On 2024/7/4 09:35, maobibo wrote:
In another thread I found that Jiaxun said he has a solution to make
LBT be a vcpu feature and still works well. However, that may take
some time and is too late for 6.11.
But we have another choice now: just remove the UAPI and vm.c parts in
this series, let the LBT main parts be upstream in 6.11, and then
solve other problems after 6.11. Even if Jiaxun's solution isn't
usable, we can still use this old vm feature solution then.
IMO this is the best approach to make some progress.
I am sure it is best if it is VM feature for LBT feature detection,
LSX/LASX feature detection uses CPU feature, we can improve it later.
Please justify the reason, we should always be serious on UAPI design
choices.
I don't really understand why the approach worked so well on Arm &
RISC-V is not working
for you.
I understand you may have some plans in your mind, please elaborate so
we can smash
them together. That's how community work.
For host cpu type or migration feature detection, I have no idea now,
also I do not think it will be big issue for me, I will do it with
scheduled time. Of source, welcome Jiaxun and you to implement host
cpu type or migration feature detection.
My concern is if you allow CPU features to have "auto" property you are
risking create
inconsistency among migration. Once you've done that it's pretty hard to
get rid of it.
Please check how RISC-V dealing with CPU features at QMP side.
I'm not meant to hinder your development work, but we should always
think ahead.
Thanks
- Jiaxun
