[Coco] GIME behavior
Darren A
mechacoco at gmail.com
Wed Apr 13 17:42:43 EDT 2011
On 4/13/11, Mark Marlette wrote:
> Darren,
>
> As you are finding out all CoCos are not created equally.
> Can you talk about your project a bit more? If not I understand.
> I have some guesses as to why one would be tracking $ff90. :)
>
> Regards,
> Mark
> Cloud-9
>
---
The project is a cartridge which containins a CPLD and some memory
(RAM and ROM).
The cartridge can interrupt software via NMI and execute "supervisor"
code located in the external RAM/ROM. On a CoCo 3, the MMU must be
disabled when the supervisor code executes to gaurantee that the
$8000-FEFF address range can be mapped to the external memory. Of
course it also needs to be restored to its previous state before
returning control back to the "user" code. This is why I need to track
the state of $FF90.
The tests I've done seem to suggest that the GIME enables its data
outputs whenever the CPU reads from $FF80-FFDF, even though it only
produces meaningful data for a small subset of that range. I will
need to allocate some other address to use for reading back the
tracked state.
Darren
More information about the Coco
mailing list