https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68865
--- Comment #2 from Bill Schmidt <wschmidt at gcc dot gnu.org> --- That's very possible. No doubt r231165 just caused it to come out of the woodwork again. However, I verified that it didn't ICE with r230600 - r231164, so something about r231165 managed to trigger the issue. Thanks for looking at it!