On Tue, Dec 19, 2017 at 02:24:47PM -0600, Scott Garfinkle wrote: > On Tue, 2017-12-19 at 20:05 +0100, Jiri Denemark wrote: > > On Tue, Dec 19, 2017 at 12:39:26 -0600, Scott Garfinkle wrote: > > > Qemu now allows case-insensitive specification of CPU models. This fixes the > > > resulting problems on POWER arch machines. I believe a similar change is needed > > > in src/cpu/cpu_x86.c but I don't have a way to test this. > > > > I believe this change is not needed or even acceptable at all. What kind > > of issue are you trying to solve? > So, the problem is that as of qemu 2.11, Power arch guests don't start, > since libvirt ppc code is looking for the string POWER to determine > whether the system supports the model at hand. > > Comment from our qemu maintainer: "There was a rework in QEMU side to > allow case-insensitive specification of CPU models. As a side-effect of > that some models no longer show up as all caps in -cpu ? / > query-cpu-definitions. This requires libvirt changes to avoid relying on > specific capitalized models." > > Are you concerned that we SHOULD be looking for case-sensitive models in > libvirt? Is this a problem absent qemu? You need to dinstinguish what libvirt does internally from what it does in the public XML/APIs. Your change affected both, so it is wrong. Any fix should only apply to libvirt's internal communication with QEMU, and not affect what's done in the public XML/APIs. Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :| -- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list
