http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55890
--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> 2013-01-07 10:03:30 UTC --- Created attachment 29094 --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=29094 prototype patch For example fixed with something like the following. Of course it's always going to produce some GIGO in builtins analyzing code this way. So the other fix would be to verify that for a builtin call the number of actual arguments matches the ones required for the builtin (though types also can differ ...). Quite some amount of code to audit against these things - I suppose we might decide to simply deal with this during next stage1 ...
