From mboxrd@z Thu Jan  1 00:00:00 1970
Date: Tue,  3 Oct 1995 20:08:47 -0400
From: forsyth@plan9.cs.york.ac.uk forsyth@plan9.cs.york.ac.uk
Subject: Standalone SPARC.
Topicbox-Message-UUID: 2a7b0c64-eac8-11e9-9e20-41e7f4b1d025
Message-ID: <19951004000847.qlz2MUhQCZzKh9Kk_CPwCW0l9GRru466qwzI_hj2m50@z>

>>look like SunOS and plan9 are counting blocks the same way.  Anyone know
>>how to reconcile the numbers? 

i think you'll find that SunOS cheats in two ways:
 - it reserves a few cylinders (`alt cyl') to provide replacement blocks
   for bad blocks
 - more seriously (in terms of the amount of space), SunOS has the notion
   of (say) a 500 Mbyte disc.  that might be implemented by a disc that
   is 535 mbytes or 520 mbytes or precisely 500 mbytes (formatted).
   it simplifies supplying and replacing discs, when the discs come
   from several manufacturers, with slightly different capacities.

Plan 9 displays the results of a SCSI `get capacity' request, which
should be precise.

see /etc/format.dat on SunOS for more details.
it isn't a bad idea, given the practical restrictions, but it does lead
to the situation you describe.

oh yes.  the `bad block replacement' area is probably not needed on
a modern system; a good SCSI drive should implement the `block replace'
function.