ErrorHandlingTest.java

/******************************************************************************
 * Copyright (c) 2006, 2010 VMware Inc., Oracle Inc.
 * All rights reserved. This program and the accompanying materials
 * are made available under the terms of the Eclipse Public License v1.0
 * and Apache License v2.0 which accompanies this distribution. 
 * The Eclipse Public License is available at 
 * http://www.eclipse.org/legal/epl-v10.html and the Apache License v2.0
 * is available at http://www.opensource.org/licenses/apache2.0.php.
 * You may elect to redistribute this code under either of these licenses. 
 * 
 * Contributors:
 *   VMware Inc.
 *   Oracle Inc.
 *****************************************************************************/

package org.eclipse.gemini.blueprint.iandt.errors;

import java.io.FilePermission;
import java.util.ArrayList;
import java.util.List;
import java.util.PropertyPermission;

import org.eclipse.gemini.blueprint.iandt.BaseIntegrationTest;
import org.osgi.framework.AdminPermission;
import org.osgi.framework.Bundle;
import org.springframework.core.io.Resource;

/**
 * @author Hal Hildebrand Date: May 29, 2007 Time: 5:07:01 PM
 */
public class ErrorHandlingTest extends BaseIntegrationTest {

	protected String getManifestLocation() {
		return null;
	}

	// Specifically do not wait
	protected boolean shouldWaitForSpringBundlesContextCreation() {
		return false;
	}

	/**
	 * While it may appear that this test is doing nothing, what it is doing is
	 * testing what happens when the OSGi framework is shutdown while the
	 * Spring/OSGi extender is still running threads which are trying to clean
	 * up after an error condition in creating the context. When running under
	 * the Felix platform, this test will produce deadlock or, if things are
	 * really going south, serious CDNFE when handling the error (e.g.
	 * java.lang.InputStream, java.lang.Integer, etc.). The reason for the
	 * latter is that the OSGi framework is shutdown and the class loaders being
	 * used by extender threads are no longer valid. The deadlock case stems
	 * from a similar condition in that the system is deadlocked around the
	 * synchronization of the context.create()/close() because the underlying
	 * framework is deadlocked around resource resolution. As we process the
	 * close of the context synchronously, we're blocked waiting for the
	 * resource resolution which is blocked because the framework is shutting
	 * down. We can't unblock because we're still waiting for the lock on the
	 * context to shutdown and unblock the event notification from the
	 * underlying framework.
	 * 
	 * Consequently, please do not remove or modify this test unless you talk to
	 * Hal. :)
	 */
	public void testErrorHandling() throws Exception {
		Resource errorResource = getLocator().locateArtifact("org.eclipse.gemini.blueprint.iandt", "error",
			getSpringDMVersion());
		assertNotNull("Error bundle resource exists", errorResource);
		Bundle errorBundle = bundleContext.installBundle(errorResource.getURL().toExternalForm());
		assertNotNull("Errro bundle exists", errorBundle);

		errorBundle.start();
	}

	protected List getTestPermissions() {
		List perms = super.getTestPermissions();
		// export package
		perms.add(new AdminPermission("*", AdminPermission.EXECUTE));
		perms.add(new AdminPermission("*", AdminPermission.LIFECYCLE));
		perms.add(new AdminPermission("*", AdminPermission.RESOLVE));
		perms.add(new FilePermission("<<ALL FILES>>", "read"));
		return perms;
	}
}