[PATCH v3 8/8] DO NOT MERGE: KVM: selftests: Reproduce nested RIP restore bug

[Yosry Ahmed]

Update svm_nested_soft_inject_test such that L1 syncs to userspace
before running L2. The test then enables single-stepping and steps
through guest code until VMRUN is execute, and saves/restores the VM
immediately after (before L2 runs).

This reproduces a bug in save/restore where L2's RIP is not used
correctly to construct the vmcb02 at the destination.

Signed-off-by: Yosry Ahmed <yosry@xxxxxxxxxx>
---
 .../testing/selftests/kvm/lib/x86/processor.c |  8 +-
 .../kvm/x86/svm_nested_soft_inject_test.c     | 74 +++++++++++++++----
 2 files changed, 65 insertions(+), 17 deletions(-)

diff --git a/tools/testing/selftests/kvm/lib/x86/processor.c b/tools/testing/selftests/kvm/lib/x86/processor.c
index fab18e9be66c9..7e0213a88697d 100644
--- a/tools/testing/selftests/kvm/lib/x86/processor.c
+++ b/tools/testing/selftests/kvm/lib/x86/processor.c
@@ -1275,6 +1275,8 @@ struct kvm_x86_state *vcpu_save_state(struct kvm_vcpu *vcpu)
 	return state;
 }
 
+#define LOAD_REGS_BEFORE_NESTED 1
+
 void vcpu_load_state(struct kvm_vcpu *vcpu, struct kvm_x86_state *state)
 {
 	vcpu_sregs_set(vcpu, &state->sregs);
@@ -1287,10 +1289,14 @@ void vcpu_load_state(struct kvm_vcpu *vcpu, struct kvm_x86_state *state)
 	vcpu_events_set(vcpu, &state->events);
 	vcpu_mp_state_set(vcpu, &state->mp_state);
 	vcpu_debugregs_set(vcpu, &state->debugregs);
-	vcpu_regs_set(vcpu, &state->regs);
+	if (LOAD_REGS_BEFORE_NESTED)
+		vcpu_regs_set(vcpu, &state->regs);
 
 	if (state->nested.size)
 		vcpu_nested_state_set(vcpu, &state->nested);
+
+	if (!LOAD_REGS_BEFORE_NESTED)
+		vcpu_regs_set(vcpu, &state->regs);
 }
 
 void kvm_x86_state_cleanup(struct kvm_x86_state *state)
diff --git a/tools/testing/selftests/kvm/x86/svm_nested_soft_inject_test.c b/tools/testing/selftests/kvm/x86/svm_nested_soft_inject_test.c
index 4bd1655f9e6d0..dfefd8eed392a 100644
--- a/tools/testing/selftests/kvm/x86/svm_nested_soft_inject_test.c
+++ b/tools/testing/selftests/kvm/x86/svm_nested_soft_inject_test.c
@@ -101,6 +101,7 @@ static void l1_guest_code(struct svm_test_data *svm, uint64_t is_nmi, uint64_t i
 		vmcb->control.next_rip = vmcb->save.rip;
 	}
 
+	GUEST_SYNC(true);
 	run_guest(vmcb, svm->vmcb_gpa);
 	__GUEST_ASSERT(vmcb->control.exit_code == SVM_EXIT_VMMCALL,
 		       "Expected VMMCAL #VMEXIT, got '0x%lx', info1 = '0x%lx, info2 = '0x%lx'",
@@ -131,6 +132,7 @@ static void l1_guest_code(struct svm_test_data *svm, uint64_t is_nmi, uint64_t i
 	/* The return address pushed on stack, skip over UD2 */
 	vmcb->control.next_rip = vmcb->save.rip + 2;
 
+	GUEST_SYNC(true);
 	run_guest(vmcb, svm->vmcb_gpa);
 	__GUEST_ASSERT(vmcb->control.exit_code == SVM_EXIT_HLT,
 		       "Expected HLT #VMEXIT, got '0x%lx', info1 = '0x%lx, info2 = '0x%lx'",
@@ -140,6 +142,24 @@ static void l1_guest_code(struct svm_test_data *svm, uint64_t is_nmi, uint64_t i
 	GUEST_DONE();
 }
 
+static struct kvm_vcpu *save_and_restore_vm(struct kvm_vm *vm, struct kvm_vcpu *vcpu)
+{
+	struct kvm_x86_state *state = vcpu_save_state(vcpu);
+
+	kvm_vm_release(vm);
+	vcpu = vm_recreate_with_one_vcpu(vm);
+	vcpu_load_state(vcpu, state);
+	kvm_x86_state_cleanup(state);
+	return vcpu;
+}
+
+static bool is_nested_run_pending(struct kvm_vcpu *vcpu)
+{
+	struct kvm_x86_state *state = vcpu_save_state(vcpu);
+
+	return state->nested.size && (state->nested.flags & KVM_STATE_NESTED_RUN_PENDING);
+}
+
 static void run_test(bool is_nmi)
 {
 	struct kvm_vcpu *vcpu;
@@ -173,22 +193,44 @@ static void run_test(bool is_nmi)
 	memset(&debug, 0, sizeof(debug));
 	vcpu_guest_debug_set(vcpu, &debug);
 
-	struct ucall uc;
-
-	alarm(2);
-	vcpu_run(vcpu);
-	alarm(0);
-	TEST_ASSERT_KVM_EXIT_REASON(vcpu, KVM_EXIT_IO);
-
-	switch (get_ucall(vcpu, &uc)) {
-	case UCALL_ABORT:
-		REPORT_GUEST_ASSERT(uc);
-		break;
-		/* NOT REACHED */
-	case UCALL_DONE:
-		goto done;
-	default:
-		TEST_FAIL("Unknown ucall 0x%lx.", uc.cmd);
+	for (;;) {
+		struct kvm_guest_debug debug;
+		struct ucall uc;
+
+		alarm(2);
+		vcpu_run(vcpu);
+		alarm(0);
+		TEST_ASSERT_KVM_EXIT_REASON(vcpu, KVM_EXIT_IO);
+
+		switch (get_ucall(vcpu, &uc)) {
+		case UCALL_SYNC:
+			/*
+			 * L1 syncs before calling run_guest(), single-step over
+			 * all instructions until VMRUN, and save+restore right
+			 * after it (before L2 actually runs).
+			 */
+			debug.control = KVM_GUESTDBG_ENABLE | KVM_GUESTDBG_SINGLESTEP;
+			vcpu_guest_debug_set(vcpu, &debug);
+
+			do {
+				vcpu_run(vcpu);
+				TEST_ASSERT_KVM_EXIT_REASON(vcpu, KVM_EXIT_DEBUG);
+			} while (!is_nested_run_pending(vcpu));
+
+			memset(&debug, 0, sizeof(debug));
+			vcpu_guest_debug_set(vcpu, &debug);
+			vcpu = save_and_restore_vm(vm, vcpu);
+			break;
+
+		case UCALL_ABORT:
+			REPORT_GUEST_ASSERT(uc);
+			break;
+			/* NOT REACHED */
+		case UCALL_DONE:
+			goto done;
+		default:
+			TEST_FAIL("Unknown ucall 0x%lx.", uc.cmd);
+		}
 	}
 done:
 	kvm_vm_free(vm);
-- 
2.53.0.414.gf7e9f6c205-goog