with help from a bunch of other people.

4 files changed, 57 insertions, 63 deletions

diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index 7e5873b89df..485f6ddc1ee 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -26,7 +26,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/executor/execMain.c,v 1.297 2007/09/07 20:59:26 tgl Exp $
+ *	  $PostgreSQL: pgsql/src/backend/executor/execMain.c,v 1.298 2007/09/20 17:56:31 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -1813,8 +1813,10 @@ lreplace:;
 	 *
 	 * Note: heap_update returns the tid (location) of the new tuple in the
 	 * t_self field.
+	 *
+	 * If it's a HOT update, we mustn't insert new index entries.
 	 */
-	if (resultRelInfo->ri_NumIndices > 0)
+	if (resultRelInfo->ri_NumIndices > 0 && !HeapTupleIsHeapOnly(tuple))
 		ExecInsertIndexTuples(slot, &(tuple->t_self), estate, false);
 
 	/* AFTER ROW UPDATE Triggers */
diff --git a/src/backend/executor/execUtils.c b/src/backend/executor/execUtils.c
index 1d478062998..790a9dccc10 100644
--- a/src/backend/executor/execUtils.c
+++ b/src/backend/executor/execUtils.c
@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/executor/execUtils.c,v 1.150 2007/08/15 21:39:50 tgl Exp $
+ *	  $PostgreSQL: pgsql/src/backend/executor/execUtils.c,v 1.151 2007/09/20 17:56:31 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -981,6 +981,10 @@ ExecCloseIndices(ResultRelInfo *resultRelInfo)
  *		stuff as it only exists here because the genam stuff
  *		doesn't provide the functionality needed by the
  *		executor.. -cim 9/27/89
+ *
+ *		CAUTION: this must not be called for a HOT update.
+ *		We can't defend against that here for lack of info.
+ *		Should we change the API to make it safer?
  * ----------------------------------------------------------------
  */
 void
@@ -1029,6 +1033,10 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 
 		indexInfo = indexInfoArray[i];
 
+		/* If the index is marked as read-only, ignore it */
+		if (!indexInfo->ii_ReadyForInserts)
+			continue;
+
 		/* Check for partial index */
 		if (indexInfo->ii_Predicate != NIL)
 		{
diff --git a/src/backend/executor/nodeBitmapHeapscan.c b/src/backend/executor/nodeBitmapHeapscan.c
index f1e30aeb8f0..87e0063a03a 100644
--- a/src/backend/executor/nodeBitmapHeapscan.c
+++ b/src/backend/executor/nodeBitmapHeapscan.c
@@ -21,7 +21,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/executor/nodeBitmapHeapscan.c,v 1.19 2007/09/12 22:10:26 tgl Exp $
+ *	  $PostgreSQL: pgsql/src/backend/executor/nodeBitmapHeapscan.c,v 1.20 2007/09/20 17:56:31 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -240,12 +240,7 @@ bitgetpage(HeapScanDesc scan, TBMIterateResult *tbmres)
 	BlockNumber page = tbmres->blockno;
 	Buffer		buffer;
 	Snapshot	snapshot;
-	Page		dp;
 	int			ntup;
-	int			curslot;
-	int			minslot;
-	int			maxslot;
-	int			maxoff;
 
 	/*
 	 * Acquire pin on the target heap page, trading in any pin we held before.
@@ -258,6 +253,13 @@ bitgetpage(HeapScanDesc scan, TBMIterateResult *tbmres)
 	buffer = scan->rs_cbuf;
 	snapshot = scan->rs_snapshot;
 
+	ntup = 0;
+
+	/*
+	 * Prune and repair fragmentation for the whole page, if possible.
+	 */
+	heap_page_prune_opt(scan->rs_rd, buffer, RecentGlobalXmin);
+
 	/*
 	 * We must hold share lock on the buffer content while examining tuple
 	 * visibility.	Afterwards, however, the tuples we have found to be
@@ -265,71 +267,51 @@ bitgetpage(HeapScanDesc scan, TBMIterateResult *tbmres)
 	 */
 	LockBuffer(buffer, BUFFER_LOCK_SHARE);
 
