https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120480
--- Comment #8 from Vladimir Makarov <vmakarov at gcc dot gnu.org> --- I think we should revert the patch for PR99015 I committed yesterday. There were to many changes in this part of LRA since gcc-13 and many of them were interdependent. It will be difficult to find all necessary changes and the changes will be risky anyway. The PR99015 test is artificial with my point of view. It is better to have PR99015 than this one (and possibly others). So I am reverting the patch.