On Fri, Mar 29, 2019 at 07:13:18AM -0500, Eric Blake wrote:
On 3/29/19 3:52 AM, Richard W.M. Jones wrote:
>> The full and null plugins should probably have a trivial .extents that
>> reports completely sparse, all of the time :)
>
> I think it would be wrong for full wouldn't it? It could mean that
> the client would skip the read, thus not getting the intended ENOSPC
> error. For null I agree.
>
The full plugin always reads as 0 successfully. It is only writes that
differ (successfully ignored vs ENOSPC) between the two plugins.
Ah yes, my mistake. In that case yes we can return extents.
I'll add the simple patch.
I already did this for the null plugin:
https://github.com/libguestfs/nbdkit/commit/8b5adddbf06a6ba785e7d04099102...
So the full plugin would be the same.
Rich.
--
Richard Jones, Virtualization Group, Red Hat
http://people.redhat.com/~rjones
Read my programming and virtualization blog:
http://rwmj.wordpress.com
virt-builder quickly builds VMs from scratch
http://libguestfs.org/virt-builder.1.html