On Fri, May 27, 2011 at 6:40 PM, Jakub Jelinek <ja...@redhat.com> wrote:
> This patch adds a few peephole2s to help optimize if (!--*x) etc.
> This is something the combiner doesn't and can't easily handle, because in
> reg0 = mem1
> reg0 {+,-,&,|,^}= x
> mem1 = reg0
> cc = compare (reg0, 0)
> reg0 is used also by both the store and compare and there is no dependence
> in between them, so for reg0 = mem1; reg0 {+,-,&,|,^}= x; mem1 = reg0
> alone it can't do anything because reg0 is still needed and mem1 {+,-,&,|,^}=
> x
> doesn't set it, and the compare, being second user of reg0, doesn't have any
> LOG_LINKS and thus try_combine isn't called for it at all.
>
> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
>
> 2011-05-27 Jakub Jelinek <ja...@redhat.com>
>
> PR rtl-optimization/49095
> * config/i386/predicates.md (plusminuslogic_operator): New predicate.
> * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0.
>
> * gcc.target/i386/pr49095.c: New test.
This is OK.
On a related note, we probably want to set TARGET_READ_MODIFY_WRITE in
ix86_option_override_internal when optimize_size is in effect.
Thanks,
Uros.
