On Fri, 2019-02-08 at 17:11 -0500, Cole Robinson wrote: [...]
@@ -4553,6 +4553,11 @@ qemuDomainRNGDefValidate(const virDomainRNGDef *def, modelIsSupported = virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VIRTIO_RNG); break; + case VIR_DOMAIN_RNG_MODEL_VIRTIO_TRANSITIONAL: + case VIR_DOMAIN_RNG_MODEL_VIRTIO_NON_TRANSITIONAL: + modelIsSupported = (virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_PCI_NON_TRANSITIONAL) || + virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY)); + break;
You still need to make sure QEMU_CAPS_DEVICE_VIRTIO_RNG is present when the model is virtio-(non-)transitional. [...]
@@ -884,7 +884,11 @@ qemuDomainDeviceCalculatePCIConnectFlags(virDomainDeviceDefPtr dev, case VIR_DOMAIN_DEVICE_RNG: switch ((virDomainRNGModel) dev->data.rng->model) { case VIR_DOMAIN_RNG_MODEL_VIRTIO: + case VIR_DOMAIN_RNG_MODEL_VIRTIO_NON_TRANSITIONAL: return virtioFlags; + case VIR_DOMAIN_RNG_MODEL_VIRTIO_TRANSITIONAL: + /* Transitional devices only work in conventional PCI slots */ + return pciFlags;
Nit, but for other devices so far you've put the _TRANSITIONAL case first, so please do the same here :) With all of the above addressed, Reviewed-by: Andrea Bolognani <abologna@redhat.com> -- Andrea Bolognani / Red Hat / Virtualization