On 01/23/2012 05:30 PM, Daniel P. Berrange wrote:
On Mon, Jan 23, 2012 at 05:23:32PM +0100, Paolo Bonzini wrote:
On 01/23/2012 05:03 PM, Daniel P. Berrange wrote:
> The qemu32/qemu64 models are weird in that the exact combination of > CPUID flags does not match any actual processor. kvm32 and kvm64 are > a better match when not using TCG. Use them when -cpu is only needed > to hardcode a 32-bit guest arch or for kvmclock.
I don't think we can do this as it means the guest CPU may change unexpectedly for existing domains. A 32b domain started on current libvirt would see qemu32, while the same domain started after this patch would see kvm32. Also, IIUC, kvm32 is a fairly newly introduced CPU type for KVM - ie most deployments of KVM won't support it.
Ok, I'll redo this patch with just kvm64. As the wrong subject in 0/3 show, it was an afterthought.
Is kvm64 actually any more widely supported than kvm32 ? I though they were all fairly new.
If I'm wrong, then we could trivially provide a kvm32, by using the kvm64 model, and subtracting the the "long mode" CPU flag.
commit d1cd4bf41906980434bee1c6656e8aeb9a132df3 Author: Andre Przywara <andre.przywara@amd.com> Date: Thu Aug 20 23:34:17 2009 +0200 introduce kvm64 CPU In addition to the TCG based qemu64 type let's introduce a kvm64 CPU type, which is the least common denominator of all KVM-capable x86-CPUs (based on Intel Pentium 4 Prescott). It can be used as a base type for migration. Signed-off-by: Andre Przywara <andre.przywara@amd.com> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com> It is in RHEL6, but let's just scrap this patch until (at least) after 0.9.10. Paolo