Subject: Re: Old mail (but relevent to SCSI drivers/Jaz/Zip disks?)
To: None <ADAMGOOD@delphi.com>
From: Colin Wood <cwood@ichips.intel.com>
List: port-mac68k
Date: 08/21/1997 00:24:38
ADAMGOOD@delphi.com wrote:

> As an adjunct to all this, can anyone please explain what the difference
> between the SCSI drivers is exactly (other than the names)?  Also, why
> is SBC named SBC and ESP named ESP?

Well, I'm probably not the best person to explain all of this, but here's
a stab at (anyone feel free to correct me where I'm wrong ;-)

First, the 68k-based Mac's have 2 possible SCSI controller chips (i.e. the
chip which actually talks to the devices on the SCSI bus).  These two are
the ncr5380 and the ncr53c96.  All '030-based Mac's and (I think) the
500-series PowerBooks have the ncr5380 chip.  All other '040-based Mac's
use the ncr53c96 chip.

For each chip, there is a machine-independent (MI) driver which
understands in what format the chip expects to receive data in order to
communicate with the bus.  In addition, there is at least one
machine-dependent (MD) driver for each chip which understands how the Mac
hardware interacts with the SCSI controller.

There are 2 MD ncr5380 drivers:  ncrscsi (the older one), and sbc (the
newer one).  I can only assume that sbc stands for something like SCSI bus
controller, but it is also likely that it relates to the name of the
company that may have originally designed the chip (whose name I can't
remember at the moment).  As far as stability of these drivers goes, both
seem to work fairly well on certain kinds of drives, and fairly poorly on
other kinds of drives.  Generally, large amounts of filesystem corruption
will clue you in to the possibility that you may need to try the other
SCSI driver.  It now appears that the most persistant bug (which causes
this filesystem damage)  is in the MI driver, not in either of the MD
drivers.

As for the nc53c96 controller, there is only 1 MD driver: esp.  I believe
that esp stands for Emulex SCSI Processor (named for the company that
originally developed the chip), but I could be wrong.  I have heard of
very few problems with stability using this driver (other than the ones
that Allen just reported using his latest changes ;-).

So, hopefully, this will explain things a bit...if not, let me (or better
yet, Allen or Scott :-) know.

Later.

-- 
Colin Wood                                 cwood@ichips.intel.com
Component Design Engineer - MD6                 Intel Corporation
-----------------------------------------------------------------
I speak only on my own behalf, not for my employer.