smbmount -> system crash ?

[Urban Widmark]

On Sat, 9 Dec 2000, Zoltan Toth wrote:

>   Dec  9 17:24:09 deotron kernel: Call Trace: [lookup_dentry+365/504]
> [do_wp_page+349/416] [open_namei+109/892] [do_fork+1634/2156]
> [filp_open+74/272] [filp_open+17/272] [sys_open+63/160] 
>   Dec  9 17:24:09 deotron kernel:        [sys_open+38/160]
> [common_interrupt+24/32] [error_code+45/64] [system_call+52/64] 

This oops may have been mangled by klogd, filp_open doesn't call do_fork
and the EIP points to the ne2k-pci module.

Can you modify your startup scripts to run 'klogd -x' and not just plain
klogd? This will cause it to dump the information in a "raw" format. Then
repeat this oops (vs 2.2.18-pre25 if you can), copy it to file and run it
through ksymoops. Send me the result.


> Additionally:
> - if I do a df, it hangs when trying to get the info about the samba
> file system, and I cannot kill it (not even with kill -9)

It probably crashed while it was holding some lock, all attempts to do
something with the fs after that will block and wait for the crashed
process. Not sure about the reboots.

> I use the kernel that came with SuSE 7.0 (2.2.16) and samba version
> 2.0.7-72.

There are some oopsable bugs in smbfs in 2.2.16 that are fixed in
2.2.18-pre. Please test the latest 2.2.18-pre version first.
(http://www.hojdpunkten.ac.se/054/samba/pre-patches.html)

/Urban