[Unionfs] Another NULL pointer problem with unionfs-2.1.4_for_2.6.23-rc7

[Erez Zadok]

In message <46F3CF80.3060502 at ccs.nrl.navy.mil>, Jesse I Pollard writes:

> Had another one:
> 
> unionfs mount:
> 
>     mount -t unionfs -o dirs=/mnt/torpedo-new=rw:/mnt/torpedo-old=ro 
> none /mnt/torpedo
> 
> torpedo-old is a 6 TB file system - NFS mounted, torpedo-new is a local 
> directory. I was not trying to copy files, just doing:
> 
> find /mnt/torpedo -depth -type d -print
> 
> I wanted to use the output to look at the directory tree, and thought it 
> would be a nice test (it has
> failed before). The find was started at about 5:30PM, the crash occured 
> almost 4AM the next
> morning (today).
> 
> df shows: (without the unionfs mount)
> 
> /dev/sda1             15227008   8710172   5730860  61% /
> tmpfs                   517312         0    517312   0% /dev/shm
> /var/odf.fs             507748      2353    479181   1% /odf
> crete:/vol/torpedo   6495730240 4274851328 2220878912  66% 

ODF?  Are you using the ODF code or the latest non-odf code as per subject
line.  Just verifying.

> /mnt/torpedo-old# df -i
> 
> and df -i:
> # df -i
> Filesystem            Inodes   IUsed   IFree IUse% Mounted on
> /dev/sda1            3932160  319076 3613084    9% /
> tmpfs                 129328       1  129327    1% /dev/shm
> /var/odf.fs           131072      49  131023    1% /odf
> crete:/vol/torpedo   31876689 9264548 22612141   30% /mnt/torpedo-old
> 
> Attached is the .config for the system.
> 
> Syslog recorded:

Jesse, can you run dmesg and cut get me the entire oops trace?  syslog may
not record everything, and may not record messages in the right order or all
of the last ones; I'm missing the rest of the oops trace, where it gives me
the actual function in which it failed and the hex offset within the
function.

Thanks,
Erez.