blob: d1e2038d14db09db2c09a6a453eb2e0e0a41e93c [file] [log] [blame]
/*******************************************************************************
* Copyright (c) 2000, 2003 IBM Corporation and others.
* All rights reserved. This program and the accompanying materials
* are made available under the terms of the Common Public License v1.0
* which accompanies this distribution, and is available at
* http://www.eclipse.org/legal/cpl-v10.html
*
* Contributors:
* IBM Corporation - initial API and implementation
*******************************************************************************/
package org.eclipse.jdi.hcr;
import com.sun.jdi.Value;
/**
* Hot code replacement extension to <code>com.sun.jdi.ThreadReference</code>.
*/
public interface ThreadReference {
/**
* Resumes the execution of this thread as if the next instruction was a return
* instruction with the given value. This causes the top stack frame to be popped with
* the given value.
* <p>
* A breakpoint instruction at the current instruction is not triggered that is, this
* operation takes precedence over breakpoints. <code>try-finally</code> blocks enclosing
* the current location will be triggered in due course.
* <p>
* The triggerFinallyAndSynchronizedBlocks option on this operation controls whether
* <code>try-finally</code> and <code>synchronized</code> blocks enclosing the current
* location should be triggered:
* <ul>
* <li>If no, the stack frame is popped, the return value is returned, and execution
* continues back in the caller. Note that <code>finally</code> blocks are not run,
* and that if the code is nested within a <code>synchronized</code> statement, the
* monitor lock is not released (however, if the method is </code>synchronized</code>
* the monitor lock will be properly released). This mechanism is sure-fire, but at
* the risk of not letting the target program clean itself up (e.g., close its files).
* <li>If yes, the VM checks to see whether there might be a <code>finally</code> or
* <code>synchronized</code> block enclosing the current instruction.
* <ul>
* <li>If there is no enclosing <code>finally</code> block, the operation reduces
* to the above case.
* <li>If there is an enclosing <code>finally</code> block, the VM creates a VM
* exception and activates the <code>finally</code> block with it. If this
* exception eventually causes the stack frame to be popped, the exception is
* caught by the VM itself, the return value is returned, and execution continues
* back in the caller.
* <ul>
* </ul>
* <p>
* Note that a <code>finally</code> block manifests itself as (and is indistinguishable
* from) a <code>catch Throwable</code> block. <code>synchronized</code> statements
* also compile to a <code> catch Throwable block<code>.The target program may inadventently
* end up catching this exception.
*
* Since the choices each have their pros and cons, making the decision
* is left to the debugger. However the later option is the recommended choice.
* <p>
* The reply to the operation contains a flag indicating whether any <code>finally</code> or
* <code>synchronized</code> blocks are enclosing the current instruction.
* <p>
* This operation is ignored if the thread was not suspended. If the thread was suspended multiple
* times, wait for the same number of resumes before executing the return instruction.
* <p>
* The returned value is ignored if the method returns void.
* <p>
* Throws an <code>OperationRefusedException</code> if the VM refused to perform this operation.
* This in recognition that the VM may be in an awkward state and unable to comply:
* <ul>
* <li>for example, execution is suspended in a native method,
* <li>for example, execution is suspended during class preparation.
* </ul>
*/
public boolean doReturn(Value returnValue, boolean triggerFinallyAndSynchronizedBlocks);
}