JVM crash with Server 1.6.0_18 (hotspot 16.0) on Win64 but not 1.6.0_16

[David Holmes]

David,

I can not see any open bug reports that match this issue. Did anyone 
even get a Review Id back from Sun when they submitted the bug report?

Often with these bugs the GC is the victim, encountering an invalid oop 
that has been corrupted by other parts of the system. Unfortunately the 
site of the detection can be very far removed from the site of the 
corruption.

A small reproducible test case is invaluable if it can be created.

Cheers,
David Holmes

David Sitsky said the following on 03/30/10 10:14:
> Hi,
> 
> I apologise in advance if posting here is not appropriate.  Posting to
> the Sun's forums and submitting a bug report 5 weeks ago has generated
> no activity at all.
> 
> We have a complex data processing application involving many threads
> that is both memory and I/O intensive.  We recently upgraded our
> application to 1.6.0_18 from 1.6.0_16, and have found the application
> can crash after just a few minutes of processing on some data sets,
> with both the 32-bit and 64-bit version of the server JVM on Win64.
> Before this upgrade, the application can run solidly for many hours or
> days without issue.
> 
> I noticed other people have reported similar findings in the forum
> post here: http://forums.sun.com/thread.jspa?threadID=5424728.
> 
> Interestingly, the client VM was stable.
> 
> I have attached the output from -verbose:gc and the crash file.  Is
> there any interest from members on what we can do to get to the bottom
> of this?  I am happy to run any diagnostics to find more information.
> 
>