-	dp = (Page) BufferGetPage(buffer);
-	maxoff = PageGetMaxOffsetNumber(dp);
-
 	/*
-	 * Determine how many entries we need to look at on this page. If the
-	 * bitmap is lossy then we need to look at each physical item pointer;
-	 * otherwise we just look through the offsets listed in tbmres.
+	 * We need two separate strategies for lossy and non-lossy cases.
 	 */
 	if (tbmres->ntuples >= 0)
 	{
-		/* non-lossy case */
-		minslot = 0;
-		maxslot = tbmres->ntuples - 1;
-	}
-	else
-	{
-		/* lossy case */
-		minslot = FirstOffsetNumber;
-		maxslot = maxoff;
-	}
-
-	ntup = 0;
-	for (curslot = minslot; curslot <= maxslot; curslot++)
-	{
-		OffsetNumber targoffset;
-		ItemId		lp;
-		HeapTupleData loctup;
-		bool		valid;
-
-		if (tbmres->ntuples >= 0)
-		{
-			/* non-lossy case */
-			targoffset = tbmres->offsets[curslot];
-		}
-		else
-		{
-			/* lossy case */
-			targoffset = (OffsetNumber) curslot;
-		}
-
 		/*
-		 * We'd better check for out-of-range offnum in case of VACUUM since
-		 * the TID was obtained.
+		 * Bitmap is non-lossy, so we just look through the offsets listed in
+		 * tbmres; but we have to follow any HOT chain starting at each such
+		 * offset.
 		 */
-		if (targoffset < FirstOffsetNumber || targoffset > maxoff)
-			continue;
+		int curslot;
 
-		lp = PageGetItemId(dp, targoffset);
+		for (curslot = 0; curslot < tbmres->ntuples; curslot++)
+		{
+			OffsetNumber offnum = tbmres->offsets[curslot];
+			ItemPointerData tid;
 
+			ItemPointerSet(&tid, page, offnum);
+			if (heap_hot_search_buffer(&tid, buffer, snapshot, NULL))
+				scan->rs_vistuples[ntup++] = ItemPointerGetOffsetNumber(&tid);
+		}
+	}
+	else
+	{
 		/*
-		 * Must check for deleted tuple.
+		 * Bitmap is lossy, so we must examine each item pointer on the page.
+		 * But we can ignore HOT chains, since we'll check each tuple anyway.
 		 */
-		if (!ItemIdIsNormal(lp))
-			continue;
+		Page		dp = (Page) BufferGetPage(buffer);
+		OffsetNumber maxoff = PageGetMaxOffsetNumber(dp);
+		OffsetNumber offnum;
 
-		/*
-		 * check time qualification of tuple, remember it if valid
-		 */
-		loctup.t_data = (HeapTupleHeader) PageGetItem((Page) dp, lp);
-		loctup.t_len = ItemIdGetLength(lp);
-		ItemPointerSet(&(loctup.t_self), page, targoffset);
+		for (offnum = FirstOffsetNumber; offnum <= maxoff; offnum++)
+		{
+			ItemId		lp;
+			HeapTupleData loctup;
 
-		valid = HeapTupleSatisfiesVisibility(&loctup, snapshot, buffer);
-		if (valid)
-			scan->rs_vistuples[ntup++] = targoffset;
+			lp = PageGetItemId(dp, offnum);
+			if (!ItemIdIsNormal(lp))
+				continue;
+			loctup.t_data = (HeapTupleHeader) PageGetItem((Page) dp, lp);
+			loctup.t_len = ItemIdGetLength(lp);
+			if (HeapTupleSatisfiesVisibility(&loctup, snapshot, buffer))
+				scan->rs_vistuples[ntup++] = offnum;
+		}
 	}
 
 	LockBuffer(buffer, BUFFER_LOCK_UNLOCK);
diff --git a/src/backend/executor/spi.c b/src/backend/executor/spi.c
index af94ad1a3b8..875e4da2914 100644
--- a/src/backend/executor/spi.c
+++ b/src/backend/executor/spi.c
@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/executor/spi.c,v 1.180 2007/08/15 19:15:46 tgl Exp $
+ *	  $PostgreSQL: pgsql/src/backend/executor/spi.c,v 1.181 2007/09/20 17:56:31 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -1407,6 +1407,7 @@ _SPI_prepare_plan(const char *src, SPIPlanPtr plan)
 		plansource->num_params = nargs;
 		plansource->fully_planned = true;
 		plansource->fixed_result = false;
+		/* no need to set search_path, generation or saved_xmin */
 		plansource->resultDesc = PlanCacheComputeResultDesc(stmt_list);
 		plansource->plan = cplan;
 
@@ -1973,6 +1974,7 @@ _SPI_copy_plan(SPIPlanPtr plan, MemoryContext parentcxt)
 		newsource->num_params = newplan->nargs;
 		newsource->fully_planned = plansource->fully_planned;
 		newsource->fixed_result = plansource->fixed_result;
+		/* no need to worry about seach_path, generation or saved_xmin */
 		if (plansource->resultDesc)
 			newsource->resultDesc = CreateTupleDescCopy(plansource->resultDesc);
 		newsource->plan = newcplan;