https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81601
--- Comment #13 from Jeffrey A. Law <law at redhat dot com> --- I realize the warnings are happening in VRP1. EVRP and VRP1 use different styles of analysis and it can be the case that one is better suited for cleaning things up than the other. But I really should have re-familiarized myself with the BZ first :-) We need to prove that the reference to chrono_type and the BIT_FIELD_REF are actually the same. That way the result of the first conditional can be used to prove the second conditional is always false making the out of bounds reference unreachable. It probably needs to be done in FRE since that's our best bet to expose the relationship between the conditionals prior to VRP1.