X-Authentication-Warning:	delorie.com: mail set sender to djgpp-bounces using -f
From:	Gianfranco Boggio-Togna <gbt AT despammed DOT com>
Newsgroups:	comp.os.msdos.djgpp
Subject:	"Protected mode not available"
Date:	Tue, 04 Jan 2005 18:32 +0100
Lines:	61
Message-ID:	<3402a2F43rr8fU1@individual.net>
X-Trace:	individual.net WKjRFvB2Z/46K33Ln78QRQqzU+3P5eBX9RsLTioJ9vjjv6w28=
To:	djgpp AT delorie DOT com
DJ-Gateway:	from newsgroup comp.os.msdos.djgpp
Reply-To:	djgpp AT delorie DOT com

This post is somewhat OT as I have no problems with the DJGPP software
(except for the minor one of not being able to run it...) but I know
that among the readers of the newsgroup are some of the best experts
on real/protected mode switching.

A couple of years ago I decided to retire my Pentium 166 and get a
newer and faster machine. I had some misgivings about the move as I
wanted to retain DOS (actually Novell DOS7) as my main system, with
Windows 98SE for accessing the Internet and for support of USB devices
(I use XOSL to switch between the two). I asked my dealer to mount
an Intel motherboard (D845GEBV2), as I thought this would offer the
best level of compatibility with the original 386 architecture. In
fact, after some considerable struggles (I had to install a second disk
and an old PCI video card), I managed to get an almost exact replica
of the configuration I had on the P166, except for one thing. I had
not been aware of the existence of ACPI and I found, to my dismay,
that the only way to set the system to power saving mode was to boot
into Windows 98 and switch to standby.

Except for this nuisance, everything went smoothly for several
months. Then, all of a sudden, I started getting "Protected mode not
available" messages whenever I tried to run a program compiled with
DJGPP or Free Pascal. This was not a quirk of CWSDPMI, as the DPMI
manager of QEMM gave the same message. Clearly the machine was not
booting into real mode. Windows 3.1, however, would start and the
DJGPP software would run normally in a DOS box. But something was
wrong with the disk cache, with a hit rate below 20%: most of the time
I run TeX (Eli Zaretskii's excellent port of Web2C) and after a few
minutes the cache hit rate should get above 95% and stay there.

Some weeks later I realized, quite by chance, that after a warm boot
(the Norton Utilities BE REBOOT) the machine restarts in real mode.
I modified CONFIG.SYS and AUTOEXEC.BAT so that I can force the warm
reboot and I now have again a more or less normal system. Having to do
a double boot is tiresome but I can live with it: it just means a few
seconds' delay. I feel, however, rather nervous about the possibility
of ACPI playing some other trick (as an old programmer, I think that
allowing software to make changes to the firmware that cannot be
nullified by removing power is sheer stupidity).

I would like to have a way of restoring the motherboard to its
original "factory" state. Reloading the BIOS would probably do the
trick, but it is far too risky. The BIOS setup menu has no such
option nor is there any relevant information in the motherboard
documentation. To be fair, Intel states quite clearly that the board
requires Microsoft Windows, in a version not earlier than 98SE.

I wonder if there is any program or procedure that I might use to
reset the board or, at least, to diagnose the problem more precisely.
Would a recent version of Linux offer any facilities in this area?
I have an old distribution (which I hardly ever use) which has
apparently no support for ACPI, but I could easily upgrade it.

Ideally, I would like to turn off power saving in Windows98. But for
this I would need a program that can enable power saving in DOS. I
have done some extensive searching and I have not come across any
mention of such a program.

-- 
   Gianfranco Boggio-Togna
   Milano (Italy)

- Raw text -