Segmentation fault when using fix massflow/mesh?

Submitted by DanielDroop on Mon, 08/26/2013 - 16:26

Dear all,
i'm trying to use the fix massflow/mesh with liggghts version 2.3.6 and get a segmentation fault after some timesteps. My setup to test and reproduce the message is a simple cube without a bottom or lid. Some atoms will be placed inside, fall down and leave the box throuth the bottom and re-enter the box from above (since there is a periodic boundry in z-axis).

I attached the testcase with my liggghts script, the logfile and two mesh files. Please remove the ".txt" extension before using it.

Many thanks in advance!

Cheers,
Daniel

Error message follows:

Step CPU Atoms KinEng
1 0 12 0.0048254863
1000 0.10135698 12 0.0048885966
2000 0.19165707 12 0.0054246465
3000 0.280653 12 0.0062570253
4000 0.38291407 12 0.0071488455
5000 0.48798299 12 0.0081001071
6000 0.58222485 12 0.0091108101
7000 0.67270207 12 0.010180955
8000 0.76361799 12 0.01131054
9000 0.86405301 12 0.012499568
10000 0.96297288 12 0.013748036
11000 1.0634809 12 0.015055946
12000 1.1519849 12 0.016423298
13000 1.2431579 12 0.017850091
[VB2:02440] *** Process received signal ***
[VB2:02440] Signal: Segmentation fault (11)
[VB2:02440] Signal code: (128)
[VB2:02440] Failing at address: (nil)
[VB2:02440] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0xfcb0) [0x7f2f6bcd6cb0]
[VB2:02440] [ 1] /lib/x86_64-linux-gnu/libc.so.6(__fprintf_chk+0x58) [0x7f2f6ba11048]
[VB2:02440] [ 2] liggghts(_ZN9LAMMPS_NS15FixMassflowMesh14post_integrateEv+0x12ae) [0x5e4b6e]
[VB2:02440] [ 3] liggghts(_ZN9LAMMPS_NS6Modify14post_integrateEv+0x43) [0x6f2603]
[VB2:02440] [ 4] liggghts(_ZN9LAMMPS_NS6Verlet3runEi+0x31f) [0x805d4f]
[VB2:02440] [ 5] liggghts(_ZN9LAMMPS_NS3Run7commandEiPPc+0x288) [0x7bc068]
[VB2:02440] [ 6] liggghts(_ZN9LAMMPS_NS5Input15execute_commandEv+0x9cb) [0x6bb2fb]
[VB2:02440] [ 7] liggghts(_ZN9LAMMPS_NS5Input4fileEv+0x528) [0x6bd058]
[VB2:02440] [ 8] liggghts(main+0x49) [0x48b379]
[VB2:02440] [ 9] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xed) [0x7f2f6b92976d]
[VB2:02440] [10] liggghts() [0x48c6c1]
[VB2:02440] *** End of error message ***
--------------------------------------------------------------------------
mpirun noticed that process rank 0 with PID 2440 on node VB2 exited on signal 11 (Segmentation fault).

AttachmentSize
Plain text icon test01.liggghts.txt2.35 KB
Plain text icon log.liggghts_1.txt3.4 KB
Plain text icon Testbox_01.STL_.txt136.76 KB
Plain text icon Testlid_01.STL_.txt33.21 KB

cstoltz | Tue, 08/27/2013 - 12:29

Daniel,

I was able to reproduce your error on my machine. I then swapped out your file Testlid_01.STL in the fix_massflow/mesh with a simple 2-facet STL file stretching across the same bounds and it ran fine. I'm not sure why it doesn't like the STL file you used. Some sort of bug maybe?

Regards,
Chris

DanielDroop | Tue, 08/27/2013 - 16:21

Chris,

thanks for your help! I replaced my Testlid_01.STL as you described with a simple STL file and it ran.
But unfortunately only on my small tastcase! On my main project the same procedure didn't work. Even when I tried the same simple STL file in my main project, the segmentation fault (or a system freeze in case of multiple processes) occured. Strange. Let me double check the settings in my main project compared to the testcase.

Regards,
Daniel

ckloss's picture

ckloss | Fri, 08/30/2013 - 15:34

Hi all,

there is a pending bugfix for fix massflow/mesh for the next version. I guess it's related to what you describe here. I'll use your case to test the bugfix before releasing the next release (presumably in 1-2 weeks)

Cheers
Christoph

ckloss's picture

ckloss | Thu, 09/05/2013 - 10:16

Hi Daniel,

the error is fixed and the new version will be released in the next couple days

Christoph

DanielDroop | Thu, 09/26/2013 - 08:20

Hi Christoph,
sorry for the late reply. With the new version 2.3.7 no more crashes occured even with my more complex geometry. Thanks to all developers for doing a great job!

Daniel

ckloss's picture

ckloss | Thu, 10/03/2013 - 16:38

Hi Daniel,

thanks for the positive feed-back!

Best wishes
Christoph