[release/gfs.v17] Fix NCO fix-now compiler warnings in GDASApp and JEDI submodules by guillaumevernieres · Pull Request #2254 · NOAA-EMC/GDASApp

guillaumevernieres · 2026-09-16T14:01:21Z

Description

Fixes NCO "fix-now" compiler warnings flagged against release/gfs.v17, in
GDASApp itself and in the JEDI submodules it builds.

GDASApp changes

utils/obsprep/Ghrsst2Ioda.h, Smap2Ioda.h, Viirsaod2Ioda.h: replace
runtime-sized stack arrays (float lat[dim0][dim1], etc. — non-standard
C++ VLAs) with std::vector buffers read via .data(), indexing with
i*ncol+j. Same values, same loop order; the NOLINT markers that were
hiding the VLA lint go away with them.
utils/soca/coupled/gdas_soca_to_fv3.h, utils/soca/diagb/gdas_soca_diagb.h,
utils/soca/diagnostics/gdas_soca_diagnostics.cc: add missing override
on execute() / appname().

Submodule bumps

Each submodule is pointed at a hotfix/warnings-<base-sha> branch cut from
the exact commit release/gfs.v17 currently uses, containing only the
warning fixes for that repo:

Submodule	Base	Hotfix branch
`sorc/fv3-jedi`	`f467912c`	`hotfix/warnings-777c784b`
`sorc/fv3-jedi-lm`	`5994c385`	`hotfix/warnings-d1826a7`
`sorc/ioda`	`724af19e`	`hotfix/warnings-release-211`
`sorc/saber`	`c8e0e3c3`	`hotfix/warnings-c8e0e3c3`
`sorc/soca`	`c9a94cb5`	`hotfix/warnings-c9a94cb`
`sorc/ufo`	`bb78f475`	`hotfix/warnings-bb78f475`

Two of the submodule fixes are behavior changes rather than pure warning
suppression and deserve a closer look on the UFO side (see the UFO PR):
a comma-operator loop bound in filters/Variables.cc that could read past
oopsvars.size(), and a missing GroupBy::RECORD_ID case in
filters/ObsAccessor.cc that made record-number grouping a no-op.

.gitmodules note: the hotfix branches for saber, ioda, ufo,
fv3-jedi-lm, and fv3-jedi live in the jcsda-internal forks, so the
submodule URLs are switched from jcsda/* to jcsda-internal/*. This is
intentional so the pinned SHAs are fetchable, but reviewers should decide
whether that's acceptable for the release branch or whether the fixes need
to land in the public jcsda/* repos first and the URLs reverted.

No change of answers is expected from this PR.

Companion PRs

NOAA-EMC/soca: hotfix/warnings-c9a94cb
JCSDA-internal/ufo: hotfix/warnings-bb78f475
JCSDA-internal/ioda: hotfix/warnings-release-211
JCSDA-internal/saber: hotfix/warnings-c8e0e3c3
JCSDA-internal/fv3-jedi: hotfix/warnings-777c784b
JCSDA-internal/fv3-jedi-linearmodel: hotfix/warnings-d1826a7

Issues

Refs NOAA-EMC/global-workflow#6178

Automated CI tests to run in Global Workflow

RussTreadon-NOAA · 2026-09-16T16:52:32Z

Conduct the following test on WCOSS2 Cactus

clone g-w dev/gfs.v17 at e271e9e
checkout hotfix/warnings-508c40a4 at df6e3d0 in sorc/gdas.cd
follow the instructions in docs/Release_Notes.md to build GFS v17

The build ran to completion.

A check of sorc/build_gdas.log finds 51 lines containing : remark and 920 lines containing : warning. Of the 920 warning lines,

677 are warning #858: type qualifier on return type is meaningless from ioda, saber, and oops
200 are /apps/prod/eckit/install-1.28.0/include/eckit/exception/Exceptions.h(189): warning #2651: attribute does not apply to any entity
22 are warning #6178: The return value of this FUNCTION has not been defined.

These warnings may not be on NCO's fix now list.

