qcow2: Move BLKDBG_EVENT out of the lock
authorKevin Wolf <kwolf@redhat.com>
Thu, 6 Dec 2012 13:32:59 +0000 (14:32 +0100)
committerKevin Wolf <kwolf@redhat.com>
Wed, 12 Dec 2012 11:33:48 +0000 (12:33 +0100)
We want to use these events to suspend requests for testing concurrent
AIO requests. Suspending requests while they are holding the CoMutex is
rather boring for this purpose.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
block/qcow2.c

index c1ff31f4827c61067ab1a51454bc76339649994b..0a08ec74c7a6b4ee2eef35b2a147f285b5cd1714 100644 (file)
@@ -835,8 +835,8 @@ static coroutine_fn int qcow2_co_writev(BlockDriverState *bs,
                 cur_nr_sectors * 512);
         }
 
-        BLKDBG_EVENT(bs->file, BLKDBG_WRITE_AIO);
         qemu_co_mutex_unlock(&s->lock);
+        BLKDBG_EVENT(bs->file, BLKDBG_WRITE_AIO);
         trace_qcow2_writev_data(qemu_coroutine_self(),
                                 (cluster_offset >> 9) + index_in_cluster);
         ret = bdrv_co_writev(bs->file,