Bug #368

Streamline plot crashes engine with h5nimrod data

Added by Kathleen Biagas about 4 years ago. Updated about 4 years ago.

Status:Pending Start:08/27/2010
Priority:Normal Due date:
Assigned to:- % Done:

0%

Category:-
Target version:-
Likelihood:3 - Occasional OS:All
Severity:4 - Crash / Wrong Results Support Group:DOE/SciDAC
Found in Version:trunk

Description

Originally reported by Guoning Chen when I asked for clarification on issue #355.

With 2.0.1 on windows, a default Streamline plot of variable B from phi.h5nimrod yields no data.
Changing to using a plane source yields a plot.

With trunk, a default Streamline plot crashes the engine.

Performed a few tests on Linux:

With trunk version on Linux, a default Streamline plot yields no data, changing to using a plane source yields no data.
Changing the plane parameters appears to hang VisIt, but checking the engine logs, the engine SEGV'd.

I haven't tested with any other data formats.

Streamline_test_disassembly_celltree_builder_split.txt (4.4 KB) Kathleen Biagas, 10/05/2010 03:41 pm

Streamline_test_debugger_output.txt (8.1 KB) Kathleen Biagas, 10/05/2010 03:42 pm

Streamline_test_callstack.txt (4.7 KB) Kathleen Biagas, 10/05/2010 03:42 pm

Station_012000.silo (189.4 KB) Kathleen Biagas, 10/05/2010 03:42 pm

History

Updated by Eric Brugger about 4 years ago

  • Status changed from New to Pending
  • Support Group changed from Any to DOE/SciDAC

Updated by Kathleen Biagas about 4 years ago

This happens with silo data, too.

On Windows, I ran through the debugger, and Visual Studio complains about a Run-Time check error, stack corruption,
occurring in avtCellLocatorBIH, celltree_builder::split routine.

Updated by Kathleen Biagas about 4 years ago

I've attached the sample silo file, and Windows debugger output, callstack and disassembly. Hope something here helps.

Also available in: Atom PDF