xfs
[Top] [All Lists]

Re: [PATCH 07/19] xfs: verify AGFL blocks as they are read from disk

To: Christoph Hellwig <hch@xxxxxxxxxxxxx>
Subject: Re: [PATCH 07/19] xfs: verify AGFL blocks as they are read from disk
From: Dave Chinner <david@xxxxxxxxxxxxx>
Date: Fri, 12 Oct 2012 08:52:00 +1100
Cc: xfs@xxxxxxxxxxx
In-reply-to: <20121011214421.GG6346@xxxxxxxxxxxxx>
References: <1349754670-32009-1-git-send-email-david@xxxxxxxxxxxxx> <1349754670-32009-8-git-send-email-david@xxxxxxxxxxxxx> <20121011214421.GG6346@xxxxxxxxxxxxx>
User-agent: Mutt/1.5.21 (2010-09-15)
On Thu, Oct 11, 2012 at 05:44:21PM -0400, Christoph Hellwig wrote:
> On Tue, Oct 09, 2012 at 02:50:58PM +1100, Dave Chinner wrote:
> > From: Dave Chinner <dchinner@xxxxxxxxxx>
> > 
> > Add an AGFL block verify callback function and pass it into the
> > buffer read functions. Add a debug only check for valid block
> > numbers in the AGFL.
> 
> The check isn't executed only for debug builds, but rather all
> the time.  I'm not sure that this actually is a good idea.

RIght, I need to update the commit message to indicate that the
debug only checks are being made unconditional. That is, we're going
to detect the AGFL being invalid on production systems now.

Realistically, I'd much prefer that we have more robust verification
of on-disk structures when we pull them into memory regardless of
CRC functionality. We don't verify the AGFL at all, and given that
it generally stays resident in memory, a small amount of iextra
overhead when they are read isn't going to be noticable.

Cheers,

Dave.
-- 
Dave Chinner
david@xxxxxxxxxxxxx

<Prev in Thread] Current Thread [Next in Thread>