Hi Salil,
On 01/09/2022 10:02, Salil Mehta wrote:
From: Salil Mehta via Linaro-open-discussions
Ideally, I should be able to share the code by COB tomorrow if I find no blockers during testing (which I do not see as of now).
Beware the WFI in user-space support that is needed for implementing CPU_SUSPEND is a little different from the previous version. From memory the change is how the vCPU gets restarted....
I see. thanks for sharing this info. It might be helpful if you have reference to lore.kernel patches where it got changed?
Aah, got it. Perhaps this patch? Yes, I can see what you mean.
[RFC PATCH 2/5] KVM: arm64: Move WFI execution to check_vcpu_requests() https://lore.kernel.org/kvm/20210608154805.216869-1-jean-philippe@linaro.org...
Yes, that was what Jean's qemu work originally supported, but since then mainline has merged 7b33a09d036f ("KVM: arm64: Add support for userspace to suspend a vCPU"), which is similar, but different.
Most of that patch from Jean got folded out in the rebase, and mainline now has some difference in the way the user-space interface behaves.
Thanks,
James