Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754013Ab3JCML3 (ORCPT ); Thu, 3 Oct 2013 08:11:29 -0400 Received: from mx1.redhat.com ([209.132.183.28]:6431 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752238Ab3JCML2 (ORCPT ); Thu, 3 Oct 2013 08:11:28 -0400 Date: Thu, 3 Oct 2013 15:11:27 +0300 From: Gleb Natapov To: Paolo Bonzini Cc: linux-kernel@vger.kernel.org, kvm@vger.kernel.org Subject: Re: [PATCH 5/7] KVM: mmu: remove argument to kvm_init_shadow_mmu and kvm_init_shadow_ept_mmu Message-ID: <20131003121126.GF17294@redhat.com> References: <1380725776-14948-1-git-send-email-pbonzini@redhat.com> <1380725776-14948-6-git-send-email-pbonzini@redhat.com> <20131003112555.GD17294@redhat.com> <524D5A2D.6070001@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <524D5A2D.6070001@redhat.com> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 5847 Lines: 151 On Thu, Oct 03, 2013 at 01:51:09PM +0200, Paolo Bonzini wrote: > Il 03/10/2013 13:25, Gleb Natapov ha scritto: > > On Wed, Oct 02, 2013 at 04:56:14PM +0200, Paolo Bonzini wrote: > >> The initialization function in mmu.c can always use walk_mmu, which > >> is known to be vcpu->arch.mmu. Only init_kvm_nested_mmu is used to > >> initialize vcpu->arch.nested_mmu. > >> > >> Signed-off-by: Paolo Bonzini > >> --- > >> arch/x86/kvm/mmu.c | 15 +++++++++------ > >> arch/x86/kvm/mmu.h | 5 ++--- > >> arch/x86/kvm/svm.c | 4 ++-- > >> arch/x86/kvm/vmx.c | 4 ++-- > >> 4 files changed, 15 insertions(+), 13 deletions(-) > >> > >> diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c > >> index 40772ef..ac598c8 100644 > >> --- a/arch/x86/kvm/mmu.c > >> +++ b/arch/x86/kvm/mmu.c > >> @@ -3742,11 +3742,13 @@ static void init_kvm_tdp_mmu(struct kvm_vcpu *vcpu) > >> update_last_pte_bitmap(vcpu, context); > >> } > >> > >> -void kvm_init_shadow_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *context) > >> +void kvm_init_shadow_mmu(struct kvm_vcpu *vcpu) > >> { > >> bool smep = kvm_read_cr4_bits(vcpu, X86_CR4_SMEP); > >> + struct kvm_mmu *context = vcpu->arch.walk_mmu; > > I'd rather use &vcpu->arch.mmu here. > > > >> + > >> ASSERT(vcpu); > >> - ASSERT(!VALID_PAGE(vcpu->arch.mmu.root_hpa)); > >> + ASSERT(!VALID_PAGE(context->root_hpa)); > >> > >> if (!is_paging(vcpu)) > >> nonpaging_init_context(vcpu, context); > >> @@ -3765,11 +3767,12 @@ void kvm_init_shadow_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *context) > >> } > >> EXPORT_SYMBOL_GPL(kvm_init_shadow_mmu); > >> > >> -void kvm_init_shadow_ept_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *context, > >> - bool execonly) > >> +void kvm_init_shadow_ept_mmu(struct kvm_vcpu *vcpu, bool execonly) > >> { > >> + struct kvm_mmu *context = vcpu->arch.walk_mmu; > >> + > >> ASSERT(vcpu); > >> - ASSERT(!VALID_PAGE(vcpu->arch.mmu.root_hpa)); > >> + ASSERT(!VALID_PAGE(context->root_hpa)); > >> > >> context->shadow_root_level = kvm_x86_ops->get_tdp_level(); > >> > >> @@ -3790,7 +3793,7 @@ EXPORT_SYMBOL_GPL(kvm_init_shadow_ept_mmu); > >> > >> static void init_kvm_softmmu(struct kvm_vcpu *vcpu) > >> { > >> - kvm_init_shadow_mmu(vcpu, vcpu->arch.walk_mmu); > >> + kvm_init_shadow_mmu(vcpu); > >> vcpu->arch.walk_mmu->set_cr3 = kvm_x86_ops->set_cr3; > >> vcpu->arch.walk_mmu->get_cr3 = get_cr3; > >> vcpu->arch.walk_mmu->get_pdptr = kvm_pdptr_read; > > And change walk_mmu to mmu here too for consistency with all other > > places. Basically if you want to initialize use mmu or nested_mmu. > > Use walk_mmu pointer only when you need to use mmu. > > Makes sense, especially considering how kvm_init_shadow_mmu initializes > vcpu->arch.mmu.base_role directly. > > Something like this (large enough that I'll probably make it a separate > patch in v2) > > diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c > index ac598c8..d1f53cf 100644 > --- a/arch/x86/kvm/mmu.c > +++ b/arch/x86/kvm/mmu.c > @@ -3702,7 +3704,7 @@ static void paging32E_init_context(struct kvm_vcpu *vcpu, > > static void init_kvm_tdp_mmu(struct kvm_vcpu *vcpu) > { > - struct kvm_mmu *context = vcpu->arch.walk_mmu; > + struct kvm_mmu *context = &vcpu->arch.mmu; > > context->base_role.word = 0; > context->page_fault = tdp_page_fault; > @@ -3745,7 +3747,7 @@ static void init_kvm_tdp_mmu(struct kvm_vcpu *vcpu) > void kvm_init_shadow_mmu(struct kvm_vcpu *vcpu) > { > bool smep = kvm_read_cr4_bits(vcpu, X86_CR4_SMEP); > - struct kvm_mmu *context = vcpu->arch.walk_mmu; > + struct kvm_mmu *context = &vcpu->arch.mmu; > > ASSERT(vcpu); > ASSERT(!VALID_PAGE(context->root_hpa)); > @@ -3759,17 +3761,17 @@ void kvm_init_shadow_mmu(struct kvm_vcpu *vcpu) > else > paging32_init_context(vcpu, context); > > - vcpu->arch.mmu.base_role.nxe = is_nx(vcpu); > - vcpu->arch.mmu.base_role.cr4_pae = !!is_pae(vcpu); > - vcpu->arch.mmu.base_role.cr0_wp = is_write_protection(vcpu); > - vcpu->arch.mmu.base_role.smep_andnot_wp > + context->base_role.nxe = is_nx(vcpu); > + context->base_role.cr4_pae = !!is_pae(vcpu); > + context->base_role.cr0_wp = is_write_protection(vcpu); > + context->base_role.smep_andnot_wp > = smep && !is_write_protection(vcpu); > } > EXPORT_SYMBOL_GPL(kvm_init_shadow_mmu); > > void kvm_init_shadow_ept_mmu(struct kvm_vcpu *vcpu, bool execonly) > { > - struct kvm_mmu *context = vcpu->arch.walk_mmu; > + struct kvm_mmu *context = &vcpu->arch.mmu; > > ASSERT(vcpu); > ASSERT(!VALID_PAGE(context->root_hpa)); > @@ -3793,11 +3795,13 @@ EXPORT_SYMBOL_GPL(kvm_init_shadow_ept_mmu); > > static void init_kvm_softmmu(struct kvm_vcpu *vcpu) > { > + struct kvm_mmu *context = &vcpu->arch.mmu; > + > kvm_init_shadow_mmu(vcpu); > - vcpu->arch.walk_mmu->set_cr3 = kvm_x86_ops->set_cr3; > - vcpu->arch.walk_mmu->get_cr3 = get_cr3; > - vcpu->arch.walk_mmu->get_pdptr = kvm_pdptr_read; > - vcpu->arch.walk_mmu->inject_page_fault = kvm_inject_page_fault; > + context->set_cr3 = kvm_x86_ops->set_cr3; > + context->get_cr3 = get_cr3; > + context->get_pdptr = kvm_pdptr_read; > + context->inject_page_fault = kvm_inject_page_fault; > } > > static void init_kvm_nested_mmu(struct kvm_vcpu *vcpu) Yes. > > How far should I go? Should I also remove the context argument from > nonpaging_init_context and friends, changing it to a local variable? > (Doesn't seem like a big improvement in clarity). > If it does not no need to do it. Hard to judge for me without trying. -- Gleb. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/