Address failures from merge of updated TriBITS (#10614, TriBITSPub/TriBITS#299) #10774

bartlettroscoe · 2022-07-19T18:20:35Z

Description

With the merge of the updated TriBITS yesterday with the merge of PR #10614, we seem to be seeing a lot of ATDM Trilinos configurations with new build failures for ATDM Trilinos testing day 2022-07-19 after the merge of PR #10614 compared to the previous ATDM Trilinos testing day 2022-07-18 before the merge of PR #10614.

In particular, we see new build failures for 20 builds:

Trilinos-atdm-ats1-hsw_intel-19.0.4_mpich-7.7.15_openmp_static_dbg
Trilinos-atdm-ats1-hsw_intel-19.0.4_mpich-7.7.15_openmp_static_opt
Trilinos-atdm-ats1-knl_intel-19.0.4_mpich-7.7.15_openmp_static_dbg
Trilinos-atdm-ats1-knl_intel-19.0.4_mpich-7.7.15_openmp_static_opt
Trilinos-atdm-ats2-gnu-7.3.1-spmpi-rolling_serial_static_dbg
Trilinos-atdm-ats2-gnu-7.3.1-spmpi-rolling_serial_static_opt
Trilinos-atdm-cee-rhel7_cuda-10.1.243_gnu-7.2.0_openmpi-4.0.3_shared_dbg
Trilinos-atdm-cee-rhel7_cuda-10.1.243_gnu-7.2.0_openmpi-4.0.3_shared_opt
Trilinos-atdm-cee-rhel7_intel-19.0.3_intelmpi-2018.4_serial_static_opt
Trilinos-atdm-cee-rhel7_intel-19.0.3_mpich2-3.2_openmp_static_opt
Trilinos-atdm-cee-rhel7_mini-no-mpi_intel-19.0.3_static_opt
Trilinos-atdm-cee-rhel7_mini_intel-19.0.3_mpich2-3.2_static_opt
Trilinos-atdm-cts1-intel-19.0.4_openmpi-4.0.3_openmp_static_dbg
Trilinos-atdm-cts1-intel-19.0.4_openmpi-4.0.3_openmp_static_opt
Trilinos-atdm-cts1empire-intel-18.0.2_openmpi-4.0.1_openmp_static_dbg
Trilinos-atdm-cts1empire-intel-18.0.2_openmpi-4.0.1_openmp_static_opt
Trilinos-atdm-sems-rhel7-cuda-10.1-Volta70-complex-shared-release-debug

All of the build errors I have looked at so far in the above list are link errors due to missing symbols. My guess is that these are due to missing TPL dependencies.

I will have to do reference builds for each of these and then builds with updated TriBITS (merged from the branch in PR #10614 carefully) and compare the link lines carefully.

NOTE: The build errors for the builds:

Trilinos-atdm-van1-tx2_arm-20.1_openmpi-4.0.5_openmp_static_dbg
Trilinos-atdm-van1-tx2_arm-20.1_openmpi-4.0.5_openmp_static_opt

were due to a a compiler license check like:

clang-9: error: Failed to check out a license. See below for more details. clang-9: note: If you need further help, provide this complete error report to your supplier or support-hpc-sw@arm.com.  - Product information location: /opt/arm/arm-linux-compiler-20.1_Generic-AArch64_RHEL-7_aarch64-linux/sw-mappings  - Toolchain location: /opt/arm/arm-linux-compiler-20.1_Generic-AArch64_RHEL-7_aarch64-linux/llvm-bin  clang-9: note:  - Checkout feature: compiler  - Feature version: 15.20200409  - ALMS error code: -114  - ALMS error message: Timed out while contacting server

NOTE: the build errors for the builds:

Trilinos-atdm-sems-rhel7-clang-7.0.1-openmp-shared-release-debug
Trilinos-atdm-sems-rhel7-clang-7.0.1-openmp-shared-release

were due to unrelated problems with the Krino package (see #10524).

NOTE: The build errors for the build:

Trilinos-atdm-tlcc2-intel-opt-openmp

are source compile errors that seem to have nothing to do with the TriBITS changes (see below).

Tasks

NOTE: Above, a task is marked as complete if the PR that fixes it is ready to merge and is just being held up by a final approval or PR testing.

Deferred Scope

Look into link errors in SEACAS due to setting HDF5_ALLOW_PACKAGE_PREFIND:BOOL=TRUE with current FindTPLHDF5.cmake module (see below) ... Going to wait to upgrade the min version of CMake to 3.22 to use modern targets from find_package(HDF5)

The text was updated successfully, but these errors were encountered:

ikalash · 2022-07-19T20:40:40Z

Just to document, I saw issues when building CISM-Albany (a land-ice modeling code) on top of Trilinos and Albany. The nature of the problem was that Trilinos_LIBRARY_DIRS, previously populated with the right path, has become blank. I fixed it by setting this in the CMakeLists.txt file of CISM-Albany (see attachment). I think this fix is OK for our needs, but please let me know if there is a better one.

Error_Fix_CALI.pdf

bartlettroscoe · 2022-07-19T23:23:08Z

but please let me know if there is a better one.

