[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: xt?



"marsh" <marsh@afn.org> writes:
> jghasler@win.bright.net wrote in article
> <[🔎] 87u3mpdj1w.fsfjgh@win.bright.net>...
> > Craig writes:
> > > No Linux will ever work on an xt or a 286.  They are missing neccessary
> > > bits of hardware called a MMU which protects the memory.  A 386SX is
> the
> > > minimum.
> > 
> > The 8088 used in the XT is lacking an MMU, but the 80286 used in the AT
> is
> > not.  Several versions of Unix were available for the AT.
>
> The differences in the memory management architechture on the 286 and the
> 386 are major.  Virtual memory paging and segments larger than 64K just two
> important things that the 386 has that are missing on the 286.  I'm not
> saying that the 286 is not a superneato processor which was quite
> impressive (16?, 17? ) years ago, but there is a reason that The Great One
> (LT) chose the 386 for his initial experiments.

More like 10 years ago;  the IBM PC was introduced 16 years ago.

> One could write something for the 286 based on the "Linux Philosophy" and
> even call it the Linux kernel, but its executable format is not going to
> look a lot like ELF.
> 
> Someone correct me if I'm wrong, but I don't think AT&T System V ever ran
> on 286s.  Xenix, Minix, and Coherent are the main derivatives I can think
> of that ran on 286s.

Okay, you are wrong:) I used to own a version of ATT System V unix
(from Microport) that I ran on a 80286 system.  Of course most people
define System V as being V.3, and that version was the original V.0
(or V.2?).  The major problem with it was that most software available
was BSD software, and of course all BSD programmers knew that int,
pointers and longs were all 32 bits.  Unfortunately the 286 version of
unix used ints as 16 bits, so many programs had to be modified.  The
system did use the 286 protected mode, so it had full memory
management.  You could use more than 64K of memory, but it was a pain
since you had to compile using Intel large model.

-- 
Carl Johnson		carlj@peak.org


Reply to: