[Bug tree-optimization/71264] [4.9/5/6 Regression] ICE in convert_move

["rguenth at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71264

--- Comment #12 from Richard Biener <rguenth at gcc dot gnu.org> ---
Ah, it has V4QI, not SImode ... and thus it triggers

          if (VECTOR_MODE_P (TYPE_MODE (gimple_expr_type (stmt))))
            {
              if (dump_enabled_p ())
                {
                  dump_printf_loc (MSG_MISSED_OPTIMIZATION, vect_location,
                                   "not vectorized: vector stmt in loop:");
                  dump_gimple_stmt (MSG_MISSED_OPTIMIZATION, TDF_SLIM, stmt,
0);
                  dump_printf (MSG_MISSED_OPTIMIZATION, "\n");
                }
              return false;
            }

I'll think about this a bit.  I don't think we have a good dejagnu target
to check here :/  (not-vector-size-4B)  Looks like even x86 has V4QI but
it is not being used here for some reason - it's never a
targetm.vector_mode_supported_p mode.