KVM: PPC: Book3S HV: Don't drop low-order page address bits

Commit caaa4c804fae ("KVM: PPC: Book3S HV: Fix physical address calculations") unfortunately resulted in some low-order address bits getting dropped in the case where the guest is creating a 4k HPTE and the host page size is 64k. By getting the low-order bits from hva rather than gpa we miss out on bits 12 - 15 in this case, since hva is at page granularity. This puts the missing bits back in.

---
Alex, please apply this to your for-3.13 branch.

Thanks,
Paul.

 arch/powerpc/kvm/book3s_hv_rm_mmu.c | 1 +
 1 file changed, 1 insertion(+)

index 1931aa3..8689e2e 100644
@@ -240,6 +240,7 @@ long kvmppc_do_h_enter(struct kvm *kvm, unsigned long flags,
 			is_io = hpte_cache_bits(pte_val(pte));
 			pa = pte_pfn(pte) << PAGE_SHIFT;
 			pa |= hva & (pte_size - 1);
+			pa |= gpa & ~PAGE_MASK;
 		}
 	}
 

1.8.4.2

Thanks, applied to for-3.13.

Alex