[linux-cifs-client] Can't mount smb shares using mount.cifs with 2.6.31 kernel

[Jeff Layton]

On Fri, 16 Oct 2009 12:06:59 -0400
Timothy Normand Miller <theosib at gmail.com> wrote:

> I'm sending you the captures privately.  The first one is just a
> mount.  The second one is a mount with an ls.
> 
> On Fri, Oct 16, 2009 at 11:43 AM, Jeff Layton <jlayton at redhat.com> wrote:
> > On Fri, 16 Oct 2009 11:35:07 -0400
> > Timothy Normand Miller <theosib at gmail.com> wrote:
> >
> >> On Fri, Oct 16, 2009 at 10:50 AM, Jeff Layton <jlayton at redhat.com> wrote:
> >> > On Fri, 16 Oct 2009 10:32:39 -0400
> >> >
> >> > A little. Can you try mounting again with "-o noserverino"? Perhaps
> >> > something is wrong with the inode numbers being generated by the server
> >> > here...
> >>
> >> That fixed the problem.  It mounts properly now.
> >>
> >> > If that doesn't help, then maybe also capture some debugging info of
> >> > you doing an "ls" in the top-level dir after doing the mount.
> >>
> >> I can't tell where one set of debug messages end and where the next
> >> set starts.  Is there something I can do to insert a dummy message
> >> into dmesg so I can tell where to start copying?
> >>
> >> > I doubt this is anything you've done. There were a number of changes in
> >> > how inodes are handled under CIFS between 2.6.30 and 2.6.31. It's quite
> >> > possible that one of those changes is causing problems with this server.
> >>
> >> Thanks for the help.  Is there anything else I can do to diagnose the
> >> cause so it can be fixed permanently?
> >>
> >>
> >
> > Yes, could you send me a wire capture of a mount on the 2.6.31 kernel
> > without the "-o noserverino" option? I need to have a look at why the
> > GetSrvInum call isn't doing the right thing here. There are
> > instructions on the troubleshooting page for doing this.
> >
> > Thanks,
> > --
> > Jeff Layton <jlayton at redhat.com>
> >
> 
> 
> 

Thanks for the info. The mount seems to be successful. The ls makes
CIFS issue a "FindFile First" which is how SMB does a READDIR on the
wire. The server sends the response and then after that there is no
further traffic. I assume that CIFS doesn't like that response for some
reason, but wireshark unfortunately does not yet implement a dissector
for the FFF infolevel in use here.

I'm going to see if I can fix that in wireshark and then may have a
little more info. What may help in the meantime is the output from
cifsFYI debugging turned up while doing an "ls" after a fresh mount
without "-o noserverino".

Thanks,
-- 
Jeff Layton <jlayton at redhat.com>