Jason Thorpe <thorpej@nas.nasa.gov> writes:

  > All I/O to SCSI disks is done via SCSI command, of course.  There are 
  > very few cases where SCSI commands themselves actually originate 
  > _outside_ the kernel.

Yes. My brain was not in gear really.

  > It's not obvious to me how the Cabletron device notifies the host that
  > it's received a packet. However, it seems as if the device would
  > fill a linear buffer with the packet, which would then be copied into
  > an mbuf chain (or cluster).  Really, in general, it seems like it could
  > basically look like a regular network interface driver with a strange way
  > of accessing the device registers :-)

It seems the only way is to poll it with scsi "read" commands. These would
presumably be scheduled using timeout(). It is a bit of a wart but it still
beats a serial port!

Given the pseudo-dma aspect of pc532 scsi-io, it would be possible to
read/write directly into the mbuf chain. However, in practice that
is probably more complication than it is worth seeing as it would need
changes to the device driver level.

Ian