[Coco] Re: Technical: PIA Help

[Brad Grier]

Correction: the interrupt start address is $c4e2, not $c4fe - I have a 
lot of numbers scribbled on the paper in front of me.

Brad Grier wrote:
> Hi Robert,
> 
> I reached my conclusions by stepping through Mocha in debug mode. I also 
> used the debugger in Jeff Vavasour's emulator (the game doesn't work 
> quite right there either - it does work in David Keil's emulator). I'm 
> surprised it doesn't work in MESS.
> 
> The strobing appears to be in the interrupt routine (both the write to 
> $ff02 and the read of $ff00). The routine starts at $c4fe - you don't 
> have to go far to see the *first* write/read of $ff02 and $ff00.
> 
> By the way, what is RTSI?
> 
> Brad
> 
> 
> 
> Robert Gault wrote:
> 
>> You correct in your understand of the keyboard strobing. As to the 
>> game, my first question would be how do you know what the game is 
>> doing? Have you disassembled the code or are you monitoring the 
>> emulated registers with a built-in debugging routine.
>>
>> There is a Tetris ROM image on RTSI which is 16K and works with a 
>> Coco3 in high res graphics and a Coco1/2 in low res graphics. A very 
>> very quick scan through the code did not find the $FF02 strobing. Does 
>> the store $FF02 with $FF occur during the interrupt, or is $FF00 read 
>> during the interrupt, or both?
>>
>> By the way, this image does not run correctly in MESS. Key presses are 
>> detected but the game responds as though several wrong keys were 
>> pressed. I'm not sure how or if I can test it on a real Coco without a 
>> ROM pack.
>>
>> Brad Grier wrote:
>>
>>> Hello all,
>>>
>>> I'm trying to get Tetris to work in Mocha but I'm having trouble 
>>> getting the emulator to recognize key presses in the game. This is 
>>> the only program I've found that exhibits this behavior.
>>>
>>> My understanding of the CoCo PIA keyboard routine is as follows: 
>>> Write a zero to the bit that corresponds to the keyboard column 
>>> you're interested in (with the rest of the bits set to 1) and poke it 
>>> into 0xff02. Next read 0xff00 and if a key is pressed you'll get a 
>>> zero in the bit corresponding to the keyboard row.
>>>
>>> Tetris seems to do all of it's keyboard polling from a field sync 
>>> interrupt routine. It's constantly poking 0xff into 0xff02 - normally 
>>> Mocha would ignore the subsequent read at 0xff00 because no bits are 
>>> zero. I've found that if I treat 0xff as a request to see if *any* 
>>> key is down regardless of column, I can at least get some keys to 
>>> kind of work. Maybe it all boils down to what does a 0xff into 0xff02 
>>> really do? I'm sure I'm overlooking something obvious...
>>>
>>> Anyway, I've been making a lot of changes to Mocha recently and this 
>>> one is last on the list for the time being. That, and I like Tetris.
>>>
>>> Thanks,
>>>
>>> Brad
>>> http://members.cox.net/javacoco/
>>>
>>>
>>
> 
>