https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117813
--- Comment #9 from Jonathan Wakely <redi at gcc dot gnu.org> --- Yup, my bisection found the same two commits. Do we want to backport r15-521-g6ad7ca1bb90573 then? Or a simpler change to undo the unwanted effects of r14-5979-g99d114c15523e0? (could the bug even still be latent on trunk?)