Re: [PATCH] FAT: optimize FSINFO writeback

["Steven J. Magnani" <steve@digidescorp.com>]

On Wed, 2009-07-01 at 09:28 +0900, OGAWA Hirofumi wrote:
> "Steven J. Magnani" <steve@digidescorp.com> writes:
> 
> >> Well, now, we are using sb->s_dirty for fsinfo, so I'm wondering why
> >> this happen frequently.
> >
> > My scenario was modifying a sector of an existing file and using
> > fdatasync() to flush it. The FSINFO sector was being updated even though
> > nothing about the FAT layout had changed.
> 
> I see. Probably, I'm missing something, or handling of sb->s_dirt may be
> buggy, or something.
> 
> If it was fixed, is this patch still needed?  I guess this patch would
> still be useful on some case though.  If you can explain, it would be
> good.

It is I who am missing something. The patch originated against a 2.6.20
kernel, where it does indeed suppress unnecessary updates. I saw the
same code in the 2.6.30 kernel and assumed the same issue was present,
and tested that with the patch present there were no unnecessary
updates. It appears that there are no unnecessary updates even _without_
the patch, so I withdraw it.

> 
> Thanks.

Thanks for looking at this so carefully.

Regards,
------------------------------------------------------------------------
 Steven J. Magnani               "I claim this network for MARS!
 www.digidescorp.com              Earthling, return my space modulator!"

 #include <standard.disclaimer>