Cards which contain memory (ROM) are easy to be made auto detectable because the memory can be programmed to output a specific identifier to the peripheral slot memory locations ie $C0XX. However it is much more difficult to detect cards that do not contain any memory chips.

The 4Play card is detectable because of luck. The unique pattern it generates by default produces $20, $20, $20, $20 for locations $C0X0 to $C0X3. So this is also easily read by the A2 to determine the slot.

However cards such as the SNES MAX do not produce any output until they are asked to do so ie poking addresses to determine what you get back. Therefore my questions are:-
Is is ok to poke each peripheral slot in an attempt to determine a cards identity?
Does this have the potential to screw up the setup on other slotted cards (I'm thinking about I/O cards because poking will be driving physical external devices)?
Was there a best practise guide between developers for doing this?
Is there consolidated information anywhere on how A2 auto slot detection of cards was performed?

Thank-you for your help.

Cheers,
Alex.

> Is is ok to poke each peripheral slot in an attempt to determine a cards identity?

No.

> Does this have the potential to screw up the setup on other slotted cards (I'm thinking about I/O cards because poking will be driving physical external devices)?

Yes.

> Was there a best practise guide between developers for doing this?

Kind of. Either the card maker offers a way to identify it, or, failing that, the user is asked in which slot the card lives.
No writing to slot space.

> Is there consolidated information anywhere on how A2 auto slot detection of cards was performed?

I don't know of any. Not the same as "no". :-)

Hi Alex,

>Is is ok to poke each peripheral slot in an attempt to determine a cards id=
>entity?
>Does this have the potential to screw up the setup on other slotted cards (=
>I'm thinking about I/O cards because poking will be driving physical extern=
>al devices)?

The only thing I know for sure is that there are opposite opinions on
this topic.

My personal opinion:

The Cx?? space is supposed to contain ROM. So you are supposed to be
allowed to read from that area without fearing to cause unwanted side
effects. In contrast the C0x? space is supposed to contain I/O ports.
So even just reading from any of those 16 addresses can potentially
cause any side effect including e.g. destroying the content of an
attached mass storage device.

An example:

None of the three known Ethernet cards (LANceGS, Uthernet, Uthernet
II) supported by IP65 has ROM allowing to cleanly identify it. So my
strong opinion is that a program needs to ask the user for the slot of
the Ethernet card to use - and then IP65 can differentiate between the
three Ethernet cards by accessing their C0x? space - because all three
Ethernet cards are well known to tolerate those accesses.

Anyhow, in contrast ADTPro runs those IP65 accesses to the C0x? space
"blindly" on all slots. In a discussion on the topic David Schmidt
told me that he has never heard of those C0x? space accesses causing
any issue.

Regards,
Oliver

On 10/3/2021 1:26 AM, Lukazi wrote:
> Cards which contain memory (ROM) are easy to be made auto detectable because the memory can be programmed to output a specific identifier to the peripheral slot memory locations ie $C0XX. However it is much more difficult to detect cards that do not contain any memory chips.
>
> The 4Play card is detectable because of luck. The unique pattern it generates by default produces $20, $20, $20, $20 for locations $C0X0 to $C0X3. So this is also easily read by the A2 to determine the slot.
>
> However cards such as the SNES MAX do not produce any output until they are asked to do so ie poking addresses to determine what you get back. Therefore my questions are:-
> Is is ok to poke each peripheral slot in an attempt to determine a cards identity?
> Does this have the potential to screw up the setup on other slotted cards (I'm thinking about I/O cards because poking will be driving physical external devices)?

Yes, I would not do that.
I can't speak for other cards but I have used the $C0xx locations for
I/O on Carte Blanche projects. Some of the locations are used to send
commands to the card. Randomly poking these locations could
theoretically cause something unwanted to happen.
In my projects, I like to put the value $CB in $C0xF to make it easy to
locate the card.

Charlie

> Was there a best practise guide between developers for doing this?
> Is there consolidated information anywhere on how A2 auto slot detection of cards was performed?
>
> Thank-you for your help.
>
> Cheers,
> Alex.
>

Charlie <charlieDOTd@verEYEzon.net> wrote:
> On 10/3/2021 1:26 AM, Lukazi wrote:
>> Cards which contain memory (ROM) are easy to be made auto detectable
>> because the memory can be programmed to output a specific identifier to
>> the peripheral slot memory locations ie $C0XX. However it is much more
>> difficult to detect cards that do not contain any memory chips.
>>
>> The 4Play card is detectable because of luck. The unique pattern it
>> generates by default produces $20, $20, $20, $20 for locations $C0X0 to
>> $C0X3. So this is also easily read by the A2 to determine the slot.
>>
>> However cards such as the SNES MAX do not produce any output until they
>> are asked to do so ie poking addresses to determine what you get back.
>> Therefore my questions are:-
>> Is is ok to poke each peripheral slot in an attempt to determine a cards identity?
>> Does this have the potential to screw up the setup on other slotted
>> cards (I'm thinking about I/O cards because poking will be driving
>> physical external devices)?
>
> Yes, I would not do that.
> I can't speak for other cards but I have used the $C0xx locations for
> I/O on Carte Blanche projects. Some of the locations are used to send
> commands to the card. Randomly poking these locations could
> theoretically cause something unwanted to happen.
> In my projects, I like to put the value $CB in $C0xF to make it easy to
> locate the card.
>
> Charlie
>
>> Was there a best practise guide between developers for doing this?
>> Is there consolidated information anywhere on how A2 auto slot detection
>> of cards was performed?
>>
>> Thank-you for your help.
>>
>> Cheers,
>> Alex.
>>
>
>

In fact, the Mockingboard (and its clones) is a pretty prevalent example of
a ROMless card that uses the Csxx space for 6522 control.

--
-michael - NadaNet 3.1 and AppleCrate II: http://michaeljmahon.com

Lukazi wrote:

> Cards which contain memory (ROM) are easy to be made auto detectable
> because the memory can be programmed to output a specific identifier to
> the peripheral slot memory locations ie $C0XX. However it is much more
> difficult to detect cards that do not contain any memory chips.

It's difficult to detect a M$ Z80 Softcard: there is no ROM, there is no
register at $C0Nx, but if you write to $CNxx you will switch off the
6502 and switch on the Z80. If there is no suitable code for the Z80
you've lost control.

Microsoft could have used a register or $C0Nx to toggle between the two
CPUs, but they did not.

Regards
Ralf