On AMD hardware, Hyper-V doesn't work nested on KVM if L1 is started with
more than one vCPU. The problem is in NMI which is heavily used by Windows
on SMP systems.
This series fixes a couple of related issues with the current NMI singlestep
logic and makes Windows with Hyper-V happy.
v1->v2:
* simplified patch 4 to call smp_send_reschedule instead of intercepting iret
(Paolo)
* added patch 5 to suppress singlestep on GIF=0 and nested exit (Paolo)
* dropped most of the cover letter blurb because we now singlestep only over
iret, as originally intended (Radim)
* still left nmi_singlestep_guest_rflags and the defensive code accessing it
in -- it shouldn't be needed if we really singlestep only over iret but it
is more robust this way (Radim)
* ran eventinj.flat, found no regression (baseline appears to be broken on
nested, I'll look into it separately) (Paolo)
Ladi Prosek (5):
KVM: SVM: introduce disable_nmi_singlestep helper
KVM: nSVM: do not forward NMI window singlestep VM exits to L1
KVM: SVM: hide TF/RF flags used by NMI singlestep
KVM: SVM: don't NMI singlestep over event injection
KVM: SVM: suppress unnecessary NMI singlestep on GIF=0 and nested exit
