The symptoms you describe could very well be a result of the ncrscsi
driver.  Other symptoms might be the sudden transformation of a whole
bunch of consecutive i-nodes into device nodes rather than files!  I
suspect that the cause is the result of missing one or more bytes during
the data transfer phase - I occassionally see the above problems, as well
as dump occassionally complaining about "SHORT READ" errors.

Yes, indeed, the sbc driver is slower than the ncrscsi driver, because it
runs in polled transfer mode, whereas the ncr driver runs in pseudo-DMA
mode.  Also, I've noticed problems (read: crashes) when using the sbc
driver with my Exabyte 8200 tape drive - that's why, despite the ncrscsi
bug I still use ncr rather than sbc.

I know that Allen Briggs was looking into the problem a while back, but I
don't think he was able to find the problem.  It seems to be related to
the "extra" chip that Apple included to enable interrupt-driven pseudo-DMA
transfers.  And, of course, Apple in their infinite wisdumb (tm) haven't
seen fit to provide much for documentation on this thing.  

A new HD might solve your problem, since the bug seems to be much more
prevalent on drives from certain manufacturers.  (I seem to recall that
Quantum drives were particularly susceptible.)  Re-installing a MacOS
partition won't have any effect.

On Thu, 13 Mar 1997 a.fowler@ic.ac.uk wrote:

>  Hi guys!  
> 	I've been having what i think now are nrscsi problems,
> in that my netbsd kernals keep segfaulting various executibals 
> on me, and get themselves corrupted time after time... :(
> 
> 	thing is, the sbc kernals are MUCH slower than the nrscsi ones.
> Does anyone know what thing it is that makes this nrscsi bug turn up, as
> i'm running on a IIcx, which I thought wouldn't have a problem, seeing as
> it's the first supported model. I have not to my knowledge got a non-standard> scsi controller.
>  
> who, if anyone is working on this bug, and would they have any idea which 
> systems show up the problem, and which don't. Is getting a new HD likely
> to remove the problem? or even re-installing the apple-partion map, which i
> trashed for the FWB one. I'd be more than happy to provide bug reports for
> anyone working on the driver. I seem to get regular occurances of the 
> problem on my system. I only have an 80MB apple badged drive, so sources and
> compiling is probably out, but binaries :) are easy to download.
> 
> 
>   	happy days are here again... 
>   			Andrew Fowler
> 
> 

------------------------------------------------------------------------------
| Paul Goyette       | PGP Public Key fingerprint:  | E-mail addresses:      |
| Network Consultant |     0E 40 D2 FC 2A 13 74 A0  |  paul@pgoyette.bdt.com |
| and kernel hacker  |     E4 69 D5 BE 65 E4 56 C6  |  paul_goyette@ins.com  |
------------------------------------------------------------------------------