On Fri, Feb 10, 2023 at 9:11 AM Sean Christopherson <seanjc@xxxxxxxxxx> wrote: > > > + gfn_t gfn = kvm_mmu_page_get_gfn(sp, iterator.index); > > + int ret = mmu->sync_spte(vcpu, sp, iterator.index); > > + > > + if (ret < 0) > > + mmu_page_zap_pte(vcpu->kvm, sp, iterator.sptep, NULL); > > + if (ret) > > + kvm_flush_remote_tlbs_with_address(vcpu->kvm, gfn, 1); > > Why open code kvm_flush_remote_tlbs_sptep()? Does it actually shave enough > cycles to be visible? Although I have read the code of sync_page() many times, I don't know why I had been having the assumption that it can possibly change the sp->gfns[] (now sp->shadowed_translation[]). I will add the following comments before calling kvm_mmu_page_get_gfn(): Get the gfn beforehand for later flushing. Although mmu->sync_spte() doesn't change it, but just avoid dependence. Or I will use kvm_flush_remote_tlbs_sptep() with comments stating that the gfn will not be changed. > > If open coding is really justified, can you rebase on one of the two branches? > And then change this to kvm_flush_remote_tlbs_gfn(). > > https://github.com/kvm-x86/linux/tree/next > https://github.com/kvm-x86/linux/tree/mmu The code was based on https://github.com/kvm-x86/linux/tree/mmu. Thanks Lai