[PATCH] match.pd: Implement missed optimization (x << c) >> c -> -(x & 1) [PR101955]

[PATCH] match.pd: Implement missed optimization (x << c) >> c -> -(x & 1) [PR101955]

[Drew Ross]

	PR middle-end/101955

gcc/ChangeLog:

	* match.pd (x << c) >> c -> -(x & 1): New simplification.

gcc/testsuite/ChangeLog:

	* gcc.dg/pr101955.c: New test.
---
 gcc/match.pd                    |  9 +++++
 gcc/testsuite/gcc.dg/pr101955.c | 62 +++++++++++++++++++++++++++++++++
 2 files changed, 71 insertions(+)
 create mode 100644 gcc/testsuite/gcc.dg/pr101955.c

diff --git a/gcc/match.pd b/gcc/match.pd
index 8543f777a28..bf63652e80f 100644
--- a/gcc/match.pd
+++ b/gcc/match.pd
@@ -3766,6 +3766,15 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
       && (wi::ltu_p (wi::to_wide (@1), element_precision (type))))
   (bit_and @0 (rshift { build_minus_one_cst (type); } @1))))
 
+/* Optimize (X << C) >> C where C = precision(type) - 1 and X is signed
+   into -(X & 1).  */
+(simplify
+ (rshift (lshift @0 INTEGER_CST@1) @@1)
+ (if (ANY_INTEGRAL_TYPE_P (type)
+      && !TYPE_UNSIGNED (type)
+      && wi::eq_p (wi::to_wide (@1), element_precision (type) - 1))
+  (negate (bit_and @0 { build_one_cst (type); }))))
+
 /* Optimize x >> x into 0 */
 (simplify
  (rshift @0 @0)
diff --git a/gcc/testsuite/gcc.dg/pr101955.c b/gcc/testsuite/gcc.dg/pr101955.c
new file mode 100644
index 00000000000..0e233269e21
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/pr101955.c
@@ -0,0 +1,62 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -fdump-tree-dse1 -Wno-psabi" } */
+
+typedef int v4si __attribute__((vector_size(16)));
+
+__attribute__((noipa)) int
+t1 (int x)
+{
+  return (x << 31) >> 31;
+}
+
+__attribute__((noipa)) int
+t2 (int x)
+{
+  int y = x << 31;
+  int z = y >> 31;
+  return z;
+}
+
+__attribute__((noipa)) int
+t3 (int x)
+{
+  int w = 31;
+  int y = x << w;
+  int z = y >> w;
+  return z;
+}
+
+__attribute__((noipa)) long long
+t4 (long long x)
+{
+  return (x << 63) >> 63;
+}
+
+__attribute__((noipa)) long long
+t5 (long long x)
+{
+  long long y = x << 63;
+  long long z = y >> 63;
+  return z;
+}
+
+__attribute__((noipa)) long long
+t6 (long long x)
+{
+  int w = 63;
+  long long y = x << w;
+  long long z = y >> w;
+  return z;
+}
+
+__attribute__((noipa)) v4si
+t7 (v4si x)
+{
+  return (x << 31) >> 31;
+}
+
+/* { dg-final { scan-tree-dump-not " >> " "dse1" } } */
+/* { dg-final { scan-tree-dump-not " << " "dse1" } } */
+/* { dg-final { scan-tree-dump-times " -" 7 "dse1" } } */
+/* { dg-final { scan-tree-dump-times " & " 7 "dse1" } } */
+
-- 
2.39.3

Re: [PATCH] match.pd: Implement missed optimization (x << c) >> c -> -(x & 1) [PR101955]

[Drew Ross]

[-- Attachment #1: Type: text/plain, Size: 3197 bytes --]

My mailer clipped off the top of the commit message:

Simplifies (x << c) >> c where x is a signed integral type of
width >= int and c = precision(type) - 1 into -(x & 1). Tested successfully
on x86_64 and x86 targets.

        PR middle-end/101955

gcc/ChangeLog:

        * match.pd (x << c) >> c -> -(x & 1): New simplification.

gcc/testsuite/ChangeLog:

        * gcc.dg/pr101955.c: New test.

