http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46416
--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-01-06 11:40:38 UTC --- There is no __real128, just __float128. Also, I think either it should use some extern inlinish quadmath.h specific macro (and it needs to care about g++ vs. gcc difference and which gcc versions support gnu_inline attribute). And also even if you have those extern inlines in the header, you still have to provide an out of line definition for them, i.e. in libquadmath sources, not headers. I'll take care of it.