v25-0008-Push-Bitmap-Table-Scan-prefetch-code-into-heap-A.patch

From b563d330ae16ce60ed3e71f2540cdfceeece7108 Mon Sep 17 00:00:00 2001
From: Melanie Plageman <melanieplageman@gmail.com>
Date: Wed, 9 Oct 2024 14:54:24 -0400
Subject: [PATCH v25 08/11] Push Bitmap Table Scan prefetch code into heap AM

In order to completely remove the layering violation in bitmap table
scan code, we must avoid using the VM for skipping prefetches as well.

To accomplish this, push prefetch code down into the heap implementation
fo the bitmap table scan AM functions. This fixes another layering
violation related to prefetching mentioned in tableam.h.

This commit moves prefetch-related members from the BitmapHeapScanState
to the BitmapHeapScanDesc and localizes the prefetch functions to heap
AM code.

heapam_scan_bitmap_next_block() no longer needs to take blockno as a
parameter because it was only being used in generic bitmap table scan
code to check if the prefetch block was falling behind the current
block.
---
 src/backend/access/heap/heapam.c          |  25 +++-
 src/backend/access/heap/heapam_handler.c  | 142 ++++++++++++++-------
 src/backend/access/table/tableam.c        |   4 +-
 src/backend/executor/nodeBitmapHeapscan.c | 147 ++++++----------------
 src/include/access/heapam.h               |  31 +++--
 src/include/access/relscan.h              |   9 +-
 src/include/access/tableam.h              |  68 +++++-----
 src/include/nodes/execnodes.h             |  14 ---
 8 files changed, 226 insertions(+), 214 deletions(-)

diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c
index 90f6816922c..b8d8d0452aa 100644
--- a/src/backend/access/heap/heapam.c
+++ b/src/backend/access/heap/heapam.c
@@ -1030,7 +1030,8 @@ TableScanDesc
 heap_beginscan(Relation relation, Snapshot snapshot,
 			   int nkeys, ScanKey key,
 			   ParallelTableScanDesc parallel_scan,
-			   uint32 flags)
+			   uint32 flags,
+			   int prefetch_maximum)
 {
 	HeapScanDesc scan;
 
@@ -1050,8 +1051,17 @@ heap_beginscan(Relation relation, Snapshot snapshot,
 	{
 		BitmapHeapScanDesc bscan = palloc(sizeof(BitmapHeapScanDescData));
 
+		bscan->rs_prefetch_blockno = InvalidBlockNumber;
 		bscan->rs_vmbuffer = InvalidBuffer;
+		bscan->rs_pvmbuffer = InvalidBuffer;
 		bscan->rs_empty_tuples_pending = 0;
+
+		/* Only used for serial BHS */
+		bscan->rs_prefetch_pages = 0;
+		bscan->rs_prefetch_target = -1;
+
+		bscan->rs_prefetch_maximum = prefetch_maximum;
+
 		scan = (HeapScanDesc) bscan;
 	}
 	else
@@ -1182,6 +1192,12 @@ heap_rescan(TableScanDesc sscan, ScanKey key, bool set_params,
 	{
 		BitmapHeapScanDesc bscan = (BitmapHeapScanDesc) scan;
 
+		bscan->rs_prefetch_blockno = InvalidBlockNumber;
+
+		/* Only used for serial BHS */
+		bscan->rs_prefetch_pages = 0;
+		bscan->rs_prefetch_target = -1;
+
 		/*
 		 * Reset empty_tuples_pending, a field only used by bitmap heap scan,
 		 * to avoid incorrectly emitting NULL-filled tuples from a previous
@@ -1194,6 +1210,11 @@ heap_rescan(TableScanDesc sscan, ScanKey key, bool set_params,
 			ReleaseBuffer(bscan->rs_vmbuffer);
 			bscan->rs_vmbuffer = InvalidBuffer;
 		}
+		if (BufferIsValid(bscan->rs_pvmbuffer))
+		{
+			ReleaseBuffer(bscan->rs_pvmbuffer);
+			bscan->rs_pvmbuffer = InvalidBuffer;
+		}
 	}
 
 	/*
@@ -1253,6 +1274,8 @@ heap_endscan(TableScanDesc sscan)
 		bscan->rs_empty_tuples_pending = 0;
 		if (BufferIsValid(bscan->rs_vmbuffer))
 			ReleaseBuffer(bscan->rs_vmbuffer);
+		if (BufferIsValid(bscan->rs_pvmbuffer))
+			ReleaseBuffer(bscan->rs_pvmbuffer);
 	}
 
 	pfree(scan);
diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c
index b4a51153d7a..dc4faa0643e 100644
--- a/src/backend/access/heap/heapam_handler.c
+++ b/src/backend/access/heap/heapam_handler.c
@@ -54,6 +54,10 @@ static bool SampleHeapTupleVisible(TableScanDesc scan, Buffer buffer,
 								   HeapTuple tuple,
 								   OffsetNumber tupoffset);
 
+static void BitmapAdjustPrefetchIterator(BitmapHeapScanDesc bscan);
+static void BitmapAdjustPrefetchTarget(BitmapHeapScanDesc bscan);
+static void BitmapPrefetch(BitmapHeapScanDesc bscan);
+
 static BlockNumber heapam_scan_get_blocks_done(HeapScanDesc hscan);
 
 
@@ -2115,7 +2119,7 @@ heapam_estimate_rel_size(Relation rel, int32 *attr_widths,
 
 static bool
 heapam_scan_bitmap_next_block(TableScanDesc scan,
-							  BlockNumber *blockno, bool *recheck,
+							  bool *recheck,
 							  uint64 *lossy_pages, uint64 *exact_pages)
 {
 	BitmapHeapScanDesc bscan = (BitmapHeapScanDesc) scan;
@@ -2131,14 +2135,15 @@ heapam_scan_bitmap_next_block(TableScanDesc scan,
 	hscan->rs_cindex = 0;
 	hscan->rs_ntuples = 0;
 
