https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117266
--- Comment #16 from H. Peter Anvin <hpa at zytor dot com> --- Well, if that is the way you feel about it. It is certainly different from the messages we get in other situations, so it is a bit confusing to me. It isn't *that* unusual that you know a priori that the upper half is zero, especially when dealing with multi-precision arithmetic. See again the referenced bug.