2.0.32-3985W (fwd)

[Mike Bilow]

Doug Ledford wrote in a message to Mike Bilow:

>Dec  4 04:23:27 lab1 kernel: (scsi0:0:1:0) Parity error during 
>phase Command.

 DL> I would guess that you are using the Nov-05 version of my
 DL> patch (this is the one where I updated the printk headers to
 DL> the style noted above).  I would first guess that your
 DL> problem is cable related.  Either termination or a bad cable
 DL> period.  The SCSI parity errors are hardware flagged by the
 DL> card and not directly controlled by the driver (other than
 DL> we tell the card to let us know when they occur).

I vote for termination or cabling as the most likely cause, but I would also
raise the possibility that some other peripheral on the bus is responsible. 
For example, NEC CD-ROM drives are notorious for doing this, even when they
should be totally quiescent on the SCSI bus.

Another issue worth checking is auto-termination in the controller.  It often
doesn't work, either because of mysterious happenings in certain ROM revisions
or because some other device on the bus, usually an image scanner, is trying to
do auto-termination at the same time.  I see little or no value to
auto-termination in the controller, since humans are much more accurately able
to judge what is plugged into where and set things accordingly.
 
-- Mike