Re: Multifunction Add-In Cards and PCI BIOS Enumeration

["Peter Marek" <peter.marek@marekmicro.de>]

Hi all,

we have successfully implemented Multi function devices in Quick Logic ESPs.
The behaviour you're seeing can be avoided. There are minor modifications to
the PCI config space VHDL module.

Regards,

Peter Marek
General Director
MarekMicro GmbH
tel.: 049-9621-9732-110
fax: 049-9621-9732-199
www.marekmicro.de

----- Original Message -----
From: "Steve Blightman" <steve@alacritech.com>
To: "Scott Davis" <sdavis@Futurex.com>
Cc: <pci-sig@znyx.com>
Sent: Friday, May 17, 2002 7:57 PM
Subject: Re: Multifunction Add-In Cards and PCI BIOS Enumeration


> I've seen this behavior before - if the BIOS sees the multi-function bit
for the
> first device, ii checks all of the other 7 possible devices.  If the
hardware
> responds with anything other than Master Abort, the BIOS assumes these
devices
> exist.  I don't know anything about the QuickLogic device, but I suspect
it is
> returning data for the other six configuration spaces (probably zeros?).
Maybe
> someone from QuickLogic can confirm this for you.  I know of no good way
around
> it other than to get your card not to respond to the nonexistent
configuration
> space accesses  (generating Master Abort).  I suspect, by the way, your
card
> will still work fine in the system - it's just an annoyance probably.
>
> Steve
> .
> Scott Davis wrote:
>
> > I've designed a multifunction PCI Add-In card that uses the QuickLogic
> > QL5130 ESP for the bus interface.  The card has two functions so I've
> > created two configuration register sets -- one for each function.  When
the
> > PC BIOS enumerates the bus, it reports that the card has been found and
then
> > lists 8 functions -- the first two functions correctly and then 6
unknown
> > functions.  I've set the MSB of the Header byte for both functions.  My
> > questions are as follows:
> >
> > Has anyone seen this type of behavior before?
> > How do I suppress the listing of the 6 unknown functions?
> > Should I be setting the MSB of both header bytes?
> >
> > Thanks,
> >
> > Scott