diff mbox

[3/3] block: Drain requests in bdrv_close

Message ID 1334140821-16247-4-git-send-email-kwolf@redhat.com
State New
Headers show

Commit Message

Kevin Wolf April 11, 2012, 10:40 a.m. UTC 
If an AIO request is in flight that refers to a BlockDriverState that
has been closed and possibly even freed, more or less anything could
happen. I have seen segfaults, -EBADF return values and qcow2 sometimes
actually catches the situation in bdrv_close() and abort()s.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
---
 block.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

Comments

Paolo Bonzini April 12, 2012, 5:23 p.m. UTC | #1 
Il 11/04/2012 12:40, Kevin Wolf ha scritto:
> @@ -816,6 +816,8 @@ void bdrv_close(BlockDriverState *bs)
>          if (bs->job) {
>              block_job_cancel_sync(bs->job);
>          }
> +        bdrv_drain_all();

Actually the bdrv_drain_all needs to go first.

This is because block_job_cancel_sync guarantees to exit only after the
block job's own I/O has finished, *but* it needs to see all I/O from the
guest.

Paolo
diff mbox

Patch

diff --git a/block.c b/block.c
index c0c90f0..4b96654 100644
--- a/block.c
+++ b/block.c
@@ -816,6 +816,8 @@  void bdrv_close(BlockDriverState *bs)
         if (bs->job) {
             block_job_cancel_sync(bs->job);
         }
+        bdrv_drain_all();
+
         if (bs == bs_snapshots) {
             bs_snapshots = NULL;
         }