Re: gcj 4.6 on OpenBSD/x86

[Kurt Miller <kurt at intricatesoftware dot com>]

On Monday 08 October 2012 04:35:19 pm David Daney wrote:
> On 10/08/2012 01:07 PM, Kurt Miller wrote:
> > Hi,
> >
> > Previously I ported gcj in gcc 4.2.4 to OpenBSD. I currently use it to build classpath-0.98 and bootstrap Oracle's 1.6 JVM using classpath and jamvm. I'm now working on gcj from gcc 4.6.3 on x86 and have run into an issue that I could use some help with.
> >
> > gjar is segfaulting while throwing a ClassCastException from frame 3. Here is some debugging output showing the state and some variables leading up to the segfault. I'm not sure where to go from here. I can't seem to find the root cause of this issue other than something is wrong with 'obj' in frame 3 and obj->getClass()->getName() fails. Any ideas on how to proceed with the debugging on this issue?
> >
> > Core was generated by `egjar'.
> > Program terminated with signal 11, Segmentation fault.
> > #0  0x0c392970 in _Jv_NewStringUTF (bytes=0xfc408b05 <Address 0xfc408b05 out of bounds>)
> >      at /usr/obj/i386/ports/gcc-4.6.3/gcc-4.6.3/libjava/java/lang/natString.cc:245
> > 245       int size = strlen (bytes);
> >
> >
> 
> I haven't studied it closely, but it would appear that 'bytes=0xfc408b05 
> <Address 0xfc408b05 out of bounds>' indicates a real problem.  If it is 
> truly a bad pointer value, you need to figure out where it came from and 
> fix it.

Indeed. I've done that already and included the details in the initial email.
Frame 3 is throwing an exception where obj->getClass()->getName() is
called. 'obj' is the likely source of the problem but the debugger believes it is
an incomplete type and I can't inspect it; see the end of initial email.

Thanks,
-Kurt