From: teunis <teunis@computersupportcentre.com>
  To  : ggi-develop@eskimo.com
  Date: Fri, 26 Feb 1999 10:27:13 -0700 (MST)

Re: Driver internal structure

On Thu, 25 Feb 1999, Brian Julin wrote:

[clip]
> P.S. Would there be any advantage to having a "bus" submodule to account
> for buswidth/timing (e.g. even just an informational submodule so the 
> chipset driver can set wait states and such) and other system-bus features 
> like where legal MMIO windows can go, or are the current facilities for a 
> generalized bus API that are being developed good enough?  Come to think
> of it, such a module may be kinda critical for those lame systems where
> the video card uses main memory; I've seen those crash due to bad timing
> configurations in the CMOS.

Yes - this would be -good-

I suspect some probs with ViRGE driver may be due to MMIO timing...  (past
bugs were and I could prove them but they are fixed now :)

Normally the BIOS/OS handles this.  Linux uses neither the BIOS nor the
provided OS-drivers such as Win98 drivers so we have to do this all over
again.  *sigh* (no biggy but it means we don't end up with that silly S3
bug that most Windows systems have that cause the system to go flaky with
older ViRGE and Trio64V+-type cards...)

G'day, eh? :)
	- Teunis