Xen users - May 2006 - JVM coredump leading to dom0/domU hanging (Xen 3.0 Test CD)

Hi,

We have an issue where a relatively simple Java + JNI program causes  
the dom0/domU to hang.  The example (http://www.kymasys.com/ 
java.jni.segv.tar.gz) consistently reproduces the issue using the  
latest Xen 3.0 Test CD running the 32 bit kernel.  Strangely, the  
issue does not happen with the PAE kernels from the same Test CD.

The code basically calls a JNI function that does a NULL pointer  
access. We expect this to dump core, but instead we never get back to  
the shell prompt and the system hangs completely in about 30 seconds.

Instructions for reproducing the issue:

(1) Install the JDK (we are using v 1.5 update 4).
(2) Untar the example tarball (the example can be downloaded from  
http://www.kymasys.com/java.jni.segv.tar.gz).
(3) from the segv directory execute the following command: "java - 
Djava.library.path=. TestClass"

Any help would be greatly appreciated.

Thanks in advance.
Sanjay



_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Hi, I can''t seem to find this in any documentation, but I''m
curious
of how Xen processor reservations work.  Do DomUs get physically  
allocated several processors to use that remain just remain a  
constant number and possibly get shared with other VMs, or do they  
really get their own processors that aren''t touched by anyone else?

Just to stack two questions, I''m also wondering how safe it is to mix  
and match PAE/non-PAE linux kernels under a PAE xen.  PAE should only  
be necessary in Linux if it''s been allocated more than 4 gigabytes  
from Xen, correct?  Thanks much for the replies in advance!
			--Greg Cymbalski

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users