Re: North bridge functions

[Marco Brambilla <marco-tpa.brambilla@st.com>]

> Question 2:
> 
> If I created a diagnostic device for the PCI bus that tried to remain
> transparent, i.e. didn't respond to config commands and didn't have a memory or
> I/O map, could it be a bus master at times.  What would happen if it asserted
> REQ#?  Would it receive a GNT# from the arbiter?

The problem may be that some motherboards remove the clock from
non-populated slots.
I saw at least one MB+BIOS combination where initially the clock was
sent to every slot, but then removed to all the connectors where the
device was not responding to config accesses, independent on the PRSNT
pins value.

I would suggest implementing a device which actually answers at least to
config accesses, perhaps with all the BARS set to all zeroes, to disable
allocating resources.

Also, if you don't implement any mem or io space then you need some
other wau (serial, usb) to send commands to your module, but I'm sure
you already know this.

Ciao, Marco.

-- 
-------------------------------------------------------
 Marco BRAMBILLA   
            
 STMicroelectronics
 Via C. Olivetti, 2          
 20041 Agrate Brianza (MI)        
 ITALY                       
 
 TPA - Wireline Communications Division
 tel   : +39 039 603.5797   (ST Agrate - TINA 050 5797)
 mailto:marco-tpa.brambilla@st.com
-------------------------------------------------------