[linux-cifs-client] Re: [PATCH] clean legacy cruft out of connect.c

[Jeff Layton]

On Wed, 6 Aug 2008 00:13:57 -0500
"Steve French" <smfrench at gmail.com> wrote:

> I did do some cleanup to the legacy session setup function somewhat similar to
> what you modified at the end of your patch (adding a goto to remove
> the need for one level of
> nesting which cleaned up all of the > 80 column lines in the old code there).
> 
> On Tue, Aug 5, 2008 at 11:57 PM, Steve French <smfrench at gmail.com> wrote:
> > Jeff,
> > I didn't merge this since patch 9 code path looks wrong when
> >
> >   pSesInfo->server->maxBuf != 0
> >
> > In that case (reconnect) then session setup
> >      rc = CIFS_SessSetup(xid, pSesInfo, first_time, nls_info);
> > would not have been called with your patch
> >
> > when maxBuf is zero we know sending a negotiate protocol is needed
> > (and also sessionsetup) but when we know that we have
> > already negotiated and don't need to resend negprot we
> > still need to send a sessionsetup
> >
> > Although it doesn't really matter (as you note it is disabled until the user
> > turns on the experimental flag)  - I would rather wait on removing the
> > NTLMSSP implementation
> > when we readd a better one based on the documentation - since some
> > servers may require
> > NTLMSSP and it is not something we should have to upcall to do
> > --
> > Thanks,
> >
> > Steve
> >
> 
> 
> 

That works. That patch is not nearly as important as getting the
setattr patches merged. My preference would be to not worry about this
one for now, and finish merging the setattr cleanup patches since those
fix a real problem.

We can reevaluate what to do with the old stuff in there after some of
the more pressing stuff is completed...

-- 
Jeff Layton <jlayton at redhat.com>