Re: [Qemu-devel] AmigaOS 4.x on QEMU

[Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>]

On 25/01/14 00:20, BALATON Zoltan wrote:

> On Sun, 12 Jan 2014, Alpha Mule wrote:
>> Hi. AmigaOS 4.x runs natively on some niche PowerPC boards. I was
>> wondering about the viability of running AmigaOS 4.x on QEMU.
>>
>> Specifically, I was wondering if there has been any development on
>> that and/or what needs to be done to get it fully working.
>>
>> Any information would be appreciated. Thanks.
>
> Looks like not much info is available on this... I think that the first
> difficulty is that AmigaOS 4.x is not freely available (AFAIK) so it's
> hard to do any testing with it.
>
> I've found that a demo version of MorphOS (which is not the same but
> runs on similar hardware) is available and runs on some Mac hardware
> that may be closer to what qemu already has support for so it may be an
> easier target to test.
>
> I've tried to boot it on qemu but haven't got very far and I think qemu
> is missing some features MorphOS needs so it may not be easy to make it
> work. Here's what I've found:
>
> 1. The CPU type MorphOS seems to prefer is G4 7447A. This CPU type may
> not be fully supported by qemu yet and specifying it with -cpu makes the
> VM hang in the openbios firmware before it even gets to boot the OS. By
> using the default G4 CPU without any -cpu option it gets further and
> hangs later in the OS's boot loader.
>
> 2. None of the video cards MorphOS can work with is emulated by qemu and
> I think those qemu has do not work with MorphOS. So even if it was
> running otherwise there would be no display. Theoretically this may be
> got around (for testing at least) with the PCI pass-through of an
> appropriate card from the host but I don't have such a video card at
> hand and don't know if this would be possible in practice at all.
>
> 3. The farthest I got is with the 'boot cd:,\mac_ppc32\boot.img' command
> line from openbios which seems to at least start to boot but hangs
> without any messages (even on the serial console) quite early and I
> don't know enough to debug this further. All I got was some debug logs
> which seem to go astray around here:
>
> IN:
> 0x00441b14: dcbtst r8,r10
> 0x00441b18: stw r5,4(r10)
> 0x00441b1c: stw r5,8(r10)
> 0x00441b20: stw r5,12(r10)
> 0x00441b24: stw r5,16(r10)
> 0x00441b28: stw r5,20(r10)
> 0x00441b2c: stw r5,24(r10)
> 0x00441b30: stw r5,28(r10)
> 0x00441b34: stwu r5,32(r10)
> 0x00441b38: bdnz+ 0x441b14
>
> Raise exception at 00441b18 => 00000002 (00)
> invalid/unsupported opcode: 00 - 00 - 00 (00000000) 00000080 0
> IN:
> 0x00000034: addi r1,r1,16
> 0x00000038: lwz r0,52(r1)
> 0x0000003c: mtlr r0
> 0x00000040: lwz r0,56(r1)
> 0x00000044: mtcr r0

FWIW I do have a MorphOS image in my OpenBIOS test suite which I run as 
part of the release process from time to time, although of course it 
never seems to get further than the bootloader. Last time I spent some 
time checking this in detail, it seemed to get stuck in an infinite loop 
whilst traversing the firmware tree but it's been a while since I 
started digging any deeper into this.

I did find some of the developers on IRC and asked a few questions about 
this a couple of years ago, however the responses I got back were 
unhelpful and rather immature so I simply gave up.


ATB,

Mark.