Attachment 26174 Details for Bug 105491 – JUnit test to reproduce the problem

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

JUnit test to reproduce the problem

TestBug105491.java (text/plain), 3.25 KB, created by John Arthorne

on 2005-08-16 18:40:39 EDT

(hide)

Description:

Filename:

MIME Type:

Creator: John Arthorne

Created: 2005-08-16 18:40:39 EDT

Size: 3.25 KB

patch

obsolete

>/*******************************************************************************
> * Copyright (c) 2005 IBM Corporation and others.
> * All rights reserved. This program and the accompanying materials
> * are made available under the terms of the Eclipse Public License v1.0
> * which accompanies this distribution, and is available at
> * http://www.eclipse.org/legal/epl-v10.html
> *
> * Contributors:
> *     IBM Corporation - initial API and implementation
> *******************************************************************************/
>package org.eclipse.ui.tests.concurrency;
>
>import java.lang.reflect.InvocationTargetException;
>import junit.framework.*;
>import org.eclipse.core.resources.*;
>import org.eclipse.core.runtime.*;
>import org.eclipse.jface.dialogs.ProgressMonitorDialog;
>import org.eclipse.jface.operation.IThreadListener;
>import org.eclipse.swt.widgets.Display;
>import org.eclipse.swt.widgets.Shell;
>import org.eclipse.ui.actions.WorkspaceModifyOperation;
>
>/**
> * Tests the following sequence of events:
> * 1) Workspace operation starts in the UI thread.  Workspace lock is taken, and a modal context thread is forked (call this modal context MC1).
> * 2) Operation transfers the workspace lock to MC1 using IJobManager.transferRule
> * 3) Operation runs in MC1, scheduling an asyncExec.
> * 4) MC1 passes the scheduling rule back to UI thread, and exits
> * 5) After passing the rule back to the UI thread, but before MC1 dies, the asyncExec is run.
> * 6) The asyncExec forks another model context (MC2), and blocks the UI thread in another event loop.
> * 7) MC2 tries to acquire the workspace lock and deadlocks, because at this point it has been transferred to the UI thread
> */
>public class TestBug105491 extends TestCase {
>	class TransferTestOperation extends WorkspaceModifyOperation implements IThreadListener {
>		public void execute(final IProgressMonitor pm) {
>			//clients assume this would not deadlock because it runs in an asyncExec
>			Display.getDefault().asyncExec(new Runnable() {
>				public void run() {
>					ProgressMonitorDialog dialog = new ProgressMonitorDialog(new Shell());
>					try {
>						dialog.run(true, false, new WorkspaceModifyOperation() {
>							protected void execute(IProgressMonitor monitor) {}
>						});
>					} catch (InvocationTargetException e) {
>						e.printStackTrace();
>						fail(e.getMessage());
>					} catch (InterruptedException e) {
>						//ignore
>					}
>				}
>			});
>		}
>
>		public void threadChange(Thread thread) {
>			Platform.getJobManager().transferRule(workspace.getRoot(), thread);
>		}
>	}
>
>	private IWorkspace workspace = ResourcesPlugin.getWorkspace();
>
>	public TestBug105491() {
>		super();
>	}
>
>	public TestBug105491(String name) {
>		super(name);
>	}
>
>	/**
>	 * Performs the test
>	 */
>	public void testBug() throws CoreException {
>		workspace.run(new IWorkspaceRunnable() {
>			public void run(IProgressMonitor monitor) {
>				ProgressMonitorDialog dialog = new ProgressMonitorDialog(new Shell());
>				try {
>					dialog.run(true, false, new TransferTestOperation());
>				} catch (InvocationTargetException e) {
>					e.printStackTrace();
>					fail(e.getMessage());
>				} catch (InterruptedException e) {
>					//ignore
>				}
>			}
>		}, workspace.getRoot(), IResource.NONE, null);
>	}
>}

Actions: View

Attachments on bug 105491: 26174