On 04.02.2025 18:35, Andrew Cooper wrote:
> On 03/02/2025 8:58 am, Guillaume wrote:
>> Oh cool, thanks a lot for the explanation.
>> I added the "vzeroupper" and Xen crashes so it looks like the CPUID
>> emulation is buggy. Also I was able to try it using a VM (same debian
>> testing) running on virt-manager+kvm and it works fine (Xen in debug
>> mode). I will have a look by printing the xstate when running on
>> virt-manager+KVM and I will also run the xen-cpuid command to see the
>> difference just by curiosity as with your test we already spotted the
>> issue.
>> Thanks again for your enlightenment. I will continue my testing later
>> today and if you need me to test something else you are welcome, just
>> ask I will do my best.
>
> It sounds like KVM has a better CPUID emulation than VirtualBox.
>
> It would be ideal to report this bug with VirtualBox.
>
> But, as you identified originally, it's not nice that Xen simply like
> this. We should see about what to for Xen, seeing as we're close to the
> line on 4.20. I'm thinking maybe making the xstate checks non-fatal in
> the cpu_has_hypervisor case. Thoughts?
In principle: Yes, that's an option. But then we need to suppress use of
xstate_{,un}compressed_size() anywhere, perhaps by disabling respective
features. Not sure whether that is what you meant to imply.
Jan
