[Coco] More sneak peeks

[Brett Gordon]

I'm good with just i/o space registers.  I dunno where... does 0xff40 -
0xff4f sound good?  Let it obey the mpi?  Maybe hook the irq up just in
case.

The cs8900a manual is indeed pretty fun.  The first 20 pages are just a
hard sell for advanced features I probably just want to avoid ;)  a couple
pages look like the baseball standings from the sports section... so many
settings!  Uhm looks like the myry#TXdeadbeef bit's ERA has slid up after
last night's game.

All funnies aside...beside endianess foul-ups, it does look pretty easy to
work with.  I'm digging through the contiki version of uip.... I think
there's a 8900 driver for the c64 floating around.  I'm starting work on
importing the latest uIP sources from contiki into fuzix and getting my
goofy TUN inferface working again.  Hmm.. hardware debugging..begs for
gratuitous Forth noodling.

-brett
On Jul 20, 2016 1:11 PM, "RETRO Innovations" <go4retro at go4retro.com> wrote:

> On 7/20/2016 6:22 AM, Brett Gordon wrote:
>
>> I finally read the cs8900 data sheet.
>>
> Stay awake now...
>
>> Is there still the option of mapping
>> in packet ram?
>>
> All of it flows through the CPLD, so it's possible, but I had actually
> planned on mapping it via IO as 16 locations in $ff40-$ff7f (any
> suggestions on where to sit it?) as that was easiest and most compact.
>
>> I must admit that the Fuzix implications of mapping out
>> 0xc000 scares me a bit...but I think there's a way to handle that - not
>> sure it's worth it just for interrupts.  I can't picture the outb inb
>> interface being horribly slower than mapped memory.
>>
> I had a similar concern.  Mapping it into external ROM space means that
> external ROM apps will have to copy themselves (a portion anyway) to RAM,
> and bank out the ROM so they can reach the memory mapped RAM on the
> cs8900.  That seemed a high bar to set.
>
> So, for now, assume it's IO mapped.  Let me know the spot you'd prefer it
> sit, and I'll write the Verilog and test and send to you.
>
> Jim
>
>
> --
> Coco mailing list
> Coco at maltedmedia.com
> https://pairlist5.pair.net/mailman/listinfo/coco
>