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

Bug 327529

Summary: ConcurrentModificationException in BranchRevisionCache
Product: [Modeling] EMF Reporter: Pascal Lehmann <pascal.lehmann>
Component: cdo.coreAssignee: Eike Stepper <stepper>
Status: CLOSED FIXED QA Contact: Eike Stepper <stepper>
Severity: normal    
Priority: P3    
Version: 4.0   
Target Milestone: ---   
Hardware: All   
OS: All   
Whiteboard:
Attachments:
Description Flags
Proposed patch none

Description Pascal Lehmann CLA 2010-10-12 04:54:32 EDT 
Build Identifier: 4.0

Just recenty stumbled over this stacktrace while testing:

Caused by: java.util.ConcurrentModificationException
	at java.util.HashMap$HashIterator.nextEntry(HashMap.java:793)
	at java.util.HashMap$EntryIterator.next(HashMap.java:834)
	at java.util.HashMap$EntryIterator.next(HashMap.java:832)
	at org.eclipse.emf.cdo.internal.common.revision.cache.branch.BranchRevisionCache.getRevisions(BranchRevisionCache.java:206)
	at org.eclipse.emf.internal.cdo.transaction.CDOTransactionImpl.createRevisionAvailabilityInfo(CDOTransactionImpl.java:371)
	at org.eclipse.emf.internal.cdo.transaction.CDOTransactionImpl.merge(CDOTransactionImpl.java:340)
... 42 more

It happened in 1 of our 10'000 random tests, couldn't reproduce since.
But when looking at BranchRevisionCache, it looks as if access to revisionLists is not consistently synchronized.

Reproducible: Sometimes
Comment 1 Pascal Lehmann CLA 2010-10-12 04:57:30 EDT 
Created attachment 180652 [details]
Proposed patch

Added synchronized access to revisionList.

1) The number of lines that you changed is smaller than 250.
confirmed.
2) You are the only author of these changed lines.
confirmed.
3) You apply the EPL to these changed lines.
confirmed.
Comment 2 Eike Stepper CLA 2010-10-13 03:13:12 EDT 
As such the synchronizations in getAllRevisions() and getRevisions(CDOBranchPoint) limit concurrency a lot if called. But these methods are really not supposed to be called in regular server operations. Currently they're only called on the client, which seems acceptable. So we probably don't need more opimized solutions ;-)

Committed to HEAD
Comment 3 Eike Stepper CLA 2011-06-23 03:40:56 EDT 
Available in R20110608-1407