Updating my reply from the other day, I found an old posting on Google from
when I was trying to reverse engineer the protocol. It's essentially what I
said, but with a couple of corrections:

The DaynaPORT thinks identifies itself as a SCSI "Processor" device. It
appears to use a couple of proprietary commands to read and write some
configuration information (mostly the MAC Id, I suspect). Actual
communications are done using SCSI 6-byte read and write commands.

I haven't been able to take apart the data yet, but I suspect that it
is doing little more than sticking a packet into the data buffer for
the command.

As for reading, the MacOS driver apparently polls the SCSI/Link every
500ms, requesting a read of (IIRC) 1559 bytes. I'm guessing that this is
the MTU plus some overhead. If there is no data, the SCSI/Link returns
a 0-byte result.

<<<john>>>

> -----Original Message-----
> From: port-mac68k-owner@netbsd.org
> [mailto:port-mac68k-owner@netbsd.org]On Behalf Of Iggy Drougge
> Sent: Sunday, May 26, 2002 10:54 AM
> To: Mattias Sandstrom
> Cc: port-mac68k@netbsd.org
> Subject: Re: Protocol of DaynaPORT SCSI/Link
>
>
> Mattias Sandstrom skrev:
>
> >found this on the net from way back. any luck? i'm actually thinking of
> >taking on this very project myself, so i was wondering if you had any
> >info on whether this device works anything like the one we have support
> >for? i have disassembled the macos driver for the dayna device, and it
> >exports all its functions by name, so the code is quite easy to read,
> >but i'd much rather try to just change the basic scsi commands and maybe
> >a buffer size or timer here and there than try to rewrite a driver from
> >scratch... ;-)