http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46590
--- Comment #27 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-08-22 11:39:56 UTC --- The issue with loop header copying is that we use incremental SSA updating with PHI insertion for each individual loop header copied. That computes dominance frontiers which is always for the whole function. I thought that loop header copying wouldn't need to insert new PHI nodes and thus can do with TODO_update_ssa_no_phi if we are in loop-closed SSA form, but appearantly that is not the case (I didn't inverstigate further here, but possibly that's just because virtual operands are not in loop-closed SSA form).