@ikalash, what is missing is linking to actual targets. I bet if you use CMake 3.23+ and set:

-D CMAKE_LINK_LIBRARIES_ONLY_TARGETS=ON

then the configure will fail and it will show you were it is trying to link against raw string names <libname> instead of a defined target. The way to fix that is to link against actual targets like <Trilinos>::all_selected_libs.

See:

https://cmake.org/cmake/help/latest/prop_tgt/LINK_LIBRARIES_ONLY_TARGETS.html#prop_tgt:LINK_LIBRARIES_ONLY_TARGETS

bartlettroscoe · 2022-07-20T14:07:46Z

There is a bug in the TriBITS-generated <tplName>Config.cmake files with TPL dependencies that get installed that are breaking the SPARC Trilinos Integration builds. The problem is that it is using <tplName>_DIR from the build tree instead of from the install tree. I will create a TriBITS GitHub issue for this and discuss how to address this (see TriBITSPub/TriBITS#500).

NOTE: My local testing of SPARC + Trilinos would not have caught this because the Trilinos build dir does not get removed before configuring SPARC. So this was only going to be caught after the merge to 'develop'.

Details for find_package(Trilinos) errors due to pointing into the build dir

Note that we are also see SPARC Trilinos integration failures starting SPARC testing day 2022-07-19. These are configure-time errors inside of find_package(Trilinos) showing errors like:

REMARK: Trilinos directory
'/projects/atdm_devops/trilinos_installs/2022-07-19/cee-rhel7_clang-9.0.1_openmpi-4.0.3_serial_serial_static_opt'
does not exist; using legacy-style directory:
'/projects/atdm_devops/trilinos_installs/2022-07-19/cee-rhel7_clang-9.0.1_openmpi-4.0.3_serial_static_opt'
-- Searching for Trilinos in Trilinos_ROOT: /projects/atdm_devops/trilinos_installs/2022-07-19/cee-rhel7_clang-9.0.1_openmpi-4.0.3_serial_static_opt
-- Enabled Kokkos devices: SERIAL
CMake Error at /projects/sparc/tools/cmake/3.23.2/x86_64/share/cmake-3.23/Modules/CMakeFindDependencyMacro.cmake:47 (find_package):
  Could not find a package configuration file provided by "METIS" with any of
  the following names:

    METISConfig.cmake
    metis-config.cmake

  Add the installation prefix of "METIS" to CMAKE_PREFIX_PATH or set
  "METIS_DIR" to a directory containing one of the above files.  If "METIS"
  provides a separate development package or SDK, be sure it has been
  installed.
Call Stack (most recent call first):
  /projects/atdm_devops/trilinos_installs/2022-07-19/cee-rhel7_clang-9.0.1_openmpi-4.0.3_serial_static_opt/lib/external_packages/ParMETIS/ParMETISConfig.cmake:27 (find_dependency)
  /projects/atdm_devops/trilinos_installs/2022-07-19/cee-rhel7_clang-9.0.1_openmpi-4.0.3_serial_static_opt/lib/cmake/Zoltan/ZoltanConfig.cmake:151 (include)
  /projects/atdm_devops/trilinos_installs/2022-07-19/cee-rhel7_clang-9.0.1_openmpi-4.0.3_serial_static_opt/lib/cmake/SEACASIoss/SEACASIossConfig.cmake:158 (include)
  /projects/atdm_devops/trilinos_installs/2022-07-19/cee-rhel7_clang-9.0.1_openmpi-4.0.3_serial_static_opt/lib/cmake/SEACAS/SEACASConfig.cmake:188 (include)
  /projects/atdm_devops/trilinos_installs/2022-07-19/cee-rhel7_clang-9.0.1_openmpi-4.0.3_serial_static_opt/lib/cmake/Trilinos/TrilinosConfig.cmake:114 (include)
  CMakeLists.txt:246 (find_package)
  CMakeLists.txt:323 (find_package_Trilinos)

That code that is failing in ParMETISConfig.cmake is:

set(METIS_DIR "/scratch/trilinos/atdm-trilinos-nightly-builds/Trilinos-atdm-cee-rhel7_clang-9.0.1_openmpi-4.0.3_serial_static_opt/SRC_AND_BUILD/BUILD/external_packages/METIS")
find_dependency(METIS REQUIRED CONFIG ${ParMETIS_SearchNoOtherPathsArgs})
unset(METIS_DIR)

Hum, that is the wrong directory location:

$ ls /scratch/trilinos/atdm-trilinos-nightly-builds/Trilinos-atdm-cee-rhel7_clang-9.0.1_openmpi-4.0.3_serial_static_opt/SRC_AND_BUILD/BUILD/external_packages/METIS
ls: cannot access /scratch/trilinos/atdm-trilinos-nightly-builds/Trilinos-atdm-cee-rhel7_clang-9.0.1_openmpi-4.0.3_serial_static_opt/SRC_AND_BUILD/BUILD/external_packages/METIS: No such file or directory

Oops, that is a bug! How did that slip through TriBITS testing?

