> This looks _exactly_ like a problem I had that was because of a kernel
> bug.  (I have an SE/30 and a RasterOps 264 card.) The fix is a
> one-liner in grf.c in the _kernel_ sources (not grf.c in dt).
> Unfortunately, fixing that bug caused dt to core dump on some other
> systems.  (I think this is due to a bug in dt that slipped by because
> of the kernel bug, but I couldn't recreate the crash on my machine.)
> Last I heard, the kernel change had been backed out, so I don't know
> exactly which kernels still have the fix in them.  The kernel I have is
> one of the netbsd.patched kernels from a few weeks ago.  I think David
> Holcomb was looking into why dt crashed with the fix, but I don't know
> what the final verdict was.
> 
> What say you, David?

The change you made to the kernel's grf.c was backed out.
The changes you made to dt's grf.c were left in.

If I can ever get my machine to power up again (it's dead
at the moment), I'll spend some more time with dt to see
why your change to the kernel's grf.c was breaking it.

--Dave Holcomb