Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!mips!spool.mu.edu!olivea!uunet!cbmvax!daveh
From: daveh@cbmvax.commodore.com (Dave Haynie)
Newsgroups: comp.sys.amiga.misc
Subject: Re: Hard-disk salvation
Message-ID: <22771@cbmvax.commodore.com>
Date: 28 Jun 91 09:26:58 GMT
References: <1991Jun24.150553.25949@eik.ii.uib.no>
Reply-To: daveh@cbmvax.commodore.com (Dave Haynie)
Distribution: comp.sys.amiga.misc
Organization: Commodore, West Chester, PA
Lines: 62

In article <1991Jun24.150553.25949@eik.ii.uib.no> larsm@eik.ii.uib.no (Lars Magnus Nordeide) writes:
>I already have DiskSalv1.3, and FixDisk. DiskSalv cannot be used on 
>hard-disks (can it?). 

Of course it can.  You're better off with V1.42, which is the latest release.
DiskSalv is not a "fix-in-place" tool, however, but a recovery tool.  It'll
get whatever it can off your troubled disk, but it won't attempt to modify, fix, 
or destroy your troubled disk.  This works real well if you have two similarly
sized disks, but it can be annoying when you need to recover your 1.23GB hard
disk (all one partition) to floppies.

>Another thing, is it normal that such an error or similar occurs if the
>machine crashes when writing to the hard-disk? 

It can happen.  It depends on what's taking place during the crash.

>If that is the case, it seems to me a bit unstable. 

Without memory protection, there's no way around this.  Non-multitasking 
systems can be a bit more stable in this respect, since nothing is running
during a write to disk.  But any non-protected system offers the chance for
some kind of corruption to crash the disk. 

>I know a crash would make the file that was written to incomplete, but 
>shouldn't that result in a corrupt file rather than a whole disk. 

It depends on what was being written at the point of the crash.  Obviously,
a partially written data block isn't a problem, but if the FileSystem is in the 
process of modifying a file header or directory block when you crash, that
file header or directory, and the surrounding disk structure, may be damaged.

>(What is the point with the validator?) 

The validator walks the directory tree to rebuild the FileSystem's bitmap.  It
can thereby reclaim any data blocks that were allocated by a file write 
operation that didn't complete.  However, it doesn't know how to deal with a
corrupt data block.  There's no easy answer, believe me, I've been hacking
around with this stuff since '85.  You really need a system of small expert
systems to determine whether an invalid block contains useful information
or not.  That's essentially what DiskSalv does, and will do even better if I
ever get time to complete V2.00 (hoping this year, but not betting heavily
on it just yet).

>Or have I got a hardware problem?

Probably not, but...

>In my case this is happening every time my Amiga is crashing while
>writing to the disk. 

What's causing the crash?  Crashes don't "just happen", something causes them.
Usually it's a bad program.  The fix, in such a case, is to find the bad 
program and stop using it.  Some programs don't cause the crash directly, but
instead trash something that's then encountered by the next program to come
along.  So examine your environment very, very carefully.  If you have
eliminated all software-influenced possibilities, you may have to consider
the chance that you have some kind of marginal hardware.

-- 
Dave Haynie Commodore-Amiga (Amiga 3000) "The Crew That Never Rests"
   {uunet|pyramid|rutgers}!cbmvax!daveh      PLINK: hazy     BIX: hazy
		"I looked for it and I found it" -R.E.M.