small files can result in an allocation size of 0 (zero)

[Jeff Layton]

On Fri, 4 Apr 2014 15:50:13 -0700
Jeremy Allison <jra at samba.org> wrote:

> On Fri, Apr 04, 2014 at 10:30:05AM +0200, Bjoern Baumbach wrote:
> > On 03/27/2014 11:41 AM, Bjoern Baumbach wrote:
> > > On 03/25/2014 05:43 PM, Bjoern Baumbach wrote:
> > >> On 03/25/2014 05:33 PM, Jeremy Allison wrote:
> > >>> But I certainly think the patch is good enough to
> > >>> go in as is for now. 
> > >>
> > >> Okay, in this case I'll provide a new patch, since this one does not
> > >> apply to master.
> > > 
> > > I've attached a new patch, which applies to master.
> > 
> > Hi Jeremy,
> > 
> > once the patch is upstream - do you think that we can use it to fix the
> > issue in 4.0 and 4.1, too?
> 
> Maybe, let log the bug and let me think about
> the UNIX extensions implications...

Would it not be best to determine whether this really is a sparse file
if you can before reporting an allocation size of 1 block? Jan's mail
that you referenced in the original email suggested using
SEEK_HOLE/SEEK_DATA to determine that.

IOW, if I have a pile of sparse files, then not issuing reads against
them could be a significant optimization...

-- 
Jeff Layton <jlayton at samba.org>