https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123084
--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> --- (In reply to Andrew Pinski from comment #1) > Specifically: > https://gcc.gnu.org/pipermail/gcc-patches/2025-December/703278.html This one looks like it does fix the regression. At least the code generation for -Ofast on aarch64 looks back to being similar. BUT I will tell you there is not much code generation difference between the 2 so it might be alignmented related.
