This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.
I was debugging and suddenly the IDE froze. I grabbed a thread dump which I will attach next. There are a lot of debugging-related threads doing things with the source model or documents. The AWT thread gets locked trying to acquire a write lock so somebody is holding a readlock while waiting. I'm not sure if it's the debugger but from the threads it looks like a promising place to start looking. This is on a trunk build fresh as of yesterday, on Ubuntu Linux: Product Version: NetBeans IDE Dev (Build 080116) Java: 1.6.0_03; Java HotSpot(TM) Client VM 1.6.0_03-b05 System: Linux version 2.6.22-14-386 running on i386; UTF-8; en_US (nb) Userdir: /home/tor/.netbeans/dev
Created attachment 55220 [details] Deadlock thread dump
This issue is considered to be a stopper for M1. Could you please evaluate this issue ASAP?
In the thread dump, I do not see anything important blocked by debugger. AWT is waiting at javax.swing.text.AbstractDocument.writeLock(). Someone is likely reading from or writting into the document, but it's not apparent who is it and where is it blocked. Can someone from editor evaluate this?
AWT waits for document write lock. The thread "ToolTip-Evaluator" (nid=0x4768) holds the document read lock (from the frame NbToolTip$Request.run(NbToolTip.java:399)) and is waiting for the JavaSource to process its request. JavaSource is currently processing "Debugger operator thread" (nid=0x46c0), which is runnable, but probably blocked somewhere deep in JDI or remote process. Document needs to be accessible, nobody should keep it locked for extended periods of time.
O.K. Thanks for the evaluation Petre, the document read lock is not visible in the dump, it's not a standard monitor. The problem seems to be in "Debugger operator thread", but that's running. It does not looks like a deadlock, but just a temporarily frozen AWT.
Tor, do you have any steps to reproduce this? How long did you wait before you've generated the thread-dump?
Martin, Tor any progress ? Jirka, could you help with reproducing this issue ? Thanks in advance.
I will change the code not to call JDI methods, that connect to remote VM inside the JavaSource task. That should help.
Hopefully fixed in trunk, please verify... /shared/data/ccvs/repository/debuggerjpda/src/org/netbeans/modules/debugger/jpda/ExpressionPool.java,v <-- ExpressionPool.java new revision: 1.6; previous revision: 1.5
Verified ... and Closing all issues resolved into NetBeans 6.7 and earlier.