Bugzilla – Full Text Bug Listing
|
Description
Brian So
Please specify the VM being used. Also, try "kill -3" (SIGQUIT) to obtain a stack dump when it is hung, and attach if available. Created attachment 9068 [details]
The attached file is a stack dump when Eclipse was hung
I still cannot find the exact way to reproduce the problem. However, I can
reproduce it by a combination of steps I mentioned in the report. The CPU usage
(via top) of the Java task after Eclipse hung was 99.5%
VM info:
java version "1.4.2_04"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.2_04-b05)
Java HotSpot(TM) Client VM (build 1.4.2_04-b05, mixed mode)
Created attachment 9328 [details]
Used "kill -3" to obtain a stack dump when Eclipse was hung again
This time Eclipse 3.08 (Solaris 9) was hung when I tried to close Eclipse
(File->Exit)- same VM. I was working on Eclipse for a whole day so I probably
cannot reproduce the problem. I hope the stack dump will provide some more info
to you. It will be easier to view it using wordPad. Do you want me to capture
the info every time Eclipse hangs?
Created attachment 9552 [details]
Used "kill -3" to obtain a stack dump when Eclipse was hung again on HPUX11.11
This time Eclipse 3.0M8 (HP-UX11.11) was hung when I tried to use Debug - same
VM. I had used Debug a fews time before it was hung. I changed the Debug
preference to prompt user to switch to a ssociate select perspective when
launching. The second time I pressed the debug button, it hung up on me.
Created attachment 9553 [details]
Used "kill -3" to obtain a stack dump when Eclipse was hung again on Solaris 9
Eclipse 3.0M8 (HP-UX11.11) was hung when I tried to use Debug - same
VM. I had used Debug a fews time before it was hung. I changed the Debug
preference to always switch to an associate select perspective when
launching. The second time I pressed the debug button, my runtime workbench
came up but the original workbench was hung. I could still close my runtime
workbench but not the original one.
Created attachment 9573 [details]
Used "kill -3" to obtain a stack dump when Eclipse was hung again on Solaris 9 when closing
This time Eclipse 3.08 (Solaris 9) was hung when I tried to close Eclipse
yesterday evening(File->Exit)- same VM. I came back in the moring and Eclipse
was still hung. I hope the stack dump will provide some more info to you.
Created attachment 9778 [details]
Used "kill -3" to obtain a stack dump when Eclipse was hung again on Solaris 9
Used "kill -3" to obtain a stack dump when Eclipse was hung again on Solaris 9
Eclipse 3.0M8 was hung when I tried to run debug on a Java application - same
VM. I was being prompted to see whether I would like to switch to Debug
perspective or not. I clicked on 'always' and Eclipse was hung.
These all appear to be hung (or infinitely looping) in SWT natives (OS.XtDestroyWidget, OS.XtManageChild). Created attachment 10213 [details]
Eclipse was hung when trying to open Java Browsing perspective on Solaris 9
Eclipse 3.0M8 (build id 200403261517)
This time I was performing stress testing. I imported 100 packages and each
package had approx. 100 classes. I then expanded all the package lists on both
the Navigation view and the Package Explorer view. When tried to open Java
Browsing perspective (Window->Open Perspective->Java Browsing), Eclipse was
hung again. It happened on both Solaris 9 and HPUX11.11.
I have only attached the dump for Solaris 9.
Created attachment 10357 [details]
Eclipse was hung when trying to exit workbench (HPUX11.11)
At first I was dragging and dropping files. I then performed some CVS testing.
I deleted a CVS project and clicked on 'X' button to close the workbench. That
was how I caused Eclipse to hang.
Created attachment 12772 [details]
Hang on HP-UX 11.23
Eclipse Hang occurs even in HP-UX 11.23. This seems to be something to do with
the code rather than a bug in the JVM
This bug is happening for all the latest versions of Eclipse viz., Eclipse 3.0M8, Eclipse 3.0 M9, Eclipse 3.0 RC1, Eclipse 3.0 RC2, Eclipse 3.0 RC3. As it hangs every now and then, development using Eclipse is becoming very difficult. The only way to restore is to kill the process and then restart Eclipse. In Display.java , there is a following snippet, // int xtContext = OS.XtDisplayToApplicationContext (xDisplay); // /* // * Bug in Xt. Under certain circumstances Xt waits // * forever looking for X events, ignoring alternate // * inputs. The fix is to never sleep forever. // */ I think the new code as a fix for this bug also has the same problem. Could this be the potential problem for this bug. Moreover, I added some System.out.printlns in the sleep() of Display.java. Whenever there is a freeze, I expected that there would be no output from these System.out.println. Well, to my suprise, it is only the UI which got frozen and there were outputs in the console from these System.out.printlns. For examples, As per my understanding Eclipse gets pinged every 1 minute (6000) in its idle state. Even during the freeze state, these pings were happening. I was able to get the output from my System.out.println. Created attachment 14321 [details]
debug configuration image
Created attachment 14322 [details]
Eclipse hang- thread dump
Hi,
I am using Eclipse 3.0 on Linux. I am also facing this eclipse hang problem. I
started debugging eclipse code to find out the root cause. I am using 'remote
java application debugging' feature available in eclipse.
I launched eclipse on my Linux box and I used eclipse on Windows as my remote
debugger
The hang very much reproducible with the following way.
1) Start Eclipse 3.0 on Linux with the following arguments
./eclipse -vmargs -Xnoagent -Xdebug -Djava.compiler=NONE -
Xrunjdwp:transport=dt_socket,address=8888,suspend=n,server=y
2) In the windows box, import all the Eclipse 3.0 plug-in sources (taken from
3.0 srcincluded.zip file)
3) Create new remote java application debug configuration shown in the
attached 'debug-configuration' image file.
4)Set a method break point of update() of org.eclipse.swt.widgets.Display(line
no 3422).After that , start debugger on Windows box.
The debugger shows all the thread flow, whenever any operation on eclipse on
Linux is performed.
I faced the eclipse hang on Linux, when I press the window->open perspective-
>other....
The same hang appears 8 out of 10 times for me. I have also attached my kill -
3 trace dump.
Hello *,
I've probably the same problem with Eclipse 3.1M1 under Linux (GTK2, SuSE 9.1,
KDE 3.2.1, X 4.3.99.902, Kernel 2.6.5-7.111.19, java 1.5.0-b64) when debugging.
What is really interesting, is that Eclipse doesn't hang forever but comes back
after about 10 minutes. Thus, to step from one line to another takes each time
10 minutes and hence makes debugging impossible.
I tried a "kill -3 {processID}", but unfortunately nothing happened. So, I
cannot provide a stack trace. I'll try it again. Should be no problem because
this happens now every time I try to debug an EJB in JBoss remotely.
The problem occured out of the blue without any changes in my code or in the
configuration. Last week everything worked fine. Really strange.
Best regards, Marco :-)
Hello again, after downloading Eclipse 3.1M4, the new lomboz and agonizing over Bug 81679 (https://bugs.eclipse.org/bugs/show_bug.cgi?id=81679) and some lomboz related problems, I finally was able to try out debugging again and it works! Seems, the problem has been solved - at least I couldn't reproduce it so far and I hope it doesn't reoccur later. Best regards, Marco :-) Moving report to triage, see http://www.eclipse.org/swt/triage.php for more info about swt triage. Solaris is no longer a supported platform. |