Re: [E1000-devel] Kernel 2.6.27.10: e1000e hangs during boot at MTU change

[Brandeburg, J. <jes...@in...>]

Hi Carsten, any update?

Carsten Aulbert wrote:
> Hi Jesse,
> 
> sorry for the late reply, too much other work to rank this question
> high enough.
> 
> I was able to work around the problem by adding netconsole module
> loading to a dhclient hook.
> 
> Brandeburg, Jesse schrieb:
>>> 
>>> 0d:00.0 0200: 8086:108c (rev 03)
>>> 0e:00.0 0200: 8086:109a
>> 
>> ah, our good friends on the supermicro motherboard, right?  please
>> make sure that the eeproms have been updated.
>> 
> 
> Well spotted. We needed to upgrade the firmware due to some IPMI
> interaction problems, which versions are recommended?
> 
> n1500:~# ethtool -i eth0
> driver: e1000e
> version: 0.3.3.3-k6
> firmware-version: 0.15-4
> bus-info: 0000:0d:00.0
> n1500:~# ethtool -i eth1
> driver: e1000e
> version: 0.3.3.3-k6
> firmware-version: 0.5-7
> bus-info: 0000:0e:00.0
> 
>> the dhcp server tells the dhcp client to change the MTU?
> 
> yes, we run with MTU 9000 everywhere on the data network.
> 
>> if magic sysrq
>> works, can you take a sysrq-t (you'll need netconsole or serial
>> console to log the output) so we can see where it is frozen?  I
>> haven't heard of any issues like this for e1000e before now.  What
>> about the watchdog timers or nmi_watchdog=1 boot option?
>> 
> 
> We will try that, I'll put one of our students on it.
> 
>> I suspect that something happens to cause an infinite loop, whether
>> it is our driver's fault needs to be isolated by getting a stack
>> dump when its hung up. 
>> 
>>> Well, now the question, how to fix it - if I've provided you with at
>>> least the initial amount of information you might need.
>> 
>> not quite sure what is wrong yet.  can you get a tcpdump of the
>> traffic leading to the hang from a mirror port on a switch (or
>> another client on a hub) or anything like that?  It would help us
>> reproduce the issue here.
> 
> Yes, we have got a tap-device which we can put in between and
> duplicate all traffic.
> 
> More (hopefully) soon
> 
> Sorry again.
> 
> Carsten