80386EX

[Paul Edwards]

I didn't have the words to describe it
before, but what I would like for
PDOS/386 is for it to be able to run on
a variant of the 80386 that only
supports 32-bit mode. Perhaps the
80386EX:

https://en.wikipedia.org/wiki/Intel_80386EX

It only supports 64 MiB of memory, enough
for me to run the GCC C compiler.

Since there will be no real mode available,
the BIOS calls that I currently use will
need to have 32-bit versions, and PDOS/386
will not do a transition to real mode to
do the interrupt. For that reason it would
be good if the 32-bit interrupts were
self-contained instead of referencing a
save area containing old and new registers,
although that way should still work.

BFN. Paul.

[Rick C. Hodgin]

AMD has a similar version that can access
the full 4 GB range, but it can also clock
down to 0 MHz, as it is a fully static part.

If you are wanting to build a machine which
can be completely customizable, including
going into sleep modes, throttling down when
it has minimum workloads, etc., then you may
want to look at AMD's version.

https://en.wikipedia.org/wiki/Am386#Am386DE_data

--
Rick C. Hodgin

[Rod Pemberton]

On Thu, 13 Sep 2018 06:46:58 -0700 (PDT)
Paul Edwards <muta...@gmail.com> wrote:

> I didn't have the words to describe it
> before, but what I would like for
> PDOS/386 is for it to be able to run on
> a variant of the 80386 that only
> supports 32-bit mode.
>

> [snip]

>
> Since there will be no real mode available,
> the BIOS calls that I currently use will
> need to have 32-bit versions, and PDOS/386
> will not do a transition to real mode to
> do the interrupt. For that reason it would
> be good if the 32-bit interrupts were
> self-contained instead of referencing a
> save area containing old and new registers,
> although that way should still work.
>

Why can't you just /not/ use RM on a normal
386 processor or later? ... This should be
the same thing, shouldn't it? Is there any
difference between the 32-bit mode on a 386
and a 386EX? Either way, you need to replace
the 16-bit BIOS (or DOS) calls with your own
routines, yes? Where would anyone get access
to a 386EX anyways? So, I think introducing
the 386EX is just confusing the issue.


Rod Pemberton
--
Bitcoin is merely a pump-and-dump scam driven by a perpetual Ponzi
scheme.

[Paul Edwards]

On Tuesday, 18 September 2018 11:42:12 UTC+10, Rod Pemberton wrote:

> Why can't you just /not/ use RM on a normal
> 386 processor or later? ... This should be
> the same thing, shouldn't it? Is there any
> difference between the 32-bit mode on a 386
> and a 386EX? Either way, you need to replace
> the 16-bit BIOS (or DOS) calls with your own
> routines, yes? Where would anyone get access
> to a 386EX anyways? So, I think introducing
> the 386EX is just confusing the issue.

Well an 80386 starts in RM. A 386EX
runs purely in 32-bit mode I believe.

I don't know if the 80386EX is still sold.

But I think my operating system should
eventually cater for a pure 32-bit CPU,
with a 32-bit BIOS.

BFN. Paul.

[Rick C. Hodgin]

On 9/17/2018 10:03 PM, Paul Edwards wrote:
> On Tuesday, 18 September 2018 11:42:12 UTC+10, Rod Pemberton wrote:
>
>> Why can't you just /not/ use RM on a normal
>> 386 processor or later? ... This should be
>> the same thing, shouldn't it? Is there any
>> difference between the 32-bit mode on a 386
>> and a 386EX? Either way, you need to replace
>> the 16-bit BIOS (or DOS) calls with your own
>> routines, yes? Where would anyone get access
>> to a 386EX anyways? So, I think introducing
>> the 386EX is just confusing the issue.
>
> Well an 80386 starts in RM. A 386EX
> runs purely in 32-bit mode I believe.
>
> I don't know if the 80386EX is still sold.

It isn't being manufactured anymore, but there
may still be some for sale from official dis-
tributors, and there are some available on eBay
from time to time.

> But I think my operating system should
> eventually cater for a pure 32-bit CPU,
> with a 32-bit BIOS.

Who will create / support the 32-bit BIOS?

--
Rick C. Hodgin

[Paul Edwards]

On Tuesday, 18 September 2018 23:32:43 UTC+10, Rick C. Hodgin wrote:

> > But I think my operating system should
> > eventually cater for a pure 32-bit CPU,
> > with a 32-bit BIOS.
>
> Who will create / support the 32-bit BIOS?