CatherineThomas-NOAA · 2026-09-17T12:33:13Z

Thanks for the PR @guillaumevernieres and thanks for testing @RussTreadon-NOAA. From the remaining warnings listed, here's how it compares to NCO's list:

warning#858: Can be addressed later
warning #2651: Not on the list at all?
warning #6178: Fix now - but @guillaumevernieres indicated that these are likely a false positive results

guillaumevernieres · 2026-09-29T14:02:51Z

The 22 #6178 warnings ("return value of this FUNCTION has not been defined") are not bugs in our code. 8 are a confirmed Intel ifort false positive; the other 14 come from a pre-built dependency. Recommend documenting as a known false positive rather than changing source.

Build: WCOSS2, ifort 19.1.3.304, RelWithDebInfo (-O2 -g -DNDEBUG -heap-arrays 32), 0 errors.

`[CSTR_PTR]` × 8: false positive

All 8 trace to f_string_to_c_dup in ioda/src/engines/ioda/fortran/f_c_string_mod.f90:44. Its result cstr_ptr is assigned unconditionally on the first executable statement (cstr_ptr = c_null_ptr) and again via c_alloc, so no path leaves it undefined. There are exactly 8 files that use f_c_string_mod: ifort reports the warning once per caller, with no line number.

`[CPTR]` × 14: in a dependency

No function with a cptr result exists in any bundled repo. The warnings land in saber, ufo, soca, fv3-jedi and oops, whose only common modules are fckit and atlas (pre-built installs). To pin it down on a 19.1.3 toolchain, compile a one-line use <module> probe per fckit/atlas module with the build flags and see which one emits #6178 ... [CPTR].

The diagnostic is compiler-version dependent

The same source with ifort 2021.13 emits zero #6178 (no -diag-disable involved; #5462 still fires in both builds). Public reports show the opposite pattern (clean on 2021.5, warns on 2021.6), and Intel forum threads on the identical #6178 [CPTR] warning conclude the diagnostic is wrong:

No official Intel erratum found.

RussTreadon-NOAA · 2026-09-29T14:50:07Z

Not sure GDASApp CI works with the release/gfs.v17 branch. Let's try and see what happens.

emcbot · 2026-09-29T15:20:12Z

Automated GW-GDASApp Testing Results:
Machine: gaeac6

Start: Tue Sep 29 11:02:03 AM EDT 2026 on gaea68
---------------------------------------------------
Build:                                  *FAILED*
Build: Failed at Tue Sep 29 11:20:11 AM EDT 2026
Build: see output at /gpfs/f6/ira-sti/scratch/role.jedipara/CI/gaeac6/GDASApp/workflow/PR/2254/global-workflow/sorc/log.build

RussTreadon-NOAA · 2026-09-29T15:48:11Z

Build failures

The build of hotfix/warnings-508c40a4 at ced8066 failed on Gaea C6 with the error:

/gpfs/f6/ira-sti/scratch/role.jedipara/CI/gaeac6/GDASApp/workflow/PR/2254/global-workflow/sorc/gdas.cd/bundle/gdas/mains/gdas.cc:12:10: fatal erro\
r: 'ufo/instantiateObsErrorFactory.h' file not found
   12 | #include "ufo/instantiateObsErrorFactory.h"
      |          ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1 error generated.
make[2]: *** [gdas/mains/CMakeFiles/gdas.x.dir/build.make:76: gdas/mains/CMakeFiles/gdas.x.dir/gdas.cc.o] Error 1

git status in sorc/gdas.cd shows that clone of hotfix/warnings-508c40a4 is not complete.

        modified:   parm/jcb-algorithms (new commits, modified content)
        modified:   parm/jcb-gdas (new commits, modified content)
        modified:   sorc/bufr-query (new commits, modified content)
        modified:   sorc/da-utils (new commits, modified content)
        modified:   sorc/gsw (modified content)
        modified:   sorc/jcb (new commits, modified content)
        modified:   sorc/jedicmake (modified content)
        modified:   sorc/land-jediincr (new commits, modified content)
        modified:   sorc/oops (new commits, modified content)
        modified:   sorc/soca (new commits, modified content)
        modified:   sorc/spoc (new commits)
        modified:   sorc/vader (new commits, modified content)

Label-based CI runs from role.jedipara on Gaea C6. Account role.jedipara can not access jscsda-internal repositories.

