On 13/08/19 23:14, Alex Williamson wrote:
>> Do we even need the call to slot_handle_all_level? The rmap update
>> should be done already by kvm_mmu_prepare_zap_page (via
>> kvm_mmu_page_unlink_children -> mmu_page_zap_pte -> drop_spte).
>>
>> Alex, can you replace it with just "flush = false;"?
> Replace the continue w/ flush = false? I'm not clear on this
> suggestion. Thanks,
diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
index 24843cf49579..382b3ee303e3 100644
--- a/arch/x86/kvm/mmu.c
+++ b/arch/x86/kvm/mmu.c
@@ -5664,7 +5668,7 @@ kvm_mmu_invalidate_zap_pages_in_memslot(struct kvm
if (list_empty(&kvm->arch.active_mmu_pages))
goto out_unlock;
- flush = slot_handle_all_level(kvm, slot, kvm_zap_rmapp, false);
+ flush = false;
for (i = 0; i < slot->npages; i++) {
gfn = slot->base_gfn + i;
