On Mon, Dec 07, 2020 at 04:40:39PM +0100, Jakub Jelinek wrote: > On Mon, Dec 07, 2020 at 09:28:19AM -0600, Segher Boessenkool wrote: > > I see. We could also allow this for 2->2 combinations (w/ the same > > restrictions we do for other 2->2 combinations, but that probably falls > > out automatically). > > In this case we actually want a 2->1 combination, where the result isn't > recognized, but the splitter splits the 1 insn into another 1. > Could be limited just to that, but yes, even 2->1 combination where > the 1 is split into 2 insns and costs say it is ok (and other restrictions > satisfied) should be ok too.
Yes, that needs exactly the same code (allow a split if we started with just two insns), and it all falls out naturally :-) Segher
