Some years ago there were a netboot problem that caused strange things
to happen. The reason was that the LANCE chip wasn't stopped after
the kernel had been loaded, and if there were broadcast traffic on
the network the chip would continue to write those received packets
into memory where the buffers were setup, but after the kernel was 
loaded anything could be at that place. It took some time to really
find that problem :-)

-- Ragge
> 
> There was a bug in the netboot code that was corrupting something in the 
> page tables. I could reproduce it on the M3100/76 by booting a compressed 
> kernel and later by netbooting any "second choice" kernel. The symptom was 
> if the first kernel it tried to netboot was not found, when it started 
> booting the one it _did_ find it would generate a lot of messages about 
> ptlen's not matching. The workaround was to make sure that it found the 
> first kernel it asked for.
> 
> If this is the same bug, then renaming netbsd to netbsd.vax (which is what 
> it looks for first) will allow it to boot. Let me know if that does indeed 
> get you up and running.
> 
> --Chuck
> 
> At 02:53 PM 10/13/01 +0200, Anders Magnusson wrote:
> > > hi,
> > > here are the results:
> > >
> >Ok, now you did get the panic later. What happens is that you get an
> >interrupt or something that isn't handled. It may be due to almost
> >anything, and not simple to find.
> >
> >-- Ragge
> 
> 
>