This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: movstrictqi/movstricthi fix for h8300


>>>>> "Bernd" == Bernd Schmidt <bernds@redhat.co.uk> writes:

    Bernd> I think it would be cleaner to represent a movstrict
    Bernd> pattern as something like

    Bernd>   (set (reg X) (ior (and X 0xFF....00) (zero_extend Y)))

    Bernd> This might even help the combiner, who knows.

    Bernd> Thoughts?

I couldn't agree more, both with the overall sentiment you express,
and also with this particular case.  If you can figure out how to get
rid of this, go for it.  If that causes slightly worse code
generation, so be it: the win will still be worth it in the long run.
It should be relatively simple to show the insn recognizers how to
take the pattern you show above and do the right thing.

--
Mark Mitchell                   mark@codesourcery.com
CodeSourcery, LLC               http://www.codesourcery.com

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]