> > @@ -969,6 +983,7 @@ static const DisplayChangeListenerOps > > display_listener_gl_ops = { > > > > .dpy_gl_scanout_disable = qemu_spice_gl_scanout_disable, > > .dpy_gl_scanout_texture = qemu_spice_gl_scanout_texture, > > + .dpy_gl_scanout_dmabuf = qemu_spice_gl_scanout_dmabuf, > > .dpy_gl_update = qemu_spice_gl_update, > > }; > > > > -- > > 2.9.3 > > > > > > Looks good. Btw, how is multi-monitor designed? Is there going to be a > scanout/dmabuf per monitor when using vfio/mdev?
Yes, that would be one QemuConsole per head (simliar to virtio-gpu). There is no multihead support right now though. Not in the drivers, and also not in the vfio API. Adding api for it without an actual test case didn't look like a good plan to me. Extending the API later is easier than dealing with a broken api in case we miss some detail and only figure later when the first driver tries to use the untested API ... cheers, Gerd