https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111754
--- Comment #10 from Richard Sandiford <rsandifo at gcc dot gnu.org> --- Yeah, the problem went latent after the fix for PR111648, but the underlying problem is still there. Prathamesh is working on a fix for that. See the thread starting at https://gcc.gnu.org/pipermail/gcc-patches/2023-October/633784.html for discussion.