On 12/02/20 20:57, Nadav Amit wrote:
> Extended HOST_EFER tests can fail with a different error code than the
> expected one, since the host address size bit is checked against
> EFER.LMA. This causes kvm-unit-tests to fail on bare metal. According
> to the SDM the errors are not ordered.
>
> Expect either "invalid control" or "invalid host state" error-codes to
> allow the tests to pass. The fix somewhat relaxes the tests, as there
> are cases when only "invalid host state" is a valid instruction error,
> but doing the fix in this manner prevents intrusive changes.
>
> Fixes: a22d7b5534c2 ("x86: vmx_tests: extend HOST_EFER tests")
> Signed-off-by: Nadav Amit <namit@xxxxxxxxxx>
> ---
> x86/vmx_tests.c | 24 +++++++++++++++++++++++-
> 1 file changed, 23 insertions(+), 1 deletion(-)
>
> diff --git a/x86/vmx_tests.c b/x86/vmx_tests.c
> index 69429e5..e69c361 100644
> --- a/x86/vmx_tests.c
> +++ b/x86/vmx_tests.c
> @@ -3407,6 +3407,27 @@ static void test_vmx_vmlaunch(u32 xerror)
> }
> }
>
> +/*
> + * Try to launch the current VMCS, and expect one of two possible
> + * errors (or success) codes.
> + */
> +static void test_vmx_vmlaunch2(u32 xerror1, u32 xerror2)
> +{
> + bool success = vmlaunch_succeeds();
> + u32 vmx_inst_err;
> +
> + if (!xerror1 == !xerror2)
> + report(success == !xerror1, "vmlaunch %s",
> + !xerror1 ? "succeeds" : "fails");
> +
> + if (!success && (xerror1 || xerror2)) {
> + vmx_inst_err = vmcs_read(VMX_INST_ERROR);
> + report(vmx_inst_err == xerror1 || vmx_inst_err == xerror2,
> + "VMX inst error is %d or %d (actual %d)", xerror1,
> + xerror2, vmx_inst_err);
> + }
> +}
> +
> static void test_vmx_invalid_controls(void)
> {
> test_vmx_vmlaunch(VMXERR_ENTRY_INVALID_CONTROL_FIELD);
> @@ -6764,7 +6785,8 @@ static void test_efer_vmlaunch(u32 fld, bool ok)
> if (ok)
> test_vmx_vmlaunch(0);
> else
> - test_vmx_vmlaunch(VMXERR_ENTRY_INVALID_HOST_STATE_FIELD);
> + test_vmx_vmlaunch2(VMXERR_ENTRY_INVALID_CONTROL_FIELD,
> + VMXERR_ENTRY_INVALID_HOST_STATE_FIELD);
> } else {
> if (ok) {
> enter_guest();
>
Queued, thanks.
Paolo
