Re: [virtio-dev] virtio-snd comments/questions

[Matias Ezequiel Vara Larsen <mvaralar@redhat.com>]

On Fri, Nov 03, 2023 at 06:17:17PM +0000, Radu Ocica wrote:
> I also think that the driver should not send a STOP command after a failed START command. There is no explicit description of any states in the virtio sound device specification. We have both adopted an interpretation that the state is determined by the last successful pcm control request and the transitions specified for pcm control requests assume success of these control requests. I wish this was clarified explicitly in the spec, to not allow room for interpretation.
>
>
I agree. I think what is missed in the spec is to define the behavior of
the PCM command cycle in case a command fails. I worked on a RFC to
improve that section, feel free to comment at
https://lists.oasis-open.org/archives/virtio-comment/202307/msg00158.html.
I think an overall solution would be to implement the PCM state machine
in the driver thus control messages would be sent only if the stream is
in the correct state.

Matias