On Fri, Jan 08, 2021 at 02:50:25AM -0500, Matt Coleman wrote:
On Nov 26, 2020, at 9:42 AM, Daniel P. Berrangé <berrange@redhat.com> wrote:
You ought to pass def->controllers[i] into this method and validate as many properties as practical. At very least validate the model and report VIR_ERR_CONFIG_UNSUPPORTED for any you can't emulate.
Hyper-V's SCSI controllers are paravirtualized ("synthetic" in Hyper-V terms). I've been omitting the model setting from my XML files.
I can think of three options for how to handle this:
1. only support VIR_DOMAIN_CONTROLLER_MODEL_SCSI_DEFAULT
2. support VIR_DOMAIN_CONTROLLER_MODEL_SCSI_DEFAULT and VIR_DOMAIN_CONTROLLER_MODEL_SCSI_AUTO
3. add VIR_DOMAIN_CONTROLLER_MODEL_SCSI_HYPERV and support VIR_DOMAIN_CONTROLLER_MODEL_SCSI_HYPERV, VIR_DOMAIN_CONTROLLER_MODEL_SCSI_DEFAULT, and VIR_DOMAIN_CONTROLLER_MODEL_SCSI_AUTO
Adding another virDomainControllerModelSCSI seems unnecessary to me because Hyper-V doesn't support emulating any SCSI controllers; it only has its paravirtualized SCSI functionality.
Which do you recommend?
IIRC "DEFAULT" is when no model=... is given at all and usually a driver would expand that into the suitable SCSI controller model for the guest. With that in mind I'd suggest (2).
Probably ought to reject any info->type which is not VIR_DOMAIN_ADDRESS_TYPE_NONE, since you're not attempting todo any device addressing at this time.
There are no address settings for Hyper-V SCSI controllers, so this will always be the case. I'll report an error if address settings are provided.
Thanks! Matt
Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|