Since last year and beyond has been talk of eventually having a Gallium3D driver for Red Hat's SPICE protocol for use with QEMU/KVM. Such a driver -- similar to that of what's available in VirtualBox and VMware -- hasn't come, but QEMU has gained VGA pass-through support (similar to Xen) so that a dedicated GPU could be assigned to a guest, but it's still not an easy setup and the support is a work-in-progress.
Finally with the Linux 3.10 kernel there is now a QXL KMS driver to provide a basic DRM kernel driver for QEMU. There's also been talk about a QXL Gallium3D wrapper, but no code had surfaced.
Now in the discussion about Oracle wanting a Mesa driver for VirtualBox, David Airlie has publicly talked about some new Mesa code he's working on. To get straight to the point, here's the relevant part of his mailing list response:
I have written some code to do something similiar [to Oracle's proposed VirtualBox Mesa driver approach], pushing all the internal gallium API over a pipe to a GL renderer on the other side of the pipe. It isn't complete, it was only a stepping stone on my research to qemu 3D work.His Mesa renderer branch can be found here and was last updated a few days ago. Basically, he's experimenting with Mesa code to be run as a guest that could then pass the OpenGL commands onto the host through QEMU to then be processed by the host's hardware and its OpenGL driver.
The biggest issue I see with doing something like that is how to make final display of things to the X server and making things like texture from pixmap work efficiently.
The stuff I've done is on the renderer-1 branch of my mesa repo but its moved on a lot from that point, and I doubt the pipe code I wrote even builds anymore, since I've mostly moved to testing it as part of qemu now. But it currently just passes TGSI shaders through and converts them into GLSL using an assembler that will look something like the one in wine I suspect.
With Red Hat's vested interest in GNOME 3.x, which now requires OpenGL acceleration for the desktop, hopefully this QEMU 3D support will materialize into a workable solution in the not so distant future.