https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123994
--- Comment #12 from Jeffrey A. Law <law at gcc dot gnu.org> --- I'd tend to lean towards the simpler fix for -16 and the more complete for -17 given the possibility of triggering new combinations. While I could argue that any regressions caused by such a change were just latent and waiting to bite us, safer seems better given we're in the latter half of Feb. But I'm not dead set against including the more complete fix now if you really wanted to go that way.
