[nSLUG] Anyone using RH 5.1 and cifs client mount?

George N. White III gnwiii at gmail.com
Sat Apr 5 13:36:13 ADT 2008


On Fri, Apr 4, 2008 at 2:57 PM, D G Teed <donald.teed at gmail.com> wrote:

> George, thanks for confirming it does work well in other distros/kernels.
>
>  I forgot to add that when I connect with smbclient, from the same client,
>  the user directory contains just the expected content, not echoes of all
>  user folders again. It looks like a kernel bug specific to RH's unique kernel
>  recipe (not the Colonel's unique Redhat recipe).
>
>  I know how to capture packets, but I don't know what to look for as
>  evidence there is a problem.  Maybe it is a problem with mount?
>  I could probably reboot with an older kernel to confirm the problem
>  goes away then (what I expect to see anyway).

The only similar debugging I've done is with NFS/RPC, where the
client was using signed int for handles that the protocol says are
unsigned.  It was easy to see that errors were occurring when
the handles were larger than the largest positive int.

Use single user mode with everything else shut down.  You need
some documentation of the protocol so you can look for specific
packets and generate a list of the steps in mounting a filesystem.
Repeating the test on a working configuration may help pin down
the differences.

If you can pinpoint the version where the problem appears you can
start checking individual patches.  Fedora is probably easier in this
respect since there are frequent updates, so only a small number
of patches to consider for each update.

-- 
George N. White III <aa056 at chebucto.ns.ca>
Head of St. Margarets Bay, Nova Scotia



More information about the nSLUG mailing list