I think that would be an extension of the
PDOS project at some point. It would be
good to have a public domain BIOS.

BFN. Paul.

[Rick C. Hodgin]

The generic abilities would be nice, but most
BIOS authors have to deal with peculiarities
of the physical hardware implementation. On
this board this port is used to access this
hardware, and because it has this chip, it
needs this software protocol, which is differ-
ent from the others.

It's something to consider. Writing a true
generic BIOS would be great (for a virtual
machine or a known piece of hardware like a
virtual machine), but writing one to replace
whatever hardware you'll find on a device,
that's what device drivers are for, and they
take the place of 32-bit BIOS, and are the
installable drivers used in a config.sys
type file.

Something to think about.

I admire our goal and focus, but the way.
It is nice to see some people so resolved
on what they want to do.

--
Rick C. Hodgin

[Paul Edwards]

On Tuesday, 18 September 2018 23:57:56 UTC+10, Rick C. Hodgin wrote:

> It's something to consider. Writing a true
> generic BIOS would be great (for a virtual
> machine or a known piece of hardware like a
> virtual machine), but writing one to replace
> whatever hardware you'll find on a device,
> that's what device drivers are for, and they
> take the place of 32-bit BIOS, and are the
> installable drivers used in a config.sys
> type file.

Thanks. I didn't think of that.

But if a machine is using the 80386EX,
it would need a 32-bit BIOS for I/O
at least until the device driver is
loaded, right?

BFN. Paul.

[Rick C. Hodgin]

Yes. And on an 80386EX, you'd be using a
BIOS design that's already established for
whatever the pmode settings are at bootup.
Just like the way BIOS ROM exists at known
addresses and the system calls a specific
address at startup, which begins executing
that code.

But remember:

The purpose of BIOS is to perform the various
tasks in a uniform way, so you'd either be
supporting the existing BIOS functions that
are well-documented and in use, or you'd be
creating a stripped down version which meets
only your needs / goals for your target sys-
tem.

For whatever design motherboard you use, you'd
have to support its quirks, and it would ess-
entially pigeonhole your PDOS design to such a
reference system. You wouldn't be able to
have it work generically with any kind of mo-
therboard that the 80386EX would support.

If that constraint is okay with you and your
goals, then it would work nicely. You'd
actually be doing a lot of what Apple does
by limiting the hardware they support, mak-
ing their systems more stable than generic
Windows systems which have to rely on 3rd
party drivers to handle things, some of
which are more buggy than others.

--
Rick C. Hodgin

[Rod Pemberton]

On Tue, 18 Sep 2018 06:47:19 -0700 (PDT)
Paul Edwards <muta...@gmail.com> wrote:

> On Tuesday, 18 September 2018 23:32:43 UTC+10, Rick wrote:

> > > But I think my operating system should
> > > eventually cater for a pure 32-bit CPU,
> > > with a 32-bit BIOS.
> >
> > Who will create / support the 32-bit BIOS?
>
> I think that would be an extension of the
> PDOS project at some point. It would be
> good to have a public domain BIOS.
>

It's probably not PD, but it's the only open-source BIOS I'm aware of.
It used to be called LinuxBIOS, but is now called coreboot. I posted
information I learned about two of their utilities over the years
(2007, 2007, 2015).

One is called FILO which is a BIOS independent 32-bit PM boot loader.
FILO was constructed from the GRUB boot loader which supports ext2,
FAT, iso9660, Multi-boot, ELF, and extended with an IDE driver, USB,
and SATA.

Another is ADLO which is a 32-bit PM ROM image loader for a 16-bit
BIOS image. ADLO enables memory shadowing, motherboard video routines,
and loads a modified Boch's BIOS image. Their BIOS image is 16-bit and
is the software only portion, i.e., no hardware initialization.

https://en.wikipedia.org/wiki/Coreboot
https://www.coreboot.org/

The Wikipedia page mentions some other open source BIOSes.

I don't see any indication that any of those BIOSes are 32-bit, although
maybe the UEFI clone is?

If you'd like to read more, this is an old post. Although, I'm sure
they've improved this project much or changed a lot since 2007.
https://groups.google.com/d/msg/alt.os.development/OxIRneBBVqw/0Szve5c0aqIJ