> So I wonder why it is necessary to track 'reverse' in gimple_match_op > at all given we bail out without optimizing as far as I can see?
Because of the valueization? If it is done, gimple_simplify returns true so the result will be synthetized from res_op by means of maybe_build_generic_op. That's what I was referring to in the opening message by "the underlying issue of the missing propagation of the flag during GIMPLE folding". -- Eric Botcazou