https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120881
--- Comment #9 from Sam James <sjames at gcc dot gnu.org> --- Thanks both. H.J.'s is slightly less pessimising because it'll only affect functions where we actually emit a call, but I don't think it really matters here, and matching the general shrinkwrap logic seems better in case that was added for other reasons. So either WFM.