Change gears and attempt to clone and build as russ.treadon on Cactus. Do the following on Cactus:

cd /lfs/h2/emc/ptmp/russ.treadon
mkdir pr2254
cd pr2254
git clone -b hotfix/warnings-508c40a4 --recursive https://github.com/noaa-emc/gdasapp.git .

Branch hotfix/warnings-508c40a4 was successfully cloned.

russ.treadon@clogin07:/lfs/h2/emc/ptmp/russ.treadon/pr2254> git branch
* hotfix/warnings-508c40a4
russ.treadon@clogin07:/lfs/h2/emc/ptmp/russ.treadon/pr2254> git status
On branch hotfix/warnings-508c40a4
Your branch is up to date with 'origin/hotfix/warnings-508c40a4'.

nothing to commit, working tree clean
russ.treadon@clogin07:/lfs/h2/emc/ptmp/russ.treadon/pr2254> git log --oneline | head -1
ced8066e Merge branch 'release/gfs.v17' into hotfix/warnings-508c40a4

However, the Cactus build fails

russ.treadon@clogin07:/lfs/h2/emc/ptmp/russ.treadon/pr2254> ./build.sh -v -f 
Start ... Tue Sep 29 15:43:28 UTC 2026
Building GDASApp on wcoss2
Resetting modules to system default. Reseting $MODULEPATH back to system default. All extra directories will be removed from $MODULEPATH.

Currently Loaded Modules:
  1) craype-x86-rome    (H)   7) craype/2.7.17      13) pnetcdf-D/1.12.2  19) sp/2.4.0       25) nco/5.2.4      31) fckit/0.13.1
  2) libfabric/1.20.1         8) cray-pals/1.3.2    14) netcdf-D/4.9.2    20) python/3.12.0  26) gsl/2.7        32) atlas/0.39.0
  3) craype-network-ofi (H)   9) git/2.29.0         15) udunits/2.2.28    21) pio-D/2.5.10   27) bufr/12.3.0    33) GDAS/wcoss2.intel
  4) envvar/1.0              10) intel/19.1.3.304   16) eigen/3.4.0       22) ve/gfs/17.0    28) fms-D/2024.01
  5) PrgEnv-intel/8.5.0      11) cray-mpich/8.1.19  17) boost/1.79.0      23) ecbuild/3.7.2  29) esmf-D/8.8.0
  6) cmake/3.27.9            12) hdf5-D/1.14.0      18) gsl-lite/v0.40.0  24) qhull/2020.2   30) eckit/1.28.0

  Where:
   H:  Hidden Module

 

Cloning into '/lfs/h2/emc/ptmp/russ.treadon/pr2254/sorc/soca'...
fatal: unable to access 'https://github.com/jcsda/soca/': SSL certificate problem: unable to get local issuer certificate

@guillaumevernieres : How do you build hotfix/warnings-508c40a4 on Cactus?

guillaumevernieres · 2026-09-29T18:09:29Z

Appologies @RussTreadon-NOAA , I'm pointing to jcsda-internal for all the jedi branches. I'll push to the NOAA-EMC fork and change the url again.

guillaumevernieres · 2026-09-29T21:17:39Z

I just pushed to my fork of the g-w with the correct submodule commits:

 git clone --recursive --branch feature/ee2-warnings-gv https://github.com/guillaumevernieres/global-workflow.git

I just tested on Ursa and the above builds.

