http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57201
--- Comment #8 from Manuel López-Ibáñez <manu at gcc dot gnu.org> 2013-05-08 14:23:07 UTC --- (In reply to comment #7) > the comment for function linemap_location_in_system_header_p > in libcpp/line-map.c says: > /* // snip > Note that this function returns 1 if LOCATION belongs to a token > that is part of a macro replacement-list defined in a system > header, but expanded in a non-system file. */ > > so this behavior is intentional? Yes. The idea is that if some piece of code gives a warning inside a macro in a system-header, we should not warn. See http://gcc.gnu.org/bugzilla/show_bug.cgi?id=7263#c8 I don't see how to reconcile this with --save-temps, unless we mark in the preprocessed code what was expanded from a system header. Interestingly, clang behaves the same, I guess following the same reasoning.
