Re: [RFC 0/4] Using a generic bus_type for virtio

[Arnd Bergmann <arnd@arndb.de>]

On Sunday 08 July 2007, Rusty Russell wrote:
>         I think it will come down to how neat PCI<->virtio is.  Can we push
> further towards PCI without screwing non-PCI?  eg. can we use
> pci_device_id?  struct pci_driver?  (Might be pushing it, but should
> probably be considered: it'd be neat if some platforms could #define
> virtio_driver_register pci_driver_register).

My idea was to have the virtio device as a child of the PCI device
instead. That way the PCI virtio host can do its own probing without
burdening the virtio bus layer with PCI specific implementation details.

>         Standardizing how to pack the info for each device into the config
> space would be especially useful.  Our drivers are going to get more
> featureful, and we're going to need a versioning/compatibility scheme
> too.

So do you like the splitting of the config space into host and driver part
into 128 bytes each?
That would split the need for versioning between host and driver.

Do you think the config space needs to be writable by the driver?
For the host part, we need to have it writable in case of a PCI host,
which finds stuff like BAR sizes using write access, but requiring that
every host has writable configuration data could make it harder for
some hosts.

	Arnd <><