https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115753
--- Comment #7 from Andrew Pinski <pinskia at gcc dot gnu.org> --- Here is a float/longlong that fails for both aarch64 and x86_64 (it works on i686 though): ``` void f(float*); void foo1 () { long long t0 = __LONG_LONG_MAX__; long long t1 = __LONG_LONG_MAX__ - 1; float tt[2]; tt[0] = t0; tt[1] = t1; f(&tt[0]); } ``` Always useful to have more testcases really :).