On Sun Nov 28 2004 at 22:22:04 +0900, Christopher SEKIYA wrote:
> On Sat, Nov 27, 2004 at 10:43:12PM +0000, dieter wrote:
> 
> > I hadn't noticed it before but when the cursor is offset ok, my card is 
> > indeed detected as revision 7. However, when I have the 
> > cursor-over-last-character problem you are seeing, it was detected as a 
> > revision 3 board.
> 
> The version probe is non-deterministic?!?

Hrm. The first thing I'd test would be adding some delays between
critical register reads/writes (or just simply adding a delay to the
register read/write functions since that's the easier solution), since
the Linux driver had some delays that I could not find reason for in the
docs which I left out of our driver.

About the cursor position bug thing: It's _very_ badly documented, if
documented at all. At least in any docs I had. Mostly I gleaned the idea
from the Linux driver, which looked somewhat incorrect itself in that
area...

Another thing that sucks with the current driver is the inability to
switch display modes, or even setup the mode properly if ARCS hasn't
setup the chip - this has to do with how "well" the video timing
generation was documented in the docs I had. I had this idea to dump the
VC2 stuff from known-good configurations and use that to setup the video
timings, but Real Life got in the way of that project... (It would,
still, be a good project for someone who has the necessary programming
skills and hardware around.)

-- 
Ilpo Ruotsalainen - <lonewolf@iki.fi> - http://www.iki.fi/lonewolf/