segfault in testOutput

[alexkaapa]

I tried to install CRPropa locally on our cluster. Up to ccmake .., everything worked fine. Afterwards, when doing "make test", the following error occured:

Running tests...
Test project /lustre/kaeaepae/Software/CRPropa/build
Start 1: testCore
1/14 Test #1: testCore ......................... Passed 0.46 sec
Start 2: testVector3
2/14 Test #2: testVector3 ...................... Passed 0.01 sec
Start 3: testModuleList
3/14 Test #3: testModuleList ................... Passed 0.38 sec
Start 4: testMagneticField
4/14 Test #4: testMagneticField ................ Passed 0.65 sec
Start 5: testAdvectionField
5/14 Test #5: testAdvectionField ............... Passed 0.36 sec
Start 6: testDINT
6/14 Test #6: testDINT ......................... Passed 0.42 sec
Start 7: testPropagation
7/14 Test #7: testPropagation .................. Passed 0.37 sec
Start 8: testBreakCondition
8/14 Test #8: testBreakCondition ............... Passed 0.37 sec
Start 9: testInteraction
9/14 Test #9: testInteraction .................. Passed 22.12 sec
Start 10: testSource
10/14 Test #10: testSource ....................... Passed 0.62 sec
Start 11: testOutput
11/14 Test #11: testOutput .......................***Exception: SegFault 0.39 sec
Start 12: testFunctionalGroups
12/14 Test #12: testFunctionalGroups ............. Passed 0.37 sec
Start 13: testAdiabaticCooling
13/14 Test #13: testAdiabaticCooling ............. Passed 0.37 sec
Start 14: testGalacticMagneticLens
14/14 Test #14: testGalacticMagneticLens ......... Passed 0.50 sec

93% tests passed, 1 tests failed out of 14

Total Test time (real) = 27.45 sec

The following tests FAILED:
11 - testOutput (SEGFAULT)
Errors while running CTest
make: *** [test] Error 8

Does anyone have a hunch what might be wrong with "testOutput"?

[adundovi]

Run it individually like ./testOutput in the build directory and c/p here the output.

[alexkaapa]

I get the following output:

Running main() from gtest_main.cc
[==========] Running 13 tests from 3 test cases.
[----------] Global test environment set-up.
[----------] 1 test from Output
[ RUN ] Output.size
[ OK ] Output.size (1 ms)
[----------] 1 test from Output (1 ms total)

[----------] 7 tests from TextOutput
[ RUN ] TextOutput.printHeader_Trajectory1D
[ OK ] TextOutput.printHeader_Trajectory1D (0 ms)
[ RUN ] TextOutput.printHeader_Event1D
[ OK ] TextOutput.printHeader_Event1D (0 ms)
[ RUN ] TextOutput.printHeader_Trajectory3D
[ OK ] TextOutput.printHeader_Trajectory3D (0 ms)
[ RUN ] TextOutput.printHeader_Event3D
[ OK ] TextOutput.printHeader_Event3D (1 ms)
[ RUN ] TextOutput.printHeader_Custom
[ OK ] TextOutput.printHeader_Custom (0 ms)
[ RUN ] TextOutput.printProperty
[ OK ] TextOutput.printProperty (0 ms)
[ RUN ] TextOutput.printHeader_Version
[ OK ] TextOutput.printHeader_Version (0 ms)
[----------] 7 tests from TextOutput (1 ms total)

[----------] 5 tests from ParticleCollector
[ RUN ] ParticleCollector.size
[ OK ] ParticleCollector.size (0 ms)
[ RUN ] ParticleCollector.fetchItem
[ OK ] ParticleCollector.fetchItem (34 ms)
[ RUN ] ParticleCollector.reprocess
[ OK ] ParticleCollector.reprocess (0 ms)
[ RUN ] ParticleCollector.dumpload
[ OK ] ParticleCollector.dumpload (3 ms)
[ RUN ] ParticleCollector.getTrajectory
Segmentation fault

[adundovi]

Tnx Alex. It seems that there is a memory management issue in the getTrajectory feature. In principle, it will not affect running CRPropa if you don't use ParticleCollector.getTrajectory().

But you can help me debug it if you want...
First, you need to recompile CRPropa with the debugging symbols enabled. In the cmake part add additional flag:

CMAKE_PREFIX_PATH=$CRPROPA_DIR cmake -DCMAKE_INSTALL_PREFIX=$CRPROPA_DIR -DCMAKE_BUILD_TYPE:STRING=Debug ..

(instead of CMAKE_PREFIX_PATH=$CRPROPA_DIR cmake -DCMAKE_INSTALL_PREFIX=$CRPROPA_DIR .. and don't forget to define CRPROPA_DIR)

After you recompile it, run GNU debugger:

gdb testOutput 

(enter start, continue and paste here the output when it segfaults).

[alexkaapa]

I copied the entire ouput:

Running main() from gtest_main.cc
[==========] Running 13 tests from 3 test cases.
[----------] Global test environment set-up.
[----------] 1 test from Output
[ RUN ] Output.size
[ OK ] Output.size (0 ms)
[----------] 1 test from Output (0 ms total)

[----------] 7 tests from TextOutput
[ RUN ] TextOutput.printHeader_Trajectory1D
[ OK ] TextOutput.printHeader_Trajectory1D (0 ms)
[ RUN ] TextOutput.printHeader_Event1D
[ OK ] TextOutput.printHeader_Event1D (1 ms)
[ RUN ] TextOutput.printHeader_Trajectory3D
[ OK ] TextOutput.printHeader_Trajectory3D (0 ms)
[ RUN ] TextOutput.printHeader_Event3D
[ OK ] TextOutput.printHeader_Event3D (0 ms)
[ RUN ] TextOutput.printHeader_Custom
[ OK ] TextOutput.printHeader_Custom (0 ms)
[ RUN ] TextOutput.printProperty
[ OK ] TextOutput.printProperty (0 ms)
[ RUN ] TextOutput.printHeader_Version
[ OK ] TextOutput.printHeader_Version (0 ms)
[----------] 7 tests from TextOutput (1 ms total)

[----------] 5 tests from ParticleCollector
[ RUN ] ParticleCollector.size
[ OK ] ParticleCollector.size (0 ms)
[ RUN ] ParticleCollector.fetchItem
[ OK ] ParticleCollector.fetchItem (4 ms)
[ RUN ] ParticleCollector.reprocess
[ OK ] ParticleCollector.reprocess (0 ms)
[ RUN ] ParticleCollector.dumpload
[ OK ] ParticleCollector.dumpload (3 ms)
[ RUN ] ParticleCollector.getTrajectory

Program received signal SIGBUS, Bus error.
0x000000000041a7a1 in crpropa::ParticleCollector_getTrajectory_Test::TestBody() ()

[adundovi]

Can you give details on what hardware and OS you run it? I cannot reproduce the error and I don't see which pointer causes the segfault from this output.

[TobiasWinchen]

I assume that this problem has been fixed by now as there has not been an update for three month. Please reopen if this is not the case.