Re: [OpenBIOS] Why protected mode?

[Timothy J. Massey]

On Thu, 15 Jul 1999 12:31:48 -0400 (EDT), Kevin P Lawton wrote:
> Also IMHO, by far their greatest mistake in history
> was when they went to a new protected mode environment
> in the 286, they carried forward the segmentation
> architecture.

But they had to.  The 286 is a 16-bit processor.  Without segmented memory,
you would be limited to 64k of memory...  Leaving protected mode out of it,
the rest of the 286 was merely an evolution of the 80186:  a change from a
20-bit memory bus to a 24-bit.  That's the only real change.  Of course,
there *is* protected mode...

Only with a 32-bit processor can you realistically eliminate segmented
memory.

Tim Massey

-
To Unsubscribe: send mail to [e-mail]
with "unsubscribe openbios" in the body of the message

Why protected mode?

[Matthias Wächter]

On Thu, 15 Jul 1999, Timothy J. Massey wrote:
> Only with a 32-bit processor can you realistically eliminate segmented
> memory.

Don't say that: There are some Xeon motherboards around (well, most of
them are sold in complete supported systems) supporting 16GB of memory and
more. Since you need more than 32 bits to address this space, they had to
design work-arounds like segmentation (I don't know anything more about
it). Even worse: The switching is done externally, by the chipset...
> Tim Massey

Sehr Wus,
- Matthias

Why protected mode?

> But they had to.  The 286 is a 16-bit processor.  Without segmented memory,
> you would be limited to 64k of memory...  Leaving protected mode out of it,
> the rest of the 286 was merely an evolution of the 80186:  a change from a
> 20-bit memory bus to a 24-bit.  That's the only real change.  Of course,
> there *is* protected mode...

It didn't have to be a 16bit processor, if they went
to a new architecture.  They could have made it 64
if they wanted to.  You don't need segmented memory
to access > 64k if your architecture is 32 or 64bit.

Should have dropped real mode, had the CPU start
out in protected mode, and today we'd all have
protected mode only BIOSs.  Oh well, such is life.

-Kevin
-
To Unsubscribe: send mail to [e-mail]
with "unsubscribe openbios" in the body of the message

Why protected mode?

[Maciej W. Rozycki]

On Thu, 15 Jul 1999, [iso-8859-1] Matthias Wächter wrote:
> Don't say that: There are some Xeon motherboards around (well, most of
> them are sold in complete supported systems) supporting 16GB of memory and
> more. Since you need more than 32 bits to address this space, they had to
> design work-arounds like segmentation (I don't know anything more about
> it). Even worse: The switching is done externally, by the chipset...

You haven't read Intel processor specifications, have you?

Since PPro ia32 processors support the so called Physical Address
Extensions which for current processors allow to address up to 64 GB of
memory.  The limitation comes from the limited number of processor's
address lines which currently support 36-bit addressing.

Using Physical Address Extensions the CPU may address memory above 4 GB
using C-mode paging which needs a special setup of page tables; in general
after setting the PAE bit of CR4, paging becomes three-level deep and CR3
becomes PDBR (Page Directory Base Register).  Page directory and table
entries are now 64-bit.  You have a choice of 4 kB and 2 MB page sizes in
this mode.  See Intel documentation for details.

Actually A-step Pentiums also included this feature but probably due to
bugs in the implementation the feature has been removed for B-step and
later chips.  A-step chips were never available to the market.

Since PII Xeon, some further extensions were added -- see appropriate
docs.