AndrewEichmann-NOAA

Nothing obviously wrong by inspection

RussTreadon-NOAA · 2026-09-30T16:43:58Z

WCOSS2 build test

Did the following on Cactus:

Clone g-w dev/gfs.v17 in /lfs/h2/emc/ptmp/russ.treadon/gfs.v17.0.0
cd sorc/gdas.cd
git checkout hotfix/warnings-508c40a4
git pull
git submodule update --init --recursive
cd ..
./build_all.sh gfs gdas gsi > log.build 2>&1 &

The gfs, gdas, and gsi builds successfully completed.

Examine sorc/logs/build_gdas.log. Exclude warnings 858, 2651, and 6178. The following warning messages remain in build_gdas.log.

russ.treadon@clogin06:/lfs/h2/emc/ptmp/russ.treadon/gfs.v17.0.0/sorc/logs> grep ": warning" build_gdas.log  | grep -v "warning #858" | grep -v "warning #2651" | grep -v "warning #6178"
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../x86_64-suse-linux/bin/ld: CMakeFiles/jediincr.dir/sorc/NoahMPdisag_module.f90.o: warning: relocation against `noahmpdisag_module._' in read-only section `.trace'
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../x86_64-suse-linux/bin/ld: warning: creating DT_TEXTREL in a shared object
: warning: relocation in read-only section `                 from /lfs/h2/emc/ptmp/russ.treadon/gfs.v17.0.0/sorc/gdas.cd/bundle/oops/src/oops/mpi/mpi.cc(11):
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../x86_64-suse-linux/bin/ld: warning: creating DT_TEXTREL in a shared object
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../x86_64-suse-linux/bin/ld: CMakeFiles/bufr_query.dir/src/bufr/DataContainer.cpp.o: warning: relocation in read-only section `.trace'
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../x86_64-suse-linux/bin/ld: warning: creating DT_TEXTREL in a shared object
/lfs/h2/emc/ptmp/russ.treadon/gfs.v17.0.0/sorc/gdas.cd/bundle/oops/src/oops/util/Stacktrace.cc(43): warning #2196: routine is both "inline" and "noinline"
/lfs/h2/emc/ptmp/russ.treadon/gfs.v17.0.0/sorc/gdas.cd/bundle/oops/src/oops/util/Stacktrace.cc(43): warning #2196: routine is both "inline" and "noinline"
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../x86_64-suse-linux/bin/ld: CMakeFiles/bufr_python.dir/DataObjectFunctions.cpp.o: warning: relocation in read-only section `.trace'
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../x86_64-suse-linux/bin/ld: warning: creating DT_TEXTREL in a shared object
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../x86_64-suse-linux/bin/ld: /apps/prod/hpc-stack/i-19.1.3.304__m-8.1.19__h-1.14.0__n-4.9.2__p-2.5.10__e-8.8.0_pnetcdf/intel-19.1.3.304/cray-mpich-8.1.19/fms/2024.01/lib/libfms_r8.a(fms_string_utils_binding.c.o): warning: relocation in read-only section `.trace'
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../x86_64-suse-linux/bin/ld: warning: creating DT_TEXTREL in a shared object
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../x86_64-suse-linux/bin/ld: CMakeFiles/ioda_engines.dir/src/ioda/Attribute.cpp.o: warning: relocation in read-only section `.trace'
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../x86_64-suse-linux/bin/ld: warning: creating DT_TEXTREL in a shared object
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../x86_64-suse-linux/bin/ld: CMakeFiles/_ioda_python.dir/py_ioda.cpp.o: warning: relocation in read-only section `.trace'
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../x86_64-suse-linux/bin/ld: warning: creating DT_TEXTREL in a shared object
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../x86_64-suse-linux/bin/ld: CMakeFiles/ioda.dir/ObsSpace.cc.o: warning: relocation in read-only section `.trace'
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../x86_64-suse-linux/bin/ld: warning: creating DT_TEXTREL in a shared object
/lfs/h2/emc/ptmp/russ.treadon/gfs.v17.0.0/sorc/gdas.cd/bundle/ufo/src/ufo/operators/scatwind/NeutralMetOffice/ufo_scatwind_neutralmetoffice_tlad_mod.F90: warning #5462: Global name too long, shortened from: ufo_scatwind_neutralmetoffice_tlad_mod_mp_ufo_scatwind_neutralmetoffice_tlad_settraj_$blk.var$876 to: fo_scatwind_neutralmetoffice_tlad_mod_mp_ufo_scatwind_neutralmetoffice_tlad_settraj_$blk.var$876
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../x86_64-suse-linux/bin/ld: /apps/prod/hpc-stack/i-19.1.3.304__m-8.1.19__h-1.14.0__n-4.9.2__p-2.5.10__e-8.8.0_pnetcdf/intel-19.1.3.304/cray-mpich-8.1.19/fms/2024.01/lib/libfms_r8.a(fms_string_utils_binding.c.o): warning: relocation in read-only section `.trace'
/usr/lib64/gcc/x86_64-suse-linux/7/../../../../x86_64-suse-linux/bin/ld: warning: creating DT_TEXTREL in a shared object

