[PATCH] tree-optimization/110278 - uns < (typeof uns)(uns != 0) is always false

[PATCH] tree-optimization/110278 - uns < (typeof uns)(uns != 0) is always false

[Richard Biener]

The following adds two patterns simplifying comparisons,
uns < (typeof uns)(uns != 0) is always false and x != (typeof x)(x == 0)
is always true.

Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.

	PR tree-optimization/110278
	* match.pd (uns < (typeof uns)(uns != 0) -> false): New.
	(x != (typeof x)(x == 0) -> true): Likewise.
---
 gcc/match.pd | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/gcc/match.pd b/gcc/match.pd
index 264f9cb8a40..48b76e6a051 100644
--- a/gcc/match.pd
+++ b/gcc/match.pd
@@ -6410,6 +6410,17 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
 	 (if (cmp == GT_EXPR)
 	  (lt (view_convert:st @0) { build_zero_cst (st); })))))))))))
 
+/* unsigned < (typeof unsigned)(unsigned != 0) is always false.  */
+(simplify
+ (lt:c @0 (convert (ne @0 integer_zerop)))
+ (if (TYPE_UNSIGNED (TREE_TYPE (@0)))
+  { constant_boolean_node (false, type); }))
+
+/* x != (typeof x)(x == 0) is always true.  */
+(simplify
+ (ne:c @0 (convert (eq @0 integer_zerop)))
+ { constant_boolean_node (true, type); })
+
 (for cmp (unordered ordered unlt unle ungt unge uneq ltgt)
  /* If the second operand is NaN, the result is constant.  */
  (simplify
-- 
2.35.3

Re: [PATCH] tree-optimization/110278 - uns < (typeof uns)(uns != 0) is always false

[Andrew Pinski]

On Fri, Jun 16, 2023 at 4:46 AM Richard Biener via Gcc-patches
<gcc-patches@gcc.gnu.org> wrote:
>
> The following adds two patterns simplifying comparisons,
> uns < (typeof uns)(uns != 0) is always false and x != (typeof x)(x == 0)
> is always true.

A few more that should be done (I will file a bug in a few minutes):
`x == (typeof x)(x == 0)` is always false.
`x == (typeof x)(x != 0)` is `(unsigned_type)x <= 1`
`x != (typeof x)(x != 0)` is `(unsigned_type)x > 1`
`uns <= (typeof uns)(uns != 0)` -> `uns <= 1`
`uns > (typeof uns)(uns != 0)` is `uns > 1`
`uns >= (typeof uns)(uns != 0)` is always true

That should be all of them I think and  I think I did it correctly.

>
> Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.
>
>         PR tree-optimization/110278
>         * match.pd (uns < (typeof uns)(uns != 0) -> false): New.
>         (x != (typeof x)(x == 0) -> true): Likewise.
> ---
>  gcc/match.pd | 11 +++++++++++
>  1 file changed, 11 insertions(+)
>
> diff --git a/gcc/match.pd b/gcc/match.pd
> index 264f9cb8a40..48b76e6a051 100644
> --- a/gcc/match.pd
> +++ b/gcc/match.pd
> @@ -6410,6 +6410,17 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
>          (if (cmp == GT_EXPR)
>           (lt (view_convert:st @0) { build_zero_cst (st); })))))))))))
>
> +/* unsigned < (typeof unsigned)(unsigned != 0) is always false.  */
> +(simplify
> + (lt:c @0 (convert (ne @0 integer_zerop)))
> + (if (TYPE_UNSIGNED (TREE_TYPE (@0)))
> +  { constant_boolean_node (false, type); }))
> +
> +/* x != (typeof x)(x == 0) is always true.  */
> +(simplify
> + (ne:c @0 (convert (eq @0 integer_zerop)))
> + { constant_boolean_node (true, type); })
> +
>  (for cmp (unordered ordered unlt unle ungt unge uneq ltgt)
>   /* If the second operand is NaN, the result is constant.  */
>   (simplify
> --
> 2.35.3

Re: [PATCH] tree-optimization/110278 - uns < (typeof uns)(uns != 0) is always false

[Andrew Pinski]

On Fri, Jun 16, 2023 at 4:14 PM Andrew Pinski <pinskia@gmail.com> wrote:
>
> On Fri, Jun 16, 2023 at 4:46 AM Richard Biener via Gcc-patches
> <gcc-patches@gcc.gnu.org> wrote:
> >
> > The following adds two patterns simplifying comparisons,
> > uns < (typeof uns)(uns != 0) is always false and x != (typeof x)(x == 0)
> > is always true.
>
> A few more that should be done (I will file a bug in a few minutes):
> `x == (typeof x)(x == 0)` is always false.
> `x == (typeof x)(x != 0)` is `(unsigned_type)x <= 1`
> `x != (typeof x)(x != 0)` is `(unsigned_type)x > 1`
> `uns <= (typeof uns)(uns != 0)` -> `uns <= 1`
> `uns > (typeof uns)(uns != 0)` is `uns > 1`
> `uns >= (typeof uns)(uns != 0)` is always true
>
> That should be all of them I think and  I think I did it correctly.

Filed as PR 110293 .

Thanks,
Andrew

>
> >
> > Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.
> >
> >         PR tree-optimization/110278
> >         * match.pd (uns < (typeof uns)(uns != 0) -> false): New.
> >         (x != (typeof x)(x == 0) -> true): Likewise.
> > ---
> >  gcc/match.pd | 11 +++++++++++
> >  1 file changed, 11 insertions(+)
> >
> > diff --git a/gcc/match.pd b/gcc/match.pd
> > index 264f9cb8a40..48b76e6a051 100644
> > --- a/gcc/match.pd
> > +++ b/gcc/match.pd
> > @@ -6410,6 +6410,17 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
> >          (if (cmp == GT_EXPR)
> >           (lt (view_convert:st @0) { build_zero_cst (st); })))))))))))
> >
> > +/* unsigned < (typeof unsigned)(unsigned != 0) is always false.  */
> > +(simplify
> > + (lt:c @0 (convert (ne @0 integer_zerop)))
> > + (if (TYPE_UNSIGNED (TREE_TYPE (@0)))
> > +  { constant_boolean_node (false, type); }))
> > +
> > +/* x != (typeof x)(x == 0) is always true.  */
> > +(simplify
> > + (ne:c @0 (convert (eq @0 integer_zerop)))
> > + { constant_boolean_node (true, type); })
> > +
> >  (for cmp (unordered ordered unlt unle ungt unge uneq ltgt)
> >   /* If the second operand is NaN, the result is constant.  */
> >   (simplify
> > --
> > 2.35.3