On 22/01/2021 18:39, Boris Ostrovsky wrote:
On 1/22/21 1:33 PM, Julien Grall wrote:
Thanks. In which case, I would suggest to move the definition of ignore_msrs in the idle
to an x86 specific structure. Maybe "arch_x86"?
I did consider this but left it in common area since there are already a bunch
of fields there that are x86-specific.
Most of them were introduced before Arm existed :).
If there is desire to create an x86 sub-struct then we should move all of them
there.
The libxl interface is meant to be stable at the source level but not
ABI stable. So we wouldn't be able to move the existing field in a new
x86 without adding a compat layer (see for instance hvm.altp2m vs
altp2m). I don't think this is worth it for them.
That said, we should try to treat each architecture equally. So I think
it is not a good idea to continue to pursue adding an field in common
area if the field is only meant to be used for a single arch.
If Ian or Wei thinks it is fine to define in common code, then so it be.
Cheers,
--
Julien Grall
