/usr/bin/ld error when building Fortran interface code about cmor

[durack1]

@strandwg is this that same "gcc was built with disable-multilib on" error we hit back in early September?

[mauzey1]

@strandwg I'm not sure how exactly you are building your Fortran code with CMOR 3.7, but looking on Google for the "unable to initialize decompress status for section .debug_info" error shows several results talking about binutils-2.32. Which version of binutils are you using in your environment?

An example of the issue: https://wiki.gentoo.org/wiki/Project:Toolchain/Binutils_2.32_upgrade_notes/elfutils_0.175:_unable_to_initialize_decompress_status_for_section_.debug_info

[durack1]

For reference - this was identified as a env/compiler conflict:

On Thu, Sep 8, 2022 at 4:35 PM Durack, Paul J. wrote:

This looks like an environment/platform/version issue to me

https://stackoverflow.com/questions/62405518/weird-error-when-compiling-32-bit-programs-with-gcc

Are you using a recent version of gfortran, and is this version aligned with the version used to compile CMOR on conda-forge (10.4.0 – see here)?

P

[strandwg]

The closest version of gnu to 10.4.0 available to me is 10.1.0, and we do have gnu 11.1.0, 11.2.0, and 12.1.0 On Tue, Oct 4, 2022 at 4:54 PM Paul J. Durack ***@***.***> wrote:

…

For reference - this was identified as a env/compiler conflict: On Thu, Sep 8, 2022 at 4:35 PM Durack, Paul J.> wrote: This looks like an environment/platform/version issue to me https://stackoverflow.com/questions/62405518/weird-error-when-compiling-32-bit-programs-with-gcc Are you using a recent version of gfortran, and is this version aligned with the version used to compile CMOR on conda-forge (10.4.0 – see here <https://app.circleci.com/pipelines/github/PCMDI/cmor/274/workflows/575575cb-a89a-4be0-8ec3-f0bd10dc9d97/jobs/4413/parallel-runs/0/steps/0-106> )? P — Reply to this email directly, view it on GitHub <#675 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AEVK36MJT6JGQBG3SD3GPNLWBSYUDANCNFSM6AAAAAAQ467WD4> . You are receiving this because you were mentioned.Message ID: ***@***.***>

-- Gary Strand, CESM Data Manager ORCID 0000-0001-9740-0104 ***@***.***

[durack1]

@strandwg any luck with gfortran 12.1.0?

[strandwg]

The gfortran version doesn't matter - I get the same set of errors:

/usr/bin/ld: /glade/work/strandwg/conda-envs/CMOR/lib/libcmor.a(cmor.o): unable to initialize decompress status for section .debug_info
/usr/bin/ld: /glade/work/strandwg/conda-envs/CMOR/lib/libcmor.a(cmor.o): unable to initialize decompress status for section .debug_info
/usr/bin/ld: /glade/work/strandwg/conda-envs/CMOR/lib/libcmor.a(cmor_fortran_interface.o): unable to initialize decompress status for section .debug_info
/usr/bin/ld: /glade/work/strandwg/conda-envs/CMOR/lib/libcmor.a(cmor_fortran_interface.o): unable to initialize decompress status for section .debug_info
/glade/work/strandwg/conda-envs/CMOR/lib/libcmor.a: error adding symbols: File format not recognized

The other interesting thing is that I get the same errors simply linking in a "Hello, world" Fortran90 program, without calling any CMOR routines at all.

[durack1]

@strandwg just to confirm have you tried to build with disable-multilib=on? see #675 (comment)

[mauzey1]

@strandwg

If you are using a conda environment to build your CMOR programs, then could you provide us a listing of packages in your environment? Use the command conda list -n <env_name>.

You might also want to try cleaning your build directory by running make distclean.

[durack1]

@strandwg this issue seems to have become stale, so will close this out - please reopen if you still have a problem that requires attention