darwin & libgcj ?
Andreas Tobler
toa@pop.agri.ch
Thu Jan 9 18:03:00 GMT 2003
Andrew Haley wrote:
> Jeff Sturm writes:
> > On Mon, 6 Jan 2003, Andrew Haley wrote:
> > > Okay. Throw_2 failures on their own usually indicate that null
> > > pointer exceptions in CNI code don't work, and that's because
> > > -fcheck-references is being used.
> > > Good call. It looks like Darwin doesn't throw SIGSEGV when deferencing a
> > null pointer, but something else:
> > > Program received signal EXC_BAD_ACCESS, Could not access memory.
> > java::lang::Double::parseDouble(java::lang::String*) (str=0x0) at
> > java/lang/String.h:50
> > 50 jint length () { return count; }
>
> Ah. Oh well, it looks like can_unwind_signal should not be set on
> this platform. There is no MD_FALLBACK_FRAME_STATE_FOR in darwin.h.
I'm on the way to build libgcj with 'can_unwind_signal = no' now, do I
have to follow some other aspects as well? (boehm-gc already 'tweaked')
I also update my dejagnu/expect chain. But so far I still get this ERROR
no main ..... Here I assume it's a(the) bug in the OS. I'll ping apple
people about it again.
Andreas
More information about the Java
mailing list