[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Carte Blanche

mdj wrote:

On Apr 3, 11:16 am, srk...@gmail.com wrote:

Maybe I should explain why this possible.

The FPGA on carte blanche is able to load its bit stream configuration
image from several sources. Carte blanche has two of these methods
implemented. The first is JTAG, with the highest priority, the second
is a generic SPI FLASH memory - an 8Mbit (1MB) FLASH device. If you
plug up a JTAG interface, the FPGA will always respond and accept data
over this interface, otherwise it will contact its dedicated SPI
interface and load a bit stream from there if its available.

Once the FPGA has booted, the dedicated SPI interface is relegated to
generic IO. This is where carte blanche can get its mits on the
interface and use the device for its own storage (the bit stream only
needs part of the device) which also allows us to reprogram the FLASH
with a new image if we wish to. On a power cycle, the FPGA once again
attempts to boot from SPI, and if a new image has been updated to
flash, it will load the new design into the FPGA.


OK. I guess the catch is then that the initial, and all subsequent
'personas' for the Carte Blanch will have to support the reprogramming
interface ?

That should work great, assuming nothing ever goes wrong ... :-)


I agree--but then that's just like a lot of things about computers.. ;-)

Sounds like the program to load a configuration into the flash RAM
is pretty trivial, as I hoped.  ;-)

It would be good, then, to design an architecture for the reprogramming
interface with the host Apple II so that it is:  1) simple, 2)fast
enough, and 3) has room for expansion in the future.  This would make
it a "module" that could be included in all card configurations.

I would recommend that an "unlock" read dance be performed on some
IOSTB (slot ROM) location--say, $CsFF, $CsFE, repeated four times
with no intervening $Csxx references.  This would allow no interference
with the address map, etc., of any configuration.

After the power cycle (shame it has to be so clumsy--maybe the last
act of the reprogramming state machine should be to trigger an on-card
reset signal), the card would reload and come up "locked".

-michael

******** Note new website URL ********

NadaNet and AppleCrate II for Apple II parallel computing!
Home page:  http://home.comcast.net/~mjmahon/

"The wastebasket is our most important design
tool--and it's seriously underused."