On Fri, Mar 26, 2021 at 4:29 PM Jason Wang <jasow...@redhat.com> wrote:
>
>
> 在 2021/3/25 下午3:15, Cindy Lu 写道:
>
> +enum virtio_config_status {
> + VIRTIO_CONFIG_SUPPORT,
> + VIRTIO_CONFIG_WORK,
> + VIRTIO_CONFIG_STOP,
> + VIRTIO_CONFIG_STATUS_UNKNOWN,
>
> Any reason for this extra state? I think we can know whether the config
> interrupt is being used through a
>
> Thanks
>
> The problem is I need to split the backend devices into 3 types,
> 1） normal device
> 2）vdpa support config interrupt. and the configur interrupt is active now
> 3）vdpa not support config interrupt.
> So I add this bit and this bit will init in vpda /vhost modules and
> qemu can check this bit to know the which behariver we will do in
> virtio bus and other modules,
>
>
> I wonder whether it's a must. We can setup guest notifiers unconditionally,
> so if it's an vhost bakcend without config interrupt support, such notifiers
> won't be used.
>
> Thanks
>
sure,That make sense I will post a new version
>
> Maybe I need to change this bit's name
> to make it more clearly
>
> Thanks
> Cindy
>
>
>
