Re: [Qemu-devel] [RFC 0/6] converting some of virtio to VMState

[Cornelia Huck]

On Wed, 24 Aug 2016 14:42:27 +0100
"Dr. David Alan Gilbert (git)" <address@hidden> wrote:

> From: "Dr. David Alan Gilbert" <address@hidden>
> 
> Hi,
>   This series converts two parts of virtio to VMState - the
> device load/save for virtio-balloon and virtio-net.
> 
> It's only been smoke tested (which it passes); but if anyone
> has a good suggestion for testing virtio-net migration
> I'd be greatful.
> 
> The first couple of patches are just general error reporting improvements
> for vmstate; and could go in straight away.
> The 3rd adds a new vmstate macro for skipping chunks of input data
> 
> The 4th wires in a call to vmstate_{load|save}_state in virtio_{load|save}
> calling the vmsd on virtio device class; this is instead of the current 
> ->load/->save
> methods.  The idea is that eventually I'll be able to kill off the 
> ->load/->save.
> Also my intention is to add the recursion into vdc->vmsd into the main 
> vmstate_virtio
> vmsd via some new macro I've not figured out yet.
> 
> The meat is in the 5th and 6th patches that do balloon and virtio-net.
> 
> My plan is to attack virtio_blk and virtio_serial_bus next; both of which
> have loops walking along lists (in subtly different ways); I intend
> to see how Jianjun Duan's loop migration code would fit with those and
> try and come up with something general that does at least all 3 cases.

I've had a quick glance at your changes and they look sane (at least up
to the balloon patch; I'll need some quiet time to look at the
virtio-net changes). I'll try to see when I can give it a bit of testing.