Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.

Bug 179494

Summary: Eclipse completely locks up due to an XML file in task context
Product: z_Archived Reporter: Jörg Thönnes <jtk499>
Component: MylynAssignee: Mik Kersten <mik.kersten>
Status: RESOLVED FIXED QA Contact:
Severity: critical    
Priority: P3    
Version: 2.0 M1   
Target Milestone: ---   
Hardware: PC   
OS: Linux   
Whiteboard:
Attachments:
Description Flags
First lockup stack trace
none
Second lockup stack trace
none
Console output showing event loop exceptions
none
Console output showing IndexOutOfBounds
none
XML possibly causing the issue
none
jstack trace from completely frozen Eclipse none

Description Jörg Thönnes CLA 2007-03-27 08:05:41 EDT
I saw this error related to some major performance impacts using my Eclipse 3.2.2 + Mylar:

- Activated task context
- Package Explorer + Outline View on XML file
- Change to another branch

This takes end-less. The event before is an unhandled event loop exception.

Are there any performance issues related to switching to another CVS branch or using an XML Outline view?

-- Error Log --
Date: Tue Mar 27 11:55:58 CEST 2007
Message: Widget is disposed
Severity: Error
Plugin ID: org.eclipse.ui
Stack Trace:
org.eclipse.swt.SWTException: Widget is disposed
at org.eclipse.swt.SWT.error(SWT.java:3374)
at org.eclipse.swt.SWT.error(SWT.java:3297)
at org.eclipse.swt.SWT.error(SWT.java:3268)
at org.eclipse.swt.widgets.Widget.error(Widget.java:434)
at org.eclipse.swt.widgets.Widget.checkWidget(Widget.java:372)
at org.eclipse.swt.widgets.Composite.getClientArea(Composite.java:501)
at org.eclipse.mylar.internal.tasks.ui.views.WorkweekProgressBar.scale(WorkweekProgressBar.java:111)
at org.eclipse.mylar.internal.tasks.ui.views.WorkweekProgressBar.reset(WorkweekProgressBar.java:90)
at org.eclipse.mylar.internal.tasks.ui.views.TaskListFilteredTree$3.run(TaskListFilteredTree.java:155)
at org.eclipse.swt.widgets.RunnableLock.run(RunnableLock.java:35)
at org.eclipse.swt.widgets.Synchronizer.runAsyncMessages(Synchronizer.java:123)
at org.eclipse.swt.widgets.Display.runAsyncMessages(Display.java:3157)
at org.eclipse.swt.widgets.Display.readAndDispatch(Display.java:2859)
at org.eclipse.ui.internal.Workbench.runEventLoop(Workbench.java:1930)
at org.eclipse.ui.internal.Workbench.runUI(Workbench.java:1894)
at org.eclipse.ui.internal.Workbench.createAndRunWorkbench(Workbench.java:422)
at org.eclipse.ui.PlatformUI.createAndRunWorkbench(PlatformUI.java:149)
at org.eclipse.ui.internal.ide.IDEApplication.run(IDEApplication.java:95)
at org.eclipse.core.internal.runtime.PlatformActivator$1.run(PlatformActivator.java:78)
at org.eclipse.core.runtime.internal.adaptor.EclipseAppLauncher.runApplication(EclipseAppLauncher.java:92)
at org.eclipse.core.runtime.internal.adaptor.EclipseAppLauncher.start(EclipseAppLauncher.java:68)
at org.eclipse.core.runtime.adaptor.EclipseStarter.run(EclipseStarter.java:400)
at org.eclipse.core.runtime.adaptor.EclipseStarter.run(EclipseStarter.java:177)
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:39)
at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:25)
at java.lang.reflect.Method.invoke(Method.java:585)
at org.eclipse.core.launcher.Main.invokeFramework(Main.java:336)
at org.eclipse.core.launcher.Main.basicRun(Main.java:280)
at org.eclipse.core.launcher.Main.run(Main.java:977)
at org.eclipse.core.launcher.Main.main(Main.java:952)
Comment 1 Jörg Thönnes CLA 2007-03-28 03:40:24 EDT
This took me half a day: On every eclipse start with the culprit task activated or activating the task sometimes locks up the UI completely.
Doing several thread dumps with jstack showed no difference, so the eclipse seemed to be really locked up. I suspect an XML file being
part of the task context (package explorer and outline view open) as the culprit.

Today I did a number of experiments with a clean workspace and the XML file alone, but it worked fine. Actually, I do not have more time to look at this, 
but it is really, really a show stopper. Therefore, I try to rise priority.