-	*blockno = InvalidBlockNumber;
 	*recheck = true;
 
+	BitmapAdjustPrefetchIterator(bscan);
+
 	do
 	{
 		CHECK_FOR_INTERRUPTS();
 
-		tbmres = tbm_iterate(&scan->st.rs_tbmiterator);
+		tbmres = tbm_iterate(&scan->st.bitmap.rs_iterator);
 
 		if (tbmres == NULL)
 			return false;
@@ -2155,7 +2160,7 @@ heapam_scan_bitmap_next_block(TableScanDesc scan,
 			 tbmres->blockno >= hscan->rs_nblocks);
 
 	/* Got a valid block */
-	*blockno = tbmres->blockno;
+	block = tbmres->blockno;
 	*recheck = tbmres->recheck;
 
 	/*
@@ -2176,8 +2181,6 @@ heapam_scan_bitmap_next_block(TableScanDesc scan,
 		return true;
 	}
 
-	block = tbmres->blockno;
-
 	/*
 	 * Acquire pin on the target heap page, trading in any pin we held before.
 	 */
@@ -2271,6 +2274,20 @@ heapam_scan_bitmap_next_block(TableScanDesc scan,
 	else
 		(*lossy_pages)++;
 
+	/*
+	 * If private, we can error out if the the prefetch block doesn't stay
+	 * ahead of the current block.
+	 */
+	if (scan->st.bitmap.rs_pstate == NULL &&
+		!tbm_exhausted(&scan->st.bitmap.rs_prefetch_iterator) &&
+		bscan->rs_prefetch_blockno < block)
+		elog(ERROR,
+			 "prefetch and main iterators are out of sync. pblockno: %d. block: %d",
+			 bscan->rs_prefetch_blockno, block);
+
+	/* Adjust the prefetch target */
+	BitmapAdjustPrefetchTarget(bscan);
+
 	/*
 	 * Return true to indicate that a valid block was found and the bitmap is
 	 * not exhausted. If there are no visible tuples on this page,
@@ -2290,6 +2307,9 @@ heapam_scan_bitmap_next_tuple(TableScanDesc scan,
 	OffsetNumber targoffset;
 	Page		page;
 	ItemId		lp;
+#ifdef USE_PREFETCH
+	ParallelBitmapHeapState *pstate = scan->st.bitmap.rs_pstate;
+#endif							/* USE_PREFETCH */
 
 	if (bscan->rs_empty_tuples_pending > 0)
 	{
@@ -2298,6 +2318,7 @@ heapam_scan_bitmap_next_tuple(TableScanDesc scan,
 		 */
 		ExecStoreAllNullTuple(slot);
 		bscan->rs_empty_tuples_pending--;
+		BitmapPrefetch(bscan);
 		return true;
 	}
 
@@ -2308,6 +2329,36 @@ heapam_scan_bitmap_next_tuple(TableScanDesc scan,
 	if (hscan->rs_cindex >= hscan->rs_ntuples)
 		return false;
 
+#ifdef USE_PREFETCH
+
+	/*
+	 * Try to prefetch at least a few pages even before we get to the second
+	 * page if we don't stop reading after the first tuple.
+	 */
+	if (!pstate)
+	{
+		if (bscan->rs_prefetch_target < bscan->rs_prefetch_maximum)
+			bscan->rs_prefetch_target++;
+	}
+	else if (pstate->prefetch_target < bscan->rs_prefetch_maximum)
+	{
+		/* take spinlock while updating shared state */
+		SpinLockAcquire(&pstate->mutex);
+		if (pstate->prefetch_target < bscan->rs_prefetch_maximum)
+			pstate->prefetch_target++;
+		SpinLockRelease(&pstate->mutex);
+	}
+#endif							/* USE_PREFETCH */
+
+	/*
+	 * We issue prefetch requests *after* fetching the current page to try to
+	 * avoid having prefetching interfere with the main I/O. Also, this should
+	 * happen only when we have determined there is still something to do on
+	 * the current page, else we may uselessly prefetch the same page we are
+	 * just about to request for real.
+	 */
+	BitmapPrefetch(bscan);
+
 	targoffset = hscan->rs_vistuples[hscan->rs_cindex];
 	page = BufferGetPage(hscan->rs_cbuf);
 	lp = PageGetItemId(page, targoffset);
@@ -2621,26 +2672,28 @@ SampleHeapTupleVisible(TableScanDesc scan, Buffer buffer,
  *	iterator in prefetch_pages. For each block the main iterator returns, we
  *	decrement prefetch_pages.
  */
-void
-BitmapAdjustPrefetchIterator(BitmapHeapScanState *node)
+static void
+BitmapAdjustPrefetchIterator(BitmapHeapScanDesc bscan)
 {
 #ifdef USE_PREFETCH
-	ParallelBitmapHeapState *pstate = node->pstate;
+	TableScanDesc scan = &(&bscan->rs_heap_base)->rs_base;
+	ParallelBitmapHeapState *pstate = scan->st.bitmap.rs_pstate;
 	TBMIterateResult *tbmpre;
 
 	if (pstate == NULL)
 	{
-		TBMIterator *prefetch_iterator = &node->prefetch_iterator;
+		TBMIterator *prefetch_iterator =
+			&scan->st.bitmap.rs_prefetch_iterator;
 
