On Mon, Apr 27, 2015 at 11:45:56AM +0200, Gerd Hoffmann wrote: > On Fr, 2015-04-10 at 14:31 +0200, Gerd Hoffmann wrote: > > Hi, > > > > Here is a series of patches for virtio-pci, with some cleanups and > > preparations for virtio-vga. > > > > Patches 1-6 clean up the modern (virtio-1.0) memory region > > initialization a bit. They are preparations for the other patches, > > but IMHO are also a useful cleanup on their own. > > > > Patches 7+8 changes initialization so virtio-vga has a chance to > > arrange things as it pleases. > > > > Patches 1-8 should apply fine on top of mst's virtio-1.0 branch, and > > I'm looking for reviews of them. > > > > Patch 9 is the showcase for patches 7+8. It has more dependencies > > though, if you wanna play with that I'd suggest to fetch > > https://www.kraxel.org/cgit/qemu/log/?h=rebase/vga-wip > > Ping? mst, had you a chance to look at this?
I'll merge this on the virtio 1 branch shortly. > What are your virtio-1.0 merge plans for qemu, now that the 2.4 tree is > open for development? I'd love to see a quick merge so I can proceed > with bits having dependencies on virtio-1.0. > > thanks, > Gerd That branch still needs some work. I hope to have things in mergeable state end of may. In particular - the way we have structured breaks migration and makes supporting it properly tricky, so I'm refactoring some code. - virtio 1 spec forbids devices from accessing memory until DRIVER_OK is set, but we also need to support legacy guests which trigger ready state by kicking some ring. Do you need some bits from master on that branch? I could rebase it if it's helpful. -- MST
