Progress with Java VM portings

[Charles-Antoine Gauthier]

Eric Norum wrote:
> 
> > Charles-Antoine Gauthier wrote:
> > > My collegues just complete the port of RTEMS to the MBX821/MBX860. The
> > > last few weeks were spent trying to figure out why the MBX821 will not
> > > work at all with the data cache turned on. We suspect that the memory
> > > controller values in the documentation are wrong and that we are getting
> > > bus errors during burst writes. Also, on the MBX860, interrupts from the
> > > serial ports do not interrupt the PowerPC core when the data cache is
> > > turned on. We can see the interrupt pending in the SIPEND register, the
> > > SIMASK register is set to pass them on, and the interrupts are enabled
> > > in the core, but no cookie :-( We submitted the code yesterday to OAR. I
> > > am sure that we missed the cutoff date for inclusion in RTEMS 4.5, but
> > > that was unavoidable.
> > >
> 
> Could this be related to the problems I've seen with the 68360?  The
> event bit is set, the mask bit is set, interrupts are enabled, but the
> bit in the `in-service' register (CISR) is also set so no interrupt
> occurs.  I was able to work around the problem by running the
> programmable interrupt timer at the same CPU32 priority as the CPM.  If
> the PIT is at a higher priority than the CPM, the CPM interrupt would
> get locked once every 30,000 to 40,000 times.
> The 860 has only one interrupt level though (recalled from my dim
> memory...), right?  So perhaps interrupt priority problems could not be
> the problem here.

I don't know. We are aware of the problems with the 68360. We are
planning some further experiments in the near future to try to narrow it
down.

> 
> > > For the first time yesterday, we had the Lego RCX driving around under
> > > control from a Java program running in the MBX821 and MBX860. C++ and
> > > Java exceptions are working. Previously, we only could use the MVME167
> > > to control the robot.
> 
> Does that mean you have a network driver working for the MVME167???? :-)

You wish :-). The student is here, but he was employed on higher
priority tasks, and continues to be. At this time, I don't know when he
might get back to the network driver. Sorry. So much code to write and
debug, so little time... FYI, the RCX is driven from the serial port at
2400bsp, 8 data bits, odd parity, 1 stop bit, through an infrared
transceiver. No networking required. It does mean that the MBX8xx and
MVME167 serial drivers can run at other settings than 9600 8N1 (for the
MBX8xx, that will be true when we submit a patch to the patch we
submitted yesterday, which should be Real Soon Now).

> 
> --
> Eric Norum                                 eric at cls.usask.ca
> Canadian Light Source                      Phone: (306) 966-6308
> University of Saskatchewan                 FAX:   (306) 966-6058
> Saskatoon, Canada.

-- 
Charles-Antoine Gauthier
Research Officer
Software Engineering Group
Institute for Information Technology
National Research Council of Canada
Ottawa, ON, Canada
K1A 0R6