Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 5/3/83; site utcsrgv.UUCP
Path: utzoo!utcsrgv!thomson
From: thomson@utcsrgv.UUCP (Brian Thomson)
Newsgroups: net.unix-wizards
Subject: Re: m and n of mkfs
Message-ID: <2763@utcsrgv.UUCP>
Date: Fri, 18-Nov-83 12:47:21 EST
Article-I.D.: utcsrgv.2763
Posted: Fri Nov 18 12:47:21 1983
Date-Received: Fri, 18-Nov-83 13:35:10 EST
References: <298@bunker.UUCP>
Organization: CSRG, University of Toronto
Lines: 20

m is the shuffle factor for disk block interleaving.
Usually m == 3, which means that the freelist is constructed
by linking together every third data block.  It should be
large enough that no disk rotations are lost between successive
io requests, and small enough that the rotational latency between
blocks is small.

n is the number of blocks in a cylinder (== number of blocks disk
can access without head movement).  The interleaving is done
on a cylinder-by-cylinder basis, so a 100 block filesystem
with m==2, n==50 would be interleaved like so:

0,2,4,6,...,48,1,3,5,...,49,50,52,54,...,96,98,51,53,55,...,97,99

Wherever I have said "block" above, I DON'T necessarily mean disk
segment (512 bytes), I mean filesystem logical block size, 1kbyte
in most 4.1BSD installations.
-- 
			Brian Thomson,	    CSRG Univ. of Toronto
			{linus,ihnp4,uw-beaver,floyd,utzoo}!utcsrgv!thomson