[TUHS] V7/8086 I'm a bit behind the times here, but...

[TUHS] V7/8086 I'm a bit behind the times here, but...

[Steve Nickolas]

I already had some ideas, and when I saw something called "v7upgrade", a weird thought came to my head...

I'm wondering if any gurus out there would be able to point me in the general direction, as far as getting V7 stuff running on an 8086, 
perhaps a full V7 system.  Something like v7upgrade but including a kernel and bootloader.  I don't know.  Just musing...

My only experience with a "real" UNIX is either SunOS via telnet or PicoBSD.  I use RH8 Linux, FreeDOS ODIN 0.31 and Win98SE at 
home.  It would be interesting to play with V7 on one of my computers. :)

BTW I do have v7upgrade running on my Linux box - sweet! :}

-uso.

kirei-na pinku-na E-MAIL-saito
___________________________________________________________
Get your own Web-based E-mail Service at http://www.zzn.com

[TUHS] V7/8086 I'm a bit behind the times here, but...

[Peter Jeremy]

On 2003-Jul-01 18:16:27 -0400, Steve Nickolas <usagi.tsukino at pinku.zzn.com> wrote:
>I already had some ideas, and when I saw something called "v7upgrade", a weird thought came to my head...
>
>I'm wondering if any gurus out there would be able to point me in the
>general direction, as far as getting V7 stuff running on an 8086,
>perhaps a full V7 system.  Something like v7upgrade but including a
>kernel and bootloader.  I don't know.  Just musing...

I've also been considering this in the background ever since "ancient"
Unix became available.  At least one other person has posted that they
did actually do a suitable port.

Downsides:
- 8086 has no hardware protection.  You'd need to go to the 80286 to
  get any inter-process protection.
- 80x86 16-bit memory management is far more primitive than the PDP-11.
  Given the requirement for data and stack addresses to be in the same
  address space (assumed by virtually all C programs), the total data+
  heap+stack space required for a process must be pre-allocated and
  there is no protection between heap and stack.

The main reason I've never proceeded is the lack of a suitable C
compiler: 16-bit 80x86, open source, able to run in 64K I+D mode, and
generating half-way decent code.  (Without the last requirement, the
kernel and some of the larger userland utilities will probably be too
large).

Supporting overlays would add the requirement to handle mixed-mode
code, support segment overrides and 'far' objects.

Peter