On Wed, May 13, 2026 at 9:14 PM Cui, Lili <[email protected]> wrote:
>
>
>
> > -----Original Message-----
> > From: Rainer Orth <[email protected]>
> > Sent: Wednesday, May 13, 2026 4:52 PM
> > To: Cui, Lili <[email protected]>
> > Cc: [email protected]; Liu, Hongtao <[email protected]>;
> > [email protected]
> > Subject: Re: [PATCH] testsuite: Restrict slp-reduc-15.c to x86_64 and
> > aarch64
> >
> > Hi Lili,
> >
> > > diff --git a/gcc/testsuite/gcc.dg/vect/slp-reduc-15.c
> > > b/gcc/testsuite/gcc.dg/vect/slp-reduc-15.c
> > > index 4745f85511b..d22114d6960 100644
> > > --- a/gcc/testsuite/gcc.dg/vect/slp-reduc-15.c
> > > +++ b/gcc/testsuite/gcc.dg/vect/slp-reduc-15.c
> > > @@ -1,6 +1,7 @@
> > > -/* { dg-do compile } */
> > > -/* { dg-require-effective-target vect_float } */
> > > -/* { dg-additional-options "-fgimple -march=x86-64-v3" { target
> > > x86_64-*-* } } */
> > > +/* { dg-do compile { target { x86_64-*-* || aarch64-*-* } } } */
> > > +/* { dg-additional-options "-fgimple" } */
> > > +/* { dg-additional-options "-march=x86-64-v3" { target x86_64-*-* } }
> > > +*/
> >
> > using unadorned target x86_64-*-* is always wrong: a multilibbed
> > i386-*-* target with -m64 also generates 64-bit code. If you want to
> > restrict
> > the test to 64-bit x86, use target { x86 && lp64 } instead.
> >
> Hi Rainer,
>
> Thank you for the feedback and the clear explanation about the target
> selector issue.
> I've updated the test to use { x86 && lp64 } instead of x86_64-*-*, which
> properly handles both i386-*-* configurations with -m64 and excludes
> x86_64-*-* with -m32.
Does the test work for -mx32? If yes, you should use
{ x86 && { ! ia32 } }
> diff --git a/gcc/testsuite/gcc.dg/vect/slp-reduc-15.c
> b/gcc/testsuite/gcc.dg/vect/slp-reduc-15.c
> index 4745f85511b..e8cce237c20 100644
> --- a/gcc/testsuite/gcc.dg/vect/slp-reduc-15.c
> +++ b/gcc/testsuite/gcc.dg/vect/slp-reduc-15.c
> @@ -1,6 +1,7 @@
> -/* { dg-do compile } */
> -/* { dg-require-effective-target vect_float } */
> -/* { dg-additional-options "-fgimple -march=x86-64-v3" { target x86_64-*-* }
> } */
> +/* { dg-do compile { target { { x86 && lp64 } || aarch64-*-* } } } */
> +/* { dg-additional-options "-fgimple" } */
> +/* { dg-additional-options "-march=x86-64-v3" { target { x86 && lp64 } } } */
> +/* { dg-additional-options "-march=armv8.2-a+sve" { target aarch64-*-* } } */
>
> /* Test that SLP reduction vectorization handles commutative operand swap
> for .COND_ADD in multi-lane SLP where the reduction operand appears
> @@ -76,6 +77,6 @@ foo (float * restrict p0, float * restrict p1,
>
> /* With the IFN commutative swap fix, these 4 reductions should be
> vectorized using SLP despite different reduc_idx values (1 vs 2). */
> -/* { dg-final { scan-tree-dump "swapped operands to match def types in"
> "vect" { target x86_64-*-* } } } */
> -/* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 4 "vect"
> { target x86_64-*-* } } } */
> +/* { dg-final { scan-tree-dump "swapped operands to match def types in"
> "vect" } } */
> +/* { dg-final { scan-tree-dump-times "vectorizing stmts using SLP" 4 "vect"
> } } */
>
> Tested on aarch64 and x86_64. OK for trunk?
>
> Thanks,
> Lili.
>
> > Rainer
> >
> > --
> > -----------------------------------------------------------------------------
> > Rainer Orth, Center for Biotechnology, Bielefeld University
--
H.J.
