According to Jukka Marin:
>
>May depend on the disk?  I don't know if an SCSI host interface does
>any error checking (other than parity, perhaps).  I could be wrong,
>I haven't read the SCSI specs that much.
>

No, actually all hard disks have built in ECC.  The code gets
validated as the data is being read off the disk.  The ECC is part of
the data that gets put in between the sectors.  Haven't you ever
wondered why the formatted capacity of a disk is lower than the
unformatted?  This is part of the reason why - the inter-sector gap
hold all sorts of information useful to the controller for peeling the
data off the disk.

Errors between the controller and disk can happen - for long busses
like the SCSI one they have parity that you can enable.  I don't
believe the ESDI, IDE, ST506 drives/controllers have anything like
this but their life is simpler because they are only designed for
short hauls inside the same computer so the likelihood of errors
occurring is much much smaller.

When I framed the response I was thinking more along the lines of disk
media errors than anything else, I must admit....

>> It should not panic - I would expect things to slow to a crawl or
>> appear to stop but the machine should not panic.  If it does then,
>> IMHO, it is a bug in the kernel.
>

AHA! That's one down ;-)

-- 
Brett Lymn, Computer Systems Administrator, AWA Defence Industries
===============================================================================
  "Upgrading your memory gives you MORE RAM!" - ad in MacWAREHOUSE catalogue.