----- Original Message -----
From: "Noji Ratzlaff"
To: "Gabor FUNK"
Sent: Wednesday, February 28, 2007 1:27 AM
Subject: Re: Fw: GA-8I945GZME-RH (this is thesuccessorottheotherone...)


So far, this is what I found, and it's not much.

When the driver loads for the SATA channel, everything appears to work
fine until the first Read request is sent to the hard drive.  The Read
request completes as expected and without problems.  After that, any
subsequent Reads in Protected Mode work fine, which probably explains
why it works in NetWare 6.5 (6.5 remains in Protected Mode all the
time).  But the system seems to hang up when the first Read takes place
in Real Mode after that (such as reading a file on the DOS partition.)
The NetWare Loader is in control of switching to and reading in Real
Mode.

When the first Real Mode read occurs, the registers seem to be set up
correctly, and the Control Register reflects the fact that a command has
been written and the controller is waiting for the interrupt to fire.
The problem is that the interrupt never fires, and so the NetWare Loader
gets into this endless loop waiting for the interrupt.  The loop has a
timeout, which allows you to jump into the debugger momentarily, but
then starts the loop over again forever.

Why the interrupt doesn't fire is a mystery.  In my experience,
interrupts don't just not fire...they DO fire, but it's possible that
something else has claimed the interrupt, and so the SATA controller is
left waiting for an interrupt that has been serviced by another program.
 NetWare doesn't have control over these Real Mode interrupts, and so
there's no way to catch the culprit in the driver.

Because DOS/BIOS is in control of the Real Mode side of things, I
decided to remove CONFIG.SYS and AUTOEXEC.BAT (and therefore all the
programs they might be running) from the picture.  That made no
difference.  Then I decided to try and reproduce the problem with DR DOS
instead of MS DOS.  The problem still exists, which seems to rule out
DOS as the culprit.

This whole issue smells like a problem with the BIOS properly handling
Real Mode, something that BIOS makers have had to worry about less and
less these days.  Where can I go to get an update for the Gigabyte
(Award) BIOS, just in case?  My system has BIOS version Award Modular
BIOS v6.00PG, and I have requested an upgrade from www.esupport.com.

Noji