Re: [Qemu-ppc] MorphOS 4.x on QEMU

[Mark Cave-Ayland]

On 26/02/14 02:43, BALATON Zoltan wrote:

> On Tue, 25 Feb 2014, BALATON Zoltan wrote:
> > On Tue, 25 Feb 2014, BALATON Zoltan wrote:
> > > overwrites part of the exception_return code of openbios so the next
> > > time it gets a DSI exception and is trying to run this code it leads
> > > to an invalid opcode exception and that's it.
> >
> > Actually if I restore the overwritten word in memory it runs further
> > until it prints "console failed".
>
> Hmm... Does anyone actually reads this list or am I talking to myself here?

Hi Zoltan,

Sorry I am around, just exceptionally busy at the moment. My efforts 
around QEMU/OpenBIOS are unsponsored and so get relegated into whatever 
spare time I have left :)

> Debugging it further shows that despite not finding a compatible Radeon
> graphics adapter it continues further and only stops after trying to
> identify/initialise the CPU. I still have no fix for the zeroing the
> 0x80 address causes an exception problem.
>
> Is is possible to use pci passthrough to make a graphics card available
> to the guest? Was it tested recently on ppc? Is there any documentation
> on how to do it? In case I can find a compatible card I might try to see
> if that would work.
>
> Also I've found that the MorphOS kernel looks for a device within /pci
> with a class-code == 0x0780 (other communication device) for a serial
> port where it may display some debug info. Unfortunately there is no
> such device on QEMU. Is there an easy way to add one? Can someone verify
> if this is present on Mac hardware or does it expect some third party
> serial card for debugging? Having some debug printout might help
> debugging this further.
>
> Any ideas anyone?

When I last spoke with the MorphOS guys, I did manage to talk to someone 
on IRC who had the source code open on his machine and would give me 
pointers in the right direction as to roughly what the bootloader was 
doing. I suggest that you'll need to ask in the forums/mailing lists/IRC 
to see if someone is willing to help you at all.

FWIW there is a known issue with PPC's next-property iterator being 
implemented differently to the standard (see Olivier's tentative patch 
at 
http://www.openfirmware.info/pipermail/openbios/2014-February/008141.html). 
It's likely that a broken device tree iteration could be the cause of 
the MorphOS bootloader not being able to locate a suitable console mode 
which could be related to your issue.

Also on SPARC there were problems with the hardware flow control causing 
serial console output from the Solaris kernel to freeze; the eventual 
solution was to add properties to the serial port node which tell the 
kernel to ignore the hardware flow control lines.


HTH,

Mark.