I just ported Basilisk II to MorphOS. It is very early release with poor user interface but more or less it works.

I reused old AmigaOS source code and ethernet/scsi/sound interfaces should work but I dont take responsibility if it wipes your HD.

The binary archive is available here (390K) (sources 780K).

(Basilisk is 68k Macintosh emulator.)

Itix

You are the MAN! THank you. Its exciting.. Have you tried any of the MacOs(s) yet?

When I have time I'll try os8..

magnetic

I booted succesfully to MacOS 7.x but I have no programs to try. Colours are reportedly wrong but will check these when I retrieve old Mac setup from my Amiga HD.

Cool deal! I can install MacOs on MorphOS now! You know if the speed is good you can run Netscape to browse the internet! I know my A3000 ran it faster then a real Mac did in the day.

Itix, this is very cool. Good work!

i've just run macos 7.5 from my old shapeshifter filedisk. cpu emulation i fast (speedomeeter gives 160% of the speed of quadra 605 - 040-25mhz). i've have one problem: emulator run nicely on window but it is only black&white color mode, when i choose fullscreen mode - emulator freeze after showing mac face.

i've g3 600mhz peg2
24 bit mode is working ok - i've tested MacWorms. the problem is only with 8bit mode - wrong colors (palette) but i've run warcraft2 (without problems and without proper colors ;)

wonderful news, good job !!!!
will try it asap !
thanks a lot man, you're the don !

Very nice! Thank you Itix!

SheepShaver shares many source files with BasiliskII, do you think that it may be possible to have SheepShaver running? Gwenole version is here. AFAIK The latest version accepts newworld roms so it should be more easier to port. Now it uses SDL so it would be a great program to have running.

I lack the knowledge to port it

Have you taken a look of latest sources?

Running MacOS9 on MorphOS may be very useful! And people wouldn't have to install Linux to run MOL

I guess Basilisk emulate 68k without JIT. AFAIR it have UAE 68k emulation engine. We can say that it is fast, but it can be maybe faster? Classic Amiga version do not emulate any processor. I mean 68k Basilisk on MorphOS will use built-in MOS emulator for MacOS.

I hope I haven't jabbered. ;)

I just found this one the other day... maybe You can make use of it

Basilisk II/JIT

Quote:

Yeah, all you'd need is an x86 emulator in addition to that ;)

Quote:

Ups, didn't think about that... too long time since I have been using UAE...

Basilisk II is able to use the native CPU, how about making use the JIT emulator of MorphOS... I think that would be cool

[ Edited by hnl_dk on 2005/4/13 13:31 ]

@mcbarlo & hnl_dk

Using the virtual 68k CPU from MorphOS to emulate 68k in Basilisk doesnt work. Basilisk patches Mac ROM by planting illegal instructions and on 68k Amiga these are catched by an exception handler. Those probably wont work on MorphOS at all.

Also some 68k FPU features are not emulated in MorphOS because they are not used on Amiga. However, I have no idea if Mac needs these.

But real killer is that Basilisk needs free access to zero page (0x0000 - 0x1fff) if real 68k is used. In MorphOS this is not possible at all.

Great work!
thank you very much!!!

itix wrote:
"Also some 68k FPU features are not emulated in MorphOS because they are not used on Amiga. However, I have no idea if Mac needs these."

on my machine speedomeeter doesn't detect fpu

If anyone can get the 7.5.3 files off the apple site and make a normal install disk out of them, or if you have a normal install disk it can be shared. Apple has given it away free for a long time now. I cant seem to find my OS 8 install disks and want to install this on my Peg!!!!

@Crumb

SheepShaver should be relatively easy to port. I've SheepShaver sources here and it is very similar to Basilisk. But due to PPC CPU emulation it wont be too fast...

@Itix:

The feature that attracts me of SheepShaver is that it uses de PPC cpu directly on PPC systems

So there would be no need to emulate the PPC.

Previous versions required pre-newworld roms, but the latest one should work with newworld roms.

AFAIK newworld roms can be allocated in any part of the memory, not like the old ones.

I don't know if the first 8KB of memory are still used by MacOS or if it can be configured to use another memory location.

Some memory management functions may come handly... I hope future MOS versions include some memory management functions to remap areas of memory, catch illegal accesses etc... Every modern OS in the world (Windows, Linux, AmigaOS4, AmigaOS3+MMU.library) has memory management functions, so why MorphOS can't?

I think that LouiSe was doing a PearPC port but I don't think he finished that port...

As you are in close communications with the MOS team maybe you could convince them to keep the first 8KB free and grant access to the ABOX to these first kilobytes. It doesn't compromise stability and it doesn't involve heavy modifications so I hope they will allow this. Come on, even CyberstormPPC included an option to keep free the first kilobytes of memory...

@ Crumb

I guess missing MMU access will not allow SheepShaver use the PPC directly, but I do not know. Same killer as it is for MOL on MOS.

Memory protection is hard to achieve without breaking the backward compatibility (apps do not ask for a memory access, they just do). That's one of the points the whole design was as chosen as it is. The unsecure ABox and the option for a secure future box (mumbling something about Q here - well, nobody knows what's going on with it, but theoretically...).

I don't know what the aproach by Hyperion, to give every task an own virtual (emulated) address space, is worth actually...

[ Edited by Zylesea on 2005/4/13 16:23 ]

Code:

I'm not talking about memory protection. I'm talking about providing Memory funcitons to the ABox that allow the coder to mark zones as illegal to raise exceptions (usefull for nallehpuh) or simply to remap memory areas (like remaping the first 8KBs in pther part of the memory)

You could do that with Thor's MMU library on Amiga 68k!

It doesn't compromise Q-Box design because you use a function and you don't use the MMU directly. You'll only be able to use the memory areas allowed by the Q-Box.

[ Edited by Crumb on 2005/4/13 21:13 ]

Quote:

Okay, bringing MP into the game I was actually one step too far, but IIRC it was said (in some MOL on MOS discussion) that the ABox won't see any MMU functions.

Yeah, Ralph Schmidt told me that he didn't want to allow MMU functions, but maybe I didn't explain correctly that I don't want direct access to the MMU but only a set of memory management functions like the ones I have told (memory context for each task/all the tasks, marking parts of memory as read-only/ilegal, remaping parts o memory).

Maybe MOS team will reconsider that decission. You will use functions so they can validate the memory areas that you mark...

The should add an option like the OS4 team to mark some memory areas as "private" so new programs can define areas that can't be overwritten by other tasks (of course, we wouldn't have full memory protection because tasks would continue using the same message system and old tasks wouldn't have their memory areas marked as private)

No.

Implementing private address spaces into ABox would be just silly. It only makes memory management more complex without gain.

Maybe low mem (0x0000 - 0x1fff) could be made writable/readable when Basilisk is running, but we will see... with current 1.4 kernels it is not possible.