bartlettroscoe · 2022-07-20T15:08:05Z

The top builds to prioritize fixing are the ones used in SPARC Trilinos Integration testing listed here and are:

ats1-hsw_intel-19.0.4_mpich-7.7.15_openmp_static_opt
ats1-knl_intel-19.0.4_mpich-7.7.15_openmp_static_opt
ats2-cuda-10.1.243-gnu-7.3.1-spmpi-rolling_static_opt
cee-rhel7_clang-9.0.1_openmpi-4.0.3_serial_static_opt
cee-rhel7_gnu-7.2.0_openmpi-4.0.3_serial_shared_opt
cee-rhel7_intel-19.0.3_intelmpi-2018.4_serial_static_opt
cee-rhel7_intel-19.0.3_mpich2-3.2_openmp_static_opt
cts1-intel-19.0.4_openmpi-4.0.3_openmp_static_opt
van1-tx2_arm-20.1_openmpi-4.0.5_openmp_static_opt

bartlettroscoe · 2022-07-20T15:34:57Z

FYI: I created the branch 10774-pre-tribits-update-ref off of the commit bdcac66 which corresponds to the version of Trilinos 'develop' for the 'atdm-nighlty' branch for testing day 2022-07-18 just before the merge of PR #10614. Details are below.

Creating the branch '10774-pre-tribits-update-ref' for ATDM Trilinos testing day 2022-07-18 details: (click to expand)

The version of that ATDM Trilinos builds for the testing day 2022-07-18 just before PR #10614 was synced was f15068c. That comes from:

$ git log-short --name-status --graph f15068c6c536560f3f30e62d5dd84c025edeb601

*   f15068c6c53 "Merge remote-tracking branch 'origin/develop' into atdm-nightly"
|\  Author: trilinos-autotester <trilinos@sandia.gov>
| | Date:   Sun Jul 17 21:05:07 2022 -0600 (2 days ago)
| |   
| *   bdcac66d1be "Merge Pull Request #10736 from iyamazaki/Trilinos/FastILU-metis"
| |\  Author: trilinos-autotester <trilinos@sandia.gov>
| | | Date:   Sun Jul 17 07:19:08 2022 -0600 (2 days ago)
| | | 
| | * d087ab4891f "FastILU : sort only not Metis, and sorted Tpetra CrsMatrix"
| | | Author: iyamazaki <ic.yamazaki@gmail.com>
| | | Date:   Wed Jul 13 21:59:22 2022 -0600 (6 days ago)
| | | 
| | | M packages/ifpack2/src/Ifpack2_Details_Filu_def.hpp

So the Trilinos version of 'develop' for the ATDM Trilinos testing day

So I create the Trilinos branch '10774-pre-tribits-update-ref' off of 'develop' at the commit bdcac66:

$ git checkout -b 10774-pre-tribits-update-ref bdcac66d1be
Switched to a new branch '10774-pre-tribits-update-ref'

$ git push -u rab-github HEAD
Total 0 (delta 0), reused 0 (delta 0), pack-reused 0
remote: 
remote: Create a pull request for '10774-pre-tribits-update-ref' on GitHub by visiting:
remote:      https://github.com/bartlettroscoe/Trilinos/pull/new/10774-pre-tribits-update-ref
remote: 
To github.com:bartlettroscoe/Trilinos
 * [new branch]              HEAD -> 10774-pre-tribits-update-ref
Branch '10774-pre-tribits-update-ref' set up to track remote branch '10774-pre-tribits-update-ref' from 'rab-github'.

…all-tree (trilinos#10774) Addresses TriBITSPub/TriBITS#500 which is part of trilinos#10774.

bartlettroscoe · 2022-07-20T19:19:28Z

FYI: The fix for the bug with installed <tplName>Config.cmake files pointing into the build tree is in PR #10784. Just need someone to approve that PR and then PR builds need to pass. In the meantime, I am going to manually merge into the 'atdm-nightly' branch.

…nual-updates (#10774) Addresses TriBITSPub/TriBITS#500

bartlettroscoe · 2022-07-20T19:35:09Z

I manually merged the PR branch for #10784 into the 'atdm-nightly-manual-updates' branch so this will get run in the ATDM Trilinos builds tomorrow and we should see (some of) the SPARC Trilinos Integration builds clear up the day after that.

Update of 'atdm-nightly-manual-updates' branch: (click to expand)

$ git checkout atdm-nightly-manual-updates 

$ git fetch pull

$ git merge --ff github/atdm-nightly

$ git merge 10774-fix-tpl-config-install-tree 
Merge made by the 'recursive' strategy.
 cmake/tribits/CHANGELOG.md                                                  | 8 ++++++++
 cmake/tribits/core/package_arch/TribitsExternalPackageWriteConfigFile.cmake | 2 +-
 cmake/tribits/core/package_arch/TribitsProcessTplsLists.cmake               | 2 +-
 cmake/tribits/ctest_driver/README                                           | 8 ++++----
 cmake/tribits/ctest_driver/TribitsGetCDashUrlsInsideCTestS.cmake            | 2 +-
 cmake/tribits/examples/MixedSharedStaticLibs/README                         | 2 +-
 6 files changed, 16 insertions(+), 8 deletions(-)