On Thu, Jul 20, 2023 at 10:47 AM Drew Ross <drross@redhat.com> wrote:

>         PR middle-end/101955
>
> gcc/ChangeLog:
>
>         * match.pd (x << c) >> c -> -(x & 1): New simplification.
>
> gcc/testsuite/ChangeLog:
>
>         * gcc.dg/pr101955.c: New test.
> ---
>  gcc/match.pd                    |  9 +++++
>  gcc/testsuite/gcc.dg/pr101955.c | 62 +++++++++++++++++++++++++++++++++
>  2 files changed, 71 insertions(+)
>  create mode 100644 gcc/testsuite/gcc.dg/pr101955.c
>
> diff --git a/gcc/match.pd b/gcc/match.pd
> index 8543f777a28..bf63652e80f 100644
> --- a/gcc/match.pd
> +++ b/gcc/match.pd
> @@ -3766,6 +3766,15 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
>        && (wi::ltu_p (wi::to_wide (@1), element_precision (type))))
>    (bit_and @0 (rshift { build_minus_one_cst (type); } @1))))
>
> +/* Optimize (X << C) >> C where C = precision(type) - 1 and X is signed
> +   into -(X & 1).  */
> +(simplify
> + (rshift (lshift @0 INTEGER_CST@1) @@1)
> + (if (ANY_INTEGRAL_TYPE_P (type)
> +      && !TYPE_UNSIGNED (type)
> +      && wi::eq_p (wi::to_wide (@1), element_precision (type) - 1))
> +  (negate (bit_and @0 { build_one_cst (type); }))))
> +
>  /* Optimize x >> x into 0 */
>  (simplify
>   (rshift @0 @0)
> diff --git a/gcc/testsuite/gcc.dg/pr101955.c
> b/gcc/testsuite/gcc.dg/pr101955.c
> new file mode 100644
> index 00000000000..0e233269e21
> --- /dev/null
> +++ b/gcc/testsuite/gcc.dg/pr101955.c
> @@ -0,0 +1,62 @@
> +/* { dg-do compile } */
> +/* { dg-options "-O2 -fdump-tree-dse1 -Wno-psabi" } */
> +
> +typedef int v4si __attribute__((vector_size(16)));
> +
> +__attribute__((noipa)) int
> +t1 (int x)
> +{
> +  return (x << 31) >> 31;
> +}
> +
> +__attribute__((noipa)) int
> +t2 (int x)
> +{
> +  int y = x << 31;
> +  int z = y >> 31;
> +  return z;
> +}
> +
> +__attribute__((noipa)) int
> +t3 (int x)
> +{
> +  int w = 31;
> +  int y = x << w;
> +  int z = y >> w;
> +  return z;
> +}
> +
> +__attribute__((noipa)) long long
> +t4 (long long x)
> +{
> +  return (x << 63) >> 63;
> +}
> +
> +__attribute__((noipa)) long long
> +t5 (long long x)
> +{
> +  long long y = x << 63;
> +  long long z = y >> 63;
> +  return z;
> +}
> +
> +__attribute__((noipa)) long long
> +t6 (long long x)
> +{
> +  int w = 63;
> +  long long y = x << w;
> +  long long z = y >> w;
> +  return z;
> +}
> +
> +__attribute__((noipa)) v4si
> +t7 (v4si x)
> +{
> +  return (x << 31) >> 31;
> +}
> +
> +/* { dg-final { scan-tree-dump-not " >> " "dse1" } } */
> +/* { dg-final { scan-tree-dump-not " << " "dse1" } } */
> +/* { dg-final { scan-tree-dump-times " -" 7 "dse1" } } */
> +/* { dg-final { scan-tree-dump-times " & " 7 "dse1" } } */
> +
> --
> 2.39.3
>
>

Re: [PATCH] match.pd: Implement missed optimization (x << c) >> c -> -(x & 1) [PR101955]

[Andrew Pinski]

