jzy3d seg faulting on window close

Hi All,

So, I have been in direct communication with Martin who developed jzy3D to help resolve some issues.  He has been very helpful pointing me to the various release and even went as far as fixing some code and posting an update.

The chart opens fine and works, but upon closing the window there is a segmentation fault detected by the JVM.

I am using jzy3D version 0.9 with JOGL libraries that Martin has kindly posted here:

http://jzy3d.org/releases/0.9alpha2/

This is on a 64-bit Linux box.  Here is the error, and thanks for any help:

#
# A fatal error has been detected by the Java Runtime Environment:
#
#  SIGSEGV (0xb) at pc=0x0000003547a7e120, pid=17785, tid=1099270464
#
# JRE version: 6.0_15-b03
# Java VM: Java HotSpot(TM) 64-Bit Server VM (14.1-b02 mixed mode linux-amd64 )
# Problematic frame:
# C  [libGL.so.1+0x7e120]
#
# An error report file with more information is saved as:
# /net/ultra-trix/export/home/ultra-trix/jason/release_test/64-bga-fuck/host/JEditor/hs_err_pid17785.log
#
# If you would like to submit a bug report, please visit:
#   http://java.sun.com/webapps/bugreport/crash.jsp
#
./runBGA: line 44: 17785 Aborted                 (core dumped) /usr/java/jre1.6.0/bin/java -server -Xmx2048M -DPCB=$PCB -DLANG=en_US -classpath .:$PCB/app-defaults/JEditor:$pcb/lib/JEditor.jar:$pcb/lib/JEditorResources.jar:$pcb/lib/JEditormain.jar:$pcb/lib/jEditorResources.jar:$pcb/lib/piccolo-1.2.jar:$pcb/lib/piccolox-1.2.jar:$pcb/lib/jfreechart-0.9.21.jar:$pcb/lib/jcommon-0.9.6.jar:$pcb/lib/jhotdraw.jar:$pcb/lib/14_os_jpedal.jar:$pcb/lib/bcprov-jdk14-119.jar:$pcb/lib/jai_codec.jar:$pcb/lib/jai_core.jar:$pcb/lib/jdom.jar:$pcb/lib/log4j-1.2.9.jar:$pcb/lib/jworkbook-0.3.0.jar:$pcb/lib/poi-2.0-pre1-20030517.jar:$pcb/lib/forms-1.0.5.jar:$pcb/lib/l2fprod-common-all.jar:$pcb/lib/dialgauge.jar:$pcb/lib/mm.jar:$pcb/lib/mydoggy-api-1.4.2.jar:$pcb/lib/mydoggy-plaf-1.4.2.jar:$pcb/lib/mydoggy-res-1.4.2.jar:$pcb/lib/TableLayout-20050920.jar:$pcb/lib/Jama-1.0.2.jar:$pcb/lib/autonetworker.jar:$pcb/lib/sqlitejdbc-v056.jar:$pcb/lib/mysql-connector-java-5.1.15-bin.jar:$pcb/lib/swingfx.jar:$pcb/lib/nativewindow.all.jar:$pcb/lib/jogl.util.jar:$pcb/lib/jogl.all.jar:$pcb/lib/gluegen-rt.jar:$pcb/lib/swt.jar:$pcb/lib/swt-debug.jar:$pcb/lib/org.jzy3d-0.9.jar:$pcb/lib/org.convexhull.jar:$pcb/lib/opencsv-2.1.jar:$pcb/lib/newt.all.jar com.visionpro.bga.gui.BGAOverview

Hi

Can you provide a very small test case? Do you reproduce this crash with the demos of jzy3d?

Hi,

Yes this is using a very slightly modified version of BuildSurfaceDemo.java.  All I have done is added some other data points.  Everything renders perfect and behaves perfectly.  It's only when I close the frame does it dump core.

Martin suggested I give more info about the architecture.

It is 8 core .. each core is (according to /proc/cpuinfo):

Intel(R) Xeon(R) CPU           E5520  @ 2.27GHz

according to /proc/version :

Linux version 2.6.18-128.el5 (mockbuild@builder10.centos.org) (gcc version 4.1.2 20080704 (Red Hat 4.1.2-44)) #1 SMP Wed Jan 21 10:41:1
4 EST 2009

Let me know of any other helpful information.

--Jason

Hi all,

Jason also told me something that may be usefull for understanding the bug:

jazzmongrel wrote

One thing, I am pushing these pixels using X server.  So I have gone
to a terminal, typed "xhost +" to allow access, then ssh'd into a
remote machine, and launched it from there.  All other grpahical
programs that I launch this way have worked (e.g. firefox).

There is a reason I have to do it this way, I only have remote access
to ONE machine that is remote that I can VNC into, then from there I
must open a terminal and ssh into any other machine that I want, but
must use the X server to get graphics from the "other" machine.

Could this possibly be the problem, it can;t figure out the display
device or something.  I am not that knowledgable with X or these
graphical details.

Let me add that I have two other java frames that pop up in addition
to the chart, and they have no problem showing the their contents
through the x server.

Thanks, for helping him!
Martin

Can you reproduce this bug without using any remote access?

OK looks like it may have something to do with accessing it remotely afterall.

I had somebody at my work (which is in another state) go to the physical box that I was ssh-ing into, run my code, and no seg fault upon window close!

I had them log in as myself and try it to rule out environment settings, still no seg fault.

I then had them ssh into the machine as themself and as me, still everything is okay.

So the only missing piece is that I am first using VNC to access the first machine, then ssh-ing into the target machine to run.

I cannot understand what VNC would have anything to do with this, because if you think about it the seg fault is happening locally on whatever machine that is running the code, then the pixels are just being pushed through the layers to me (through X server, then through VNC).   But the origin of the error still resides only on the target computer, because that's where the libraries and code live.  

I'll have to think about it more but it seems to work good enough for now.

Thanks for all your help guys, especially Martin.

Cheers
Jason