-		if (node->prefetch_pages > 0)
+		if (bscan->rs_prefetch_pages > 0)
 		{
 			/* The main iterator has closed the distance by one page */
-			node->prefetch_pages--;
+			bscan->rs_prefetch_pages--;
 		}
 		else if (!tbm_exhausted(prefetch_iterator))
 		{
 			tbmpre = tbm_iterate(prefetch_iterator);
-			node->prefetch_blockno = tbmpre ? tbmpre->blockno :
+			bscan->rs_prefetch_blockno = tbmpre ? tbmpre->blockno :
 				InvalidBlockNumber;
 		}
 		return;
@@ -2655,9 +2708,10 @@ BitmapAdjustPrefetchIterator(BitmapHeapScanState *node)
 	 * Note that moving the call site of BitmapAdjustPrefetchIterator()
 	 * exacerbates the effects of this bug.
 	 */
-	if (node->prefetch_maximum > 0)
+	if (bscan->rs_prefetch_maximum > 0)
 	{
-		TBMIterator *prefetch_iterator = &node->prefetch_iterator;
+		TBMIterator *prefetch_iterator =
+			&scan->st.bitmap.rs_prefetch_iterator;
 
 		SpinLockAcquire(&pstate->mutex);
 		if (pstate->prefetch_pages > 0)
@@ -2681,7 +2735,7 @@ BitmapAdjustPrefetchIterator(BitmapHeapScanState *node)
 			if (!tbm_exhausted(prefetch_iterator))
 			{
 				tbmpre = tbm_iterate(prefetch_iterator);
-				node->prefetch_blockno = tbmpre ? tbmpre->blockno :
+				bscan->rs_prefetch_blockno = tbmpre ? tbmpre->blockno :
 					InvalidBlockNumber;
 			}
 		}
@@ -2697,33 +2751,34 @@ BitmapAdjustPrefetchIterator(BitmapHeapScanState *node)
  * page/tuple, then to one after the second tuple is fetched, then
  * it doubles as later pages are fetched.
  */
-void
-BitmapAdjustPrefetchTarget(BitmapHeapScanState *node)
+static void
+BitmapAdjustPrefetchTarget(BitmapHeapScanDesc bscan)
 {
 #ifdef USE_PREFETCH
-	ParallelBitmapHeapState *pstate = node->pstate;
+	TableScanDesc scan = &(&bscan->rs_heap_base)->rs_base;
+	ParallelBitmapHeapState *pstate = scan->st.bitmap.rs_pstate;
 
 	if (pstate == NULL)
 	{
-		if (node->prefetch_target >= node->prefetch_maximum)
+		if (bscan->rs_prefetch_target >= bscan->rs_prefetch_maximum)
 			 /* don't increase any further */ ;
-		else if (node->prefetch_target >= node->prefetch_maximum / 2)
-			node->prefetch_target = node->prefetch_maximum;
-		else if (node->prefetch_target > 0)
-			node->prefetch_target *= 2;
+		else if (bscan->rs_prefetch_target >= bscan->rs_prefetch_maximum / 2)
+			bscan->rs_prefetch_target = bscan->rs_prefetch_maximum;
+		else if (bscan->rs_prefetch_target > 0)
+			bscan->rs_prefetch_target *= 2;
 		else
-			node->prefetch_target++;
+			bscan->rs_prefetch_target++;
 		return;
 	}
 
 	/* Do an unlocked check first to save spinlock acquisitions. */
-	if (pstate->prefetch_target < node->prefetch_maximum)
+	if (pstate->prefetch_target < bscan->rs_prefetch_maximum)
 	{
 		SpinLockAcquire(&pstate->mutex);
-		if (pstate->prefetch_target >= node->prefetch_maximum)
+		if (pstate->prefetch_target >= bscan->rs_prefetch_maximum)
 			 /* don't increase any further */ ;
-		else if (pstate->prefetch_target >= node->prefetch_maximum / 2)
-			pstate->prefetch_target = node->prefetch_maximum;
+		else if (pstate->prefetch_target >= bscan->rs_prefetch_maximum / 2)
+			pstate->prefetch_target = bscan->rs_prefetch_maximum;
 		else if (pstate->prefetch_target > 0)
 			pstate->prefetch_target *= 2;
 		else
@@ -2736,19 +2791,21 @@ BitmapAdjustPrefetchTarget(BitmapHeapScanState *node)
 /*
  * BitmapPrefetch - Prefetch, if prefetch_pages are behind prefetch_target
  */
-void
-BitmapPrefetch(BitmapHeapScanState *node, TableScanDesc scan)
+static void
+BitmapPrefetch(BitmapHeapScanDesc bscan)
 {
 #ifdef USE_PREFETCH
-	ParallelBitmapHeapState *pstate = node->pstate;
+	TableScanDesc scan = &(&bscan->rs_heap_base)->rs_base;
+	ParallelBitmapHeapState *pstate = scan->st.bitmap.rs_pstate;
 
 	if (pstate == NULL)
 	{
-		TBMIterator *prefetch_iterator = &node->prefetch_iterator;
+		TBMIterator *prefetch_iterator =
+			&scan->st.bitmap.rs_prefetch_iterator;
 
 		if (!tbm_exhausted(prefetch_iterator))
 		{
-			while (node->prefetch_pages < node->prefetch_target)
+			while (bscan->rs_prefetch_pages < bscan->rs_prefetch_target)
 			{
 				TBMIterateResult *tbmpre = tbm_iterate(prefetch_iterator);
 				bool		skip_fetch;
@@ -2759,8 +2816,8 @@ BitmapPrefetch(BitmapHeapScanState *node, TableScanDesc scan)
 					tbm_end_iterate(prefetch_iterator);
 					break;
 				}
-				node->prefetch_pages++;
-				node->prefetch_blockno = tbmpre->blockno;
+				bscan->rs_prefetch_pages++;
+				bscan->rs_prefetch_blockno = tbmpre->blockno;
 
