Re: [PATCH 0/7] KVM: SVM: baby steps towards nested state migration

[Krish Sadhukhan]

On 5/15/20 10:41 AM, Paolo Bonzini wrote:
> Here are some refactorings to prepare for an SVM implementation of
> KVM_SET_NESTED_STATE.  It's a prerequisite for that to eliminate
> exit_required, moving exceptions to svm_check_nested_events.  However:
>
> - I might work on that soon, because it's needed to handle RSM when
> the L1 hypervisor wants to get it from #UD rather than the specific
> RSM intercept
>
> - this should be enough to get a quick prototype, that I need in order to
> debug a particularly crazy bug and figure out its reproducibility.
>
> So, I am getting these patches out of my todo list for now.
>
> Thanks,
>
> Paolo
>
> Paolo Bonzini (7):
>    KVM: SVM: move map argument out of enter_svm_guest_mode
>    KVM: SVM: extract load_nested_vmcb_control
>    KVM: SVM: extract preparation of VMCB for nested run
>    KVM: SVM: save all control fields in svm->nested
>    KVM: nSVM: remove HF_VINTR_MASK
>    KVM: nSVM: do not reload pause filter fields from VMCB
>    KVM: SVM: introduce data structures for nested virt state
>
>   arch/x86/include/asm/kvm_host.h |   1 -
>   arch/x86/include/uapi/asm/kvm.h |  26 +++++++-
>   arch/x86/kvm/svm/nested.c       | 115 +++++++++++++++++---------------
>   arch/x86/kvm/svm/svm.c          |  11 ++-
>   arch/x86/kvm/svm/svm.h          |  28 +++++---
>   5 files changed, 116 insertions(+), 65 deletions(-)
Reviewed-by: Krish Sadhukhan <krish.sadhukhan@xxxxxxxxxx>