https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120165
--- Comment #19 from John Paul Adrian Glaubitz <glaubitz at debian dot org> --- I have had an LLM look at this problem and while it could not resolve this problem, it has suggested that there is a bug in the m68k backend rather the generic LRA code. In particular, it mentioned the following three possibilities for the bug: - how m68k calculates INITIAL_ELIMINATION_OFFSET - how m68k legitimizes addresses during elimination - how indexed addressing modes ((disp, An, Xi)) are handled during register elimination
