[Coco] Fwd: IP packets on my coco

[Dave Philipsen]

I might be a little worried about using a chip that has these kinds of 
problems.  The interrupts may appear to work for awhile but they will 
eventually fail?  What's up with that?  Sounds like they made lots of 
addendums/notations after the fact.  It reminds me of a cheesy UART chip 
that we used back in the early 90s (I won't mention the manufacturer or 
which chip it was).  It was advertised to support a 9-bit address 
wake-up mode that would have been useful for multi-drop RS485 
applications.  The 9-bit mode never worked the way it was supposed to 
and we had to come up with a less than ideal firmware fix because we had 
already committed to a board design with the chip.  The guy I was 
working with couldn't stop muttering curses under his breath toward the 
company.

Dave


On 6/9/2016 9:34 PM, John W. Linville wrote:
> On Thu, Jun 09, 2016 at 08:19:43PM -0500, RETRO Innovations wrote:
>> On 6/9/2016 12:36 PM, John W. Linville wrote:
>>> https://www.cirrus.com/jp/pubs/proDatasheet/CS8900A_F5.pdf
>>>
>>>   From page 77:
>>>
>>> 4.10.9  Basic I/O Mode Receive
>>>
>>> I/O Mode receive operations occur in the following order (In this
>>> example, interrupts are enabled to signal the presence of a valid
>>> receive frame):
>>>
>>> 1)  A frame is received by the CS8900A, triggering an enabled
>>> interrupt.
>>>
>>> <snip>
>>>
>>> That looks fairly clear to me.  Is the datasheet wrong?
>> I don't think it is wrong, but maybe it is incomplete.  Numerous developers
>> in the 64 side have lamented the lack of IRQ capability in IO mode.  I will
>> check to see where they got their information.
> Well, it looks like the datasheet might be wrong...
>
> Application Note 181 from Cirrus Logic says:
>
> "Unsupported functions in 8 bit mode
>
> - Interrupts are not supported. Polled mode must
>    be used.
>
> - The DMA engine only uses 16 bit memory ac-
>    cesses and does not support 8 bit transfers.
>
> - The packet page pointer has an auto increment
>    feature that cannot be used in 8 bit mode.
>
> - An EEPROM is not supported.  Most 8 bit de-
>    signs should not require one and can eliminate
>    the added cost."
>
> And Application note 205 says:
>
> "1) Application Note 181 says that interrupts are not supported
> in 8-bit mode. Is this really true?
>
> Yes, interrupts are NOT supported in 8-bit mode. They might appear
> to work for a while but will eventually fail."
>
> So apparently the 8-bit mode is a bit unreliable in the CS8900A?
>
> With that said, the usefulness of interrupts for servicing an Ethernet
> NIC on <2MHz CPU is debatable...
>
> John