$ git log-short --name-status HEAD --not github/develop github/atdm-nightly

35735a8b7c9 "Merge branch '10774-fix-tpl-config-install-tree' into atdm-nightly-manual-updates (#10774)"
Author: Roscoe A. Bartlett <rabartl@sandia.gov>
Date:   Wed Jul 20 13:29:52 2022 -0600 (45 seconds ago)

4038856c026 "Merge branch 'tribits_github_snapshot' into 10774-fix-tpl-config-install-tree (#10774)"
Author: Roscoe A. Bartlett <rabartl@sandia.gov>
Date:   Wed Jul 20 13:12:54 2022 -0600 (18 minutes ago)

f7f6ed3495f "Automatic snapshot commit from tribits at 5d4696b2"
Author: Roscoe A. Bartlett <rabartl@sandia.gov>
Date:   Wed Jul 20 13:10:37 2022 -0600 (20 minutes ago)

M       cmake/tribits/CHANGELOG.md
M       cmake/tribits/core/package_arch/TribitsExternalPackageWriteConfigFile.cmake
M       cmake/tribits/core/package_arch/TribitsProcessTplsLists.cmake
M       cmake/tribits/ctest_driver/README
M       cmake/tribits/ctest_driver/TribitsGetCDashUrlsInsideCTestS.cmake
M       cmake/tribits/examples/MixedSharedStaticLibs/README

$ git push
Enumerating objects: 20, done.
Counting objects: 100% (18/18), done.
Delta compression using up to 32 threads
Compressing objects: 100% (8/8), done.
Writing objects: 100% (8/8), 1.03 KiB | 350.00 KiB/s, done.
Total 8 (delta 6), reused 0 (delta 0), pack-reused 0
remote: Resolving deltas: 100% (6/6), completed with 4 local objects.
To github.com:trilinos/Trilinos.git
   a0cabff009f..35735a8b7c9  atdm-nightly-manual-updates -> atdm-nightly-manual-updates

bartlettroscoe · 2022-07-21T16:32:19Z

Looking over all of the errors in the builds listed above for ATDM Trilinos testing day 2022-07-19 falls into a few different categories:

1) Link errors for libcgns missing hdf5 symbols

These errors look like:

/projects/sparc/tpls/cee-rhel7/cgns-c09a5cd/a80e7a6e298e722aef76764ff37c4fc0aec5fabc/cee-cpu_intel-19.0.3_intelmpi-2018.4/lib/libcgns.a(ADFH.c.o): In function `get_str_att':
ADFH.c:(.text+0x259): undefined reference to `H5Aopen_name'
/projects/sparc/tpls/cee-rhel7/cgns-c09a5cd/a80e7a6e298e722aef76764ff37c4fc0aec5fabc/cee-cpu_intel-19.0.3_intelmpi-2018.4/lib/libcgns.a(ADFH.c.o): In function `delete_children':
ADFH.c:(.text+0x58a): undefined reference to `H5Aopen_name'
ADFH.c:(.text+0x8af): undefined reference to `H5Aopen_name'

This includes the builds:

Trilinos-atdm-ats2-gnu-7.3.1-spmpi-rolling_serial_static_dbg
Trilinos-atdm-ats2-gnu-7.3.1-spmpi-rolling_serial_static_opt
Trilinos-atdm-cee-rhel7_intel-19.0.3_intelmpi-2018.4_serial_static_opt
Trilinos-atdm-cee-rhel7_intel-19.0.3_mpich2-3.2_openmp_static_opt
Trilinos-atdm-cts1-intel-19.0.4_openmpi-4.0.3_openmp_static_opt
Trilinos-atdm-cts1empire-intel-18.0.2_openmpi-4.0.1_openmp_static_opt
Trilinos-atdm-cts1-intel-19.0.4_openmpi-4.0.3_openmp_static_dbg
Trilinos-atdm-cts1empire-intel-18.0.2_openmpi-4.0.1_openmp_static_dbg
Trilinos-atdm-cee-rhel7_intel-19.0.3_intelmpi-2018.4_serial_static_opt
Trilinos-atdm-cee-rhel7_intel-19.0.3_mpich2-3.2_openmp_static_opt
Trilinos-atdm-cee-rhel7_mini-no-mpi_intel-19.0.3_static_opt
Trilinos-atdm-cee-rhel7_mini_intel-19.0.3_mpich2-3.2_static_opt

I think these can all be fixed by adding FindTPLCGNSDependencies.cmake and adding a dependency of CGNS on HDF5.

ToDo: Characterize the other failures!

This should fix the build errors for the SEACAS exectuable SEACASIoss_Utst_structured_decomp.exe on a bunch of ATDM Trilinos builds (see

…rilinos#10774) A defect added as part of work on TriBITSPub/TriBITS#299.

…rilinos/Trilinos#10774) THe logic for constructing the full path for the FindTPL<tplName>Dependencies.cmake file is not correct when <tplName>_FINDMOD is an absolute path instead of a relative path. This causes all of the 'TribitsExampleProject2_find_tpl_parts' tests to fail.

