http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45978
--- Comment #2 from Richard Guenther <rguenth at gcc dot gnu.org> 2010-11-15 12:00:20 UTC --- Hmm, I'm not sure how to address this - we warn about the address-taking operation which only at VRP time get's folded to &MEM[(struct Y *)&<retval>].ar[4] via maybe_fold_stmt_addition (which I only preserved to avoid some regressions I don't remember anymore during mem-ref development). Clearly even the first vectorized access spans more than the array (but we could set TREE_NO_WARNING on the memory reference itself).
