Check bufHdr->cntxDirty and call StartBufferIO in BufferSync()

*before* acquiring shlock on buffer context. This way we should be
protected against conflicts with FlushRelationBuffers.
(Seems we never do excl lock and then StartBufferIO for the same
buffer, so there should be no deadlock here, - but we'd better
check this very soon).

src/backend/storage/buffer/bufmgr.c

@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $Header: /cvsroot/pgsql/src/backend/storage/buffer/bufmgr.c,v 1.107 2001/02/18 04:39:42 tgl Exp $
+ *	  $Header: /cvsroot/pgsql/src/backend/storage/buffer/bufmgr.c,v 1.108 2001/03/21 10:13:29 vadim Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -727,7 +727,6 @@ BufferSync()
 	RelFileNode	rnode;
 	XLogRecPtr	recptr;
 	Relation	reln = NULL;
-	bool		dirty = false;
 
 	for (i = 0, bufHdr = BufferDescriptors; i < NBuffers; i++, bufHdr++)
 	{
@@ -741,16 +740,44 @@ BufferSync()
 		}
 
 		/*
-		 * Pin buffer and ensure that no one reads it from disk
+		 * We can check bufHdr->cntxDirty here *without* holding any lock
+		 * on buffer context as long as we set this flag in access methods
+		 * *before* logging changes with XLogInsert(): if someone will set
+		 * cntxDirty just after our check we don't worry because of our
+		 * checkpoint.redo points before log record for upcoming changes
+		 * and so we are not required to write such dirty buffer.
+		 */
+		if (!(bufHdr->flags & BM_DIRTY) && !(bufHdr->cntxDirty))
+		{
+			SpinRelease(BufMgrLock);
+			continue;
+		}
+
+		/*
+		 * IO synchronization. Note that we do it with unpinned buffer
+		 * to avoid conflicts with FlushRelationBuffers.
 		 */
-		PinBuffer(bufHdr);
-		/* Synchronize with BufferAlloc */
 		if (bufHdr->flags & BM_IO_IN_PROGRESS)
+		{
 			WaitIO(bufHdr, BufMgrLock);
+			if (!(bufHdr->flags & BM_VALID) ||
+				(!(bufHdr->flags & BM_DIRTY) && !(bufHdr->cntxDirty)))
+			{
+				SpinRelease(BufMgrLock);
+				continue;
+			}
+		}
+
+		/*
+		 * Here: no one doing IO for this buffer and it's dirty.
+		 * Pin buffer now and set IO state for it *before* acquiring
+		 * shlock to avoid conflicts with FlushRelationBuffers.
+		 */
+		PinBuffer(bufHdr);
+		StartBufferIO(bufHdr, false);		/* output IO start */
 
 		buffer = BufferDescriptorGetBuffer(bufHdr);
 		rnode = bufHdr->tag.rnode;
-		dirty = bufHdr->flags & BM_DIRTY;
 
 		SpinRelease(BufMgrLock);
 
@@ -764,17 +791,6 @@ BufferSync()
 		 */
 		LockBuffer(buffer, BUFFER_LOCK_SHARE);
 
-		if (!dirty && !(bufHdr->cntxDirty))
-		{
-			LockBuffer(buffer, BUFFER_LOCK_UNLOCK);
-			SpinAcquire(BufMgrLock);
-			UnpinBuffer(bufHdr);
-			SpinRelease(BufMgrLock);
-			if (reln != (Relation) NULL)
-				RelationDecrementReferenceCount(reln);
-			continue;
-		}
-
 		/*
 		 * Force XLOG flush for buffer' LSN
 		 */
@@ -782,73 +798,61 @@ BufferSync()
 		XLogFlush(recptr);
 
 		/*
-		 * Now it's safe to write buffer to disk
-		 * (if no one else already)
+		 * Now it's safe to write buffer to disk. Note that no one else
+		 * should not be able to write it while we were busy with locking
+		 * and log flushing because of we setted IO flag.
 		 */
 		SpinAcquire(BufMgrLock);
-		if (bufHdr->flags & BM_IO_IN_PROGRESS)
-			WaitIO(bufHdr, BufMgrLock);
+		Assert(bufHdr->flags & BM_DIRTY || bufHdr->cntxDirty);
+		bufHdr->flags &= ~BM_JUST_DIRTIED;
+		SpinRelease(BufMgrLock);
 
