https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118770
--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> --- (In reply to ymity.me from comment #4) > I would expect it to take longer in such a case, it's almost instant on my > laptop, and quite random on my tower pc. If i modify it just a little bit it > would continue endlessly. well if you call a fucntion recusively in many cases you could get a stack overflow and that is what is happening with respect to this with push_macro/pop_macro inside a _Pragma.