[kaffe] Re: config/i386/common.h vs md.h - Patch Attached
zuse at libero.it
Thu Jul 29 01:25:42 PDT 2004
> Yes, ppc shares its 'sysdepCallMethod.h' between several os'es. But
> if you once look into the code, there are so many 'ifdef' clauses.
> And also, I am wondering how many os'es for ppc is now currently
> supported. Maybe only linux and darwin are active. Are there any
> active developers for aix/machten/netbsd1 (on ppc)?
I don't count as a developer :-)
I tried netbsd2 and it worked very well the last time I tried. I would
expect netbsd1 to fail with pthreads (there is no real distinction among
netbsd's in kaffe's code)
I don't actively use Machten any longer but if there is interest I could
put it on an external disk and give it a spin.
I'd have more interest in Aix, which I have. But I have the predecessor
of the PowerPC processor, the original POWER. If kaffe's code is generic
enough, it will probably run, since a subset of both POWER and PowerPC
code is binary compatible. GCC code generated without arch.
specification satisfies this.
I should hack configure stuff to recognize my platform as supported, but
I am no expert with this.
>>Why is m68k so different?
>>From ABI's viewpoint, linux and netbsd are same. But they do different
> packing for structure and so on, and some misterious behavior (at
> least for me) of inlined sysdepCallMethod on netbsd makes it hard
> to keep only one sysdepCallMethod. Much worse, amigaos uses its
> own ABI. If Tony Wyatt is still on this list, he can add more comment
> about that.
I wonder how OpenBSD falls in the line...
PS: Kyio when do you submit your improvments? I'm eager to test them on
OpenBSD and Linux
More information about the kaffe