-		if (bufHdr->flags & BM_DIRTY || bufHdr->cntxDirty)
+		if (reln == (Relation) NULL)
 		{
-			bufHdr->flags &= ~BM_JUST_DIRTIED;
-			StartBufferIO(bufHdr, false);		/* output IO start */
-
-			SpinRelease(BufMgrLock);
-
-			if (reln == (Relation) NULL)
-			{
-				status = smgrblindwrt(DEFAULT_SMGR,
-									bufHdr->tag.rnode,
-									bufHdr->tag.blockNum,
-									(char *) MAKE_PTR(bufHdr->data),
-									true);	/* must fsync */
-			}
-			else
-			{
-				status = smgrwrite(DEFAULT_SMGR, reln,
+			status = smgrblindwrt(DEFAULT_SMGR,
+								bufHdr->tag.rnode,
 								bufHdr->tag.blockNum,
-								(char *) MAKE_PTR(bufHdr->data));
-			}
+								(char *) MAKE_PTR(bufHdr->data),
+								true);	/* must fsync */
+		}
+		else
+		{
+			status = smgrwrite(DEFAULT_SMGR, reln,
+							bufHdr->tag.blockNum,
+							(char *) MAKE_PTR(bufHdr->data));
+		}
 
-			if (status == SM_FAIL)	/* disk failure ?! */
-				elog(STOP, "BufferSync: cannot write %u for %s",
-					 bufHdr->tag.blockNum, bufHdr->blind.relname);
+		if (status == SM_FAIL)	/* disk failure ?! */
+			elog(STOP, "BufferSync: cannot write %u for %s",
+				 bufHdr->tag.blockNum, bufHdr->blind.relname);
 
-			/*
-			 * Note that it's safe to change cntxDirty here because of
-			 * we protect it from upper writers by share lock and from
-			 * other bufmgr routines by BM_IO_IN_PROGRESS
-			 */
-			bufHdr->cntxDirty = false;
+		/*
+		 * Note that it's safe to change cntxDirty here because of
+		 * we protect it from upper writers by share lock and from
+		 * other bufmgr routines by BM_IO_IN_PROGRESS
+		 */
+		bufHdr->cntxDirty = false;
 
-			/*
-			 * Release the per-buffer readlock, reacquire BufMgrLock.
-			 */
-			LockBuffer(buffer, BUFFER_LOCK_UNLOCK);
-			BufferFlushCount++;
+		/*
+		 * Release the per-buffer readlock, reacquire BufMgrLock.
+		 */
+		LockBuffer(buffer, BUFFER_LOCK_UNLOCK);
+		BufferFlushCount++;
 
-			SpinAcquire(BufMgrLock);
+		SpinAcquire(BufMgrLock);
 
-			bufHdr->flags &= ~BM_IO_IN_PROGRESS;	/* mark IO finished */
-			TerminateBufferIO(bufHdr);				/* Sync IO finished */
+		bufHdr->flags &= ~BM_IO_IN_PROGRESS;	/* mark IO finished */
+		TerminateBufferIO(bufHdr);				/* Sync IO finished */
 
-			/*
-			 * If this buffer was marked by someone as DIRTY while
-			 * we were flushing it out we must not clear DIRTY
-			 * flag - vadim 01/17/97
-			 */
-			if (!(bufHdr->flags & BM_JUST_DIRTIED))
-				bufHdr->flags &= ~BM_DIRTY;
-			UnpinBuffer(bufHdr);
-			SpinRelease(BufMgrLock);
-		}
-		else
-		{
-			UnpinBuffer(bufHdr);
-			SpinRelease(BufMgrLock);
-			LockBuffer(buffer, BUFFER_LOCK_UNLOCK);
-		}
+		/*
+		 * If this buffer was marked by someone as DIRTY while
+		 * we were flushing it out we must not clear DIRTY
+		 * flag - vadim 01/17/97
+		 */
+		if (!(bufHdr->flags & BM_JUST_DIRTIED))
+			bufHdr->flags &= ~BM_DIRTY;
+		UnpinBuffer(bufHdr);
+		SpinRelease(BufMgrLock);
 
 		/* drop refcnt obtained by RelationNodeCacheGetRelation */
 		if (reln != (Relation) NULL)
@@ -2079,6 +2083,12 @@ LockBuffer(Buffer buffer, int mode)
 		buf->w_lock = true;
 		*buflock |= BL_W_LOCK;
 
+		/*
+		 * This is not the best place to set cntxDirty flag (eg indices
+		 * do not always change buffer they lock in excl mode). But please
+		 * remember that it's critical to set cntxDirty *before* logging
+		 * changes with XLogInsert() - see comments in BufferSync().
+		 */
 		buf->cntxDirty = true;
 
 		if (*buflock & BL_RI_LOCK)