Path: utzoo!mnetor!uunet!husc6!hao!gatech!purdue!i.cc.purdue.edu!j.cc.purdue.edu!pur-ee!iuvax!inuxc!ihnp4!homxb!antique!cjp
From: cjp@antique.UUCP (Charles Poirier)
Newsgroups: comp.sys.amiga
Subject: Re: Disk Errors. AmigaDOS file system.
Message-ID: <2020@antique.UUCP>
Date: 5 Jan 88 21:36:48 GMT
References: <1338@sugar.UUCP> <37823@sun.uucp>
Reply-To: cjp@vax135.UUCP (Charles Poirier)
Organization: AT&T Bell Labs, Holmdel, NJ
Lines: 33
Summary: Bad Block side effects

In article <37823@sun.uucp> cmcmanis@sun.UUCP (Chuck McManis) writes:
>In article <1338@sugar.UUCP> peter@sugar.UUCP (Peter da Silva) writes:
>>While I'm on this subject, how about dumping the "one sector per track"
>>business and going to a conventional track/sector type arrangement? What
>
>Uh, Peter, we do have multiple sectors/track. About 11 last time I checked.
>I agree that writing to a disk with a bad block should not be fatal, however
>with the current filesystem it would cause problems with conventional disk
>interfaces (like using a WD or NEC chip) as well. The most you should lose 
>is one block of one file, everything else can be recovered from information
>on the disk. As a Request For Enhancement, I think it would be nice to add
>a 'bad block' bitmap....

First, it seems to me that Peter is referring to the "no inter-sector
gaps" track layout when he says "one sector".  I think I see a potential
problem  with this system.  What I recall (sorry if this is all wet) is
that to write a block, a track is read in, the proper block of it is
found and modified, and the track is then blasted back out.  The Amiga
doesn't sync the drive to a particular point before writing the track.
So, if a bad spot develops on the disk, that spot may appear in any of
the 11 blocks after a write to any of those blocks.  The more of those
blocks that are written to, the more of them will be corrupted by being
dropped back on top of the oxide hole.

What is needed, then, is a bad-track bitmap, not just bad blocks.  Upon
detecting an error, the file system should move all blocks of the
track, which *are* readable, to other track[s].  (Something like that
anyway.)

-- 
	Charles Poirier   (decvax,ihnp4,attmail)!vax135!cjp

   "Docking complete...       Docking complete...       Docking complete..."