I only see it as being a set of parts,

Kernel module (primary binder)
68K Emulator (First native threaded object)
Picasso96 UAE driver backend (VNC Framebuffer)
BootImage (single unit block driver image for booting AOS)

additional items would have to provide a firmware module for the KS
along with the Kernel module for the backend

Ideas for these would be Bsdsockets/AHI(OSS/ALSA and NAS for network
forwarding)

breaking it up simplifies matters along with using existing Firmware
modules
(UAE's Picasso96 driver) (UAE's Bsdsockets Patch)...

The only real change is needing to have the scsi.device in Kickstart
replaced
otherwise I would need a minimal emulation of Custom Hardware as a
second
kernel thread with events triggering across the threads using some kind
of signalling.

using a kernel framebuffer output I can dig out the code for making use
of a VT
and have the Amiga Display run against the Linux Kernels Framebuffer as
AMIthlon did
which would actually allow some pretty fancy graphics updates...
along with provide an option for some nifty 3D hacking too (would need
a library)

by keeping it simple and down to bare minimum "KS patch" for onload
patching
of known kickstarts (patches become distributable) along with the
kernel module
and a document for usage and extending it...

which is kind of the point... make use of Linux AND AmigaOS...
which is something I have been trying to keep doing for a long time
apart
from one drawback I still have a bundle of original CD's that I want to
make use of
(this year I lost a HDD with original AOS install)

and doing the kernel module with a front-end like VirtualBox seems kinda
fun
since I would start with an existing app (qemu-m68k) and modify to suit
from there
with the various features

I already have prior experience with kernel modules and was looking at
the KVM and
Xen kernel modules for how to support VM'ing the AOS properly...

if I can do it for a PPC edition and buy AOS4 and keep all the software
I currently use
I think all the better

Jeremy

On Sun, 2008-12-28 at 17:35 +0100, Mario wrote:
> Hi Jeremy
>
> There is a lot of work needed to implement your idea. Especially that
> only one person will be involved in development. Maybe you should
> consider a little different idea, like: transparent AOS 3.x 68k
> emulation inside Linux (running Amiga's apps next to Linux ones just
> like they were all native and build for one OS) or only PPC emulation
> in your VM project ("buy and run MOS on PC")?
> I think you need to judge your goals first, just not to waste your
> time. Is making another Amithlon reasonable? Even with more hardware
> support? New 68k programs are rare now (and I don't think it will
> change in next years). How about getting involved in AROS?
>
> Regards,
> Mario
>
>
>
>
>