diff mbox series

[07/30] KVM: nVMX: always update CR3 in VMCS

Message ID	20200529153934.11694-8-pbonzini@redhat.com (mailing list archive)
State	New, archived
Headers	show Return-Path: <SRS0=rOkw=7L=vger.kernel.org=kvm-owner@kernel.org> From: Paolo Bonzini <pbonzini@redhat.com> To: linux-kernel@vger.kernel.org, kvm@vger.kernel.org Subject: [PATCH 07/30] KVM: nVMX: always update CR3 in VMCS Date: Fri, 29 May 2020 11:39:11 -0400 Message-Id: <20200529153934.11694-8-pbonzini@redhat.com> In-Reply-To: <20200529153934.11694-1-pbonzini@redhat.com> References: <20200529153934.11694-1-pbonzini@redhat.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Sender: kvm-owner@vger.kernel.org Precedence: bulk
Series	KVM: nSVM: event fixes and migration support \| expand [v3,00/28] KVM: nSVM: event fixes and migration support [01/30] KVM: x86: track manually whether an event has been injected [02/30] KVM: x86: enable event window in inject_pending_event [03/30] KVM: nSVM: inject exceptions via svm_check_nested_events [04/30] KVM: nSVM: remove exit_required [05/30] KVM: nSVM: correctly inject INIT vmexits [06/30] KVM: SVM: always update CR3 in VMCB [07/30] KVM: nVMX: always update CR3 in VMCS [08/30] KVM: nSVM: move map argument out of enter_svm_guest_mode [09/30] KVM: nSVM: extract load_nested_vmcb_control [10/30] KVM: nSVM: extract preparation of VMCB for nested run [11/30] KVM: nSVM: move MMU setup to nested_prepare_vmcb_control [12/30] KVM: nSVM: clean up tsc_offset update [13/30] KVM: nSVM: pass vmcb_control_area to copy_vmcb_control_area [14/30] KVM: nSVM: remove trailing padding for struct vmcb_control_area [15/30] KVM: nSVM: save all control fields in svm->nested [16/30] KVM: nSVM: restore clobbered INT_CTL fields after clearing VINTR [17/30] KVM: nSVM: synchronize VMCB controls updated by the processor on every vmexit [18/30] KVM: nSVM: remove unnecessary if [19/30] KVM: nSVM: extract svm_set_gif [20/30] KVM: SVM: preserve VGIF across VMCB switch [21/30] KVM: nSVM: synthesize correct EXITINTINFO on vmexit [22/30] KVM: nSVM: remove HF_VINTR_MASK [23/30] KVM: nSVM: remove HF_HIF_MASK [24/30] KVM: nSVM: split nested_vmcb_check_controls [25/30] KVM: nSVM: leave guest mode when clearing EFER.SVME [26/30] KVM: MMU: pass arbitrary CR0/CR4/EFER to kvm_init_shadow_mmu [27/30] selftests: kvm: introduce cpu_has_svm() check [28/30] selftests: kvm: add a SVM version of state-test [29/30] selftests: kvm: fix smm test on SVM [30/30] KVM: nSVM: implement KVM_GET_NESTED_STATE and KVM_SET_NESTED_STATE

Commit Message

Paolo Bonzini May 29, 2020, 3:39 p.m. UTC

vmx_load_mmu_pgd is delaying the write of GUEST_CR3 to prepare_vmcs02 as
an optimization, but this is only correct before the nested vmentry.
If userspace is modifying CR3 with KVM_SET_SREGS after the VM has
already been put in guest mode, the value of CR3 will not be updated.
Remove the optimization, which almost never triggers anyway.

Fixes: 04f11ef45810 ("KVM: nVMX: Always write vmcs02.GUEST_CR3 during nested VM-Enter")
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 arch/x86/kvm/vmx/vmx.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

diff mbox series

Patch

diff --git a/arch/x86/kvm/vmx/vmx.c b/arch/x86/kvm/vmx/vmx.c
index b3a41645e157..7b55dc6230a9 100644
--- a/arch/x86/kvm/vmx/vmx.c
+++ b/arch/x86/kvm/vmx/vmx.c
@@ -3085,10 +3085,7 @@  void vmx_load_mmu_pgd(struct kvm_vcpu *vcpu, unsigned long pgd)
 			spin_unlock(&to_kvm_vmx(kvm)->ept_pointer_lock);
 		}
 
-		/* Loading vmcs02.GUEST_CR3 is handled by nested VM-Enter. */
-		if (is_guest_mode(vcpu))
-			update_guest_cr3 = false;
-		else if (!enable_unrestricted_guest && !is_paging(vcpu))
+		if (!enable_unrestricted_guest && !is_paging(vcpu))
 			guest_cr3 = to_kvm_vmx(kvm)->ept_identity_map_addr;
 		else if (test_bit(VCPU_EXREG_CR3, (ulong *)&vcpu->arch.regs_avail))
 			guest_cr3 = vcpu->arch.cr3;