On Thu, 31 Aug 1995, Daniel Carosone wrote:

(on dumb digiboards)
> However, while writing and experimenting with the driver, I discovered
> how the status register is used. Instead of doing something sensible
> like using the 8 bits as flags, one for each UART, they set the status
> register to the binary value of the number of the uart that last
> interrupted!

The idea is that one can daisychain upto 64-ports from same cards
and have them all serviced under the same irq/status port.

> Digiboard also make a `smart' multiserial card, like any number of
> other brands these have some shared memory and a host cpu on the card
> to offload a lot of the work. Programming information for these things
> is only available under NDA, if at all, for all manufacturers I've
> come across.

I thought I saw some messages from people on the usenet that digiboard 
might be reconsidering it's position on the programming docs.

> > There seems to be Linux driver that one could use as a hw-docs,
> > would it catch the gnu-public virus that way ?  
> 
> Depends.. I'd leave that determination up to your friendly
> neighbourhood core team member.

The linux-smart-digiboard driver has fep.h (program for the card
processor) and bios.h that would have to be used asis, they are just chunk
of hex-bytes that are downloaded to the board. Disassembling (on board
cpu is i186) and analyzing that code could be an option, but that is
beyond me. 

> For a dumb card, go with a boca. If you've got the money, consider a
> terminal server.

I think that i'll go with boca, dumb digiboard would also be an option but
they cost about 3 times what boca does. 

-- 
 Jarkko.Torppa@cute.fi     +358-0-648090          Cute Communications Oy