On Thu, Oct 12, 2023 at 2:40 PM Paul Szczepanek <paul.szczepa...@arm.com> wrote:
>
>
> On 09/10/2023 10:53, Juraj Linkeš wrote:
> > The aarch32 with gcc13 fails with:
> >
> > Compiler for C supports arguments -march=armv8-a: NO
> >
> > ../config/arm/meson.build:714:12: ERROR: Problem encountered: No
> > suitable armv8 march version found.
> >
> > This is because we test -march=armv8-a alone (without the -mpfu option),
> > which is no longer supported in gcc13 aarch32 builds.
> >
> > The most recent recommendation from the compiler team is to build with
> > -march=armv8-a+simd -mfpu=auto, which should work for compilers old and
> > new. The suggestion is to first check -march=armv8-a+simd and only then
> > check -mfpu=auto.
> >
> > To address this, add a way to force the architecture (the value of
> > the -march option).
> >
> > Signed-off-by: Juraj Linkeš <juraj.lin...@pantheon.tech>
> > Acked-by: Ruifeng Wang <ruifeng.w...@arm.com>
> > ---
> > config/arm/meson.build | 12 +++++++++---
> > 1 file changed, 9 insertions(+), 3 deletions(-)
> >
> > diff --git a/config/arm/meson.build b/config/arm/meson.build
> > index 3f22d8a2fc..5303d0e969 100644
> > --- a/config/arm/meson.build
> > +++ b/config/arm/meson.build
> > @@ -43,7 +43,9 @@ implementer_generic = {
> > },
> > 'generic_aarch32': {
> > 'march': 'armv8-a',
> > - 'compiler_options': ['-mfpu=neon'],
> > + 'force_march': true,
> > + 'march_features': ['simd'],
> > + 'compiler_options': ['-mfpu=auto'],
> > 'flags': [
> > ['RTE_ARCH_ARM_NEON_MEMCPY', false],
> > ['RTE_ARCH_STRICT_ALIGN', true],
> > @@ -711,7 +713,11 @@ if update_flags
> > endif
> > endforeach
> > if candidate_march == ''
> > - error('No suitable armv8 march version found.')
> > + if part_number_config.get('force_march', false)
> > + candidate_march = part_number_config['march']
> > + else
> > + error('No suitable armv8 march version found.')
> > + endif
> This section is only used when no candidate is found, this would make it
> not really be a forced arch but more a fallback arch. If we want the
> user to be able to really force the march string we'd need to put the
> "is forced?" check higher. Am I reading the code right?
Yes, you are right. The name should be a bit different to really reflect this.
The question now is what logic do we want. Either this "fallback after
fallback" when the regular fallback doesn't work OR a real forced
march where the regular fallback won't be used at all.
> > endif
> > if candidate_march != part_number_config['march']
> > warning('Configuration march version is ' +
> > @@ -741,7 +747,7 @@ if update_flags
> > # apply supported compiler options
> > if part_number_config.has_key('compiler_options')
> > foreach flag: part_number_config['compiler_options']
> > - if cc.has_argument(flag)
> > + if cc.has_multi_arguments(machine_args + [flag])
> > machine_args += flag
> > else
> > warning('Configuration compiler option ' +
