Fix bugs in set_msr_mce() where KVM returns '-1' instead of '1' when
rejecting a WRMSR, which results in -EPERM getting returned to userspace.

Patches 2 and 3 are cleanups to make {g,s}et_msr_mce a bit less wierd
(especially with respect to Jue's CMCI series) and a bit less magical.

Sean Christopherson (3):
  KVM: x86: Signal #GP, not -EPERM, on bad WRMSR(MCi_CTL/STATUS)
  KVM: x86: Use explicit case-statements for MCx banks in
    {g,s}et_msr_mce()
  KVM: x86: Add helpers to identify CTL and STATUS MCi MSRs

 arch/x86/kvm/x86.c | 84 +++++++++++++++++++++++++++-------------------
 1 file changed, 50 insertions(+), 34 deletions(-)


base-commit: 2764011106d0436cb44702cfb0981339d68c3509
-- 
2.36.0.550.gb090851708-goog