VM crash on Linux after removing System.exit from JMESwingTest

Troubleshooting General Help
1

Hi there,



(my first post, new here…)



I'm currently working on embedding jMonkey in an Eclipse RCP application (succeeded). Currently, the easiest way I found was by using the SWT AWT bridge that comes with SWT.



I also used the specialized SWTRenderer Monkeyworld3d uses (succeeded) but that involves overriding a lot of classes in JME. I preferred to keep JME intact is much as possible (no changes were required…). It also gave a lot of trouble with the specialized classloader in Eclipse, so I had to do some extra changes in JME.



I used the JMESwingTest (in jmetest.util) example to setup an example scene. I noticed that my SWT Example application and my Eclipse RCP application crash when closing the window containing the JME window.



Going back to JMESwingTest I noticed the System.exit(0) call, when the windows is closed. Removing that call gives me the same VM crash.



I'm running Ubuntu Linux (x86) with Java VM: Java HotSpot™ Server VM (1.5.0_04-b05 mixed mode)



On Windows removing the System.exit from JMESwingTest doesn't produce a VM crash.



Does anyone know a solution to this? I think that if the problem is on the OpenGL side, it's probably in LWJGL, but I'm not sure…



Regards,

Erwin.














2

Oops, I forgot the report from the VM:



#

An unexpected error has been detected by HotSpot Virtual Machine:

#

#  SIGSEGV (0xb) at pc=0x8d6e3198, pid=15543, tid=2374196144

#

Java VM: Java HotSpot™ Server VM (1.5.0_04-b05 mixed mode)

Problematic frame:

C  0x8d6e3198

#

An error report file with more information is saved as hs_err_pid15543.log

#

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

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

#



I don't think attaching the whole hs_err_pid15543.log file will appreciated, but will be done on request…



Erwin.

3

Well not sure with Your custom renderer but if this was on LWJGL i'd say:

Check if the display is closed before quitting the app.

This is the bit from BaseGame:



/**

  * Quits the program abruptly using <code>System.exit</code>.

  * @see AbstractGame#quit()

  */

  protected void quit() {

  if (display != null)

  display.close();

    System.exit(0);

  }

4

Sorry for the confusion, but I'm not using a custom renderer.



The crash happens using a standard jmetest.util.JMESwingTest where I have removed System.exit(0). This to simulate the behavior in my program, where I want to close and open JME windows.



I tried the suggestion to call display.close(), so where JMESwingTest has System.exit(0), I put:

DisplaySystem.getDisplaySystem().close();



This unfortunately still crashes the VM :(.



Well, I'll keep digging. Instead of the LWJGL libs from JME CVS, I'll try to compile them myself. This way I can debug them…



Erwin.

5

Well, don't know if anybody is reading this, but here is my progress:



Checked out LWJGL from CVS and managed to compile both Java and C parts…

JMESwingTest still crashes with the System.exit(0) removed… :frowning:



I'm starting to consider to do what the crash report says: send it to Sun Microsystems.

6

I'm sorry that I haven't replied, but I just wouldn't be able to offer much. I really don't know what is causing the problem as I haven't seen it. I am reading this thread though. :slight_smile: So if you determine it's something we are doing, we'll fix it.

7

I've seen this a lot in the past when using RTT but not with Swing… I'll keep an eye out as well.

8

I managed to reproduce the error in a pure LWJGL and AWT implementation. So JME is innocent :slight_smile:



I'll head over to the LWJGL forums…



Erwin.