Re: [Qemu-devel] [PATCH V3 2/7] qcow2: free allocated cluster on fail in qcow2_write_snapshots()

[Stefan Hajnoczi]

On Mon, Sep 09, 2013 at 10:57:57AM +0800, Wenchao Xia wrote:
> Signed-off-by: Wenchao Xia <address@hidden>
> ---
>  block/qcow2-snapshot.c |    7 +++++--
>  1 files changed, 5 insertions(+), 2 deletions(-)
> 
> diff --git a/block/qcow2-snapshot.c b/block/qcow2-snapshot.c
> index 40393b2..9e2d695 100644
> --- a/block/qcow2-snapshot.c
> +++ b/block/qcow2-snapshot.c
> @@ -186,7 +186,7 @@ static int qcow2_write_snapshots(BlockDriverState *bs)
>      }
>      ret = bdrv_flush(bs);
>      if (ret < 0) {
> -        return ret;
> +        goto fail;
>      }
>  
>      /* The snapshot list position has not yet been updated, so these clusters
> @@ -194,7 +194,7 @@ static int qcow2_write_snapshots(BlockDriverState *bs)
>      ret = qcow2_pre_write_overlap_check(bs, QCOW2_OL_DEFAULT, offset,
>                                          s->snapshots_size);
>      if (ret < 0) {
> -        return ret;
> +        goto fail;
>      }
>  
>  
> @@ -278,6 +278,9 @@ static int qcow2_write_snapshots(BlockDriverState *bs)
>      return 0;
>  
>  fail:
> +    /* free the new snapshot table */
> +    qcow2_free_clusters(bs, snapshots_offset, snapshots_size,
> +                        QCOW2_DISCARD_ALWAYS);

It is safer to skip qcow2_free_clusters() when the final
bdrv_pwrite_sync() fails.  We don't know if the header was updated on
disk.  If it was updated, then freeing the clusters may allow them to be
reallocated later (this will cause corruption).