Re: a JNI backtrace

[Bryce McKinlay <bryce at albatross dot co dot nz>]

"Marcus G. Daniels" wrote:

> >>>>> "BM" == Bryce McKinlay <bryce@albatross.co.nz> writes:
>
> MD> Any idea what this means?  I can see it makes it in and out of
> MD> the stub for the SwarmCImpl create method ok..
>
> BM> Your c++filt is too old and it isn't demangling the names
> BM> correctly. Make sure it finds the new c++filt that gets put in your
> BM> gcc install directory on the path first, and it should be a lot
> BM> more readable.
>
> BM> Err... get a new GDB I mean ;-)
>
> Ok thanks..  I'm building a new version now.
>
> Anyway, I'm wondering why this hash from PopSystemFrame is crashing.
> (Not the meaning of the mangling..)

It looks like a JNI bug to me. I suspect _Jv_JNI_PopLocalFrame is possibly passing some sort of invalid pointer (but
not null) to unmark_for_gc(), which is subsequently calling Hashtable.get() with a garbage pointer. You may need to
use your new GDB to poke around and figure out if this theory is correct and if so, where the garbage in env->locals
comes from.

regards

  [ bryce ]