Most of these warnings are from the linker (/bin/ld). The non-linker warnings are

: warning: relocation in read-only section `                 from /lfs/h2/emc/ptmp/russ.treadon/gfs.v17.0.0/sorc/gdas.cd/bundle/oops/src/oops/mpi/mpi.cc(11):
/lfs/h2/emc/ptmp/russ.treadon/gfs.v17.0.0/sorc/gdas.cd/bundle/oops/src/oops/util/Stacktrace.cc(43): warning #2196: routine is both "inline" and "noinline"
/lfs/h2/emc/ptmp/russ.treadon/gfs.v17.0.0/sorc/gdas.cd/bundle/oops/src/oops/util/Stacktrace.cc(43): warning #2196: routine is both "inline" and "noinline"
/lfs/h2/emc/ptmp/russ.treadon/gfs.v17.0.0/sorc/gdas.cd/bundle/ufo/src/ufo/operators/scatwind/NeutralMetOffice/ufo_scatwind_neutralmetoffice_tlad_mod.F90: warning #5462: Global name too long, shortened from: ufo_scatwind_neutralmetoffice_tlad_mod_mp_ufo_scatwind_neutralmetoffice_tlad_settraj_$blk.var$876 to: fo_scatwind_neutralmetoffice_tlad_mod_mp_ufo_scatwind_neutralmetoffice_tlad_settraj_$blk.var$876

These remaining warning messages may not be on NCO's fix-now compiler warning list.

CatherineThomas-NOAA · 2026-09-30T17:29:40Z

Thanks for the hash updates @guillaumevernieres and for that test and checking the logs @RussTreadon-NOAA.

Of the warnings Russ listed and how they compare to NCO's list:
warning #2196: routine is both "inline" and "noinline" --> Can be addressed later
warning #5462: Global name too long --> Can be addressed later

Seems we're in pretty good shape. I've also confirmed that I'm able to clone/build this branch properly now with the NOAA-EMC hashes. I'm running g-w CI to make sure results reproduce, but I'm pretty sure that they will given Russ's previous test results.

apchoiCMD · 2026-10-01T00:30:51Z

For this GDASApp PR

mindo.choi@clogin09:/lfs/h2/emc/da/noscrub/mindo.choi/expts> gw_cistat -r C96C48mx500_S2SW_cyc_gfs/
####################### C96C48mx500_S2SW_cyc_gfs #######################
   CYCLE         STATE           ACTIVATED              DEACTIVATED     
202112201200        Done    Sep 30 2026 19:10:37    Sep 30 2026 19:30:32
202112201800        Done    Sep 30 2026 19:10:37    Sep 30 2026 22:11:02
202112210000        Done    Sep 30 2026 19:10:37    Sep 30 2026 22:35:50
202112211800        Done    Sep 30 2026 19:35:42    Sep 30 2026 22:45:57

guillaumevernieres added 7 commits June 10, 2026 19:24

fixed gdas-utils warnings but still wip

1c913e4

fv3-jedi-lm optional

89c0acf

Merge branch 'release/gfs.v17' into hotfix/warnings-508c40a4

1c5329c

point to fixed-warning branches

a30b255

point to jcsda-internal

a121d26

Revert fv3-jedi-lm optional build switch

922a113

revert jedi data fetching

df6e3d0

CatherineThomas-NOAA mentioned this pull request Sep 17, 2026

JEDI WARNING: nco review NOAA-EMC/GFS#69

Open

Merge branch 'release/gfs.v17' into hotfix/warnings-508c40a4

ced8066

guillaumevernieres marked this pull request as ready for review September 29, 2026 14:14

guillaumevernieres requested review from CatherineThomas-NOAA and RussTreadon-NOAA September 29, 2026 14:14

