Subject: Re: New scsi driver & Fujitsu M2512 MO drive
To: Allen Briggs <briggs@puma.bevd.blacksburg.va.us>
From: Paul Goyette <paul@pgoyette.bdt.com>
List: port-mac68k
Date: 10/15/1995 20:46:57
> > Having done some more experiments, I can say that only larger files are
> > affected (read: '> 20K'). They look as if the file system had mixed up some
> > block references: sometimes fragments of other source files are inserted,
> > sometimes simply junk, and I even saw emacs command names. This is so
> > strange to me I don't know where to look first...
>
> Yowch!  That looks bad to me, too, and I'm now more confused
> than ever as to what the problem might be.  Has anyone seen this
> on another kind of disk?

This is exactly the kind of crap I was seeing a week or so ago on my 
plain old hard drive (5400 rpm 3.5gb DEC RZ74C).  The various changes to 
the pending_5380_count and pending_5380_data stuff took care of it for 
me.  It turns out that the drive was dropping its DRQ request in the 
middle of a large (>MIN_PHYS) transfer and then restarting that transfer 
at the wrong buffer address.  Make sure you got the latest version of 
sys/arch/mac68k/dev/mac68k5380.c:

/*	$NetBSD: mac68k5380.c,v 1.14 1995/10/01 05:10:20 briggs Exp $	*/