https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124175
--- Comment #2 from Kang-Che Sung <Explorer09 at gmail dot com> --- Well, my issue report of this isn't strictly about the 'restrict' qualifier. The qualifer was just one way to tell the compiler that the constraint should hold (so that the jump can be optimized). The print_ancestor_dirs() example showed another way for the constraint can hold (via explicit value assignment). I think that could be two issues in one report. (I'm not sure if it's worth splitting the second part into a separate report though.)
