eclipse crashes on current/amd64

[Brandon Gooch]

On Mon, May 9, 2011 at 10:35 AM, Christian Gusenbauer <c47g at gmx.at> wrote:
> Hi!
>
> I'm quite new to the Java world, but I'd like to learn Java, so I thought, I
> give eclipse a chance and compiled/installed it. After starting eclipse I
> clicked on the "javadoc" tab in the bottom half of the eclipse window and
> after a few seconds eclipse crashed :-(.

I'm a newb myself, and just began working with eclipse (and Java). I'm
experiencing similar crashes.

I haven't done any debugging of the issue, but I found a work-around:

$ eclipse -vmargs -Dorg.eclipse.swt.browser.XULRunnerPath=/dev/null

or setting it in eclipse.ini (somewhere after the -vmargs line):

cat  /usr/local/lib/eclipse/eclipse.ini
[SNIP]
-vmargs
-Dorg.eclipse.swt.browser.XULRunnerPath=/dev/null
[SNIP]

...works for me. What it does EXACTLY is unknown to me at this point,
although I assume it's disabling whatever functionality it would
otherwise obtain from libxul (perhaps using some other method of
detecting the components required, or using something internally).

> The core dump contains hundreds of
>
> #178 0x0000000847550595 in NS_GetComponentManager_P ()
>   from /usr/local/lib/libxul/libxul.so
> #179 0x0000000847550595 in NS_GetComponentManager_P ()
>   from /usr/local/lib/libxul/libxul.so
> #180 0x0000000847550595 in NS_GetComponentManager_P ()
>   from /usr/local/lib/libxul/libxul.so
> #181 0x0000000847550595 in NS_GetComponentManager_P ()
>   from /usr/local/lib/libxul/libxul.so
>
> entries and the top of the call stack looks like this:
>
> #1631 0x00007fffffbfd568 in ?? ()
> #1632 0x000000080000000c in ?? ()
> #1633 0x0000000806292080 in ?? ()
> #1634 0x0000000802f3b5a0 in ?? ()
> #1635 0x00007fffffbfd4e8 in ?? ()
> #1636 0x00007fffffbfd3d0 in ?? ()
> #1637 0x000000080184dbc2 in JavaCalls::call_helper ()
>   from /data/local/openjdk6/jre/lib/amd64/server/libjvm.so
> Previous frame inner to this frame (corrupt stack?)
>
> So what's up with eclipse/libxul?
>
> Thanks,
> Christian.

I'm actually seeing the JVM crash (java.core is generated), so perhaps
the subtleties of our issues differ a bit, but let me know if the
above suggestion works.

Thanks!

-Brandon