Summary: The sst-core-test test_StatisticsBasic is failing on Ubuntu 24.04 when sst is compiled with address sanitizer enabled. However, the test passes when running standalone sst which implies there may be an issue, unrelated to the address sanitizer build, with the test harness.
To reproduce built sst with
CXXFLAGS="-g3 -O0 -fsanitize=address -fno-omit-frame-pointer" ../configure --enable-debug --prefix=$SST_CORE_HOME
Then run:
sst-test-core -e test_StatisticsBasic -d -z
Result:
Traceback (most recent call last):
File "/home/ken/work/sst-core-tcl/tests/testsuite_default_StatisticsComponent.py", line 42, in test_StatisticsBasic
self.Statistics_test_template("basic")
File "/home/ken/work/sst-core-tcl/tests/testsuite_default_StatisticsComponent.py", line 97, in Statistics_test_template
subprocess.run(["h5diff",ref_group_stat_file_h5,out_group_stat_file_h5],check=True)
File "/usr/lib/python3.12/subprocess.py", line 548, in run
with Popen(*popenargs, **kwargs) as process:
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/lib/python3.12/subprocess.py", line 1026, in __init__
self._execute_child(args, executable, preexec_fn, close_fds,
File "/usr/lib/python3.12/subprocess.py", line 1955, in _execute_child
raise child_exception_type(errno_num, err_msg, err_filename)
FileNotFoundError: [Errno 2] No such file or directory: 'h5diff'
If I compare the sst_test_outputs with the normal (passing) run I see there is a lingering *.h5 file in the failing run and a missing h5diff. The timestamps indicate that the test time is much less for the failing case which seems to indicate a silent failure.
diff -r sst_test_outputs ../build.good/sst_test_outputs/
Only in sst_test_outputs/run_data: test_StatisticsComponent_basic_group_stats.h5
diff -r sst_test_outputs/run_data/test_StatisticsComponent_basic.out ../build.good/sst_test_outputs/run_data/test_StatisticsComponent_basic.out
61,66d60
< Using Marsaglia Random Number Generator with seeds m_z = 1078, m_w = 1472
< REGISTER CLOCK #1 at 1 ns
< Using Marsaglia Random Number Generator with seeds m_z = 1079, m_w = 1473
< REGISTER CLOCK #1 at 1 ns
< Using Marsaglia Random Number Generator with seeds m_z = 1080, m_w = 1474
< REGISTER CLOCK #1 at 1 ns
diff -r sst_test_outputs/xml_data/testsuite_default_StatisticsComponent.xml ../build.good/sst_test_outputs/xml_data/testsuite_default_StatisticsComponent.xml
2,6c2,4
< <testsuites disabled="0" errors="1" failures="0" tests="1" time="2.151984453201294" name="SST TESTING FRAMEWORKS">
< <testsuite disabled="0" errors="1" failures="0" name="testsuite_default_StatisticsComponent" skipped="0" tests="1" time="2.151984453201294">
< <testcase name="test_StatisticsBasic" time="2.151984" timestamp="2026_0212_19:22:12.697031 utc" classname="testcase_StatisticComponent">
< <error type="error" message="FileNotFoundError: [Errno 2] No such file or directory: 'h5diff' "/>
< </testcase>
---
> <testsuites disabled="0" errors="0" failures="0" tests="1" time="0.5170490741729736" name="SST TESTING FRAMEWORKS">
> <testsuite disabled="0" errors="0" failures="0" name="testsuite_default_StatisticsComponent" skipped="0" tests="1" time="0.5170490741729736">
> <testcase name="test_StatisticsBasic" time="0.517049" timestamp="2026_0212_16:13:17.152151 utc" classname="testcase_StatisticComponent"/>
Before digging further into the debug, perhaps someone can suggest where else I might find more debug information.
Summary: The sst-core-test
test_StatisticsBasicis failing on Ubuntu 24.04 when sst is compiled with address sanitizer enabled. However, the test passes when running standalone sst which implies there may be an issue, unrelated to the address sanitizer build, with the test harness.To reproduce built sst with
CXXFLAGS="-g3 -O0 -fsanitize=address -fno-omit-frame-pointer" ../configure --enable-debug --prefix=$SST_CORE_HOMEThen run:
sst-test-core -e test_StatisticsBasic -d -zResult:
If I compare the sst_test_outputs with the normal (passing) run I see there is a lingering
*.h5file in the failing run and a missingh5diff. The timestamps indicate that the test time is much less for the failing case which seems to indicate a silent failure.Before digging further into the debug, perhaps someone can suggest where else I might find more debug information.