iDrac 6 shared net link hang with CentOS 6.7 and OMSA 8.1 from DSU

Hi,

I am currently investigating installing OMSA 8.1 on our servers using 
the DSU repository. My test setup is an R410 with an iDrac6 express. The 
iDrac is sharing the network interface with the OS, but indeed on a 
different VLAN. The OS is CentOS 6.7 (actually CentOS 6.6 with the 6.7 
updates as CentOS 6.7 doesn't exist as such yet). The kernel is 
2.6.32-573.1.1.el6.x86_64.

This setup is working fine with OMSA 7.4.1 from the DLR repository.
The OS boots fine and the ssh connection is reliable. I can see the OS 
booting to the login prompt through the iDrac console and then I can log 
in and stay connected reliably.

I have then installed OMSA 8.1 with the DSU repository (after manually 
removing OMSA 7.4).
Just like with OMSA 7.4, the OS boots fine and the ssh connection is 
reliable. I can see the OS booting to the login prompt through the iDrac 
console and then log in.
However, after a few minutes, the iDrac console would just hang, while 
the ssh connection to the server is still reliable. At this point, iDrac 
is not answering to ping anymore.
Just before the iDrac net link goes done, I have the following lines in 
the kernel log :
bnx2 0000:01:00.1: firmware: requesting bnx2/bnx2-mips-09-6.2.1b.fw
bnx2 0000:01:00.1: firmware: requesting bnx2/bnx2-rv2p-09-6.0.17.fw
bnx2 0000:01:00.1: em2: using MSIX
ADDRCONF(NETDEV_UP): em2: link is not ready

I have tried the disable_msi driver parameter, but it doesn't help.

Reverting to the latest CentOS 6.6 kernel (2.6.32-504.30.3.el6.x86_64) 
or any other CentOS 6.6 kernel fixes the issue. The other CentOS 6.7 
kernel (2.6.32-573.el6.x86_64) also exhibits the issue.

I noticed the bnx2 driver was updated from version 2.2.4 to version 
2.2.5 between RHEL/CentOS 6.6 and 6.7, although this change is not 
documented in the technical notes. I guess this is likely the culprit.

Another note, all firmware (including bios 1.12.0, net 7.12.17 and idrac 
1.99) have been updated in an attempt to fix the issue, but it turned 
out later this doesn't change anything.

Does anyone have any insight on what is going wrong ?
Can anyone reproduce the hang either on CentOS or RHEL 6 ?

I'll now try to reproduce this on an R710 with an iDrac6 Express.

Regards,
Xavier