https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126207
--- Comment #3 from Vladislav Semykin <vladislav.semykin at gmail dot com> --- Drea Pinski is already answered, thank you! But I can suggest a little refactor: add an optional int* nonnull out-parameter to typeid_evaluated_p, so build_typeid gets the value from the same call instead of invoking resolves_to_fixed_type_p a second time. That would remove the Coverity noise by construction rather than relying on the call-order reasoning. Happy to send a patch if that seems worthwhile, or leave it as is since the current code is correct.
