We recently moved anything related to PC updates into the guest entry
code to help with the protected KVM effort. However, I missed a
critical point: userspace needs to be able to observe state changes
when the vcpu exits. Otherwise, live migration is a bit broken and
vcpu reset can fail (as reported by Zenghui). Not good.
These two patches aim at fixing the above, and carry a Cc stable.
* From v1:
- Sanitized flag checking
- Added comment about relying on __kvm_adjust_pc() being preempt-safe
- Dropped superfluous includes
Marc Zyngier (2):
KVM: arm64: Move __adjust_pc out of line
KVM: arm64: Commit pending PC adjustemnts before returning to
userspace
arch/arm64/include/asm/kvm_asm.h | 3 +++
arch/arm64/kvm/arm.c | 10 ++++++++++
arch/arm64/kvm/hyp/exception.c | 18 +++++++++++++++++-
arch/arm64/kvm/hyp/include/hyp/adjust_pc.h | 18 ------------------
arch/arm64/kvm/hyp/nvhe/hyp-main.c | 8 ++++++++
arch/arm64/kvm/hyp/nvhe/switch.c | 3 +--
arch/arm64/kvm/hyp/vhe/switch.c | 3 +--
7 files changed, 40 insertions(+), 23 deletions(-)
--
2.29.2
