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

Bug 353541

Summary:	VerifyError after workspace refresh after crash
Product:	[Tools] Objectteams	Reporter:	Stephan Herrmann <stephan.herrmann>
Component:	OTEquinox	Assignee:	Stephan Herrmann <stephan.herrmann>
Status:	RESOLVED WORKSFORME	QA Contact:
Severity:	normal
Priority:	P3
Version:	2.0
Target Milestone:	2.1 M1
Hardware:	Other
OS:	Linux
Whiteboard:

Description Stephan Herrmann

2011-08-02 04:05:08 EDT

I keep seeing exceptions like

java.lang.VerifyError: (class: org/eclipse/jdt/internal/corext/refactoring/reorg/ReorgPolicyFactory$SubCuElementReorgPolicy, method: _OT$getDestinationNode$chain signature: ([Lorg/objectteams/ITeam;[IIII[Ljava/lang/Object;Lorg/eclipse/jdt/core/IJavaElement;Lorg/eclipse/jdt/core/dom/CompilationUnit;)Ljava/lang/Object;) Expecting to find integer on stack
	at org.eclipse.objectteams.otdt.internal.refactoring.adaptor.ReorgAdaptor._OT$registerAtBases(ReorgAdaptor.java)
	at org.objectteams.Team.doRegistration(Team.java:371)
	at org.objectteams.Team.activate(Team.java:164)
	at org.objectteams.TeamThreadManager.newThreadStarted(TeamThreadManager.java:66)
	at org.eclipse.objectteams.otequinox.internal.JobAndThreadListener.aboutToRun(JobAndThreadListener.java:42)
	at org.eclipse.core.internal.jobs.JobListeners$1.notify(JobListeners.java:29)
	at org.eclipse.core.internal.jobs.JobListeners.doNotify(JobListeners.java:96)
	at org.eclipse.core.internal.jobs.JobListeners.aboutToRun(JobListeners.java:142)
	at org.eclipse.core.internal.jobs.JobManager.startJob(JobManager.java:1485)
	at org.eclipse.core.internal.jobs.WorkerPool.startJob(WorkerPool.java:221)
	at org.eclipse.core.internal.jobs.Worker.run(Worker.java:50)

In other stacks a callin binding tries to trigger failing with the
same verify error.

The first message in the log says 

Warning
Tue Aug 02 01:58:58 CEST 2011
The workspace exited with unsaved changes in the previous session; refreshing workspace to recover changes.


After a clean restart of eclipse the problem no longer occurred.
Unsure if this is reproducable by any means.

Comment 1 Stephan Herrmann

2011-08-13 11:15:15 EDT

Sounds like the OTRE in one particular situation produced illegal code.
I see three possibilities:
- we received corrupt bytes as the input
- the attribute was corrupted giving inconsistent instructions to the OTRE
- the OTRE messed up itself
For any of these a very rare race condition gives a good hypothesis.

I've no idea how we could ever reproduce.