On 2/19/2014 8:03 AM, Keith Amling wrote:
I may be barking up the wrong tree e-mailing an OpenJDK list when I've
only reproduced this against an Oracle JDK (1.7.0_25, and in particular
it doesn't reproduce against 1.7.0_21), but the attached file produces
NPEs [1] via EventQueue.isDispatchThread() fairly regularly with an
argument of 10 [threads] and can even produce it with just 2 (although
less often).
This is an interesting exception. I am able to reproduce it with
the JDK 7u51 but
not able to reproduce it with the JDK 8.
Could you check that the issue is not reproduced on your side with
the latest JDK 8:
https://jdk8.java.net/download.html
We've had even weirder NPEs [2] in the code behind that but I have yet
to distill them to a repro unfortunately.
We have several reports about this NPE like:
https://bugs.openjdk.java.net/browse/JDK-8019272
It should have been already fixed in JDK 7u60 and in the JDK 8.
Thanks,
Alexandr.
Are either of these known? Is there something we should be doing in our
rather multi-threaded environment to protect this code?
Keith
[1] E.g.:
java.lang.NullPointerException
at java.awt.EventQueue.isDispatchThread(EventQueue.java:1014)
at EQNPE$1.run(EQNPE.java:23)
[2] Ending in:
java.lang.NullPointerException
at sun.awt.SunToolkit.getSystemEventQueueImplPP(SunToolkit.java:1011)
at sun.awt.SunToolkit.getSystemEventQueueImplPP(SunToolkit.java:1007)
at
sun.awt.HeadlessToolkit.getSystemEventQueueImpl(HeadlessToolkit.java:352)
at java.awt.Toolkit.getSystemEventQueue(Toolkit.java:1717)
<our code>
