Azureus crashes both openjdk and sun java.

Tested the version in karmic, which I believe is current with upstream. (sun) crash log attached.

I added the karmic sources to /etc/apt/sources.list, created:
$ cat /etc/apt/apt.conf
APT::Default-Release "9.04";

And dist-upgraded. Only packages from karmic on my otherwise jaunty system are azureus and java-wrappers.

$ lsb_release -rd
Description: Ubuntu 9.04
Release: 9.04

$ apt-cache policy azureus
azureus:
  Installed: 4.2.0.8-1

$ apt-cache policy sun-java6-jre
sun-java6-jre:
  Installed: 6-16-0ubuntu1.9.04

A useful workaround may be adding "-XX:CompileCommand=exclude,com/aelitis/net/udp/uc/impl/PRUDPPacketHandlerImpl$5,runSupport" to the java incantation. For example, in karmic's version of azureus, change /usr/bin/azrueus from:

run_java -Dazureus.install.path="$HOME/.azureus" $UI \
    org.gudy.azureus2.ui.common.Main "$@"

to

run_java -Dazureus.install.path="$HOME/.azureus" $UI \
    -XX:CompileCommand=exclude,com/aelitis/net/udp/uc/impl/PRUDPPacketHandlerImpl$5,runSupport \
    org.gudy.azureus2.ui.common.Main "$@"

It runs. But it hasn't been long enough to be sure it prevents the crashing.

The above command line option corresponds to a line occurring in all of my java crash logs, both Sun and openjdk:

hs_err_pid13280.log:C2: 69% ! com.aelitis.net.udp.uc.impl.PRUDPPacketHandlerImpl$5.runSupport()V @ 36 (385 bytes)
hs_err_pid14566.log:C2:100% ! com.aelitis.net.udp.uc.impl.PRUDPPacketHandlerImpl$5.runSupport()V @ 36 (385 bytes)
hs_err_pid18945.log:C2: 72% ! com.aelitis.net.udp.uc.impl.PRUDPPacketHandlerImpl$5.runSupport()V @ 36 (385 bytes)
hs_err_pid22751.log:C2: 92% ! com.aelitis.net.udp.uc.impl.PRUDPPacketHandlerImpl$5.runSupport()V @ 36 (385 bytes)
hs_err_pid27619.log:C2: 86% ! com.aelitis.net.udp.uc.impl.PRUDPPacketHandlerImpl$5.runSupport()V @ 36 (385 bytes)
hs_err_pid31409.log:C2: 92% ! com.aelitis.net.udp.uc.impl.PRUDPPacketHandlerImpl$5.runSupport()V @ 36 (385 bytes)
hs_err_pid6077.log:C2:101% ! com.aelitis.net.udp.uc.impl.PRUDPPacketHandlerImpl$5.runSupport()V @ 36 (385 bytes)

Previous workaround was WRONG. Un-escaped "$". This may work:

run_java -Dazureus.install.path="$HOME/.azureus" $UI \
    -XX:CompileCommand=exclude,com/aelitis/net/udp/uc/impl/PRUDPPacketHandlerImpl\$5,runSupport \
    org.gudy.azureus2.ui.common.Main "$@"

Oh, this fix / workaround works.

Upstream bug: http://forum.vuze.com/thread.jspa?threadID=85427

Darxus, unless I'm mistaken in what you want, you probably need to subscribe (not assign) motu-release team to this bug, since at this time in the development cycle you'll need an FFE. You can also add [FFE] at the beginning of the bug title, since that would make it more clear.

Loïc,

This does not need motu-release approval. It is a simple bug fix revision (not new version).

Well, although I'm unable to reproduce this bug running the same versions of the JVM and azureus, a little chat with upstream devs pointed me that the bug isn't specific to azureus, and it's probably something wrong with the hotspot compiler. The issue has happened before with eclipse and other software (a quick google search gave a lot of similar results), and it was supposedly fixed sometime around java 1.5, but as far as I can see, it's happening again. The workaround was reported by The_8472 on freenode, and seems to be working. That said, I'll probably try to tight what really is going on into that thread, that makes it SIGSEV.
Best Regards,
Adrian Perez

Darxus, I just sponsored your debdiff into Karmic. Thanks for your contribution to Ubuntu.

This bug was fixed in the package azureus - 4.2.0.8-1ubuntu1

---------------
azureus (4.2.0.8-1ubuntu1) karmic; urgency=low

  * Avoid bug in both java implimentations by using a command line
    argument to avoid JIT compilation of the problematic class.
    (LP: #428514)

 -- Darxus <email address hidden> Mon, 21 Sep 2009 18:06:46 -0400