So the suitable way of doing this will be checking for `VIR_CPU_TYPE_HOST` and perform a comparison in this case and still return IDENTICAL for all other cases right? Then the upper layer caller(like OpenStack Nova) will have to pass source host cpu info as a parameter. BR, Zhenyu On Mon, Sep 7, 2020 at 7:16 PM Jiri Denemark <jdenemar@redhat.com> wrote:
On Mon, Sep 07, 2020 at 09:21:02 +0800, Zhenyu Zheng wrote:
Thanks alot for the reply,
This sounds like a valid use case (not sure it is that useful), but we
need to be careful. But we should make sure implementing this does not break anything. This means, we need to do different things depending on the type of the CPU definition we are asked to compare. Guest CPU definitions should keep the old behaviour (return IDENTICAL) and host CPU definitions can be compared to the host CPU. But when doing so, don't get too influenced by x86 code, which I believe is way too complicated for ARM. Specifically you don't need to create armCompute and mess with guestData and other stuff there as all you want to do is compare the two CPU definitions. In x86 the same function is used for several things, but that's not the case for ARM.
for this, what I have in mind now is that we can check `VIR_CPU_MODE_HOST_PASSTHROUGH` and if that is the case, we compare CPU vendors and models to allow only identical definitions to pass, like the implementation of
https://gitlab.com/libvirt/libvirt/-/blob/master/src/cpu/cpu_ppc64.c#L505 ,
This doesn't sound like a good idea. PPC is quite special in the way it uses host-model and host-passthrough. But anyway, it's actually easier to just return IDENTICAL for host-passthrough CPUs (and all other guest CPUs) than copying the host CPU from capabilities and comparing it to itself.
(this is because when a VM is in host-passthrough mode, its' CPU xml reflects the original host CPU definition, and we actually compare the source host and destination host CPU definitions,
The CPU definition of a domain with host-passthrough remains the same. That is, it doesn't reflect the original host CPU definition, it's still just host-passthrough. The only way to compare source and destinations host CPUs is by taking the host CPU def from one host and passing it to cpu-compare called on the other host.
Jirka