[PATCH v2 1/6] KVM: nSVM: refactor the CR3 reload on migration

[Maxim Levitsky]

Document the actual reason why we need to do it
on migration and move the call to svm_set_nested_state
to be closer to VMX code.

To avoid loading the PDPTRs from possibly not up to date memory map,
in nested_svm_load_cr3 after the move, move this code to
.get_nested_state_pages.

Signed-off-by: Maxim Levitsky <mlevitsk@xxxxxxxxxx>
---
 arch/x86/kvm/svm/nested.c | 32 +++++++++++++++++++++++++-------
 1 file changed, 25 insertions(+), 7 deletions(-)

diff --git a/arch/x86/kvm/svm/nested.c b/arch/x86/kvm/svm/nested.c
index 540d43ba2cf4..ceb37ed1dc98 100644
--- a/arch/x86/kvm/svm/nested.c
+++ b/arch/x86/kvm/svm/nested.c
@@ -385,12 +385,12 @@ static inline bool nested_npt_enabled(struct vcpu_svm *svm)
  * if we are emulating VM-Entry into a guest with NPT enabled.
  */
 static int nested_svm_load_cr3(struct kvm_vcpu *vcpu, unsigned long cr3,
-			       bool nested_npt)
+			       bool nested_npt, bool reload_pdptrs)
 {
 	if (CC(kvm_vcpu_is_illegal_gpa(vcpu, cr3)))
 		return -EINVAL;
 
-	if (!nested_npt && is_pae_paging(vcpu) &&
+	if (reload_pdptrs && !nested_npt && is_pae_paging(vcpu) &&
 	    (cr3 != kvm_read_cr3(vcpu) || pdptrs_changed(vcpu))) {
 		if (CC(!load_pdptrs(vcpu, vcpu->arch.walk_mmu, cr3)))
 			return -EINVAL;
@@ -576,7 +576,7 @@ int enter_svm_guest_mode(struct kvm_vcpu *vcpu, u64 vmcb12_gpa,
 	nested_vmcb02_prepare_save(svm, vmcb12);
 
 	ret = nested_svm_load_cr3(&svm->vcpu, vmcb12->save.cr3,
-				  nested_npt_enabled(svm));
+				  nested_npt_enabled(svm), true);
 	if (ret)
 		return ret;
 
@@ -806,7 +806,7 @@ int nested_svm_vmexit(struct vcpu_svm *svm)
 
 	nested_svm_uninit_mmu_context(vcpu);
 
-	rc = nested_svm_load_cr3(vcpu, svm->vmcb->save.cr3, false);
+	rc = nested_svm_load_cr3(vcpu, svm->vmcb->save.cr3, false, true);
 	if (rc)
 		return 1;
 
@@ -1291,6 +1291,19 @@ static int svm_set_nested_state(struct kvm_vcpu *vcpu,
 	    !nested_vmcb_valid_sregs(vcpu, save))
 		goto out_free;
 
+	/*
+	 * While the nested guest CR3 is already checked and set by
+	 * KVM_SET_SREGS, it was set when nested state was yet loaded,
+	 * thus MMU might not be initialized correctly.
+	 * Set it again to fix this.
+	 */
+
+	ret = nested_svm_load_cr3(&svm->vcpu, vcpu->arch.cr3,
+				  nested_npt_enabled(svm), false);
+	if (WARN_ON_ONCE(ret))
+		goto out_free;
+
+
 	/*
 	 * All checks done, we can enter guest mode. Userspace provides
 	 * vmcb12.control, which will be combined with L1 and stored into
@@ -1343,9 +1356,14 @@ static bool svm_get_nested_state_pages(struct kvm_vcpu *vcpu)
 	if (WARN_ON(!is_guest_mode(vcpu)))
 		return true;
 
-	if (nested_svm_load_cr3(&svm->vcpu, vcpu->arch.cr3,
-				nested_npt_enabled(svm)))
-		return false;
+	if (!nested_npt_enabled(svm) && is_pae_paging(vcpu))
+		/*
+		 * Reload the guest's PDPTRs since after a migration
+		 * the guest CR3 might be restored prior to setting the nested
+		 * state which can lead to a load of wrong PDPTRs.
+		 */
+		if (CC(!load_pdptrs(vcpu, vcpu->arch.walk_mmu, vcpu->arch.cr3)))
+			return false;
 
 	if (!nested_svm_vmrun_msrpm(svm)) {
 		vcpu->run->exit_reason = KVM_EXIT_INTERNAL_ERROR;
-- 
2.26.2