Subject: Re: 12 Guage
To: Neville Duguid <nduguid@eastwind.livewire.com.au>
From: Michael L. Hitch <osymh@gemini.oscs.montana.edu>
List: amiga
Date: 09/17/1995 11:44:11
On Sep 17, 11:06pm, Neville Duguid wrote:
> On Wed, 13 Sep 1995, James Cheseborough wrote:
> >     I have a 12 Guage too. I can't even install NetBSD with it. Did you 
> > get a:
> > 
> > no rdb found
> > 
> > error? I can't get my ZIP drive or Quantum to respond correctly. Did you 
> > do anything different or have that error? Thanks! 
> >
> No, but my Hard Drive is on the A1200's internal SCSI which uses
> Commodore's scsi.device.  The CDROM which I can't access under NetBSD
> is on the 12-Gauge. I have had a few problems with the 12-Gauge SCSI
> port, but have been blaming the NEC CDR-25 CDROM drive. It works ok
> on the Amiga side. I have heard others complain about the CDR-25, but
> maybe because I am unlucky (lucky?) enough not to have a second hard
> drive, I have not experienced problems myself until I tried to use
> the CDR with MS-DOS and NetBSD.

  The A1200 has an internal SCSI?  As far as I am aware, the A1200 only
has an internal IDE drive.  The driver name is scsi.device, which
emulates SCSI operations on the IDE drive(s).  I took the same approach
on NetBSD to keep things consistent.

  The 12-Gauge driver in NetBSD does have a bug in the pseudo-DMA code
that prevents it from working.  The initial device probing is done
without using the pseudo-DMA code, so any devices will be recognized. 
Once the kernel has configured everthing and tries to mount the disks,
the driver will be using the pseudo-DMA routines for reading the disk. 
This doesn't work, so the RDB blocks aren't found and the mount fails.

  It should work OK if you get the binpatch program and use it to patch
the kernel to force the driver to use programmed-I/O transfers only.  The
command for this is:
	binpatch -s _sci_no_dma=1 netbsd
for the newer version of binpatch - if that command fails, try
	binpatch -s _sci_no_dma -r 1 netbsd
Running the command again will show you if the patch was actually made.

  The current kernel sources now include a fix for the bug, but I have
no way to test them, so I'm not certain the problem is actually fixed
yet.

Michael

-- 
Michael L. Hitch			INTERNET:  osymh@montana.edu
Computer Consultant
Office of Systems and Computing Services
Montana State University	Bozeman, MT	USA