Redhat comments with upcoming release of 2.2.1

Shirish Kalele kalele at
Thu May 17 17:26:55 GMT 2001

> On Thu, 17 May 2001, Shirish Kalele wrote:
> > > btw....I'm curious why the path check will be slower even with
> > > host msdfs = no.  I'll dig through the code and look when I get a
> > > In my mind, the cost should be conditional to the 'host msdfs' being
> > > enabled.  Otherwise the symlinks will be treated normally.
> > >
> >
> > Yes, it is conditional to "host msdfs" being set, but even if it is
> > set to no, we still need to make this extra check (for "host msdfs")
> > that wouldn't be there without any dfs compiled in. Now that I think
> > about it, I wonder if this boolean check counts for much though..
> I thought you were saying that the extra check was caused by an extra
> lstat or something.  I'm sure the if (lp_hosts_msdfs()) cost for somthing,
> but we would need to profile it to see how much it adds up to.
No, the superfluous lstat check is if you have local directories or files in
a dfs root share.

Looking at the code, I don't really do an explicit lp_host_mdfs() check. I
check to see if the DFS_PATHNAME flag is set in the smb header. This implies
that DFS was successfully negotiated which can happen only if "host msdfs =

While we're on the subject, I've also been thinking for a while now, about
having dfs links in subdirectories of the dfs root. Currently, you can only
have dfs links in the dfs root directory itself, although you can have
multiple dfs roots on one machine.

- SK

More information about the samba-technical mailing list