TestParcours Access Violation

Troubleshooting physics
1

When i try to run TestParcour i always get an access violation error.

I don't know how to get more information about the error, i guess i would need a debugable ODE library?


An unexpected error has been detected by HotSpot Virtual Machine:

#

#  EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x038898e2, pid=2212, tid=2228

#

Java VM: Java HotSpot™ Client VM (1.5.0_10-b03 mixed mode, sharing)

Problematic frame:

C  [odejava.dll+0x498e2]

#

An error report file with more information is saved as hs_err_pid2212.log

#

If you would like to submit a bug report, please visit:

#  http://java.sun.com/webapps/bugreport/crash.jsp

#



[error occurred during error reporting, step 270, id 0xc0000005]

2
Core-Dump said:

# An error report file with more information is saved as hs_err_pid2212.log


Stack: [0x00030000,0x00070000),  sp=0x0006e8d0,  free space=250k
Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code)
C  [odejava.dll+0x498e2]
C  [odejava.dll+0x4992d]
C  [odejava.dll+0x49979]

Java frames: (J=compiled Java code, j=interpreted, Vv=VM code)
j  org.odejava.ode.OdeJNI.spaceCollide(J)I+0
j  org.odejava.ode.Ode.spaceCollide(Lorg/odejava/ode/SWIGTYPE_p_dSpaceID;)I+4
j  org.odejava.collision.JavaCollision.collide(Lorg/odejava/Space;)V+33
j  com.jmex.physics.impl.ode.OdePhysicsSpace.computeTimeStep()V+8
j  com.jmex.physics.impl.ode.OdePhysicsSpace.update(F)V+115
j  com.jmex.physics.util.SimplePhysicsGame.update(F)V+56
j  com.jme.app.BaseGame.start()V+40
j  com.jmetest.physics.TestParcours.main([Ljava/lang/String;)V+19
v  ~StubRoutines::call_stub
3

ok it stops crashing when not generating TriangleAccurate Physics for the static node.

staticNode.generatePhysicsGeometry( true );


to

staticNode.generatePhysicsGeometry( false );



Now it doesn't crash, but it doesn't work as intended anymore :)

This solution workaround was inspired by ...
http://www.jmonkeyengine.com/jmeforum/index.php?topic=6610.0;topicseen
4

perick already explained in that thread why it cannot be a solution to switch off triangle accuracy.



The mesh seems to confuse ODE (e.g. because there are anomalies in it). Though I have no idea why it works here and does not at your machine :expressionless: - if I get the chance I'll test it on some more machines and see if there are any troubles with it (next year ;)) - what OS are you testing it on?

5

hehe i didn't mean it as a solution, just wanted to point to the other thread :slight_smile:

i have a WinXp PC (amd, nvidia) and Xp Notebook (intel,ati) and it crashes on both.

6

I didn't want to mean that generating physics without triangle accuracy was a solution either  ://.



Just pointed out that sometimes triangle accuracy is making ODE crash showing that Exception.



Just in case it helps, and this is me guessing, I suspect this can be happening when some vertex are too close to each other to the point that transforming them could result in a different topology. Sort of (if you allow me to put it this way) ode "z-fighting". I say this because I am only observing that error in very small models, for which I can observe graphical z-fighting too.



I don't know enough about jME / jMEPhyscis / ODE as to know if that could really be a problem. Maybe someone can confirm or dismiss this :?.