 				/*
 				 * If we expect not to have to actually read this heap page,
@@ -2770,9 +2827,9 @@ BitmapPrefetch(BitmapHeapScanState *node, TableScanDesc scan)
 				 */
 				skip_fetch = (!(scan->rs_flags & SO_NEED_TUPLES) &&
 							  !tbmpre->recheck &&
-							  VM_ALL_VISIBLE(node->ss.ss_currentRelation,
+							  VM_ALL_VISIBLE(scan->rs_rd,
 											 tbmpre->blockno,
-											 &node->pvmbuffer));
+											 &bscan->rs_pvmbuffer));
 
 				if (!skip_fetch)
 					PrefetchBuffer(scan->rs_rd, MAIN_FORKNUM,
@@ -2785,7 +2842,8 @@ BitmapPrefetch(BitmapHeapScanState *node, TableScanDesc scan)
 
 	if (pstate->prefetch_pages < pstate->prefetch_target)
 	{
-		TBMIterator *prefetch_iterator = &node->prefetch_iterator;
+		TBMIterator *prefetch_iterator =
+			&scan->st.bitmap.rs_prefetch_iterator;
 
 		if (!tbm_exhausted(prefetch_iterator))
 		{
@@ -2818,14 +2876,14 @@ BitmapPrefetch(BitmapHeapScanState *node, TableScanDesc scan)
 					break;
 				}
 
-				node->prefetch_blockno = tbmpre->blockno;
+				bscan->rs_prefetch_blockno = tbmpre->blockno;
 
 				/* As above, skip prefetch if we expect not to need page */
 				skip_fetch = (!(scan->rs_flags & SO_NEED_TUPLES) &&
 							  !tbmpre->recheck &&
-							  VM_ALL_VISIBLE(node->ss.ss_currentRelation,
+							  VM_ALL_VISIBLE(scan->rs_rd,
 											 tbmpre->blockno,
-											 &node->pvmbuffer));
+											 &bscan->rs_pvmbuffer));
 
 				if (!skip_fetch)
 					PrefetchBuffer(scan->rs_rd, MAIN_FORKNUM,
diff --git a/src/backend/access/table/tableam.c b/src/backend/access/table/tableam.c
index bd8715b6797..3dafcea9649 100644
--- a/src/backend/access/table/tableam.c
+++ b/src/backend/access/table/tableam.c
@@ -117,7 +117,7 @@ table_beginscan_catalog(Relation relation, int nkeys, struct ScanKeyData *key)
 	Snapshot	snapshot = RegisterSnapshot(GetCatalogSnapshot(relid));
 
 	return relation->rd_tableam->scan_begin(relation, snapshot, nkeys, key,
-											NULL, flags);
+											NULL, flags, 0);
 }
 
 
@@ -184,7 +184,7 @@ table_beginscan_parallel(Relation relation, ParallelTableScanDesc pscan)
 	}
 
 	return relation->rd_tableam->scan_begin(relation, snapshot, 0, NULL,
-											pscan, flags);
+											pscan, flags, 0);
 }
 
 
diff --git a/src/backend/executor/nodeBitmapHeapscan.c b/src/backend/executor/nodeBitmapHeapscan.c
index a3381a08790..fcdbbb821ab 100644
--- a/src/backend/executor/nodeBitmapHeapscan.c
+++ b/src/backend/executor/nodeBitmapHeapscan.c
@@ -37,8 +37,6 @@
 
 #include <math.h>
 
-/* XXX: temporary include so prefetch lift and shift works */
-#include "access/heapam.h"
 #include "access/relscan.h"
 #include "access/tableam.h"
 #include "access/visibilitymap.h"
@@ -84,17 +82,26 @@ BitmapHeapNext(BitmapHeapScanState *node)
 	 * If we haven't yet performed the underlying index scan, do it, and begin
 	 * the iteration over the bitmap.
 	 *
-	 * For prefetching, we use *two* iterators, one for the pages we are
-	 * actually scanning and another that runs ahead of the first for
-	 * prefetching.  node->prefetch_pages tracks exactly how many pages ahead
-	 * the prefetch iterator is.  Also, node->prefetch_target tracks the
-	 * desired prefetch distance, which starts small and increases up to the
-	 * node->prefetch_maximum.  This is to avoid doing a lot of prefetching in
-	 * a scan that stops after a few tuples because of a LIMIT.
+	 * While prefetching is the responsibility of the table AM, the prefetch
+	 * iterator is set up here along with the main iterator. Two iterators are
+	 * used -- one for the pages actually being scanned now and the other to
+	 * run ahead and inform the storage layer what blocks to prefetch.
 	 */
 	if (!node->initialized)
 	{
 		TBMIterator tbmiterator;
+		int			prefetch_max;
+		Relation	rel = node->ss.ss_currentRelation;
+#ifdef USE_PREFETCH
+		TBMIterator prefetch_iterator = {0};
+#endif							/* USE_PREFETCH */
+
+		/*
+		 * Maximum number of prefetches for the tablespace if configured,
+		 * otherwise the current value of the effective_io_concurrency GUC.
+		 */
+		prefetch_max =
+			get_tablespace_io_concurrency(rel->rd_rel->reltablespace);
 
 		if (!pstate)
 		{
@@ -104,15 +111,6 @@ BitmapHeapNext(BitmapHeapScanState *node)
 				elog(ERROR, "unrecognized result from subplan");
 
 			node->tbm = tbm;
-
-#ifdef USE_PREFETCH
-			if (node->prefetch_maximum > 0)
-				node->prefetch_iterator =
-					tbm_begin_iterate(node->tbm, dsa,
-									  pstate ?
-									  pstate->prefetch_iterator :
-									  InvalidDsaPointer);
-#endif							/* USE_PREFETCH */
 		}
 		else if (BitmapShouldInitializeSharedState(pstate))
 		{
@@ -135,7 +133,7 @@ BitmapHeapNext(BitmapHeapScanState *node)
 			pstate->tbmiterator = tbm_prepare_shared_iterate(tbm);
 