Origin repo remote tracking branch: 'github/master' Origin repo remote repo URL: 'github = git@github.com:TriBITSPub/TriBITS.git' Git describe: Vera4.0-RC1-start-1219-g8b3872ed At commit: commit 4b26997a2b19c29cbc6deaba5ad303b2336b63e6 Author: Roscoe A. Bartlett <rabartl@sandia.gov> Date: Thu Jul 21 10:35:22 2022 -0600 Summary: Add dependency of CGNS on HDF5 (#10774)

…ilinos#10774) Brings in PR TriBITSPub/TriBITS#502

bartlettroscoe · 2022-07-21T22:30:01Z

Continuing from above, the next failure seen in many different builds is:

2) CUDA build failures due to duplicate CUDA library function definitions when creating libIntercept.so.13.5

These errors look like:

/projects/sierra/linux_rh7/SDK/compilers/nvidia/cuda_10.1.243/bin/..//lib64/libcudart_static.a(libcudart_static.a.o): In function `cudaMemcpy2DFromArray':
(.text+0x2c300): multiple definition of `cudaMemcpy2DFromArray'
CMakeFiles/Intercept.dir/Intercept.cpp.o:tmpxft_000073c0_00000000-5_Intercept.cudafe1.cpp:(.text+0x830): first defined here
/projects/sierra/linux_rh7/SDK/compilers/nvidia/cuda_10.1.243/bin/..//lib64/libcudart_static.a(libcudart_static.a.o): In function `cudaMemcpy3D':
(.text+0x2bc20): multiple definition of `cudaMemcpy3D'
CMakeFiles/Intercept.dir/Intercept.cpp.o:tmpxft_000073c0_00000000-5_Intercept.cudafe1.cpp:(.text+0xcd0): first defined here
/projects/sierra/linux_rh7/SDK/compilers/nvidia/cuda_10.1.243/bin/..//lib64/libcudart_static.a(libcudart_static.a.o): In function `cudaMemcpy2DArrayToArray':
(.text+0x2bdc0): multiple definition of `cudaMemcpy2DArrayToArray'
CMakeFiles/Intercept.dir/Intercept.cpp.o:tmpxft_000073c0_00000000-5_Intercept.cudafe1.cpp:(.text+0x700): first defined here
...
<and many more>
...

We see these in the CUDA builds:

ToDo: Characterize the rest of the failures!

…stall-tree Fix <tplName>Config.cmake files to not point into build dir (#10774) We are force merging this PR because all of the PR tests passed except the new CUDA 11 build, which had 2 failing tests. There were 2 Tpetra tests that failed that have been randomly failing for other PRs and the failures are not related to these changes.

bartlettroscoe · 2022-07-22T01:03:34Z

Continuing from above, the next failure seen in many different builds is:

3) Link errors missing '__dlopen'

These errors look like:

/usr/lib/../x86_64-suse-linux/bin/ld: packages/kokkos/core/src/libkokkoscore.a(Kokkos_Profiling.cpp.o): in function `Kokkos::Tools::initialize(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&)':
[CTest: warning matched] Kokkos_Profiling.cpp:(.text+0x2994): warning: Using 'dlopen' in statically linked applications requires at runtime the shared libraries from the glibc version used for linking
/usr/lib/../x86_64-suse-linux/bin/ld: /usr/lib/../lib64/libdl.a(dlopen.o): in function `dlopen':
/home/abuild/rpmbuild/BUILD/glibc-2.26/dlfcn/dlopen.c:30: undefined reference to `__dlopen'
/usr/lib/../x86_64-suse-linux/bin/ld: /usr/lib/../lib64/libdl.a(dlsym.o): in function `dlsym':
/home/abuild/rpmbuild/BUILD/glibc-2.26/dlfcn/dlsym.c:29: undefined reference to `__dlsym'
/usr/lib/../x86_64-suse-linux/bin/ld: /usr/lib/../lib64/libdl.a(dlerror.o): in function `dlerror':
/home/abuild/rpmbuild/BUILD/glibc-2.26/dlfcn/dlerror.c:33: undefined reference to `__dlerror'

These errors start in Sacado for the 'ats1' opt builds on 'mutrino':

and these errors start in SEACAS for the 'ats1' dbg builds on 'mutrino'

That is all of the systematic errors that impact all of the builds.

bartlettroscoe · 2022-07-22T01:12:25Z

4) Compile errors in Kokkos for coming from gtest.h

The last new set of build errors are just for the build:

Trilinos-atdm-tlcc2-intel-opt-openmp

and they look like:

