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.

Bug 41776 - "Deadlock" during startup
Summary: "Deadlock" during startup
Status: CLOSED DUPLICATE of bug 33398
Alias: None
Product: platform
Classification: Unclassified
Component: -- Other -- (show other bugs)
Version: 4.x
Hardware: All All
: P2 blocker (vote)
Assignee: Petr Nejedly
URL:
Keywords: THREAD
Depends on:
Blocks:
 
Reported: 2004-04-07 08:53 UTC by Marian Petras
Modified: 2008-12-22 19:37 UTC (History)
0 users

See Also:
Issue Type: DEFECT
Exception Reporter:


Attachments
thread-dump (13.32 KB, text/plain)
2004-04-07 08:53 UTC, Marian Petras
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Marian Petras 2004-04-07 08:53:20 UTC
I encountered a "deadlock" during NetBeans
startup. The strange thing is that the CPU usage
was at 100% (and not near 0% which is typical for
deadlocks).

I had a build made from NetBeans sources updated
on 6 April at about 3:00 PM CEST (1:00 PM GMT).

I am attaching a full thread-dump.
Comment 1 Marian Petras 2004-04-07 08:53:59 UTC
Created attachment 14311 [details]
thread-dump
Comment 2 pzajac 2004-04-07 09:38:20 UTC
probably problem in JVM (wating on condition)
Comment 3 Petr Nejedly 2004-04-07 10:54:00 UTC
I won't blame JVM so fast.
waiting on condition is there because of allocation.
JVM was probably doing GC just before the thread dump was taken.

If CPU is 100% then it's not a deadlock.
It may be some infinite loop keeping the threads around busy
or some kind of livelock (e.g. two threads pinging each other).

I'd need more dumps few seconds apart, every time CPU is 100%

Anyway, I have strong indication that this is our long-lasting
livelock in Mutex, which is finally easy to reproduce (I've just
reproduced it) thanks to few buildsystem aspects:
1. It remembers all opened nodes in the projects view
2. It computes the package view by touching _all_ the data objects
3. You had several nodes opened before shutdown

This together stresses Mutex enough to livelock.
Comment 4 Petr Nejedly 2004-04-07 14:52:57 UTC
OK, tkis is actually a dup of #33398, but one that will finally help
us to solve that long standing problem. Thanks much for the report.


*** This issue has been marked as a duplicate of 33398 ***
Comment 5 Marian Mirilovic 2005-07-15 07:43:52 UTC
closed