On 10/20/2016 08:57 AM, zhanghailiang wrote:
> Without this callback, there will be an error reports in the primary side:
> "qemu-system-x86_64: Couldn't determine the cluster size of the target image,
> which has no backing file: Operation not supported
> Aborting, since this may create an unusable destination image"
> 
> For nbd driver, it doesn't have cluster size, so here we return
> a fake value for it.
> 
> Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
> Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
> ---
>  block/nbd.c         | 12 ++++++++++++
>  block/replication.c |  6 ++++++
>  2 files changed, 18 insertions(+)
> 
> diff --git a/block/nbd.c b/block/nbd.c
> index 6bc06d6..96d7023 100644
> --- a/block/nbd.c
> +++ b/block/nbd.c
> @@ -40,6 +40,8 @@
>  
>  #define EN_OPTSTR ":exportname="
>  
> +#define NBD_FAKE_CLUSTER_SIZE 512

Why 512?  NBD allows byte-addressable operations (even if it is more
efficient on aligned I/O); and I've been working hard to convert things
to the point that NBD does not enforce alignment on other layers.
Wouldn't 1 be better?

> +static int nbd_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
> +{
> +    bdi->cluster_size  = NBD_FAKE_CLUSTER_SIZE;

I also have patches written (but waiting for NBD write zeroes support to
be reviewed first) that add support for the experimental NBD block info,
that lets a server advertise actual sizes to the client rather than
having to guess.  Here's the last time I posted a preview of it:

https://lists.gnu.org/archive/html/qemu-devel/2016-04/msg03567.html

It would be nice to use that instead of just faking things.

-- 
Eric Blake   eblake redhat com    +1-919-301-3266
Libvirt virtualization library http://libvirt.org