RussTreadon-NOAA assigned guillaumevernieres Sep 29, 2026

RussTreadon-NOAA added gaeac6-GW-RT and removed gaeac6-GW-RT labels Sep 29, 2026

RussTreadon-NOAA added the gaeac6-GW-RT label Sep 29, 2026

emcbot added gaeac6-GW-RT-Running and removed gaeac6-GW-RT labels Sep 29, 2026

emcbot added gaeac6-GW-RT-Failed and removed gaeac6-GW-RT-Running labels Sep 29, 2026

Update submodule URLs to NOAA-EMC repositories

704942c

guillaumevernieres requested a review from AndrewEichmann-NOAA September 30, 2026 14:36

guillaumevernieres requested review from JohnSteffen-NOAA and apchoiCMD September 30, 2026 14:37

AndrewEichmann-NOAA approved these changes Sep 30, 2026

View reviewed changes

Navigation Menu

[release/gfs.v17] Fix NCO fix-now compiler warnings in GDASApp and JEDI submodules - #2254

[release/gfs.v17] Fix NCO fix-now compiler warnings in GDASApp and JEDI submodules#2254
guillaumevernieres wants to merge 9 commits into
release/gfs.v17from
hotfix/warnings-508c40a4

guillaumevernieres commented Sep 16, 2026

Uh oh!

RussTreadon-NOAA commented Sep 16, 2026

Uh oh!

CatherineThomas-NOAA commented Sep 17, 2026

Uh oh!

guillaumevernieres commented Sep 29, 2026

Uh oh!

RussTreadon-NOAA commented Sep 29, 2026

Uh oh!

emcbot commented Sep 29, 2026

Uh oh!

RussTreadon-NOAA commented Sep 29, 2026

Uh oh!

guillaumevernieres commented Sep 29, 2026

Uh oh!

guillaumevernieres commented Sep 29, 2026

Uh oh!

AndrewEichmann-NOAA left a comment

Uh oh!

RussTreadon-NOAA commented Sep 30, 2026

Uh oh!

CatherineThomas-NOAA commented Sep 30, 2026

Uh oh!

apchoiCMD commented Oct 1, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

Conversation

guillaumevernieres commented Sep 16, 2026

Description

GDASApp changes

Submodule bumps

Companion PRs

Issues

Automated CI tests to run in Global Workflow

Uh oh!

RussTreadon-NOAA commented Sep 16, 2026

Uh oh!

CatherineThomas-NOAA commented Sep 17, 2026

Uh oh!

guillaumevernieres commented Sep 29, 2026

[CSTR_PTR] × 8: false positive

[CPTR] × 14: in a dependency

The diagnostic is compiler-version dependent

Uh oh!

RussTreadon-NOAA commented Sep 29, 2026

Uh oh!

emcbot commented Sep 29, 2026

Uh oh!

RussTreadon-NOAA commented Sep 29, 2026

Uh oh!

guillaumevernieres commented Sep 29, 2026

Uh oh!

guillaumevernieres commented Sep 29, 2026

Uh oh!

AndrewEichmann-NOAA left a comment

Choose a reason for hiding this comment

Uh oh!

RussTreadon-NOAA commented Sep 30, 2026

Uh oh!

CatherineThomas-NOAA commented Sep 30, 2026

Uh oh!

apchoiCMD commented Oct 1, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

`[CSTR_PTR]` × 8: false positive

`[CPTR]` × 14: in a dependency