 #ifdef USE_PREFETCH
-			if (node->prefetch_maximum > 0)
+			if (prefetch_max > 0)
 			{
 				pstate->prefetch_iterator =
 					tbm_prepare_shared_iterate(tbm);
@@ -150,14 +148,12 @@ BitmapHeapNext(BitmapHeapScanState *node)
 										pstate ?
 										pstate->tbmiterator :
 										InvalidDsaPointer);
-
 #ifdef USE_PREFETCH
-		if (node->prefetch_maximum > 0)
-			node->prefetch_iterator =
-				tbm_begin_iterate(tbm, dsa,
-								  pstate ?
-								  pstate->prefetch_iterator :
-								  InvalidDsaPointer);
+		if (prefetch_max > 0)
+			prefetch_iterator = tbm_begin_iterate(tbm, dsa,
+												  pstate ?
+												  pstate->prefetch_iterator :
+												  InvalidDsaPointer);
 #endif							/* USE_PREFETCH */
 
 		/*
@@ -181,14 +177,17 @@ BitmapHeapNext(BitmapHeapScanState *node)
 
 			scan = table_beginscan_bm(node->ss.ss_currentRelation,
 									  node->ss.ps.state->es_snapshot,
+									  pstate,
 									  0,
 									  NULL,
-									  need_tuples);
+									  need_tuples,
+									  prefetch_max);
 
 			node->ss.ss_currentScanDesc = scan;
 		}
 
-		scan->st.rs_tbmiterator = tbmiterator;
+		scan->st.bitmap.rs_iterator = tbmiterator;
+		scan->st.bitmap.rs_prefetch_iterator = prefetch_iterator;
 		node->initialized = true;
 
 		goto new_page;
@@ -204,37 +203,6 @@ BitmapHeapNext(BitmapHeapScanState *node)
 
 			CHECK_FOR_INTERRUPTS();
 
-#ifdef USE_PREFETCH
-
-			/*
-			 * Try to prefetch at least a few pages even before we get to the
-			 * second page if we don't stop reading after the first tuple.
-			 */
-			if (!pstate)
-			{
-				if (node->prefetch_target < node->prefetch_maximum)
-					node->prefetch_target++;
-			}
-			else if (pstate->prefetch_target < node->prefetch_maximum)
-			{
-				/* take spinlock while updating shared state */
-				SpinLockAcquire(&pstate->mutex);
-				if (pstate->prefetch_target < node->prefetch_maximum)
-					pstate->prefetch_target++;
-				SpinLockRelease(&pstate->mutex);
-			}
-#endif							/* USE_PREFETCH */
-
-			/*
-			 * We issue prefetch requests *after* fetching the current page to
-			 * try to avoid having prefetching interfere with the main I/O.
-			 * Also, this should happen only when we have determined there is
-			 * still something to do on the current page, else we may
-			 * uselessly prefetch the same page we are just about to request
-			 * for real.
-			 */
-			BitmapPrefetch(node, scan);
-
 			/*
 			 * If we are using lossy info, we have to recheck the qual
 			 * conditions at every tuple.
@@ -257,31 +225,14 @@ BitmapHeapNext(BitmapHeapScanState *node)
 
 new_page:
 
-		BitmapAdjustPrefetchIterator(node);
-
 		/*
 		 * Returns false if the bitmap is exhausted and there are no further
 		 * blocks we need to scan.
 		 */
-		if (!table_scan_bitmap_next_block(scan, &node->blockno,
-										  &node->recheck,
+		if (!table_scan_bitmap_next_block(scan, &node->recheck,
 										  &node->stats.lossy_pages,
 										  &node->stats.exact_pages))
 			break;
-
-		/*
-		 * If private, we can error out if the the prefetch block doesn't stay
-		 * ahead of the current block.
-		 */
-		if (node->pstate == NULL &&
-			!tbm_exhausted(&node->prefetch_iterator) &&
-			node->prefetch_blockno < node->blockno)
-			elog(ERROR,
-				 "prefetch and main iterators are out of sync. pfblockno: %d. blockno: %d",
-				 node->prefetch_blockno, node->blockno);
-
-		/* Adjust the prefetch target */
-		BitmapAdjustPrefetchTarget(node);
 	}
 
 	/*
@@ -353,30 +304,22 @@ ExecReScanBitmapHeapScan(BitmapHeapScanState *node)
 		/*
 		 * End iteration on iterators saved in scan descriptor.
 		 */
-		tbm_end_iterate(&scan->st.rs_tbmiterator);
+		tbm_end_iterate(&scan->st.bitmap.rs_iterator);
+
+		/* If we did not already clean up the prefetch iterator, do so now. */
+		if (!tbm_exhausted(&scan->st.bitmap.rs_prefetch_iterator))
+			tbm_end_iterate(&scan->st.bitmap.rs_prefetch_iterator);
 
 		/* rescan to release any page pin */
 		table_rescan(node->ss.ss_currentScanDesc, NULL);
 	}
 
-	/* If we did not already clean up the prefetch iterator, do so now. */
-	if (!tbm_exhausted(&node->prefetch_iterator))
-		tbm_end_iterate(&node->prefetch_iterator);
-
 	/* release bitmaps and buffers if any */
 	if (node->tbm)
 		tbm_free(node->tbm);
-	if (node->pvmbuffer != InvalidBuffer)
-		ReleaseBuffer(node->pvmbuffer);
 	node->tbm = NULL;
 	node->initialized = false;
-	node->pvmbuffer = InvalidBuffer;
 	node->recheck = true;
-	/* Only used for serial BHS */
-	node->blockno = InvalidBlockNumber;
-	node->prefetch_blockno = InvalidBlockNumber;
-	node->prefetch_pages = 0;
-	node->prefetch_target = -1;
 
 	ExecScanReScan(&node->ss);
 
