[regression] hardware wifi switch kills eth0 as well

[Rolf Leggewie]

Binary package hint: network-manager

This used to work fine on my Asus 1001P netbook until about a month ago. Since then, whenever I push Fn+F2 switch to disable the wifi (and conserve power) the eth0 network connection is dropped as well. It's then gone from both "sudo ifconfig" and network-manager. Unticking "enable wireless" from the network-manager status icon works fine (but I assume that saves no power). This must be some kind of regression recently introduced in lucid. I'm always at most a few days behind for the latest packages in lucid. I had a look around /var/log/syslog, but did not spot anything unusual around the time the switch was pressed.

[Pedro Villavicencio (pedro)]

Thanks for the report, with which version of network-manager and Ubuntu are you getting this problem?

[Rolf Leggewie (r0lf)]

If you post these kind of generic questions, please be sure to actually read the original report carefully enough so that you don't miss that the requested data is already given there.

release: lucid
lucid has network-manager 0.8-0ubuntu3

[Mathieu Trudel-Lapierre (cyphermox)]

Hi Rolf,

Would you please be able to attach the contents of /var/log/syslog? Following that we may be able to understand how and why the killswitch also touches the ethernet interface (which clearly shouldn't happen).

[Rolf Leggewie (r0lf)]

Mathieu, thank you for taking a look. Here is what gets logged in syslog when I press the wifi switch (Alt+F2) while both eth0 and wlan0 have an active connection (to the same router in this case).

[Mathieu Trudel-Lapierre (cyphermox)]

Rolf,

Looking at the logs and what you are describing (and also checking NM code and discussing this with Dan), it looks like the problem isn't in NM either but rather in how the drivers for wired, wireless and killswitch interact. This bug should be reassigned to linux.

Can you attach as much as possible from syslog so that there is a bit more to work with? I suspect there may be additional kernel messages to help debugging. I'll see if I can get access to a similar Eeepc system.

[Rolf Leggewie (r0lf)]

Mathieu, thank you for taking a look. Indeed, this seems to be a kernel problem and one that is fixed in the most recent lucid kernel. 2.6.32-24-generic still exhibits the problem while 2.6.32-25-generic doesn't any longer.