On 11/28/2017 03:01 PM, Cornelia Huck wrote: > On Tue, 28 Nov 2017 14:25:08 +0100 > Christian Borntraeger <borntrae...@de.ibm.com> wrote: > >> On 11/28/2017 02:17 PM, Halil Pasic wrote: > >>> In the meanwhile I strongly prefer option 1 (at the ccw devices). I've just >>> sent a v2, and IMHO it shows the limitations of machine properties very >>> well. >> >> option 1 is still fine with me. > > I still dislike it a lot. Machine properties have their own set of > problems, but I think devices are really the wrong place for a global > property. > >> >>> >>>>> >>>>> Halil, do you see any chance to do this for 2.12? There's plenty of >>>>> time left, and I don't think it's too hard. If not, we don't have any >>>>> other option than proposal 3, even though I don't like it a lot. >>>>> >>> >>> I do think we have enough time to do this right. And of course I'm willing >>> to do it right. IMHO the 3 options summarized by Connie are not the only >>> options. But if we go for reworking our QOM composition tree, it will take >>> a lot of discussion. I'm not sure all the required people have enough spare >>> time for that. >>> >>> Halil >> >> I am actually not sure if we really want to have a "user-definable default >> css" anytime >> soon. I think it might really open a can of worms in regard to management >> tooling. > > Fair enough, there's really no short-term use case. > >> What I want now is to enable vfio-ccw for libvirt and Linux guests and for >> that >> we need to lift the restriction of having vfio not in FE. > > This, however, worries me. I don't really consider vfio-ccw ready for > prime time yet. We still need to figure out path handling at the very > least. And I'm still not sure that our non-handling of halt/clear won't > cause major issues with e.g. a storage server error recovery. > > Can we flag vfio-ccw as experimental or such in libvirt?
We should have flagged vfio-ccw experimental in QEMU then. e.g. by using x-vfio-ccw or whatever.I dont think we can do that after the facts. I am not that deep into vfio-ccw, but I was under the impression that the missing features should not affect the vfio interface that is created by libvirt. After all this should just be a "-device vfio-ccw,....." statement and some setup steps beforehand (unbind + setup of vfio-ccw) If your concern is the serviceability I think it would be valid for a RHEL KVM to disable vfio-ccw in the kernel. Maybe we could even provide a config for QEMU? PS: I learned from the PCI and CCW experience that I do not want to upstream kernel/qemu code unless I have a working libvirt code that shows that the thing will work.