This patch makes some improvements to the rtree index implementation:

(1) Keep a pin on the scan's current buffer and mark buffer. This
avoids the need to do a ReadBuffer() for each tuple produced by the
scan. Since ReadBuffer() is expensive, this is a significant win.

(2) Convert a ReleaseBuffer(); ReadBuffer() pair into
ReleaseAndReadBuffer(). Surely not a huge win, but it saves a lock
acquire/release...

(3) Remove a bunch of duplicated code in rtget.c; make rtnext() handle
both the "initial result" and "subsequent result" cases.

(4) Add support for index tuple killing

(5) Remove rtscancache(): it is dead code, for the same reason that
gistscancache() is dead code (an index scan ought not be invoked with
NoMovementScanDirection).

The end result is about a 10% improvement in rtree index scan perf,
according to contrib/rtree_gist/bench.

1 files changed, 3 insertions, 7 deletions

diff --git a/src/backend/access/rtree/rtree.c b/src/backend/access/rtree/rtree.c
index c2dd2bb8c77..b1762bfe1fb 100644
--- a/src/backend/access/rtree/rtree.c
+++ b/src/backend/access/rtree/rtree.c
@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/access/rtree/rtree.c,v 1.85 2004/12/31 21:59:26 pgsql Exp $
+ *	  $PostgreSQL: pgsql/src/backend/access/rtree/rtree.c,v 1.86 2005/01/18 23:25:47 neilc Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -280,12 +280,8 @@ rtdoinsert(Relation r, IndexTuple itup, RTSTATE *rtstate)
 
 	do
 	{
-		/* let go of current buffer before getting next */
-		if (buffer != InvalidBuffer)
-			ReleaseBuffer(buffer);
-
-		/* get next buffer */
-		buffer = ReadBuffer(r, blk);
+		/* release the current buffer, read in the next one */
+		buffer = ReleaseAndReadBuffer(buffer, r, blk);
 		page = (Page) BufferGetPage(buffer);
 
 		opaque = (RTreePageOpaque) PageGetSpecialPointer(page);