[arm/testsuite]: Skip pr97969.c if -mthumb is not compatible [PR target/97969]

Christophe Lyon christophe.lyon@linaro.org
Thu Jun 3 08:38:44 GMT 2021


On Mon, 15 Mar 2021 at 17:03, Christophe Lyon
<christophe.lyon@linaro.org> wrote:
>
> On Wed, 3 Mar 2021 at 15:00, Richard Earnshaw
> <Richard.Earnshaw@foss.arm.com> wrote:
> >
> > On 02/03/2021 18:35, Christophe Lyon wrote:
> > > On Tue, 2 Mar 2021 at 19:18, Richard Earnshaw
> > > <Richard.Earnshaw@foss.arm.com> wrote:
> > >>
> > >> On 02/03/2021 18:14, Richard Earnshaw via Gcc-patches wrote:
> > >>> On 02/03/2021 18:10, Christophe Lyon wrote:
> > >>>> On Tue, 2 Mar 2021 at 17:25, Richard Earnshaw
> > >>>> <Richard.Earnshaw@foss.arm.com> wrote:
> > >>>>>
> > >>>>> On 02/03/2021 16:19, Richard Earnshaw via Gcc-patches wrote:
> > >>>>>> On 01/03/2021 15:26, Christophe Lyon via Gcc-patches wrote:
> > >>>>>>> Ping?
> > >>>>>>>
> > >>>>>>> On Wed, 3 Feb 2021 at 10:01, Christophe Lyon <christophe.lyon@linaro.org> wrote:
> > >>>>>>>>
> > >>>>>>>> Ping?
> > >>>>>>>> I guess that's obvious enough?
> > >>>>>>>>
> > >>>>>>>> On Wed, 27 Jan 2021 at 10:03, Christophe Lyon
> > >>>>>>>> <christophe.lyon@linaro.org> wrote:
> > >>>>>>>>>
> > >>>>>>>>> Depending on how the toolchain is configured or how the testsuite is
> > >>>>>>>>> executed, -mthumb may not be compatible. Like for other tests, skip
> > >>>>>>>>> pr97969.c in this case.
> > >>>>>>>>>
> > >>>>>>>>> For instance arm-linux-gnueabihf and -march=armv5t in RUNTESTFLAGS.
> > >>>>>>>>>
> > >>>>>>>>> 2021-01-27  Christophe Lyon  <christophe.lyon@linaro.org>
> > >>>>>>>>>
> > >>>>>>>>> gcc/testsuite/
> > >>>>>>>>> PR target/97969
> > >>>>>>>>> * gcc.target/arm/pr97969.c: Skip if thumb mode is not available.
> > >>>>>>>>>
> > >>>>>>>>> diff --git a/gcc/testsuite/gcc.target/arm/pr97969.c
> > >>>>>>>>> b/gcc/testsuite/gcc.target/arm/pr97969.c
> > >>>>>>>>> index 714a1d1..0b5d07f 100644
> > >>>>>>>>> --- a/gcc/testsuite/gcc.target/arm/pr97969.c
> > >>>>>>>>> +++ b/gcc/testsuite/gcc.target/arm/pr97969.c
> > >>>>>>>>> @@ -1,4 +1,5 @@
> > >>>>>>>>>  /* { dg-do compile } */
> > >>>>>>>>> +/* { dg-skip-if "" { ! { arm_thumb1_ok || arm_thumb2_ok } } } */
> > >>>>>>>>>  /* { dg-options "-std=c99 -fno-omit-frame-pointer -mthumb -w -Os" } */
> > >>>>>>>>>
> > >>>>>>>>>  typedef a[23];
> > >>>>>>
> > >>>>>> I'm working on a patch to make this sort of change unnecessary (I hope).
> > >>>>>>  Just running some final checks.
> > >>>>>>
> > >>>>>> R.
> > >>>>>>
> > >>>>>
> > >>>>> Ah, wait.  This one already has an explicit -mthumb, so my patch won't
> > >>>>> affect this.  But why is -mthumb needed for this test anyway?  It's just
> > >>>>> a compilation test, so why not drop that and we'll generally get better
> > >>>>> coverage all round.
> > >>>>>
> > >>>>
> > >>>> For instance I see the test fail for target arm-none-linux-gnueabihf
> > >>>> --with-mode arm --with-cpu cortex-a9 --with-fpu vfp
> > >>>> and running the tests with -march=armv5t
> > >>>>
> > >>>> We get the famous thumb-1 + hard-float ABI not supported.
> > >>>>
> > >>>> I guess -mthumb is inherited from the bug report?
> > >>>>
> > >>>> Christophe
> > >>>>
> > >>>
> > >>> dropping the -mthumb should fix that though?
> > >>>
> > >>> In fact, I'd drop -Os as well, it's not needed as -Os is just one of the
> > >>> many options that are used to build this test already.
> > >>>
> > >>> R.
> > >>>
> > >>
> > >> But maybe then we need to change dg-options into dg-add-options.
> > >>
> > >
> > > Not sure to follow: the test is compiled only once, with:
> > > -std=c99 -fno-omit-frame-pointer -mthumb -w -Os
> > > in my logs
> > >
> >
> > I think it's only run the once /because/ the test sets dg-options rather
> > than dg-add-options.
> >
>
> Hi, sorry for the delay...
> I guess you mean dg-additional-options ?
> I did try that, to be sure, but the tests in gcc.target/arm are only
> compiled once.
>
> Back to the original discussion, if we drop -mthumb, which is required
> according to the PR (?), how do we ensure coverage? Sure I'm running
> the testsuite with various RUNTESTFLAGS settings, but wouldn't it be
> better to test what the PR reports by default?
>

Hi,

I'm resurrecting this discussion since Vladimir backported his patch
to gcc-9, and I just received a new failure warning from validation on
that branch.

Richard, any update?

Thanks

Christophe

> Thanks
>
> Christophe
> .


More information about the Gcc-patches mailing list