[linux-cifs-client] CIFS stopped working yesterday - "ls" reports "Invalid argument"

[Paul Risenhoover]

Hi All,

I am writing because I have a Linux server (Fedora Core 5) that acts as
a client to two other machines, one a Windows 2000 machine, and the
other a Buffalo.   Both the Windows and the Buffalo have SMB file
shares, and my Linux box mounts those two file shares and then
redistributes them to the rest of the world in various ways.

The problem is that I upgraded all the software on that machine
yesterday ("yum upgrade"), and after the upgrade, it is no longer able
to see a lot of the directories on the file shares.

The Linux box connects to these machines WITH Credentials -- these are
not anonymous connections like have been reported elsewhere.

Here are some examples:

Windows file share "fileroot" is mounted at /mnt/nas/fileroot.  A
listing of a directory

[root at carbon ~]# ll /mnt/nas/fileroot/1265
ls: reading directory /mnt/nas/fileroot/1265: Invalid argument
total 0

Note this directory has 253 files, as evidenced by a similar command
executed on a different machine:

[pauly at argon ~]$ ls -l /mnt/nas/fileroot/1265 | wc -l
253

[root at carbon ~]# ll /mnt/nas/fileroot/upload/
total 1836568
-rw-rw-rw- 1 jboss nas 161136640 Nov  7 16:26 a8Ro_0HIERnOdkQSdW4Z9Q==
-rw-rw-rw- 1 jboss nas     13795 Nov  7 12:15 APTDemo.jar

Likewise, the linux machine reports the same error when attempting to
connect to the buffalo:

[root at carbon ~]# ll /mnt/buffalo/share/assets/1032
ls: reading directory /mnt/buffalo/share/assets/1032: Invalid argument
total 0


If anybody can lend any advice on this I would greatly appreciate it.
This is a mission-critical system and it's been down for a day already.
Things are going to get really hot around here.

Thanks,
Paul