In file included from /gpfs/jenkins/skybridge-slave/workspace/Trilinos-atdm-tlcc2-intel-opt-openmp/SRC_AND_BUILD/Trilinos/packages/kokkos/core/unit_test/category_files/TestOpenMP_Category.hpp(48),
                 from packages/kokkos/core/unit_test/openmp/TestOpenMP_AtomicOperations_unsignedlongint.cpp(1):
/gpfs/jenkins/skybridge-slave/workspace/Trilinos-atdm-tlcc2-intel-opt-openmp/SRC_AND_BUILD/Trilinos/packages/kokkos/tpls/gtest/gtest/gtest.h(4588): error: no instance of function template "testing::internal::ElemFromListImpl<testing::internal::IndexSequence<I...>>::Apply [with I=<0UL>]" matches the argument list
            argument types are: (bool (*)(), bool (*)())
        decltype(ElemFromListImpl<typename MakeIndexSequence<N>::type>::Apply(
                 ^
[CTest: warning matched] /gpfs/jenkins/skybridge-slave/workspace/Trilinos-atdm-tlcc2-intel-opt-openmp/SRC_AND_BUILD/Trilinos/packages/kokkos/tpls/gtest/gtest/gtest.h(4582): note: this candidate was rejected because arguments do not match
    static R Apply(Ignore<0 * I>..., R (*)(), ...);
             ^
          detected during:
            instantiation of class "testing::internal::ElemFromList<N, T...> [with N=1UL, T=<bool, bool>]" at line 4602
            instantiation of class "testing::internal::FlatTupleElemBase<testing::internal::FlatTuple<T...>, I> [with T=<bool, bool>, I=1UL]" at line 4615
            instantiation of class "testing::internal::FlatTupleBase<testing::internal::FlatTuple<T...>, testing::internal::IndexSequence<Idx...>> [with Idx=<0UL, 1UL>, T=<bool, bool>]" at line 4655
            instantiation of class "testing::internal::FlatTuple<T...> [with T=<bool, bool>]" at line 8741
            instantiation of class "testing::internal::ValueArray<Ts...> [with Ts=<bool, bool>]" at line 9059

compilation aborted for packages/kokkos/core/unit_test/openmp/TestOpenMP_AtomicOperations_unsignedlongint.cpp (code 2)

and they are all in Kokkos. I am very curious to see what is causing this.

But from this query it seems these build errors started on testing day 2022-05-21 which was the first since testing day 2022-05-16 as shown below:

So these errors are unrelated to the TriBITS update (which is good because I have no idea how the TriBITS updates could be causing this).

…os#10774, TriBITSPub/TriBITS#516) A TriBITS update is exporting package cache vars to the <Package>Config.cmake file and you can't have a local var with the same name as a cache var with different values. In this case, it was just lucky that no downstream package was reading this var (through the cache var) because they would have gotten the wrong value. It seems that only code in CMakeLists.txt files under packages/pliris/ were reading this var.

…os#10774)

bartlettroscoe · 2022-08-25T03:02:32Z

FYI: The failure to tentatively enable the BinUtils TPL shown above for the build Trilinos-atdm-ats1-knl_intel-19.0.4_mpich-7.7.15_openmp_static_opt was occurring before the merge of PR #10614 on 2022-07-18. If you go back to the build Trilinos-atdm-ats1-knl_intel-19.0.4_mpich-7.7.15_openmp_static_opt on testing day 2022-07-17 you will see:

Processing enabled TPL: BinUtils (enabled explicitly, disable with -DTPL_ENABLE_BinUtils=OFF)
-- BinUtils_LIBRARY_NAMES='bfd;iberty'
-- TPL_BinUtils_LIBRARIES='/usr/lib64/libbfd.a;-lz;/usr/lib64/libiberty.a'
-- Must find at least one header in each of the header sets "link.h;bfd.h"
-- Searching for headers in BinUtils_INCLUDE_DIRS=''
-- Searching for a header file in the set "link.h":
--   Searching for header 'link.h' ...
--     Found header '/usr/include/link.h'
-- Searching for a header file in the set "bfd.h":
--   Searching for header 'bfd.h' ...
--     Found header '/usr/include/bfd.h'
-- Found TPL 'BinUtils' include dirs '/usr/include'
-- TPL_BinUtils_INCLUDE_DIRS='/usr/include'
-- Performing Test HAS_TPL_BINUNTILS_STACKTRACE
-- Performing Test HAS_TPL_BINUNTILS_STACKTRACE - Failed
-- Extended attempt to enable tentatively enabled TPL 'BinUtils' failed!  Setting TPL_ENABLE_BinUtils=OFF
-- TPL_ENABLE_BinUtils='OFF'

And looking at one of the recent builds today for Trilinos-atdm-ats1-knl_intel-19.0.4_mpich-7.7.15_openmp_static_opt you can see:

...

-- Tentatively enabling TPL 'DLlib'

...

Processing enabled TPL: DLlib (enabled explicitly, disable with -DTPL_ENABLE_DLlib=OFF)
-- Attempting to tentatively enable TPL 'DLlib' ...
-- DLlib_LIBRARY_NAMES='dl'
-- TPL_DLlib_LIBRARIES='-ldl'
-- Attempt to tentatively enable TPL 'DLlib' passed!

...

Therefore, the failure to tentatively enable the BinUtils TPL has nothing to do with the TriBITS upgrade merged in PR #10614.

bartlettroscoe · 2022-08-25T03:18:04Z

With the post of PR #10930, all of these issues should be addressed. Putting this issue In Review.

…os#10774, TriBITSPub/TriBITS#522)

Automatically Merged using Trilinos Pull Request AutoTester PR Title: Final fixes for TriBITS upgrade to modern CMake targets (#10614, #10774) PR Author: bartlettroscoe

bartlettroscoe · 2022-08-26T02:08:20Z

With the merge of PR #10930, this is (finally) complete!

