> I recently switched from an ide to an adaptec 1542c scsi controler.
> Until that change I had never seen anything resembling this:
> 
> Sep 20 02:02:07 sasquatch /netbsd: sd0: not queued
> Sep 20 02:02:07 sasquatch /netbsd: aha0: DMA beyond end of ISA
> Sep 20 02:02:07 sasquatch /netbsd: sd0: not queuedaha0: DMA beyond end of ISA
> 
> 
> The machine is a 486 66, genuine Intel, 16 megs of ram with a Conner
> cfa540s drive in it.  I am running NetBSD 1.0 Beta built from Sept 9
> source.  The error messages are intermitent, I sometimes can go for a
> day or so with out seeing them.  They are not crashing the machine.  Any
> suggestions are welcomed.  Other than this 1.0 beta is running very
> smoothly.

This error is caused when the buffer for I/O is beyond the address
range that the ISA bus can reach.  With 16M you should be okay,
however, some motherboards do reclaim all or part of the "lost" 384K
(from the I/O "hole" from A0000-FFFFF) and put it just beyond the end
of the rest of the memory, so you actually get 16M plus a little bit.

Was your IDE controller a VL Bus controller?  Then it would be able to reach
all of this memory.  Or maybe the wd driver just doesn't check?

The fix is bounce buffers.  Has anyone done this for the aha driver?  I've
got a 1542CF that I would switch to (from a 6360 PIO controller) except I
don't want to loose the extra 4M of memory.

The fix?  Either turn off the reclaiming of the extra memory (most
motherboards that do this allow you to disable it), hack machdep.c to
force the physical memory used to 16M (anyone got a specific patch for this?
I've tried it but haven't quite gotten it right), or use a 32 bit bus
(EISA, VL or PCI) controller.

-Mark