6 tests failed out of 243

[yurivict]

The following tests FAILED:
	164 - unit_test_aobasis (Failed)
	171 - unit_test_polarsegment (Failed)
	172 - unit_test_qmmolecule (Failed)
	180 - unit_test_ecpaobasis (Failed)
	185 - unit_test_hdf5 (Failed)
	201 - unit_test_segment (Failed)
Errors while running CTest

$ ./work/.build/xtp/src/tests/unit_test_aobasis
Running 3 test cases...
unknown location(0): fatal error: in "aobasis_test/Serializing": memory access violation at address: 0x4731322d3308: no mapping at fault address
/disk-samsung/freebsd-ports/science/votca/work/votca-2021.2/xtp/src/tests/test_aobasis.cc(97): last checkpoint: "Serializing" test entry

*** 1 failure is detected in the test module "aobasis_test"

$ ./work/.build/xtp/src/tests/unit_test_polarsegment
Running 5 test cases...
unknown location(0): fatal error: in "polarsegment_test/readwritehdf": memory access violation at address: 0x3167657308: no mapping at fault address
/disk-samsung/freebsd-ports/science/votca/work/votca-2021.2/xtp/src/tests/test_polarsegment.cc(140): last checkpoint: "readwritehdf" test entry

*** 1 failure is detected in the test module "polarsegment_test"

$ ./work/.build/xtp/src/tests/unit_test_qmmolecule
Running 3 test cases...
unknown location(0): fatal error: in "qmmolecule_test/readwritehdf": memory access violation at address: 0x3167657308: no mapping at fault address
/disk-samsung/freebsd-ports/science/votca/work/votca-2021.2/xtp/src/tests/test_qmmolecule.cc(74): last checkpoint: "readwritehdf" test entry

*** 1 failure is detected in the test module "qmmolecule_test"

$ ./work/.build/xtp/src/tests/unit_test_ecpaobasis
Running 1 test case...
unknown location(0): fatal error: in "ecpaobasis_test/Serializing": memory access violation at address: 0x803b3e0a4: object specific hardware error
/disk-samsung/freebsd-ports/science/votca/work/votca-2021.2/xtp/src/tests/test_ecpaobasis.cc(39): last checkpoint: "Serializing" test entry

*** 1 failure is detected in the test module "ecpaobasis_test"

$ ./work/.build/xtp/src/tests/unit_test_hdf5
Running 7 test cases...
unknown location(0): fatal error: in "test_hdf5/checkpoint_file_test": memory access violation at address: 0x00000030: no mapping at fault address
/disk-samsung/freebsd-ports/science/votca/work/votca-2021.2/xtp/src/tests/test_hdf5.cc(38): last checkpoint: "checkpoint_file_test" test entry

*** 1 failure is detected in the test module "test_hdf5"

$ ./work/.build/xtp/src/tests/unit_test_segment
Running 2 test cases...
unknown location(0): fatal error: in "segment_test/writehdf5": memory access violation at address: 0x656e6f00: no mapping at fault address
/disk-samsung/freebsd-ports/science/votca/work/votca-2021.2/xtp/src/tests/test_segment.cc(62): last checkpoint: "writehdf5" test entry

*** 1 failure is detected in the test module "segment_test"

Version: 2021.2
clang-11
OS: FreeBSD 13

[junghans]

Thanks for reporting!

@JensWehner @rubengerritsen these are all in xtp.

[JensWehner]

they are all hdf5 problems.

[JensWehner]

Could you post your whole CMakeOutput? Votca is not officially supported on BSD

[yurivict]

https://people.freebsd.org/~yuri/votca-CMakeOutput.log

[JensWehner]

@junghans how do we debug BSD?

[junghans]

No idea! @yurivict do you know?

@yurivict did v2021.1 package successful?

[rubengerritsen]

Apparently the problem is the same on Mac OS. We are looking into it at the moment, probably a problem with how hdf5 is setup/compiled

[yurivict]

No idea! yurivict do you know?

1. Install it into VirtualBox (https://download.freebsd.org/ftp/releases/amd64/amd64/ISO-IMAGES/13.0/FreeBSD-13.0-RELEASE-amd64-disc1.iso)
2. pkg install votca (this installs the old version of votca with all dependencies)
3. pkg delete votca
4. update the ports tree in /usr/ports through cloning of https://git.FreeBSD.org/ports.git
5. cd /usr/ports/science/votca && make test

Please let me know if you have any questions.

[JensWehner]

@rubengerritsen can you elaborate on the hdf5 problem?

[rubengerritsen]

No unfortunately no new info, Bjoern and I tried different versions of hdf5 with different environment variables and compile options, but nothing fixed the issue so far.

The only extra info we have is that writing to an orb file during a normal dft calculation results in a segfault, that is how Bjoern discovered that it doesn't work on the mac mini (ARM based) as well.

[baumeier]

So the initial segfault I saw before we were checking the unit tests, was at the end of the DFT calculation from dftgwbse. It does write the orb file (at least partially, not sure if complete), but do we close it properly? Not sure how HDF5 works.

[yurivict]

Rev. fbe18d9 doesn't have this problem any more.