…s:develop' (8906842). * trilinos-develop: (128 commits) Intrepid2: update TensorData.setFirstComponentExtentInDimension0 to modify extents_[0] (trilinos#10929) Tpetra: Adding configure option to disable Kokkos integration test Automatic snapshot commit from tribits at 142e5362 Disable Pliris tests in ATS2 GenConfig builds (trilinos#10931) Force disable Pliris in ATS2 builds (trilinos#10931) Automatic snapshot commit from tribits at ab419429 Change cmake_minimum_required() from 3.17.1 to 3.0 (TriBITSPub/TriBITS#522) Pliris: Remove local var hiding cache var Pliris_ENABLE_DREAL (trilinos#10774, TriBITSPub/TriBITS#516) Remove printing of vars that are now empty (TriBITSPub/TriBITS#299) Panzer: move periodic helper typedefs into namespace Revert incorrect fix in previous commit Fix typos in some docs fix scratch typos STK: Snapshot 08-22-22 12:44 Phalanx: remove cuda compiler warnings and add test for new use case for vov changed a double to a scalar_type to compile for complex arith MueLu: Fix signed vs unsigned comparison in Aggregates_kokkos.cpp Amesos2 : trying to fix MKL header including issues MueLu: Add Aggregates_kokkos.ComputeNodesInAggregate Testing on Geminga: Do not disable Kokkos in Epetra build ...

…s:develop' (8906842). * trilinos-develop: (130 commits) Intrepid2: update TensorData.setFirstComponentExtentInDimension0 to modify extents_[0] (trilinos#10929) Tpetra: Adding configure option to disable Kokkos integration test MueLu: Allow to print Kokkos config when default node type is used Automatic snapshot commit from tribits at 142e5362 Disable Pliris tests in ATS2 GenConfig builds (trilinos#10931) Force disable Pliris in ATS2 builds (trilinos#10931) Automatic snapshot commit from tribits at ab419429 Change cmake_minimum_required() from 3.17.1 to 3.0 (TriBITSPub/TriBITS#522) Pliris: Remove local var hiding cache var Pliris_ENABLE_DREAL (trilinos#10774, TriBITSPub/TriBITS#516) Remove printing of vars that are now empty (TriBITSPub/TriBITS#299) Panzer: move periodic helper typedefs into namespace Revert incorrect fix in previous commit Fix typos in some docs fix scratch typos STK: Snapshot 08-22-22 12:44 Phalanx: remove cuda compiler warnings and add test for new use case for vov changed a double to a scalar_type to compile for complex arith MueLu: Fix signed vs unsigned comparison in Aggregates_kokkos.cpp Amesos2 : trying to fix MKL header including issues MueLu: Add Aggregates_kokkos.ComputeNodesInAggregate ...

bartlettroscoe · 2022-09-06T16:30:48Z

Shoot, reopening. Seems that there was some delayed testing with Nalu-Wind and just got flagged in #10954. Repopening this story and adding #10954 to the list of tasks.

…os#10774, TriBITSPub/TriBITS#516) A TriBITS update is exporting package cache vars to the <Package>Config.cmake file and you can't have a local var with the same name as a cache var with different values. In this case, it was just lucky that no downstream package was reading this var (through the cache var) because they would have gotten the wrong value. It seems that only code in CMakeLists.txt files under packages/pliris/ were reading this var.

bartlettroscoe · 2022-10-04T12:26:36Z

With the PRs posted #11093 and #11099, all of the known issues for the TriBITS upgrade should be resolved. Moving to "In Review".

bartlettroscoe · 2022-10-06T19:06:55Z

With the merges of PRs #11093 and #11099, the last of these issues should be resolved.

Closing again!

bartlettroscoe self-assigned this Jul 19, 2022

bartlettroscoe added type: bug The primary issue is a bug in Trilinos code or tests ATDM Config Issues that are specific to the ATDM configuration settings labels Jul 19, 2022

This was referenced Jul 19, 2022

compilation failure for exotxt sandialabs/seacas#323

Closed

TriBITS: Switch to modern cmake targets (TriBITSPub/TriBITS#299) #10614

Merged

bartlettroscoe mentioned this issue Jul 20, 2022

Fix installed <tplName>Config.cmake files to not point into build tree TriBITSPub/TriBITS#500

Closed

2 tasks

bartlettroscoe added a commit to bartlettroscoe/Trilinos that referenced this issue Jul 20, 2022

Merge branch 'tribits_github_snapshot' into 10774-fix-tpl-config-inst…

4038856

…all-tree (trilinos#10774) Addresses TriBITSPub/TriBITS#500 which is part of trilinos#10774.

bartlettroscoe mentioned this issue Jul 20, 2022

Fix <tplName>Config.cmake files to not point into build dir (#10774) #10784

Merged

bartlettroscoe added a commit that referenced this issue Jul 20, 2022

Merge branch '10774-fix-tpl-config-install-tree' into atdm-nightly-ma…

35735a8

…nual-updates (#10774) Addresses TriBITSPub/TriBITS#500

bartlettroscoe added a commit to bartlettroscoe/Trilinos that referenced this issue Jul 21, 2022

