Process crashed -- Simulation process exited with code 0

[rfranke@…]

The model listed below uses the same variables as outputs and states (like some models from the MSL, see Modelica.Blocks.Continuous.Integrator).

Opened and simulated with OMEdit, the compilation raises a lot of warnings and the simulation crashes.

The observation has been made with the current nightly build 1.9.1+dev (r21649).

within ;
model DIC "Double Integrator Continuous-time"
  parameter Real p = 1 "gain for input";
  parameter Real y1_start = 1 "start value for first state";
  parameter Real y2_start = 0 "start value for second state";
  input Real u(start = -2);
  output Real y1, y2;
initial equation
  y1 = y1_start;
  y2 = y2_start;
equation
  der(y1) = p*u;
  der(y2) = y1;
end DIC;

[Lennart Ochel]

This seems to be an issue with the garbage collector. Adrian, can you have a look?

==15946== Conditional jump or move depends on uninitialised value(s)
==15946==    at 0x6339B0F: GC_push_all_eager (in trunk/build/lib/omc/libgc.so.1.0.3)
==15946==    by 0x6343A3D: GC_with_callee_saves_pushed (in trunk/build/lib/omc/libgc.so.1.0.3)
==15946==    by 0x633B16D: GC_push_roots (in trunk/build/lib/omc/libgc.so.1.0.3)
==15946==    by 0x633A9A8: GC_mark_some (in trunk/build/lib/omc/libgc.so.1.0.3)
==15946==    by 0x633232C: GC_stopped_mark (in trunk/build/lib/omc/libgc.so.1.0.3)
==15946==    by 0x63325DA: GC_try_to_collect_inner (in trunk/build/lib/omc/libgc.so.1.0.3)
==15946==    by 0x633D24D: GC_init (in trunk/build/lib/omc/libgc.so.1.0.3)
==15946==    by 0x4E5A0DF: mmc_init (mmc_gc.h:79)
==15946==    by 0x660D3FF: pthread_once (pthread_once.S:104)
==15946==    by 0x405871: main (in DIC)

[Martin Sjölund]

No. If you run valgrind you need to suppress GC errors, like so:

[martin@mega tmp]$ cat ~/bin/valgrind-boehm && valgrind-boehm ./DIC 
#!/bin/bash
valgrind --suppressions="$HOME/.valgrind/boehm-gc.suppressions" $*
==1956== Memcheck, a memory error detector
==1956== Copyright (C) 2002-2013, and GNU GPL'd, by Julian Seward et al.
==1956== Using Valgrind-3.10.0.SVN and LibVEX; rerun with -h for copyright info
==1956== Command: ./DIC
==1956== 
==1956== Invalid write of size 4
==1956==    at 0x4050F1: DIC_initialAnalyticJacobianA (in /home/martin/tmp/DIC)
==1956==    by 0x4E93FF4: dassl_initial (dassl.c:235)
==1956==    by 0x4EA1238: initializeSolverData (solver_main.c:184)
==1956==    by 0x4EA22D1: solver_main (solver_main.c:570)
==1956==    by 0x4ED2BA2: callSolver (simulation_runtime.cpp:695)
==1956==    by 0x4ED143D: startNonInteractiveSimulation (simulation_runtime.cpp:548)
==1956==    by 0x4ED3FB8: _main_SimulationRuntime (simulation_runtime.cpp:914)
==1956==    by 0x403C4C: main (in /home/martin/tmp/DIC)
...

[Lennart Ochel]

Replying to sjoelund.se:

No. If you run valgrind you need to suppress GC errors, like so:

Ahh – good to know.

[Willi Braun]

fixed in r21660.