Hi Dan,

[ trace ]
> >Is there anything below this?
> 
> No, that was the whole trace. Sometimes it doesn't even give that much,
> just the first 4 lines.

Sigh, that won;t help me much...

> >For the Falcon, I somehow suspect the DMA controller. Did you try running
> >NetBSD from the IDE drive? I don't know if this is an option for you. If
> 
> Not really - me IDE is only 65Mb and is full of the TOS stuff since my
> SCSI drive got cleard for NetBSD. ;) At most I could move my swap
> partition to the IDE drive (would it work from there?)

Yes. But you'll need a custom kernel to do that because you need to say:
	config netbsd root on sd0a swap on wd0b
or something close to that ;-) 

> >this helps, it makes sense to go diff-ing the SCSI-driver. I can also try
> >building a 1.2E kernel with an 1.1 SCSI-driver.
> 
> I'm tempted to suspect that too, since most times it happens there
> is intense disk activity going on - e.g. compiling something, or
> very annoyingly while fsck is fixing the filesystem after the last crash.
> 
> Using the 1.1 SCSI-driver sounds attractive to me since I never
> had problems with that one. If it didn't take 16 hours (assuming no
> crashes) to build my own kernels I'd experiment a lot more myself.

I managed to merge in the 1.1 SCSI driver in the 1.2E BOOT kernel. I'll
send it to you. _If_ this works, are you willing to track down what
bit of changes triggers this bug? I mean, the first (full) compile might
take 16 hours, the next compile - changing 1 source file - should be faster ;-)
Besides, I'm not in a hurry. I _do_ however, want to get rid of these
panics!

Leo.