PR 91731 - Configure error on building MPICH

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

PR 91731 - Configure error on building MPICH

Damian Rouson-4
The following is the trailing output of configuring MPICH 3.2, 3.2.1,
or 3.3.1 with the current GCC trunk:

configure: error: The Fortran compiler /opt/gnu/10.0.0/bin/gfortran
will not compile files that call the same routine with arguments of
different types.

I got the same message when I configured in a way that passes gfortran
the -fallow-argument-mismatch argument enabled by the patch posted at
https://gcc.gnu.org/ml/fortran/2019-08/msg00118.html so I assume that
patch hasn't hit the trunk yet. Does GCC 9 have the same issue.  If
so, could the patch be backported to the 9 branch after it is
committed to the trunk?  Also, is the above error message indicative
of non-standard code in MPICH?  If so, I'll submit an issue to the
MPICH developers.

Damian
Reply | Threaded
Open this post in threaded view
|

Re: PR 91731 - Configure error on building MPICH

Steve Kargl
On Tue, Sep 10, 2019 at 06:32:25PM -0700, Damian Rouson wrote:

> The following is the trailing output of configuring MPICH 3.2, 3.2.1,
> or 3.3.1 with the current GCC trunk:
>
> configure: error: The Fortran compiler /opt/gnu/10.0.0/bin/gfortran
> will not compile files that call the same routine with arguments of
> different types.
>
> I got the same message when I configured in a way that passes gfortran
> the -fallow-argument-mismatch argument enabled by the patch posted at
> https://gcc.gnu.org/ml/fortran/2019-08/msg00118.html so I assume that
> patch hasn't hit the trunk yet. Does GCC 9 have the same issue.  If
> so, could the patch be backported to the 9 branch after it is
> committed to the trunk?  Also, is the above error message indicative
> of non-standard code in MPICH?  If so, I'll submit an issue to the
> MPICH developers.
>

-fallow-argument-mismatch should be in trunk.  You probably need
to look in config.log to see if MPICH cannot deal with the warning.

If the above "configure: error:" message is correct, then
whomever is responsible for MPICH needs to fix its expectations
to use an INTERFACE with TYPE(*) or pass the -w option along
with -fallow-argument-mismatch.

--
Steve