Dell PowerEdge 1400 server and some other Dual-Processor PIII 1GHz.
Windows 2000 Server

I have 2 3COM PCI faxmodems that I have installed into them. Both do the
same: The Blue Screen on startup.

I have disabled COM1, 2 and LPT1 in my server's BIOS, to allow enough IRQ's
to be available for two 3COM PCI internal modems in the machine. In the
BIOS this appears correctly and it reports the devices. However, when W2K
(fully patched with all SP's and hotfixes) starts it detects com1, com2 and
lpt1 anyway, and then reports that the second modem has a problem. When I
force the IRQ's, after disabling the incorrectly detected com and lpt ports,
the machine crashes with a BSOD.

I moved the modems to a totally different server (a dual 1GHz PIII server)
and the problem is exactly the same, so my suspicion that the BIOS is the
problem, was proven wrong. It's actually windows 2000 server that stuffs
this up.

So my question is: Is it possible to tell W2K that no COM or LPT ports are
installed? (or alternatively I don't want auto-detection). I think then
I'd be able to manually config the modems to work correctly. Of course then
I would not want the modems autodetected either.

I have googled and searches microsoft for anything on this behaviour, but
come up emptyhanded.

Please help!

regards

Roland Giesler

Roland,

I would reset the bios config for the COM and LPT ports to default.
I would also check for BIOS updates - it is / used to be common on 'fast to
market' computers for Disable functions in the BIOS to not be implemented
correctly particularly for on board devices.

You will get a BSOD if anything interferes with parallel.sys touching its
LPT ports - IE if the driver believes there is an LPT1 and it is not
accessible you will get a BSOD - Try a Bootlog, Try disabling parallel.sys.

I would remove the 3COM cards and get the system to boot correctly. Go to
the 3COM site and check for W2K compatability, drivers, then double check
installation instructions. Assuming that they do not say that W2K is
unsupported, I would add 1 card at a time as per their instructions. Get the
first card going correctly then check the method - it may need adjusting -
for adding the second card.

The design of PCI is such that IRQ's can be shared. This is normal. There
are more myths about IRQ sharing than facts, so it is easy to summarise. If
pure PCI (no ISA slots) system has an issue that can be proven to be IRQ
related then there is either a defect in the design of the PCI card (rare),
or bugs in the device driver for the card. Simple as that. The trouble with
this is that many people in their urgency in getting systems to work will
willfully work around buggy device drivers by forcing separate IRQ's and
specific IRQ assignments - this does no one any favours as it just
encourages the product vendros to persist in releasing buggy drivers.

What is the STOP code you get and what are the device drivers visible at the
top of the screen?

- Tim



"Roland Giesler" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> Dell PowerEdge 1400 server and some other Dual-Processor PIII 1GHz.
> Windows 2000 Server
>
> I have 2 3COM PCI faxmodems that I have installed into them. Both do the
> same: The Blue Screen on startup.
>
> I have disabled COM1, 2 and LPT1 in my server's BIOS, to allow enough
> IRQ's to be available for two 3COM PCI internal modems in the machine. In
> the BIOS this appears correctly and it reports the devices. However, when
> W2K (fully patched with all SP's and hotfixes) starts it detects com1,
> com2 and lpt1 anyway, and then reports that the second modem has a
> problem. When I force the IRQ's, after disabling the incorrectly detected
> com and lpt ports, the machine crashes with a BSOD.
>
> I moved the modems to a totally different server (a dual 1GHz PIII server)
> and the problem is exactly the same, so my suspicion that the BIOS is the
> problem, was proven wrong. It's actually windows 2000 server that stuffs
> this up.
>
> So my question is: Is it possible to tell W2K that no COM or LPT ports
> are installed? (or alternatively I don't want auto-detection). I think
> then I'd be able to manually config the modems to work correctly. Of
> course then I would not want the modems autodetected either.
>
> I have googled and searches microsoft for anything on this behaviour, but
> come up emptyhanded.
>
> Please help!
>
> regards
>
> Roland Giesler
>
>
>

"Tim" <Tim@NoSpam> wrote in message
news:uctgnX$(E-Mail Removed)...
>
> The design of PCI is such that IRQ's can be shared. This is normal. There
> are more myths about IRQ sharing than facts, so it is easy to summarise. If
> pure PCI (no ISA slots) system has an issue that can be proven to be IRQ
> related then there is either a defect in the design of the PCI card (rare),
> or bugs in the device driver for the card. Simple as that.

Or defects or limitations in the motherboard design. Some of the earlier
motherboards with more then 4 PCI slots share some of the other control lines
(such as bus mastering) and have conflicts between some slots. On one
motherboard I have, conflicts between slots make configuring boards very
chalenging. It's currently running with cards in all of its slots and there's
only one configuration which works. Swap any cards and something will stop
working.

--
-GJC [MS Windows SDK MVP]
-Software Consultant (Embedded systems and Real Time Controls)
- http://www.mvps.org/ArcaneIncantations/consulting.htm
-(E-Mail Removed)

Gary,

Not surprising.
What make, model, year and so on was the motherbaord?

I've been very lucky over the years - none of my customers have made *bad*
hardware purchases - some not so hot, and in one case it was one machine
that was a Big Name Brand POS................ that I would never have
recommended.

- Tim


"Gary Chanson" <(E-Mail Removed)> wrote in message
news:%(E-Mail Removed)...
>
> "Tim" <Tim@NoSpam> wrote in message
> news:uctgnX$(E-Mail Removed)...
>>
>> The design of PCI is such that IRQ's can be shared. This is normal. There
>> are more myths about IRQ sharing than facts, so it is easy to summarise.
>> If
>> pure PCI (no ISA slots) system has an issue that can be proven to be IRQ
>> related then there is either a defect in the design of the PCI card
>> (rare),
>> or bugs in the device driver for the card. Simple as that.
>
> Or defects or limitations in the motherboard design. Some of the
> earlier
> motherboards with more then 4 PCI slots share some of the other control
> lines
> (such as bus mastering) and have conflicts between some slots. On one
> motherboard I have, conflicts between slots make configuring boards very
> chalenging. It's currently running with cards in all of its slots and
> there's
> only one configuration which works. Swap any cards and something will
> stop
> working.
>
> --
> -GJC [MS Windows SDK MVP]
> -Software Consultant (Embedded systems and Real Time Controls)
> - http://www.mvps.org/ArcaneIncantations/consulting.htm
> -(E-Mail Removed)
>

"Tim" <Tim@NoSpam> wrote in message
news:%(E-Mail Removed)...
> Gary,
>
> Not surprising.
> What make, model, year and so on was the motherbaord?

It's a motherboard from several years ago, an ABit BX-133. I expect that
you'd find similar situations with other motherboards of the same vintage
because the limitations were set by the chipsets available at the time (not
enough control lines).

> I've been very lucky over the years - none of my customers have made *bad*
> hardware purchases - some not so hot, and in one case it was one machine
> that was a Big Name Brand POS................ that I would never have
> recommended.

I avoid big name brand PCs. I always find them to be more problematic
then the generic clones. They make poor component choices and add too many
proprietary features.

--
-GJC [MS Windows SDK MVP]
-Software Consultant (Embedded systems and Real Time Controls)
- http://www.mvps.org/ArcaneIncantations/consulting.htm
-(E-Mail Removed)