Merge branch 'kvm-arm64/vncr-fixes' into next

* kvm-arm64/vncr-fixes:
  : Fixes for handling L1 VNCR mappings, courtesy of Marc Zyngier
  :
  :  - Recompute the VNCR software TLB when the MMU is toggled at stage-1
  :
  :  - Improve TLB invalidation intersection to handle TLBIs affecting the
  :    end of the VA space
  :
  :  - Fix race to invalidate the VNCR fixmap between TLBI emulation and
  :    vcpu_put()
  :
  :  - Add missing sign extension for computing TLBI ranges
  :
  :  - Make VNCR invalidation participate in the MMU notifier seqcount,
  :    preventing a concurrent VNCR TLB fill from consuming a stale
  :    translation
  KVM: arm64: Correctly cap TLBI Range to the architural limit
  KVM: arm64: Add VNCR TLB tracking again
  KVM: arm64: Make VNCR invalidation participate in MMU invalidation retry
  KVM: arm64: Sign-extend VA for range-based TLBI invalidation
  KVM: arm64: Handle VNCR TLB invalidation race with vcpu_put() VNCR unmapping
  KVM: arm64: Correctly handle end of VA space TLBI invalidation
  KVM: arm64: Consider SCTLR_EL2.M when mapping the L1 VNCR page
  KVM: arm64: Handle negative S1 walk levels in VNCR TLB size evaluation
  KVM: arm64: Remove VM-wide VNCR mapping counter

Signed-off-by: Oliver Upton <oupton@kernel.org>