https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121095
--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> --- (In reply to lucier from comment #4) > Thanks for the explanation. > > Now that I understand the issue better, this whole approach---hijacking and > hacking the PRE infrastructure to elide unnecessary reads and writes of a > new, very special use, status register even when -fno-gcse is > specified---seems unnecessarily heavy-handed and ill judged. > > But that's just me. So it turns out it was not hard to detect if the hard register was a live ever and a new patch was submitted: https://gcc.gnu.org/pipermail/gcc-patches/2025-July/689758.html