Hi On Monday 13 June 2005 05:09 am, anarkhos@xxxxxxxxxxx wrote: > I've spared myself knowledge of such differences for the most part. The > whole situation is ridiculous IMO. Apple took a short-cut to avoid changing > their build tools and we're still paying for it. It doesn't even use the > proper PPC SYSV ABI. I always wondered why they hung on to the Mach-O binary format. About the only advantage it is has over ELF is support for fat binaries. Of course, now its obvious why they wanted fat binaries. ;-) > It certainly hasn't made it easy for the guys trying to port QEMU's ABI > emulation layer. Hmm. Yes. Rather them than me. ;-) > Like what? :) Well, the problem is how to call AmigaOS functions from the emulator. The solution at the moment is the 'stack-magic' code, which creates a new stack frame and thus makes it possible to swap to a new stack context to execute an AmigaOS call and then swap the stack back to resume the emulator code. It's kinds of like setjmp/longjmp with knobs on. ;-) A more portable method to do this, at least SYSV-compatible operating systems, might be to use makecontext/swapcontext. A better idea all together would be make the CPU emulation re-entrant, so all this nonsense shouldn't be necessary. There's still some way to go to achieve this yet. > BTW: You have less than a year to add PowerUp support ;) Hehe. It's on my list of things to do. > I'm sorry if I haven't been working on e-uae OS X, but I've been busy > working on Darwine since Apple's announcement. Learning some nitty-gritty > about OS X's event handling and graphics. Not too dissimilar to what e-uae > needs to squeeze out some better performance and compatibility. That's an interesting project. Good luck with it. > Anyway, thanks a lot Richard. I'll find some time to test your changes > later this week. You're welcome. Cheers, Rich