[PATCH] Remove unused lgetxattr support from smbd

[Jeremy Allison]

On Tue, Apr 03, 2012 at 02:59:42PM +1000, Andrew Bartlett wrote:
> On Mon, 2012-04-02 at 14:51 +0200, Björn JACKE wrote:
> > On 2012-04-02 at 13:18 +1000 Andrew Bartlett sent off:
> > > I was reviewing our xattr vfs stack in preparation for supporting the
> > > posix:eadb format of xattr-in-a-tdb in smbd.  
> > > 
> > > This led me to check if I needed to implement all the calls, and a grep
> > > and subsequent patches indicates that lgetxattr isn't called anywhere.
> > > The attached patch removes this unused code. 
> > > 
> > > Can anyone think of a reason why we should keep this?  (I'm not really
> > > bothered either way, but just wanted to reduce complexity before I
> > > started my implementation work). 
> > 
> > I think linux does not support EAs or ACLs on symlinks but afaik *BSD systems
> > do. So it would be nice if we'd be able to support this via unix extensions,
> > then we would need them.  
> 
> That's sort of what I was asking:  For something that isn't supported in
> our TDB-based xattr databases, isn't supported on linux and would only
> be supported on some platforms, is it likely that we will come up with a
> use for this (more than just that it could be done in theory)? 
> 
> (The some platforms thing is important:  protocol extensions that only
> work sometimes have far, far less utility than things that can be relied
> on).  
> 
> > But as now we don't have that function in use
> > currently, then it seems currently safe to remove the symlink related
> > functions.
> 
> Thanks,

+1 from me. These were added within the VFS when the whole
POSIX interface was being virtualized within Samba, but
there are no cases where a client would use these (symlinks
are handled solely on the client in the POSIX extensions
other than readlink).

This code does simplify the interface, so yeah - let's push
for 4.0.x.

Cheers,

	Jeremy.