On Wed, Nov 20, 2024 at 12:16:26 +0000, Daniel P. Berrangé wrote:
On Tue, Nov 19, 2024 at 07:49:47PM +0100, Jiri Denemark wrote:
> The signatures in the CPU map are used for matching physical CPUs and
> thus we need to cover all possible real world variants we know about.
> When adding a new version of an existing CPU model, we should copy the
> signature(s) of the existing model rather than replacing it with the
> signature that QEMU uses.
Also sanity check that the signature QEMU uses is part of
libvirt's existing recorded signature list ?
I don't think this is needed. CPU model versions in QEMU do not have the
ability to change the signature so they inherit it from the base model.
And the base model is either new to libvirt in which case we will use
the signature from QEMU or it is an existing model with signatures
configured according to real world CPUs and it doesn't really matter
whether QEMU agrees with it.
Remember the CPU signature is only used for selecting the best CPU model
for describing a host CPU and the relation to the numbers used by QEMU
is irrelevant. Well, unless you are nested and want to detect a host CPU
model for the "host" CPU created by QEMU. But anyway, I'm pretty sure
libvirt signatures include the one used by QEMU for all models in our
CPU map.
Jirka