Scan the buffer pool just once, not once per fork, during relation drop.

This provides a speedup of about 4X when NBuffers is large enough.
There is also a useful reduction in sinval traffic, since we
only do CacheInvalidateSmgr() once not once per fork.

Simon Riggs, reviewed and somewhat revised by Tom Lane

2 files changed, 81 insertions, 1 deletions

diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c
index b178eee2214..d46faaf958d 100644
--- a/src/backend/storage/buffer/bufmgr.c
+++ b/src/backend/storage/buffer/bufmgr.c
@@ -2020,7 +2020,7 @@ BufferIsPermanent(Buffer buffer)
  *		DropRelFileNodeBuffers
  *
  *		This function removes from the buffer pool all the pages of the
- *		specified relation that have block numbers >= firstDelBlock.
+ *		specified relation fork that have block numbers >= firstDelBlock.
  *		(In particular, with firstDelBlock = 0, all pages are removed.)
  *		Dirty pages are simply dropped, without bothering to write them
  *		out first.	Therefore, this is NOT rollback-able, and so should be
@@ -2090,6 +2090,46 @@ DropRelFileNodeBuffers(RelFileNodeBackend rnode, ForkNumber forkNum,
 }
 
 /* ---------------------------------------------------------------------
+ *		DropRelFileNodeAllBuffers
+ *
+ *		This function removes from the buffer pool all the pages of all
+ *		forks of the specified relation.  It's equivalent to calling
+ *		DropRelFileNodeBuffers once per fork with firstDelBlock = 0.
+ * --------------------------------------------------------------------
+ */
+void
+DropRelFileNodeAllBuffers(RelFileNodeBackend rnode)
+{
+	int			i;
+
+	/* If it's a local relation, it's localbuf.c's problem. */
+	if (rnode.backend != InvalidBackendId)
+	{
+		if (rnode.backend == MyBackendId)
+			DropRelFileNodeAllLocalBuffers(rnode.node);
+		return;
+	}
+
+	for (i = 0; i < NBuffers; i++)
+	{
+		volatile BufferDesc *bufHdr = &BufferDescriptors[i];
+
+		/*
+		 * As in DropRelFileNodeBuffers, an unlocked precheck should be safe
+		 * and saves some cycles.
+		 */
+		if (!RelFileNodeEquals(bufHdr->tag.rnode, rnode.node))
+			continue;
+
+		LockBufHdr(bufHdr);
+		if (RelFileNodeEquals(bufHdr->tag.rnode, rnode.node))
+			InvalidateBuffer(bufHdr);	/* releases spinlock */
+		else
+			UnlockBufHdr(bufHdr);
+	}
+}
+
+/* ---------------------------------------------------------------------
  *		DropDatabaseBuffers
  *
  *		This function removes all the buffers in the buffer cache for a
diff --git a/src/backend/storage/buffer/localbuf.c b/src/backend/storage/buffer/localbuf.c
index 63c14f7300c..46eeaf742d8 100644
--- a/src/backend/storage/buffer/localbuf.c
+++ b/src/backend/storage/buffer/localbuf.c
@@ -331,6 +331,46 @@ DropRelFileNodeLocalBuffers(RelFileNode rnode, ForkNumber forkNum,
 }
 
 /*
+ * DropRelFileNodeAllLocalBuffers
+ *		This function removes from the buffer pool all pages of all forks
+ *		of the specified relation.
+ *
+ *		See DropRelFileNodeAllBuffers in bufmgr.c for more notes.
+ */
+void
+DropRelFileNodeAllLocalBuffers(RelFileNode rnode)
+{
+	int			i;
+
+	for (i = 0; i < NLocBuffer; i++)
+	{
+		BufferDesc *bufHdr = &LocalBufferDescriptors[i];
+		LocalBufferLookupEnt *hresult;
+
+		if ((bufHdr->flags & BM_TAG_VALID) &&
+			RelFileNodeEquals(bufHdr->tag.rnode, rnode))
+		{
+			if (LocalRefCount[i] != 0)
+				elog(ERROR, "block %u of %s is still referenced (local %u)",
+					 bufHdr->tag.blockNum,
+					 relpathbackend(bufHdr->tag.rnode, MyBackendId,
+									bufHdr->tag.forkNum),
+					 LocalRefCount[i]);
+			/* Remove entry from hashtable */
+			hresult = (LocalBufferLookupEnt *)
+				hash_search(LocalBufHash, (void *) &bufHdr->tag,
+							HASH_REMOVE, NULL);
+			if (!hresult)		/* shouldn't happen */
+				elog(ERROR, "local buffer hash table corrupted");
+			/* Mark buffer invalid */
+			CLEAR_BUFFERTAG(bufHdr->tag);
+			bufHdr->flags = 0;
+			bufHdr->usage_count = 0;
+		}
+	}
+}
+
+/*
  * InitLocalBuffers -
  *	  init the local buffer cache. Since most queries (esp. multi-user ones)
  *	  don't involve local buffers, we delay allocating actual memory for the