Internal compiler error gfortran

[hsorby]

When compiling Iron on Ubuntu 18.04 I am getting the following error:

/home/cmiss/tmp/opencmiss/src/iron/src/fieldml_output_routines.F90:182:0:

       FIELDML_OUTPUT_IMPORT_FML = Fieldml_AddImport( FML_HANDLE, IMPORT_INDEX, cchar(REMOTE_NAME), cchar(REMOTE_NAME) )

internal compiler error: Segmentation fault

this is the gfortran compiler version that comes with Ubutnu 18.04

GNU Fortran (Ubuntu 7.4.0-1ubuntu1~18.04.1) 7.4.0

[sgenie68]

You need to get GCC updated to 8+ version

Gene

-- Scientific Software Specialist Auckland Bioengineering Institute The University of Auckland e: g.soudlenkov@auckland.ac.nz p: +64 9 3737599 ext 89834 c: +64 21 840 825 w: https://www.auckland.ac.nz/en/abi.html

On 20/06/19 13:33, Hugh Sorby wrote:

When compiling Iron on Ubuntu 18.04 I am getting the following error:

|/home/cmiss/tmp/opencmiss/src/iron/src/fieldml_output_routines.F90:182:0: FIELDML_OUTPUT_IMPORT_FML = Fieldml_AddImport( FML_HANDLE, IMPORT_INDEX, cchar(REMOTE_NAME), cchar(REMOTE_NAME) ) internal compiler error: Segmentation fault |

this is the gfortran compiler version that comes with Ubutnu 18.04

|GNU Fortran (Ubuntu 7.4.0-1ubuntu1~18.04.1) 7.4.0 |

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/OpenCMISS/iron/issues/181?email_source=notifications&email_token=AAGJIDVHVYE6HEKF7P6QZZ3P3LM7FA5CNFSM4HZORWQ2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4G2RAEUQ, or mute the thread https://github.com/notifications/unsubscribe-auth/AAGJIDXZVT5MFGN46HIBCOLP3LM7FANCNFSM4HZORWQQ.

[hsorby]

I am also getting the same error on gfortran version 8.3

[sgenie68]

Are you sure you used 8.3?

Cheers Gene

Sent from my iPhone

On 21/06/2019, at 5:03 PM, Hugh Sorby notifications@github.com wrote:

I am also getting the same error on gcc version 8.3

— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.

[sgenie68]

Hugh, where did you try compiling it? bioeng49.uoa.auckland.ac.nz had gcc/gfortran reverted back to v7 for some reason - I set up aliases for cmiss user to point to 8.3

Cheers, Gene

-- Scientific Software Specialist Auckland Bioengineering Institute The University of Auckland e: g.soudlenkov@auckland.ac.nz p: +64 9 3737599 ext 89834 c: +64 21 840 825 w: https://www.auckland.ac.nz/en/abi.html

On 21/06/19 17:03, Hugh Sorby wrote:

I am also getting the same error on gcc version 8.3

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/OpenCMISS/iron/issues/181?email_source=notifications&email_token=AAGJIDRTCUWWLYMZS673UTTP3ROI3A5CNFSM4HZORWQ2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODYHNXUA#issuecomment-504290256, or mute the thread https://github.com/notifications/unsubscribe-auth/AAGJIDT4VFTEH37LTL7Y4A3P3ROI3ANCNFSM4HZORWQQ.

[chrispbradley]

I've got 8.3 on my Fedora 29 system and 8.3 works fine and doesn't give a compiler internal error.

[hsorby]

cmiss user on bioeng49 /home/cmiss/tmp directory.

[sgenie68]

Yeah - I requested Tiong to set gcc-8 as a default gcc but yesterday it was reset back to gcc-7. But cmniss user now has aliases set to gcc-8

Gene

-- Scientific Software Specialist Auckland Bioengineering Institute The University of Auckland e: g.soudlenkov@auckland.ac.nz p: +64 9 3737599 ext 89834 c: +64 21 840 825 w: https://www.auckland.ac.nz/en/abi.html

On 21/06/19 20:22, Hugh Sorby wrote:

cmiss user on bioeng49 tmp directory.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/OpenCMISS/iron/issues/181?email_source=notifications&email_token=AAGJIDVCHE63UJGSJW6TPRTP3SFUVA5CNFSM4HZORWQ2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODYHZSGA#issuecomment-504338712, or mute the thread https://github.com/notifications/unsubscribe-auth/AAGJIDTGRY5RUCBX3V7VUK3P3SFUVANCNFSM4HZORWQQ.

[hsorby]

So the update-alternatives was not done correctly and gfortran wasn't updated to match the gcc version. I have now corrected this and things will now hopefully carry on as before. I am currently waiting to see if this has corrected the problem. Which I think it will.

And please do let us know when you are upgrading software on the build slaves as there are repercussions of these changes.