Serious e1000e driver issue on Intrepid

[Tim Gardner]

Daniel J Blueman wrote:
> Hi Tim,
> 
> [cross-posting to LKML]
> 
> On Tue, Sep 23, 2008 at 2:16 PM, Tim Gardner <tim.gardner at canonical.com> wrote:
>> Folks,
>>
>> There are beginning to be an increasing number of reports regarding a
>> serious regression in Intrepid with regard to the e1000e ethernet driver
>> wherein it can be rendered permanently inoperable. The original Bugzilla
>> report is here:
>>
>> http://bugzilla.kernel.org/show_bug.cgi?id=11382
>>
>> SuSE has also made an announcement regarding this issue:
>>
>> http://news.opensuse.org/2008/09/22/serious-e1000e-driver-issue-in-sle-11-beta-1-and-opensuse-111-beta-1
>>
>> Given the severity of this issue, I'm inclined to blacklist the driver
>> by default in modutils until such time as Intel develops a fix.
>>
>> How do we inform users when their ethernet suddenly stops working after
>> an update?
> 
> The 2.6.27-rc7 changelog shows this problem has been fixed for e1000:
> http://www.kernel.org/pub/linux/kernel/v2.6/testing/ChangeLog-2.6.27-rc7
> 
> and this patch in the -mm series fixes it for e1000e:
> http://marc.info/?l=linux-mm-commits&m=122038324200305&w=2
> 
> Clearly, we need to ensure the second patch makes it into 2.6.27
> final; the corruption is reported to occur during suspend/resume
> cycles.
> 
> Thanks,
>   Daniel

LKML is aware of the issue, so I don't think cross-posting is necessary.
Furthermore, the patch in the -mm series is against the e1000 driver,
not to be confused with the e1000e. The ICH8 hardware has a method for
hardware level exclusion that is used by the e1000e driver, so the issue
is likely more complex then it seems at first blush.

rtg
-- 
Tim Gardner tim.gardner at canonical.com