[Coco] GIME behavior

[Darren A]

On 4/14/11, Boisy G. Pitre wrote:
>
> I'm not a hardware guy, but couldn't you do your own decoding of the
> address, clock and R/W lines to track writes to $FF90, storing that value
> somewhere as well as writing it to the "true" $FF90 address, then
> intercepting the read call and returning the value then?  Is there a
> hardware scheme to mask an address to do this?
>

Boisy,

That is exactly what I was doing.  Storing the value elsewhere while
it's also being written to the "true" $FF90 isn't a problem.  Trying
to read the stored value using $FF90 as the IO address is a problem
because, although the register is write only, the GIME does activate
its data outputs when you read from there (supplying bogus data). This
causes contention where two devices are trying to drive the data bus
at the same time. I will need to decode an additional IO address
(outside of the GIME domain) for reading.

Darren