https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86418
--- Comment #5 from Eric Gallager <egallager at gcc dot gnu.org> --- (In reply to Martin Sebor from comment #4) > Confirmed. My preference would be to resolve pr82922 and diagnose all calls > to functions without a prototype. Agreed. > Short of that, this could be handled by saving the type of the first call to a > function without a prototype in a translation unit, comparing it to the type > of > each subsequent call, and warning if they don't match. What about with LTO?
