From: Jim Ursetto <ursetto@uiuc.edu>
  To  : ggi-develop@eskimo.com
  Date: Sun, 13 Sep 1998 19:04:45 -0500

Re: Multi-Adaptor Compatibilty Matrix

At 06:26 PM on 1998 September 13, Brian Julin did write:

>                                                     Also, when you 
> have a chipset in a laptop and plug it into a docking station with 
> another VGA, sometimes the laptop VGA is kept active but is residing 
> at the monochrome IO ports; most of these cases (the WD drivers will 
> have multiheading by exploiting this) aren't going to be using the 
> IBM/vga driver but a more specific driver tho.  You have to dance
> circles around resource conmtentions for the 0x3cx PIO range.
> 
> See VGADOC's VGAREGS.TXT and look at register 0x3c2 bit 0.

Assume you have two vga-compatible PCI cards with linear framebuffers which
cannot relocate their i/o ports (no MMIO), and the BIOS disables one.  If
you wake it up, does it crash the computer?  What if both cards have a way
to ignore writes and reads to their I/O ports, i.e.  turning off the i/o
space?  And what if you have drivers which shut off the i/o space and
enable it only long enough to set the registers, for example during a mode
set.  _And_, what if you map one card into the monochrome I/O space, so
the ports only need to be disabled when writing to the 0x3cx range?  It's
farfetched, but I was curious to know if there would actually be a
hardware problem such as bus contention when doing this, or if it's even
remotely possible.

Or have you another way to avoid contention for the 0x3cx range?

Jim