@@ -435,7 +378,11 @@ ExecEndBitmapHeapScan(BitmapHeapScanState *node)
 		/*
 		 * End iteration on iterators saved in scan descriptor.
 		 */
-		tbm_end_iterate(&scanDesc->st.rs_tbmiterator);
+		tbm_end_iterate(&scanDesc->st.bitmap.rs_iterator);
+
+		/* If we did not already clean up the prefetch iterator, do so now. */
+		if (!tbm_exhausted(&scanDesc->st.bitmap.rs_prefetch_iterator))
+			tbm_end_iterate(&scanDesc->st.bitmap.rs_prefetch_iterator);
 
 		/*
 		 * close table scan
@@ -443,17 +390,11 @@ ExecEndBitmapHeapScan(BitmapHeapScanState *node)
 		table_endscan(scanDesc);
 	}
 
-	/* If we did not already clean up the prefetch iterator, do so now. */
-	if (!tbm_exhausted(&node->prefetch_iterator))
-		tbm_end_iterate(&node->prefetch_iterator);
-
 	/*
 	 * release bitmaps and buffers if any
 	 */
 	if (node->tbm)
 		tbm_free(node->tbm);
-	if (node->pvmbuffer != InvalidBuffer)
-		ReleaseBuffer(node->pvmbuffer);
 }
 
 /* ----------------------------------------------------------------
@@ -486,18 +427,13 @@ ExecInitBitmapHeapScan(BitmapHeapScan *node, EState *estate, int eflags)
 	scanstate->ss.ps.ExecProcNode = ExecBitmapHeapScan;
 
 	scanstate->tbm = NULL;
-	scanstate->pvmbuffer = InvalidBuffer;
 
 	/* Zero the statistics counters */
 	memset(&scanstate->stats, 0, sizeof(BitmapHeapScanInstrumentation));
 
-	scanstate->prefetch_pages = 0;
-	scanstate->prefetch_target = -1;
 	scanstate->initialized = false;
 	scanstate->pstate = NULL;
 	scanstate->recheck = true;
-	scanstate->blockno = InvalidBlockNumber;
-	scanstate->prefetch_blockno = InvalidBlockNumber;
 
 	/*
 	 * Miscellaneous initialization
@@ -537,13 +473,6 @@ ExecInitBitmapHeapScan(BitmapHeapScan *node, EState *estate, int eflags)
 	scanstate->bitmapqualorig =
 		ExecInitQual(node->bitmapqualorig, (PlanState *) scanstate);
 
-	/*
-	 * Maximum number of prefetches for the tablespace if configured,
-	 * otherwise the current value of the effective_io_concurrency GUC.
-	 */
-	scanstate->prefetch_maximum =
-		get_tablespace_io_concurrency(currentRelation->rd_rel->reltablespace);
-
 	scanstate->ss.ss_currentRelation = currentRelation;
 
 	/*
diff --git a/src/include/access/heapam.h b/src/include/access/heapam.h
index 232f80821a0..cbda064c612 100644
--- a/src/include/access/heapam.h
+++ b/src/include/access/heapam.h
@@ -22,8 +22,6 @@
 #include "access/table.h"		/* for backward compatibility */
 #include "access/tableam.h"
 #include "nodes/lockoptions.h"
-/* XXX: temporary include so prefetch lift and shift works */
-#include "nodes/execnodes.h"
 #include "nodes/primnodes.h"
 #include "storage/bufpage.h"
 #include "storage/dsm.h"
@@ -101,10 +99,29 @@ typedef struct HeapScanDescData
 }			HeapScanDescData;
 typedef struct HeapScanDescData *HeapScanDesc;
 
+
+/*
+* For prefetching, we use *two* iterators, one for the pages we are
+* actually scanning and another that runs ahead of the first for
+* prefetching. prefetch_pages tracks exactly how many pages ahead
+* the prefetch iterator is.  Also, prefetch_target tracks the
+* desired prefetch distance, which starts small and increases up to the
+* prefetch_maximum.  This is to avoid doing a lot of prefetching in
+* a scan that stops after a few tuples because of a LIMIT.
+*/
 typedef struct BitmapHeapScanDescData
 {
 	HeapScanDescData rs_heap_base;
 
+	/* used to validate prefetch block stays ahead of current block */
+	BlockNumber rs_prefetch_blockno;
+	/* maximum value for prefetch_target */
+	int			rs_prefetch_maximum;
+	/* Current target for prefetch distance */
+	int			rs_prefetch_target;
+	/* # pages prefetch iterator is ahead of current */
+	int			rs_prefetch_pages;
+
 	/*
 	 * These fields are only used for bitmap scans for the "skip fetch"
 	 * optimization. Bitmap scans needing no fields from the heap may skip
@@ -115,6 +132,8 @@ typedef struct BitmapHeapScanDescData
 
 	/* page of VM containing info for current block */
 	Buffer		rs_vmbuffer;
+	/* page of VM containing info for prefetch block */
+	Buffer		rs_pvmbuffer;
 	int			rs_empty_tuples_pending;
 }			BitmapHeapScanDescData;
 typedef struct BitmapHeapScanDescData *BitmapHeapScanDesc;
@@ -300,7 +319,8 @@ typedef enum
 extern TableScanDesc heap_beginscan(Relation relation, Snapshot snapshot,
 									int nkeys, ScanKey key,
 									ParallelTableScanDesc parallel_scan,
-									uint32 flags);
+									uint32 flags,
+									int prefetch_maximum);
 extern void heap_setscanlimits(TableScanDesc sscan, BlockNumber startBlk,
 							   BlockNumber numBlks);
 extern void heap_prepare_pagescan(TableScanDesc sscan);
@@ -426,11 +446,6 @@ extern bool HeapTupleHeaderIsOnlyLocked(HeapTupleHeader tuple);
 extern bool HeapTupleIsSurelyDead(HeapTuple htup,
 								  struct GlobalVisState *vistest);
 
