While thinking during idle moments this past weekend (a 2+ hour each way drive) I was thinking perhaps some of the issues I'm facing with cimtest and associating this ControllerRASD is that there's no logical device Controller like there is for Input, Graphics, Console, etc. devices. Additionally, there's a pools and rasds association which I just haven't quite got right. So, I started looking into CIM_Controller and thought well - isn't that more of what we're talking about for a libvirt controller device? Other than a 'ccid' and 'virtio-serial' types the other types in the libvirt list can be correlated to a CIM_Controller "ProtocolSupported" value: ide == IDE fdc == Flexbile Diskette scsi == [one of the] SCSI * types sata == Serial ATA ccid == ?Our own value? virtio-serial == Our own value usb = Universal Serial Bus So I'm headed off in that direction, but before going to far - I wanted to be sure I wasn't off in the weeds too far. Still trying to nail down some minor particulars such as where is the DeviceID value set (eg dev->id), but "so far" for the most part things seem to be better 'cimtest' wise than going with the RASD type. Given that this is the right direction - I'm not sure a ControllerRASD is necessary, is it? The one thing that does seem to be necessary is when a bus is provided for some device, we have to "find" or "ensure" that KVM/CIM_Controller device exists. How the association there will work I haven't yet figured out, but theoretically it seems to be the right path. John