On Sep 26, 2013, at 8:34 AM, Jonathan Katz <jon at jonworld.com> wrote:
OK, I'll give up. However, that screenshot which
someone had from
"virtual PC" was interesting. Maybe I'll ferret that guy out.
After I stopped working on AIX PS/2, I recall that there was some effort
to make it run on AT class machines. At this point I don't know if that
ever was shipped or what version (it could have been after 1.3) it was.
On Thu, Sep 26, 2013 at 11:24 AM, Guy Sotomayor <ggs at shiresoft.com> wrote:
On Sep 26, 2013, at 6:49 AM, Jonathan Katz <jon at jonworld.com> wrote:
On Thu, Sep 26, 2013 at 8:13 AM, Joost van de
Griek <gyorpb at gmail.com> wrote:
> How hardware-specific was it? Is there a
chance of it running in a VM?
<http://en.wikipedia.org/wiki/File:Aixps2.gif>
"Challenge Accepted!"
I'm playing with VirtualBox but not getting anywhere. Kernel will
load, install disk 1 will load, but depending upon my setup (ESDI boot
disk with IDE controller, or SCSI boot disk with SCSI controller) I
get to different places. With ESDI/IDE combo I get to a "broken"
single-user shell that doesn't have a filesystem at all. With the SCSI
setup I get errors about lack of token ring, lack of devices, and it
halts/panics.
I'm wondering if the Virtual PC setup in the screenshot had some of
the more specific devices emulated (or patched through) which the
kernel is looking for (model of SCSI controller, etc.)
I bumped the RAM down to 32M and have a 1GB virtual disk.
It's going to be a real challenge...it's likely that you'd have to add a lot
of support to the VM for the disks. The ESDI and SCSI subsystems
were not anything like what the rest of the PC industry did in that space.
I don't remember exactly what the ESDI interface looked like except
that it provided an LBA interface. I don't recall what it did about bad
blocks.
The SCSI interface was based upon IBM's SCB (sub-system control
block) architecture. There were some "funnies" in the SCSI interface
but it followed the SCB model. All control blocks were built in memory
and you gave the HW the address of the first control block to start with.
The control blocks could be chained and you could specify which
control blocks would generate interrupts when they completed. All status
and error information was also contained within the control blocks.
TTFN - Guy
--
-Jon
Jonathan Katz, Indianapolis, IN.