On 7/6/22 00:39, Rob Clark wrote: > On Tue, Jul 5, 2022 at 10:02 AM Dmitry Osipenko > <dmitry.osipenko@xxxxxxxxxxxxx> wrote: >> >> On 7/5/22 18:45, Gerd Hoffmann wrote: >>> Hi, >>> >>>>> Also note that pci is not the only virtio transport we have. >>>> >>>> The VirtIO indeed has other transports, but only PCI is really supported >>>> in case of the VirtIO-GPU in kernel and in Qemu/crosvm, AFAICT. Hence >>>> only the PCI transport was tested. >>> >>> qemu -M microvm \ >>> -global virtio-mmio.force-legacy=false \ >>> -device virtio-gpu-device >>> >>> Gives you a functional virtio-gpu device on virtio-mmio. >>> >>> aarch64 virt machines support both pci and mmio too. >>> s390x has virtio-gpu-ccw ... >> >> Gerd, thank you very much! It's was indeed unclear to me how to test the >> MMIO GPU, but yours variant with microvm works! I was looking for trying >> aarch64 in the past, but it also was unclear how to do it since there is >> no DT support for the VirtIO-GPU, AFAICS. > > just a drive-by note, IME on aarch64 kernels, at least with crosvm, > virtgpu is also a pci device.. the non-pci things in the guest kernel > use dt, but devices on discoverable busses like pci don't need dt > nodes (which is true also in the non-vm case) Sure, I was only looking how to test MMIO GPU on aarch64. Since I haven't a found a way to test MMIO back then, I concluded that the MMIO case wasn't really well supported. -- Best regards, Dmitry
