Merge branch kvm-arm64/virtual-cache-geometry into kvmarm/next

* kvm-arm64/virtual-cache-geometry:
  : Virtualized cache geometry for KVM guests, courtesy of Akihiko Odaki.
  :
  : KVM/arm64 has always exposed the host cache geometry directly to the
  : guest, even though non-secure software should never perform CMOs by
  : Set/Way. This was slightly wrong, as the cache geometry was derived from
  : the PE on which the vCPU thread was running and not a sanitized value.
  :
  : All together this leads to issues migrating VMs on heterogeneous
  : systems, as the cache geometry saved/restored could be inconsistent.
  :
  : KVM/arm64 now presents 1 level of cache with 1 set and 1 way. The cache
  : geometry is entirely controlled by userspace, such that migrations from
  : older kernels continue to work.
  KVM: arm64: Mark some VM-scoped allocations as __GFP_ACCOUNT
  KVM: arm64: Normalize cache configuration
  KVM: arm64: Mask FEAT_CCIDX
  KVM: arm64: Always set HCR_TID2
  arm64/cache: Move CLIDR macro definitions
  arm64/sysreg: Add CCSIDR2_EL1
  arm64/sysreg: Convert CCSIDR_EL1 to automatic generation
  arm64: Allow the definition of UNKNOWN system register fields

Signed-off-by: Oliver Upton <oliver.upton@linux.dev>

1 files changed, 0 insertions, 4 deletions

diff --git a/arch/arm64/include/asm/kvm_emulate.h b/arch/arm64/include/asm/kvm_emulate.h
index 193583df2d9c..b1134af843bd 100644
--- a/arch/arm64/include/asm/kvm_emulate.h
+++ b/arch/arm64/include/asm/kvm_emulate.h
@@ -88,10 +88,6 @@ static inline void vcpu_reset_hcr(struct kvm_vcpu *vcpu)
 	if (vcpu_el1_is_32bit(vcpu))
 		vcpu->arch.hcr_el2 &= ~HCR_RW;
 
-	if (cpus_have_const_cap(ARM64_MISMATCHED_CACHE_TYPE) ||
-	    vcpu_el1_is_32bit(vcpu))
-		vcpu->arch.hcr_el2 |= HCR_TID2;
-
 	if (kvm_has_mte(vcpu->kvm))
 		vcpu->arch.hcr_el2 |= HCR_ATA;
 }