Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/3/84; site cygnet.CYGNETSYSTEMS
Path: utzoo!watmath!clyde!cbosgd!ihnp4!pesnta!hplabs!cygnet!karl
From: karl@cygnet.CYGNETSYSTEMS (Karl Danz)
Newsgroups: net.unix
Subject: Re: Formatting Fujitsu Eagles under 4.2 BSD Unix
Message-ID: <532@cygnet.CYGNETSYSTEMS>
Date: Sat, 2-Mar-85 02:00:27 EST
Article-I.D.: cygnet.532
Posted: Sat Mar  2 02:00:27 1985
Date-Received: Sun, 3-Mar-85 04:57:45 EST
References: <232@resonex.UUCP> <478@mcvax.UUCP>
Reply-To: karl@cygnet.UUCP (Karl Danz)
Distribution: net
Organization: Cygnet Systems -- Sunnyvale, CA
Lines: 27

[]
>Also, adding new bad blocks to the list is a real pain -- the list of
>bad blocks is kept in sorted order, so when you add a block to it, all
>the ones after it must be shifted down one (i.e., you need to copy the
>contents of the forwarded block to the next physical block on the disk).

We were bitten by this same problem (on an Eagle as well).  I retched at
the thought of all that manual shifting and instead made the inclosed
one line change to /sys/vax/dkbad.c (4.2BSD).  I suppose it doubles the
average time spent looking up replacement blocks, but somehow I doubt
that's our major performance bottleneck!

-------------------------------------------------------
RCS file: RCS/dkbad.c,v
retrieving revision 1.1
diff  -r1.1 dkbad.c
25c25
< 		if (blk < bblk || bblk < 0)
---
> 		if (bblk < 0)
-------------------------------------------------------
-- 

					Karl Danz
					Cygnet Systems
					...!hplabs!cygnet!karl
					(408) 773-0770