Jeff Layton wrote:
> On Fri, 09 Nov 2007 14:17:38 +0300
> Vasily Averin <vvs@sw.ru> wrote:
> 
>> Hello Steve,
>>
>> Could you please take a look at the following oops? It looks like it
>> is related to cifs:
>> 1) Code line is corrupted (21 byte?!?), It looks like part of CIFS
>> smb_hdr with CIFS magic:
>>  00 00 50 ff 53 4d 42 32 00 00 00 00 80 41 c0 <00> 00 00 00 00 00
>>           FF S  M  B
> 
> Looks like the random memory corruption I was chasing around a month
> ago. It's a nasty bug. This patch from the cifs-2.6 git tree will
> probably fix it:
> 
> commit c18c732ec6bf372aa959ca6534cbfc32e464defd
> Author: Steve French <sfrench@us.ibm.com>
> Date:   Wed Oct 17 18:01:11 2007 +0000
> 
>     [CIFS] fix bad handling of EAGAIN error on kernel_recvmsg in cifs_demultiplex_thread

Jeff,
thank you for this hint.
However I would pay your attention that our kernel is based on RHEL4, and
obviously it is vulnerabled too.

thank you,
	Vasily Averin