Re: [PATCH RFC v4 00/17] linux: towards virtio-1 guest support

[Rusty Russell <rusty@ozlabs.org>]

"Michael S. Tsirkin" <mst@redhat.com> writes:
> Based on patches by Cornelia Rusty and others, but
> with an API that should allow better static checking of code,
> and slightly more concervative changes in vring,net and blk.
>
> Based on patches by Cornelia and others, but
> with an API that should allow better static checking of code,
> slightly more concervative changes in vring and drivers,
> and compatibility for existing drivers so that
> this series be applied before all drivers are converted.
>
> virtio drivers now pass sparse without warnings.

That's good!  Some comments:

1) Patch order.  You need to increase feature bits to 64, then define
   VIRTIO_F_VERSION_1, then use it in your __virtio conversion macros.

2) You need to enhance virtio_check_driver_offered_feature to look
   at the legacy bits, too.
   [ Oh, I see you did that in a followup ]

3) virtio_has_feature() is a bit heavy-weight, you probably want to add
   __virtio_has_feature() or just opencode vdev->features & (1ULL <<
   VIRTIO_F_VERSION_1);

Obviously, patch correctness don't matter until we enable the
VIRTIO_F_VERSION_1 feature in a transport.  That's what makes me
nervous, and I'll be waiting for Cornelia's ACK on that, since
it looks like s390 is ready to go?

Thanks,
Rusty.