This is the mail archive of the
java@gcc.gnu.org
mailing list for the Java project.
Re: Interpreting bytecode on a powerpc 405 EP
- From: Andrew Haley <aph at gcc dot gnu dot org>
- To: "Patrick Olinet" <patrick dot olinet at gmail dot com>
- Cc: java at gcc dot gnu dot org
- Date: Thu, 19 Apr 2007 11:53:21 +0100
- Subject: Re: Interpreting bytecode on a powerpc 405 EP
- References: <30f6090e0704190317i4f3d72a4i64716a598015ee79@mail.gmail.com>
Patrick Olinet writes:
> Hello all,
>
> I'm a newbie on a project using gcj 3.4.3. We execute java code on a
> powerpc 405 EP with a linux 2.4.20 kernel. Our cross-platform
> toolchain was build two years ago by an other team and works fine.
> Today we want to interpret some java bytecode so as to load
> dynamically jar libraries. But, the executable fails when we try to
> load a simple java class. The sample below fails with an illegal
> instruction whereas it works well on a X86 platform:
>
> Class c = cc.loadClass("InterpretedClasse");
> System.out.println(" we load: " + InterpretedClasse.class.getName());
> InterfaceClasse obj2= (InterfaceClasse)c.newInstance(); // GCJ crashes
> System.out.println(" instance created:" + obj2.name());
>
> Has someone a similar experience with this kind of platform ?
> I don't know how to investigate?
I'd run it under gdb. Have a look at that illegal instruction.
gcj 3.4.3 is very old. I don't know if interpreted code on PPC was
supported back then; maybe, maybe not.
> Maybe using the latest version of gcj will solve my problem?
Probably.
> I will have a look at the libffi which could be the one to blame?
libffi has its own testsuite. It would be a good idea to run it.
Andrew.