[210] in linux-scsi channel archive
Re: SCSI testing at boot
daemon@ATHENA.MIT.EDU (Paul Gortmaker)
Wed May 24 20:49:29 1995
From: Paul Gortmaker <gpg109@rsphy1.anu.edu.au>
To: ericc@Arco.COM (Chang Eric Y (214)509-3562)
Date: Thu, 25 May 1995 10:08:21 +1000 (EST)
Cc: linux-scsi@vger.rutgers.edu
In-Reply-To: <9505242026.AA15691@Arco.COM> from "Chang Eric Y (214)509-3562" at May 24, 95 03:26:47 pm
>
> Hi. I have a strange problem with the installation of Linux on a
> Compaq Deskpro 386/25 w/ a Future Domain TMC-830 SCSI card with a
> single hard drive. It comes up with 2 or 3 extra SCSI devices on
> boot, which cause a kernel panic. DOS, however, only finds one
> device, and it boots fine. In addition, DOS diagnostics show the
> drive to be OK. How does the SCSI checking differ? Why is is
> more strict? Thanks, Eric
You have the memory area of the TMC-950 chip set as cacheable.
(Probably only 64k cache on the old 386 board) -- The card goes
to look for the device at the next ID and finds the stale data
from the previous INQUIRY command. This can also manifest itself
as worm drives and printers appearing.
Even if you get past the probe stage, you will still not be able
to xfer data between the PC and card until you get this resolved.
Also don't forget that the dain-bramaged tmc-950 chips take ID 6
(not 7) for themselves. So don't jumper the disk for ID 6.
Paul.
