https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80005
--- Comment #10 from Florian Weimer <fw at gcc dot gnu.org> --- It only affects RISC-V, and the use is not in an installed header, so I think the glibc case is rather harmless. (But that's only because I watched out for this particular issue and requested changes from __has_include__ to __has_include in submitted patches. Which is why I don't want __has_include__ to come back.)
