Bug#1057081: mumps: FTBFS with LTO (link time optimization) enabled

Graham Inggs ginggs at debian.org
Wed Nov 29 11:43:38 GMT 2023


Source: mumps
Version: 5.6.2-1
Severity: minor
User: debian-gcc at lists.debian.org
Usertags: ftbfs-lto

Hi Maintainer

MUMPS currently fails to build (at least on the amd64
architecture) with link time optimizations enabled.  For a background
for LTO please see

https://wiki.debian.org/ToolChain/LTO

The goal is to enable this optimization by default in an upcoming
Debian release in dpkg-buildflags for 64-bit architectures.  The goal
is to get this package to build with link time optimizations, or to
explicitly disable link time optimizations for this package build.

To reproduce the build failure, enable the lto optimization in
testing/unstable by adding "optimize=+lto" to DEB_BUILD_MAINT_OPTIONS
in the debian/rules file, or if this macro is unset, just set it:

export DEB_BUILD_MAINT_OPTIONS = optimize=+lto

Please try to fix the build with lto enabled, fixing the packaging or
forwarding the issue upstream. If the issue cannot be fixed,
explicitly disallow building the package with lto by adding to your
rules file:

export DEB_BUILD_MAINT_OPTIONS = optimize=-lto

or adding that string to your existing setting of DEB_BUILD_MAINT_OPTIONS.

I've copied what I hope is the relevant part of the log below.

Regards
Graham


../libseq/mpi.f:648:35: note: previously defined here
../libseq/mpi.f:668:35: error: ‘mumps_copy_logical’ has already been defined
  668 |       SUBROUTINE MUMPS_COPY_LOGICAL( S, R, N )
      |                                   ^
../libseq/mpi.f:668:35: note: previously defined here
../libseq/mpi.f:688:32: error: ‘mumps_copy_real’ has already been defined
  688 |       SUBROUTINE MUMPS_COPY_REAL( S, R, N )
      |                                ^
../libseq/mpi.f:688:32: note: previously defined here
../libseq/mpi.f:708:44: error: ‘mumps_copy_double_precision’ has
already been defined
  708 |       SUBROUTINE MUMPS_COPY_DOUBLE_PRECISION( S, R, N )
      |                                            ^
../libseq/mpi.f:708:44: note: previously defined here
../libseq/mpi.f:718:35: error: ‘mumps_copy_complex’ has already been defined
  718 |       SUBROUTINE MUMPS_COPY_COMPLEX( S, R, N )
      |                                   ^
../libseq/mpi.f:718:35: note: previously defined here
../libseq/mpi.f:728:42: error: ‘mumps_copy_double_complex’ has already
been defined
  728 |       SUBROUTINE MUMPS_COPY_DOUBLE_COMPLEX( S, R, N )
      |                                          ^
../libseq/mpi.f:728:42: note: previously defined here
../libseq/mpi.f:698:45: error: ‘mumps_copy_2double_precision’ has
already been defined
  698 |       SUBROUTINE MUMPS_COPY_2DOUBLE_PRECISION( S, R, N )
      |                                             ^
../libseq/mpi.f:698:45: note: previously defined here
../libseq/mpi.f:678:36: error: ‘mumps_copy_2integer’ has already been defined
  678 |       SUBROUTINE MUMPS_COPY_2INTEGER( S, R, N )
      |                                    ^
../libseq/mpi.f:678:36: note: previously defined here
../libseq/mpi.f:658:36: error: ‘mumps_copy_integer8’ has already been defined
  658 |       SUBROUTINE MUMPS_COPY_INTEGER8( S, R, N )
      |                                    ^
../libseq/mpi.f:658:36: note: previously defined here
../libseq/mpi.f:739:40: error: ‘mumps_is_in_place’ has already been defined
  739 |       LOGICAL FUNCTION MUMPS_IS_IN_PLACE( SENDBUF, CNT )
      |                                        ^
../libseq/mpi.f:739:40: note: previously defined here
../libseq/mpi.f:333:25: error: ‘mpi_init’ has already been defined
  333 |       SUBROUTINE MPI_INIT(IERR)
      |                         ^
../libseq/mpi.f:333:25: note: previously defined here
../libseq/mpi.f:279:29: error: ‘mpi_finalize’ has already been defined
  279 |       SUBROUTINE MPI_FINALIZE( IERR )
      |                             ^
../libseq/mpi.f:279:29: note: previously defined here
lto1: fatal error: errors during merging of translation units
compilation terminated.



More information about the debian-science-maintainers mailing list