http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52754

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-03-28 
16:48:51 UTC ---
Seems ref_at_iteration is buggy, when iter is negative, but ARRAY_REF's index
is TYPE_UNSIGNED smaller than size of pointer, computing the ARRAY_REF's index
in the smaller unsigned type isn't a good idea, because then we end up with
those
0xfffffffdU etc. indexes while we want -3L.
Wonder if it is generally ok to do the computation in a wider type (if any),
or if we should just handle the simple cases (integer_onep (iv.step) ? ) and
punt otherwise.

Reply via email to