On Thu, Jul 20, 2023 at 7:47 AM Drew Ross via Gcc-bugs
<gcc-bugs@gcc.gnu.org> wrote:
>
>         PR middle-end/101955
>
> gcc/ChangeLog:
>
>         * match.pd (x << c) >> c -> -(x & 1): New simplification.
>
> gcc/testsuite/ChangeLog:
>
>         * gcc.dg/pr101955.c: New test.
> ---
>  gcc/match.pd                    |  9 +++++
>  gcc/testsuite/gcc.dg/pr101955.c | 62 +++++++++++++++++++++++++++++++++
>  2 files changed, 71 insertions(+)
>  create mode 100644 gcc/testsuite/gcc.dg/pr101955.c
>
> diff --git a/gcc/match.pd b/gcc/match.pd
> index 8543f777a28..bf63652e80f 100644
> --- a/gcc/match.pd
> +++ b/gcc/match.pd
> @@ -3766,6 +3766,15 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
>        && (wi::ltu_p (wi::to_wide (@1), element_precision (type))))
>    (bit_and @0 (rshift { build_minus_one_cst (type); } @1))))
>
> +/* Optimize (X << C) >> C where C = precision(type) - 1 and X is signed
> +   into -(X & 1).  */
> +(simplify
> + (rshift (lshift @0 INTEGER_CST@1) @@1)
> + (if (ANY_INTEGRAL_TYPE_P (type)
> +      && !TYPE_UNSIGNED (type)
> +      && wi::eq_p (wi::to_wide (@1), element_precision (type) - 1))
> +  (negate (bit_and @0 { build_one_cst (type); }))))

I think the lshift does not need to be in the same type but could be
in the unsigned type so you should add a nop_convert? operand there
That is:
(simplify
 (rshift (nop_convert? (lshift @0 INTEGER_CST@1)) @@1)
...
 (negate (bit_and (convert @0) { build_one_cst (type); }))))

Note I see you use INTEGER_CST and then ANY_INTEGRAL_TYPE_P. If you
are going to support vectors, you should use uniform_integer_cst_p
instead of INTEGER_CST and then use uniform_integer_cst_p to get the
constant.

Thanks,
Andrew


> +
>  /* Optimize x >> x into 0 */
>  (simplify
>   (rshift @0 @0)
> diff --git a/gcc/testsuite/gcc.dg/pr101955.c b/gcc/testsuite/gcc.dg/pr101955.c
> new file mode 100644
> index 00000000000..0e233269e21
> --- /dev/null
> +++ b/gcc/testsuite/gcc.dg/pr101955.c
> @@ -0,0 +1,62 @@
> +/* { dg-do compile } */
> +/* { dg-options "-O2 -fdump-tree-dse1 -Wno-psabi" } */
> +
> +typedef int v4si __attribute__((vector_size(16)));
> +
> +__attribute__((noipa)) int
> +t1 (int x)
> +{
> +  return (x << 31) >> 31;
> +}
> +
> +__attribute__((noipa)) int
> +t2 (int x)
> +{
> +  int y = x << 31;
> +  int z = y >> 31;
> +  return z;
> +}
> +
> +__attribute__((noipa)) int
> +t3 (int x)
> +{
> +  int w = 31;
> +  int y = x << w;
> +  int z = y >> w;
> +  return z;
> +}
> +
> +__attribute__((noipa)) long long
> +t4 (long long x)
> +{
> +  return (x << 63) >> 63;
> +}
> +
> +__attribute__((noipa)) long long
> +t5 (long long x)
> +{
> +  long long y = x << 63;
> +  long long z = y >> 63;
> +  return z;
> +}
> +
> +__attribute__((noipa)) long long
> +t6 (long long x)
> +{
> +  int w = 63;
> +  long long y = x << w;
> +  long long z = y >> w;
> +  return z;
> +}
> +
> +__attribute__((noipa)) v4si
> +t7 (v4si x)
> +{
> +  return (x << 31) >> 31;
> +}
> +
> +/* { dg-final { scan-tree-dump-not " >> " "dse1" } } */
> +/* { dg-final { scan-tree-dump-not " << " "dse1" } } */
> +/* { dg-final { scan-tree-dump-times " -" 7 "dse1" } } */
> +/* { dg-final { scan-tree-dump-times " & " 7 "dse1" } } */
> +
> --
> 2.39.3
>

Re: [PATCH] match.pd: Implement missed optimization (x << c) >> c -> -(x & 1) [PR101955]

[Andrew Pinski]

