e1000 events lost after suspend/resume (was Re: Two issues with 0.3.5)

[Derek Atkins <warlord MIT EDU>]

Derek Atkins <warlord MIT EDU> writes:

> Peter Jones <pjones redhat com> writes:
>
>> On Thu, 2005-07-14 at 13:00 -0400, Derek Atkins wrote:
>>> Hi,
>>> 
>>> I've got two issues with 0.3.5 right now (well, STABLE_0_3 branch):
>>> 
>>> 1) it's not noticing eth0.  I've got kernel log messages that the
>>>    cable is plugged in, but NM never seems to switch to the ethernet
>>>    device.  I'm wondering if dbus or hal is wedged.  Is there some
>>>    reasonable way to test this?
>>
>> What kind of card?  What driver?  Does ethtool work with it?
>
> I'm not completely convinced this is a NM issue.  I just had
> this issue again earlier today where NM wouldn't notice either
> my e1000 OR my atheros card..  It just didn't notice ANYTHING.
> I tried restarting NM, messagebus, and haldaemon and it still
> failed to find anything...

FYI, this error just recurred.  I was able to fix it by stopping and
then restarting messagebus, haldaemon, and NetworkManager.  I suspect
that either HAL or dbus is getting lodged after a suspend/resume
sequence and no longer notices the e1000 events.  It ONLY happens
after a suspend/resume.  After I restarted the services the network
came up just fine (although I lost a lot of other user state and had
to logout and login again).

So..  Restarting hal, dbus, and NM corrected the behavior..

My next test is to see whether removing/reloading the e1000 driver on
suspend/resume will fix the problem, too.

-derek

-- 
       Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
       Member, MIT Student Information Processing Board  (SIPB)
       URL: http://web.mit.edu/warlord/    PP-ASEL-IA     N1NWH
       warlord MIT EDU                        PGP key available