-/* in heapam_handler.c */
-extern void BitmapAdjustPrefetchIterator(BitmapHeapScanState *node);
-extern void BitmapAdjustPrefetchTarget(BitmapHeapScanState *node);
-extern void BitmapPrefetch(BitmapHeapScanState *node, TableScanDesc scan);
-
 /*
  * To avoid leaking too much knowledge about reorderbuffer implementation
  * details this is implemented in reorderbuffer.c not heapam_visibility.c
diff --git a/src/include/access/relscan.h b/src/include/access/relscan.h
index 8ca8f789617..f14db173660 100644
--- a/src/include/access/relscan.h
+++ b/src/include/access/relscan.h
@@ -43,8 +43,13 @@ typedef struct TableScanDescData
 	 */
 	union
 	{
-		/* Iterator for Bitmap Table Scans */
-		TBMIterator rs_tbmiterator;
+		/* State for Bitmap Table Scans */
+		struct
+		{
+			struct ParallelBitmapHeapState *rs_pstate;
+			TBMIterator rs_iterator;
+			TBMIterator rs_prefetch_iterator;
+		}			bitmap;
 
 		/*
 		 * Range of ItemPointers for table_scan_getnextslot_tidrange() to
diff --git a/src/include/access/tableam.h b/src/include/access/tableam.h
index dbcea659fbc..df7b02be9ca 100644
--- a/src/include/access/tableam.h
+++ b/src/include/access/tableam.h
@@ -324,12 +324,16 @@ typedef struct TableAmRoutine
 	 * the scan's behaviour (ScanOptions's SO_ALLOW_*, several may be
 	 * specified, an AM may ignore unsupported ones) and whether the snapshot
 	 * needs to be deallocated at scan_end (ScanOptions's SO_TEMP_SNAPSHOT).
+	 *
+	 * `prefetch_maximum` is the maximum prefetch distance for use by bitmap
+	 * table scans as needed.
 	 */
 	TableScanDesc (*scan_begin) (Relation rel,
 								 Snapshot snapshot,
 								 int nkeys, struct ScanKeyData *key,
 								 ParallelTableScanDesc pscan,
-								 uint32 flags);
+								 uint32 flags,
+								 int prefetch_maximum);
 
 	/*
 	 * Release resources and deallocate scan. If TableScanDesc.temp_snap,
@@ -780,9 +784,10 @@ typedef struct TableAmRoutine
 	 */
 
 	/*
-	 * Prepare to fetch / check / return tuples from `blockno` as part of a
-	 * bitmap table scan. `scan` was started via table_beginscan_bm(). Return
-	 * false if the bitmap is exhausted and true otherwise.
+	 * Prepare to fetch / check / return tuples from the next block yielded by
+	 * the iterator as part of a bitmap table scan. `scan` was started via
+	 * table_beginscan_bm(). Return false if the bitmap is exhausted and true
+	 * otherwise.
 	 *
 	 * This will typically read and pin the target block, and do the necessary
 	 * work to allow scan_bitmap_next_tuple() to return tuples (e.g. it might
@@ -797,28 +802,13 @@ typedef struct TableAmRoutine
 	 * always need to be rechecked, but some non-lossy pages' tuples may also
 	 * require recheck.
 	 *
-	 * `blockno` is the current block and is set by the table AM. The table AM
-	 * is responsible for advancing the main iterator, but the bitmap table
-	 * scan code still advances the prefetch iterator. `blockno` is used by
-	 * bitmap table scan code to validate that the prefetch block stays ahead
-	 * of the current block.
-	 *
-	 * XXX: Currently this may only be implemented if the AM uses md.c as its
-	 * storage manager, and uses ItemPointer->ip_blkid in a manner that maps
-	 * blockids directly to the underlying storage. nodeBitmapHeapscan.c
-	 * performs prefetching directly using that interface.  This probably
-	 * needs to be rectified at a later point.
-	 *
-	 * XXX: Currently this may only be implemented if the AM uses the
-	 * visibilitymap, as nodeBitmapHeapscan.c unconditionally accesses it to
-	 * perform prefetching.  This probably needs to be rectified at a later
-	 * point.
+	 * Prefetching future blocks indicated in the bitmap is left to the table
+	 * AM.
 	 *
 	 * Optional callback, but either both scan_bitmap_next_block and
 	 * scan_bitmap_next_tuple need to exist, or neither.
 	 */
 	bool		(*scan_bitmap_next_block) (TableScanDesc scan,
-										   BlockNumber *blockno,
 										   bool *recheck,
 										   uint64 *lossy_pages,
 										   uint64 *exact_pages);
@@ -916,7 +906,8 @@ table_beginscan(Relation rel, Snapshot snapshot,
 	uint32		flags = SO_TYPE_SEQSCAN |
 		SO_ALLOW_STRAT | SO_ALLOW_SYNC | SO_ALLOW_PAGEMODE;
 
-	return rel->rd_tableam->scan_begin(rel, snapshot, nkeys, key, NULL, flags);
+	return rel->rd_tableam->scan_begin(rel, snapshot, nkeys, key,
+									   NULL, flags, 0);
 }
 
 /*
@@ -945,7 +936,8 @@ table_beginscan_strat(Relation rel, Snapshot snapshot,
 	if (allow_sync)
 		flags |= SO_ALLOW_SYNC;
 
-	return rel->rd_tableam->scan_begin(rel, snapshot, nkeys, key, NULL, flags);
+	return rel->rd_tableam->scan_begin(rel, snapshot, nkeys, key,
+									   NULL, flags, 0);
 }
 
