[PATCH v7 7/9] drm/virtio: Improve DMA API usage for shmem BOs

[Gerd Hoffmann]

On Wed, Jul 06, 2022 at 10:22:52AM +0300, Dmitry Osipenko wrote:
> On 7/6/22 10:13, Gerd Hoffmann wrote:
> >   Hi,
> > 
> >> 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.
> > 
> > aarch64 uses acpi by default (can be disabled via 'qemu -no-acpi').
> > Not fully sure about arm(v7).
> > 
> > Even with DT it should work because DT only describes the virtio-mmio
> > 'slots', not the actual virtio devices.
> > 
> >> There is no virgl support because it's a virtio-gpu-device and not
> >> virtio-gpu-device-gl that is PCI-only in Qemu. Hence everything seems good.
> > 
> > It's named 'virtio-gpu-gl-device'
> 
> Ah, thanks again! Just quickly tested virtio-gpu-gl-device and
> everything works too for MMIO GPU on microvm, including virgl and Xorg
> (glamor).
> 
> [drm] features: +virgl +edid -resource_blob -host_visible
> [drm] features: -context_init
> [drm] number of scanouts: 1
> [drm] number of cap sets: 2
> [drm] cap set 0: id 1, max-version 1, max-size 308
> [drm] cap set 1: id 2, max-version 2, max-size 696
> [drm] Initialized virtio_gpu 0.1.0 0 for LNRO0005:01 on minor 0
> virtio-mmio LNRO0005:01: [drm] drm_plane_enable_fb_damage_clips() not called

Cool.  Havn't found the time to try s390x, so I'm taking that as good
enough test that we don't have any unnoticed dependencies on pci.

Queued up.  I'll go over a few more pending patches, and assuming no
issues show up in testing this should land in drm-misc-next in a few
hours.

take care,
  Gerd