On Thu, Jul 20, 2023 at 8:17 AM Andrew Pinski <pinskia@gmail.com> wrote:
>
> On Thu, Jul 20, 2023 at 7:47 AM Drew Ross via Gcc-bugs
> <gcc-bugs@gcc.gnu.org> wrote:
> >
> >         PR middle-end/101955
> >
> > gcc/ChangeLog:
> >
> >         * match.pd (x << c) >> c -> -(x & 1): New simplification.
> >
> > gcc/testsuite/ChangeLog:
> >
> >         * gcc.dg/pr101955.c: New test.
> > ---
> >  gcc/match.pd                    |  9 +++++
> >  gcc/testsuite/gcc.dg/pr101955.c | 62 +++++++++++++++++++++++++++++++++
> >  2 files changed, 71 insertions(+)
> >  create mode 100644 gcc/testsuite/gcc.dg/pr101955.c
> >
> > diff --git a/gcc/match.pd b/gcc/match.pd
> > index 8543f777a28..bf63652e80f 100644
> > --- a/gcc/match.pd
> > +++ b/gcc/match.pd
> > @@ -3766,6 +3766,15 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
> >        && (wi::ltu_p (wi::to_wide (@1), element_precision (type))))
> >    (bit_and @0 (rshift { build_minus_one_cst (type); } @1))))
> >
> > +/* Optimize (X << C) >> C where C = precision(type) - 1 and X is signed
> > +   into -(X & 1).  */
> > +(simplify
> > + (rshift (lshift @0 INTEGER_CST@1) @@1)
> > + (if (ANY_INTEGRAL_TYPE_P (type)
> > +      && !TYPE_UNSIGNED (type)
> > +      && wi::eq_p (wi::to_wide (@1), element_precision (type) - 1))
> > +  (negate (bit_and @0 { build_one_cst (type); }))))
>
> I think the lshift does not need to be in the same type but could be
> in the unsigned type so you should add a nop_convert? operand there
> That is:
> (simplify
>  (rshift (nop_convert? (lshift @0 INTEGER_CST@1)) @@1)
> ...
>  (negate (bit_and (convert @0) { build_one_cst (type); }))))
>
> Note I see you use INTEGER_CST and then ANY_INTEGRAL_TYPE_P. If you
> are going to support vectors, you should use uniform_integer_cst_p
> instead of INTEGER_CST and then use uniform_integer_cst_p to get the
> constant.

Note I just noticed you sent the patch to gcc-bugs@, it really should
go to gcc-patches@ .

Thanks,
Andrew

>
> Thanks,
> Andrew
>
>
> > +
> >  /* Optimize x >> x into 0 */
> >  (simplify
> >   (rshift @0 @0)
> > diff --git a/gcc/testsuite/gcc.dg/pr101955.c b/gcc/testsuite/gcc.dg/pr101955.c
> > new file mode 100644
> > index 00000000000..0e233269e21
> > --- /dev/null
> > +++ b/gcc/testsuite/gcc.dg/pr101955.c
> > @@ -0,0 +1,62 @@
> > +/* { dg-do compile } */
> > +/* { dg-options "-O2 -fdump-tree-dse1 -Wno-psabi" } */
> > +
> > +typedef int v4si __attribute__((vector_size(16)));
> > +
> > +__attribute__((noipa)) int
> > +t1 (int x)
> > +{
> > +  return (x << 31) >> 31;
> > +}
> > +
> > +__attribute__((noipa)) int
> > +t2 (int x)
> > +{
> > +  int y = x << 31;
> > +  int z = y >> 31;
> > +  return z;
> > +}
> > +
> > +__attribute__((noipa)) int
> > +t3 (int x)
> > +{
> > +  int w = 31;
> > +  int y = x << w;
> > +  int z = y >> w;
> > +  return z;
> > +}
> > +
> > +__attribute__((noipa)) long long
> > +t4 (long long x)
> > +{
> > +  return (x << 63) >> 63;
> > +}
> > +
> > +__attribute__((noipa)) long long
> > +t5 (long long x)
> > +{
> > +  long long y = x << 63;
> > +  long long z = y >> 63;
> > +  return z;
> > +}
> > +
> > +__attribute__((noipa)) long long
> > +t6 (long long x)
> > +{
> > +  int w = 63;
> > +  long long y = x << w;
> > +  long long z = y >> w;
> > +  return z;
> > +}
> > +
> > +__attribute__((noipa)) v4si
> > +t7 (v4si x)
> > +{
> > +  return (x << 31) >> 31;
> > +}
> > +
> > +/* { dg-final { scan-tree-dump-not " >> " "dse1" } } */
> > +/* { dg-final { scan-tree-dump-not " << " "dse1" } } */
> > +/* { dg-final { scan-tree-dump-times " -" 7 "dse1" } } */
> > +/* { dg-final { scan-tree-dump-times " & " 7 "dse1" } } */
> > +
> > --
> > 2.39.3
> >