 /*
@@ -956,15 +948,21 @@ table_beginscan_strat(Relation rel, Snapshot snapshot,
  */
 static inline TableScanDesc
 table_beginscan_bm(Relation rel, Snapshot snapshot,
-				   int nkeys, struct ScanKeyData *key, bool need_tuple)
+				   struct ParallelBitmapHeapState *pstate,
+				   int nkeys, struct ScanKeyData *key, bool need_tuple,
+				   int prefetch_maximum)
 {
+	TableScanDesc result;
 	uint32		flags = SO_TYPE_BITMAPSCAN | SO_ALLOW_PAGEMODE;
 
 	if (need_tuple)
 		flags |= SO_NEED_TUPLES;
 
-	return rel->rd_tableam->scan_begin(rel, snapshot, nkeys, key,
-									   NULL, flags);
+	result = rel->rd_tableam->scan_begin(rel, snapshot, nkeys,
+										 key, NULL, flags,
+										 prefetch_maximum);
+	result->st.bitmap.rs_pstate = pstate;
+	return result;
 }
 
 /*
@@ -989,7 +987,8 @@ table_beginscan_sampling(Relation rel, Snapshot snapshot,
 	if (allow_pagemode)
 		flags |= SO_ALLOW_PAGEMODE;
 
-	return rel->rd_tableam->scan_begin(rel, snapshot, nkeys, key, NULL, flags);
+	return rel->rd_tableam->scan_begin(rel, snapshot, nkeys, key,
+									   NULL, flags, 0);
 }
 
 /*
@@ -1002,7 +1001,8 @@ table_beginscan_tid(Relation rel, Snapshot snapshot)
 {
 	uint32		flags = SO_TYPE_TIDSCAN;
 
-	return rel->rd_tableam->scan_begin(rel, snapshot, 0, NULL, NULL, flags);
+	return rel->rd_tableam->scan_begin(rel, snapshot, 0, NULL,
+									   NULL, flags, 0);
 }
 
 /*
@@ -1015,7 +1015,7 @@ table_beginscan_analyze(Relation rel)
 {
 	uint32		flags = SO_TYPE_ANALYZE;
 
-	return rel->rd_tableam->scan_begin(rel, NULL, 0, NULL, NULL, flags);
+	return rel->rd_tableam->scan_begin(rel, NULL, 0, NULL, NULL, flags, 0);
 }
 
 /*
@@ -1094,7 +1094,8 @@ table_beginscan_tidrange(Relation rel, Snapshot snapshot,
 	TableScanDesc sscan;
 	uint32		flags = SO_TYPE_TIDRANGESCAN | SO_ALLOW_PAGEMODE;
 
-	sscan = rel->rd_tableam->scan_begin(rel, snapshot, 0, NULL, NULL, flags);
+	sscan = rel->rd_tableam->scan_begin(rel, snapshot, 0, NULL,
+										NULL, flags, 0);
 
 	/* Set the range of TIDs to scan */
 	sscan->rs_rd->rd_tableam->scan_set_tidrange(sscan, mintid, maxtid);
@@ -1966,16 +1967,11 @@ table_relation_estimate_size(Relation rel, int32 *attr_widths,
  * `recheck` is set by the table AM to indicate whether or not the tuples
  * from this block should be rechecked.
  *
- * `blockno` is the current block and is set by the table AM and is used by
- * bitmap table scan code to validate that the prefetch block stays ahead of
- * the current block.
- *
  * Note, this is an optionally implemented function, therefore should only be
  * used after verifying the presence (at plan time or such).
  */
 static inline bool
 table_scan_bitmap_next_block(TableScanDesc scan,
-							 BlockNumber *blockno,
 							 bool *recheck,
 							 uint64 *lossy_pages,
 							 uint64 *exact_pages)
@@ -1989,7 +1985,7 @@ table_scan_bitmap_next_block(TableScanDesc scan,
 		elog(ERROR, "unexpected table_scan_bitmap_next_block call during logical decoding");
 
 	return scan->rs_rd->rd_tableam->scan_bitmap_next_block(scan,
-														   blockno, recheck,
+														   recheck,
 														   lossy_pages,
 														   exact_pages);
 }
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 7cd4b57509e..53fcb9d33a4 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -1833,18 +1833,11 @@ typedef struct SharedBitmapHeapInstrumentation
  *
  *		bitmapqualorig	   execution state for bitmapqualorig expressions
  *		tbm				   bitmap obtained from child index scan(s)
- *		pvmbuffer		   buffer for visibility-map lookups of prefetched pages
  *		stats			   execution statistics
- *		prefetch_iterator  iterator for prefetching ahead of current page
- *		prefetch_pages	   # pages prefetch iterator is ahead of current
- *		prefetch_target    current target prefetch distance
- *		prefetch_maximum   maximum value for prefetch_target
  *		initialized		   is node is ready to iterate
  *		pstate			   shared state for parallel bitmap scan
  *		sinstrument		   statistics for parallel workers
  *		recheck			   do current page's tuples need recheck
- *		blockno			   used to validate pf and current block stay in sync
- *		prefetch_blockno   used to validate pf stays ahead of current block
  * ----------------
  */
 typedef struct BitmapHeapScanState
@@ -1852,18 +1845,11 @@ typedef struct BitmapHeapScanState
 	ScanState	ss;				/* its first field is NodeTag */
 	ExprState  *bitmapqualorig;
 	TIDBitmap  *tbm;
-	Buffer		pvmbuffer;
 	BitmapHeapScanInstrumentation stats;
-	TBMIterator prefetch_iterator;
-	int			prefetch_pages;
-	int			prefetch_target;
-	int			prefetch_maximum;
 	bool		initialized;
 	ParallelBitmapHeapState *pstate;
 	SharedBitmapHeapInstrumentation *sinstrument;
 	bool		recheck;
-	BlockNumber blockno;
-	BlockNumber prefetch_blockno;
 } BitmapHeapScanState;
 
 /* ----------------
-- 
2.45.2