I would really appreciate if the Mylar team could look at this. Possibly, its an WST or Eclipse core issue. So please feel free to forward this bug.
My knowledge of internals is too limited.
Comment 2 Jörg Thönnes CLA 2007-03-28 03:45:00 EDT
Created attachment 62182 [details]
First lockup stack trace

Doing several jmap <pid> calls always showed this stack trace.
No differences --> lockup.
Comment 3 Jörg Thönnes CLA 2007-03-28 03:46:09 EDT
Created attachment 62183 [details]
Second lockup stack trace

Doing several jmap <pid> calls always showed this stack trace.
No differences --> lockup.

Second example, maybe the pattern is different (or the same).
Comment 4 Jörg Thönnes CLA 2007-03-28 03:48:15 EDT
Created attachment 62184 [details]
Console output showing event loop exceptions

Some the lockup breaks and an exception log appears at stdout.

This is an exception log showing an event loop.
Comment 5 Jörg Thönnes CLA 2007-03-28 03:51:10 EDT
Created attachment 62185 [details]
Console output showing IndexOutOfBounds

This console output shows an IndexOutOfBounds error.
If I remember correctly, the lockup broke here.
Comment 6 Jörg Thönnes CLA 2007-03-28 03:53:15 EDT
Finally, I cleaned up by deleting the context of the culprit task and checked whether none of the Mylar XML contains
references to the "poisoned" XML file. But not sure whether XML file is the culprit.

Do you need my lockup Mylar context, too? It may contain proprietary infos, so would like to clean up before.
Comment 7 Jörg Thönnes CLA 2007-03-28 03:56:02 EDT
Created attachment 62186 [details]
XML possibly causing the issue

This XML file possibly caused this issue. It switched to another CVS branch on it with Package Explorer and Outline View enabled.

But I could not reproduce using this file alone. So I may be wrong and a more complex interaction is the reason.
Comment 8 Mik Kersten CLA 2007-03-28 19:51:02 EDT
I fixed this last week on Eclipse 3.3 and still need to merge it back to Eclipse 3.2.  A 3.2 dev build will be available this evening from the 3.2 update site: http://www.eclipse.org/mylar/dl.php
Comment 9 Mik Kersten CLA 2007-03-28 19:51:24 EDT
If you still see this bug with the dev build please reopen.
Comment 10 Jörg Thönnes CLA 2007-03-29 03:31:10 EDT
 (In reply to comment #8)
> I fixed this last week on Eclipse 3.3 and still need to merge it back to Eclipse
> 3.2.  A 3.2 dev build will be available this evening from the 3.2 update site:
> http://www.eclipse.org/mylar/dl.php

Thank you very much, Mik. This actually led me to the point were I considered stop using Mylar for the moment.
This would be very sad, since I love to use it.

Just curious: Could you provide some background on the bug?

 (In reply to comment #9)
> If you still see this bug with the dev build please reopen.

Will do.
Comment 11 Mik Kersten CLA 2007-03-29 10:35:51 EDT
What was happening is that a refresh job for the progress bar on the Task List was running whether or not the workbench was running.  

Please note that a 3.2 build is not available yet, but will be available later today.
Comment 12 Jörg Thönnes CLA 2007-03-29 15:17:05 EDT
 (In reply to comment #11)
> What was happening is that a refresh job for the progress bar on the Task List
> was running whether or not the workbench was running.

I will import the Mylar export with the culprit task context and see what happens...

> Please note that a 3.2 build is not available yet, but will be available later
> today.

Hmmm, are you talking about 

(a) 2.0M2 on http://download.eclipse.org/technology/mylar/update-site/e3.2

or

(b) http://download.eclipse.org/technology/mylar/update-site/dev/e3.2

How save is it to use option (b)?

Cheers, Jörg


Comment 13 Mik Kersten CLA 2007-03-29 17:15:55 EDT
When I refer to dev builds I refer to option (b).  Using dev builds is generally as safe as using releases: http://wiki.eclipse.org/index.php/Mylar_User_Guide#Releases  Fyi Mylar committers tend to work straight from HEAD.

Comment 14 Jörg Thönnes CLA 2007-03-30 16:03:30 EDT
Created attachment 62553 [details]
jstack trace from completely frozen Eclipse

Today (before the next Mylar dev build) I got again a lock-up.
Looking at the stack trace, do you think it is related?
Comment 15 Mik Kersten CLA 2007-03-30 18:14:29 EDT
Joerg: no, this does not look related to me., it looks like a bug that we fixed shortly after that dev build went out.  We'll have a dev build available again in about 20min.