Add dependency of CGNS on HDF5 (trilinos#10774)

d740429

This should fix the build errors for the SEACAS exectuable SEACASIoss_Utst_structured_decomp.exe on a bunch of ATDM Trilinos builds (see

bartlettroscoe added a commit to bartlettroscoe/Trilinos that referenced this issue Jul 21, 2022

WIP: Correctly handle abs dir for findmod for TPL dependencies file (t…

07bae56

…rilinos#10774) A defect added as part of work on TriBITSPub/TriBITS#299.

bartlettroscoe mentioned this issue Jul 21, 2022

Fix abs path issue with TPL dependencies, add CGNS/HDF5 dep (#299, trilinos/Trilinos#10744) TriBITSPub/TriBITS#502

Merged

bartlettroscoe added a commit to bartlettroscoe/Trilinos that referenced this issue Jul 21, 2022

Merge branch 'tribits_github_snapshot' into 10774-fix-atdm-builds (tr…

c651bff

…ilinos#10774) Brings in PR TriBITSPub/TriBITS#502

This was referenced Jul 21, 2022

IMPLICITLY MERGED: Fix ATDM Trilinos builds broken by TriBITS update (#10774) #10791

Closed

Move to modern CMake breaking Trilinos backward compatibility by changing include dirs from -I to -isystem #10160

Closed

bartlettroscoe added a commit to bartlettroscoe/Trilinos that referenced this issue Aug 25, 2022

Merge branch 'tribits_github_snapshot' into 10774-final-fixes (trilin…

a4e5427

…os#10774)

bartlettroscoe mentioned this issue Aug 25, 2022

Final fixes for TriBITS upgrade to modern CMake targets (#10614, #10774) #10930

Merged

bartlettroscoe added a commit to bartlettroscoe/Trilinos that referenced this issue Aug 25, 2022

Merge branch 'tribits_github_snapshot' into 10774-final-fixes (trilin…

e3e9012

…os#10774, TriBITSPub/TriBITS#522)

bartlettroscoe closed this as completed Aug 26, 2022

tasmith4 mentioned this issue Aug 30, 2022

TriBITS: Nalu-Wind Cuda builds won't configure #10954

Closed

bartlettroscoe reopened this Sep 6, 2022

bartlettroscoe closed this as completed Oct 6, 2022

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Address failures from merge of updated TriBITS (#10614, TriBITSPub/TriBITS#299) #10774

Address failures from merge of updated TriBITS (#10614, TriBITSPub/TriBITS#299) #10774

bartlettroscoe commented Jul 19, 2022 •

edited

Loading

ikalash commented Jul 19, 2022

bartlettroscoe commented Jul 19, 2022

bartlettroscoe commented Jul 20, 2022 •

edited

Loading

bartlettroscoe commented Jul 20, 2022

bartlettroscoe commented Jul 20, 2022

bartlettroscoe commented Jul 20, 2022

bartlettroscoe commented Jul 20, 2022

bartlettroscoe commented Jul 21, 2022 •

edited

Loading

bartlettroscoe commented Jul 21, 2022 •

edited

Loading

bartlettroscoe commented Jul 22, 2022

bartlettroscoe commented Jul 22, 2022

bartlettroscoe commented Aug 25, 2022

bartlettroscoe commented Aug 25, 2022

bartlettroscoe commented Aug 26, 2022

bartlettroscoe commented Sep 6, 2022

bartlettroscoe commented Oct 4, 2022

bartlettroscoe commented Oct 6, 2022

Address failures from merge of updated TriBITS (#10614, TriBITSPub/TriBITS#299) #10774

Address failures from merge of updated TriBITS (#10614, TriBITSPub/TriBITS#299) #10774

Comments

bartlettroscoe commented Jul 19, 2022 • edited Loading

Description

Tasks

Deferred Scope

ikalash commented Jul 19, 2022

bartlettroscoe commented Jul 19, 2022

bartlettroscoe commented Jul 20, 2022 • edited Loading

bartlettroscoe commented Jul 20, 2022

bartlettroscoe commented Jul 20, 2022

bartlettroscoe commented Jul 20, 2022

bartlettroscoe commented Jul 20, 2022

bartlettroscoe commented Jul 21, 2022 • edited Loading

1) Link errors for libcgns missing hdf5 symbols

bartlettroscoe commented Jul 21, 2022 • edited Loading

2) CUDA build failures due to duplicate CUDA library function definitions when creating libIntercept.so.13.5

bartlettroscoe commented Jul 22, 2022

3) Link errors missing '__dlopen'

bartlettroscoe commented Jul 22, 2022

4) Compile errors in Kokkos for coming from gtest.h

bartlettroscoe commented Aug 25, 2022

bartlettroscoe commented Aug 25, 2022

bartlettroscoe commented Aug 26, 2022

bartlettroscoe commented Sep 6, 2022

bartlettroscoe commented Oct 4, 2022

bartlettroscoe commented Oct 6, 2022

bartlettroscoe commented Jul 19, 2022 •

edited

Loading

bartlettroscoe commented Jul 20, 2022 •

edited

Loading

bartlettroscoe commented Jul 21, 2022 •

edited

Loading

bartlettroscoe commented Jul 21, 2022 •

edited

Loading