Subject: Re: more info about "wd0: (uncorrectable data error)"
To: Miles Nordin <carton@Ivy.NET>
From: Manuel Bouyer <bouyer@antioche.eu.org>
List: port-sparc64
Date: 01/10/2007 19:45:02
On Sat, Jan 06, 2007 at 07:57:45PM -0500, Miles Nordin wrote:
> >>>>> "jc" == Joel CARNAT <joel@carnat.net> writes:
> 
>     jc> where: 
> 
>     jc> - Windows just goes blue screen 
> 
>     jc> - Linux froze the shell where the access to the disk was
>     jc> started (cd, cp, or such)
> 
>     jc> - NetBSD dropped thousand of messages
> 
> IMHO, if the error is in a data block the kernel should return EIO to
> whatever was reading.

This is actually what's happens. But the drive may take some time before
returning error (with IDE it can take up to 30s, that's what's in the specs)
Also the kernel or application sometimes insists on reading the same sector
again and again, which of course slows the whole thing down

> If the error is in metadata it should remount
> the filesystem read-only and return EIO as needed.

I dissagree. Linux does this and a single erro on one block causes
all sort of pain. 

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 26 ans d'experience feront toujours la difference
--