Bug 725842 – sys-cluster/mpich-3.3 USE=fortran - configure: error: The Fortran compiler x86_64-pc-linux-gnu-gfortran will not compile files that call the same routine with arguments of different types.

• Status: RESOLVED FIXED
• Product: Gentoo Linux

[Description Agostino Sarubbo 2020-05-28 07:30:33 UTC]

This is an auto-filed bug because sys-cluster/mpich fails to compile.
The issue was originally discovered on amd64, but it may be reproducible on other arches as well.
If you think that a different summary clarifies the issue better, feel free to change it.
Attached build log and emerge --info.

NOTE:
If you need further logs, feel free to ask.

[Comment 1 Agostino Sarubbo 2020-05-28 07:30:38 UTC]

Created attachment 642306 [details]
build.log

build log and emerge --info

[Comment 2 Agostino Sarubbo 2020-05-28 07:30:41 UTC]

Created attachment 642308 [details]
other.tar.bz2

other logs

[Comment 3 Pavan Balaji 2020-06-04 16:38:46 UTC]

Hello,

mpich would generate a collection of config.log files for various subcomponents.  Is there a way to view those for this bug?

  -- Pavan

[Comment 4 Agostino Sarubbo 2020-06-16 15:01:13 UTC]

Created attachment 644972 [details]
config.log.bz2

config.log.bz2

[Comment 5 Jose 2020-11-29 15:05:54 UTC]

It seems there is a fix for this bug:
https://bugzilla.redhat.com/show_bug.cgi?id=1795817
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91731

By adding "-fallow-argument-mismatch" to gfortran the error is replaced by a warning.

[addCollapseLink(6, 1, 'Toggle comment display'); Comment 6 Sam James 2021-01-02 12:01:29 UTC Comment hidden (obsolete)]

*** This bug has been marked as a duplicate of bug 725722 ***

[Comment 7 Oleg Korsak 2021-01-15 14:46:32 UTC]

same here

[Comment 8 Oleg Korsak 2021-01-15 17:05:45 UTC]

"-fallow-argument-mismatch" in CFLAGS doesn't help. build still fails

[Comment 9 Oleg Korsak 2021-02-23 16:26:38 UTC]

anybody? :/

[Comment 10 Attila Tóth 2021-03-07 11:26:08 UTC]

It must be added to FFLAGs, not to CFLAGs.
There is a separate flag-o-matic routine for that: append-fflags.

[Comment 11 Oleg Korsak 2021-03-08 14:47:04 UTC]

(In reply to Attila Tóth from comment #10)
> It must be added to FFLAGs, not to CFLAGs.
> There is a separate flag-o-matic routine for that: append-fflags.

same with FFLAGS="-fallow-argument-mismatch"

[Comment 12 Attila Tóth 2021-03-08 23:06:18 UTC]

(In reply to Oleg Korsak from comment #11)
> (In reply to Attila Tóth from comment #10)
> > It must be added to FFLAGs, not to CFLAGs.
> > There is a separate flag-o-matic routine for that: append-fflags.
> 
> same with FFLAGS="-fallow-argument-mismatch"

Unfortunately I must confirm your observation. The flag wouldn't show up where it's necessary. I see several points it can be introduced into the build system. I planned to look into this more extensively but I hadn't have enough time for this over the weekend. Will do it later.

[Comment 13 Fredrik Lingvall 2021-04-08 06:03:54 UTC]

I have also hit this issue when using:

sys-devel/gcc-10.2.0-r5

but not when using:

sys-devel/gcc-9.3.0-r2


# FFLAGS=-fallow-argument-mismatch emerge =sys-cluster/mpich-3.3-r1

does not work due to the line:

unset CFLAGS CPPFLAGS CXXFLAGS FFLAGS FCFLAGS LDFLAGS

in the sys-cluster/mpich/mpich-3.3-r1.ebuild file.

Adding something like:

FFLAGS="${FFLAGS} -fallow-argument-mismatch"
export MPICHLIB_CFLAGS="${CFLAGS}"
export MPICHLIB_CPPFLAGS="${CPPFLAGS}"
export MPICHLIB_CXXFLAGS="${CXXFLAGS}"
export MPICHLIB_FFLAGS="${FFLAGS}"
export MPICHLIB_FCFLAGS="${FCFLAGS}"
export MPICHLIB_LDFLAGS="${LDFLAGS}"
#unset CFLAGS CPPFLAGS CXXFLAGS FFLAGS FCFLAGS LDFLAGS 

to the relevant part of the ebuild made it build with gcc-10.2.0-r5

[Comment 14 Attila Tóth 2021-04-09 08:42:13 UTC]

(In reply to Fredrik Lingvall from comment #13)
> does not work due to the line:
> 
> unset CFLAGS CPPFLAGS CXXFLAGS FFLAGS FCFLAGS LDFLAGS

Nice catch! I've missed the unset line...

[Comment 15 Judge Dredd 2021-06-01 01:29:09 UTC]

> # FFLAGS=-fallow-argument-mismatch emerge =sys-cluster/mpich-3.3-r1
> 
> does not work due to the line:
> 
> unset CFLAGS CPPFLAGS CXXFLAGS FFLAGS FCFLAGS LDFLAGS
> 
> in the sys-cluster/mpich/mpich-3.3-r1.ebuild file.

Thanks. I had set package.env and thought I was going insane. Why would they do that?

[Comment 16 Larry the Git Cow 2021-09-23 20:31:30 UTC]

The bug has been closed via the following commit(s):

https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=14d08304513db0d19e6b4720d747fbce302074c3

commit 14d08304513db0d19e6b4720d747fbce302074c3
Author:     Sam James <[e-mail]>
AuthorDate: 2021-09-23 20:20:51 +0000
Commit:     Sam James <[e-mail]>
CommitDate: 2021-09-23 20:31:21 +0000

    sys-cluster/mpich: workaround build failure with GCC 10
    
    Closes: https://bugs.gentoo.org/725842
    Signed-off-by: Sam James <[e-mail]>

 sys-cluster/mpich/mpich-3.3-r1.ebuild | 15 +++++++++++----
 1 file changed, 11 insertions(+), 4 deletions(-)

[Comment 17 Larry the Git Cow 2021-09-23 20:36:56 UTC]

The bug has been referenced in the following commit(s):

https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=146fff709a4168bd9011831dfb64ff3a021855e9

commit 146fff709a4168bd9011831dfb64ff3a021855e9
Author:     Sam James <[e-mail]>
AuthorDate: 2021-09-23 20:36:31 +0000
Commit:     Sam James <[e-mail]>
CommitDate: 2021-09-23 20:36:49 +0000

    sys-cluster/mpich: workaround build failure with GCC 10 (stable)
    
    Bug: https://bugs.gentoo.org/725842
    Signed-off-by: Sam James <[e-mail]>

 sys-cluster/mpich/mpich-3.0.4.ebuild | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)