[Coco] Re: Technical: PIA Help

[Robert Gault]

You will never determine what is going on without looking at the code in 
question. Part of what you report comes from

$cf01 extends to about $cffe
pshs d,x,y,u
leas -5,s
clr $ff02
lda $ff03
anda #%11111011   data direction register
sta $ff03
lda #$ff
sta $ff02         normal, all output
lda $ff00
coma
anda #%011111111  ignore joystick comparator
bne $cf2f         go if key press
set some data and at $cfde deactivate data direction

$cf2f
leau $da23,pcr
ldx #$601
clr 2,s
clr 4,s
lda #%10000000
sta $ff02         only MSB is output
lda $ff00
bita #%01000000   check for shift keys
bne $cf48
inc 4,s
$cf48 lda #1
sta $ff02
10 nops to wait
lda $ff00
coma
anda #%01111111
ldb 2,s
cmpb #7
beq $cf6b
cmpb #4
beq $cf6b
cmpb #3
bne $cf6d
$cf6b anda #%00111111
$cf6b sta ,s
eora ,x
anda ,s
beq $cfb0
ldb #1
$cf77 lsra
bcs $cf7d
aslb
bra $cf77
$cf7d andb ,s
stb ,s
lda #15
ldb 2,s
mul
leau d,u
lda ,s
lbsr $cc7c
ldb #3
mul
leau d,u
lda 4,s
lda a,u
etc.



There does not seem to be any advantage to activate data direction and 
this is normally only done for $FF02 after a cold start. I think that 
the entire business is an error and more likely is that bit3 was 
intended. That would change the MSB of the MUX line which might make 
some sense.

Brad Grier wrote:
> I added a few println calls to Mocha at the memory locations in 
> question. This is what happens at those addresses during a Tetris 
> interrupt.
> 
> I'm guessing some of the peek return values aren't right but I'm 
> displaying what Mocha thinks the result should be.
> 
> The big question: What's going on with the peek and pokes at $ff03?
> 
> fire interrupt
> poke ff02: ff
> peek ff00= ff
> poke ff02: 0
> peek ff03= b5
> poke ff03: b1 (ff02 data direction bit=0)
> poke ff02: ff
> peek ff00= ff
> poke ff02: ff
> peek ff03= 31
> poke ff03: 35 (ff02 data direction bit=1)
> poke ff02: ff
> peek ff02= resets field sync IRQ flag
> fire interrupt
> etc, etc, etc, etc
> 
> Maybe I need to hook up my coco and figure out how to use EDTASM again.
> 
> Brad
> 
> 
> Torsten Dittel wrote:
> 
>> Hi Brad,
>>
>> just some thoughts produced by a blurry remembrance (might be pure
>> nonsense): I guess before writing and reading the keyboard you have to
>> programm the corresponding PIA registers as inputs or outputs. I guess
>> it's possible to "inverse" the input/output configuration. Maybe one
>> even could mix it and make some PIA lines read and others write (no idea
>> what this would be good for).
>>
>> I remember darkly on first Motorola "CoCo" board designs (using two PIAS
>> 6821) it was even possible to programm both PIAS as outputs and "burn"
>> the PIA lines by pressing the corresponding keys (kind of short
>> circuit). Those lines would work only in one direction after being
>> killed that way. Some software would still work (e.g. BASIC) but other
>> using the "inverse" PIA setting wouldn't recognize the keys any longer).
>> AFAIR this was a know problem and Motorola replaced the 6821 by the 6822
>> which had kind of protection against this (I wonder if it's possible to
>> kill PIAs in one of the CoCo1 emulators, it would win a price for
>> accurate emulation... ;-)).
>>
>> I'd enjoy some comments by all you gurus out there! :-)
>>
>> Best regards,
>> Torsten
>>
>>
> 
>