[PLUG] Controling Network Interfaces

Mike Connors mconnors1 at gmail.com
Thu Apr 12 08:26:30 UTC 2012


>
> I could agree with this if the Sony's behavior was seen on the other 6
> portables on which I've run linux and Slackware. This Sony Vaio is unique
> in
> the persistence with which it refuses to connect to wireless networks. :-)
>
>   In the next couple of days, I'll de-activate wicd and install, configure,
> and make active at boot network manager. Gotta' try that first.
>
>   My suspicion is that the wireless chip is still new enough that the
> driver
> does not handle it under all circumstances, even with the 3.2.7 kernel.
> While kernel versions pop out regularly not all drivers are updated at the
> same frequency. I have the 3.2.11 kernel tarball here and can install it or
> a higher number kernel and see if that helps.
>
>   This chip has caused issues with different brands and models of portables
> running different linux ditributions (arch and ubuntu show up frequently in
> web searches). Some have been resolved by upgrading the kernel to 2.6.28 so
> it may well be a driver issue. But, in that case, upgrading to the 3.2.7
> kernel should have fixed the problem. Shrug.
>

To be clear the wireless NIC driver should play no role at all in the
up/down state of eth0. What if you wanted to be mult-homed to many
networks? That same Linux routing engine and tcp/ip stack is used in
high-end routers. It makes no sense to say that eth0 has to come down when
wlan0 comes up.

As I mentioned before, when I test this on my Lenovo laptop running Debian
Sid with a static config for eth0, the iface is up and there's a def gwy
listed in the routing table. Yet, I have no issues routing packets via the
wlan0 interface.

"As I understand it, the network management software should, on transition
from signal to no-signal (i.e. the cable is unplugged) should take the
interface down, and it should remain down until another transition."

=> Agree. However, the behavior is that eth0 is up and injecting a def gwy
into the routing table without a cable connected and it being in a
"running" state. This is due to the static config in rc.inet1.conf. Which
shouldn't be a problem. The problem is that ip packets coming in through
wlan0 from the WAP are attempting to go out eth0. They never make it back
to the WAP and hence Rich never gets to the T&C web page.

They way routing should work is that ip packets go back out the interface
they came in on but something is mucking up the works.

Rich - There are some settings for WICD conf (
http://linux.die.net/man/5/wicd-manager-settings.conf) file that might
help. I'm curious what the following settings are and if they're the same
on your Dell.

*link_detect_tool =* *<0|1|2>
*         0 = autodetect
         1 = ethtool
         2 = mii-tool*wired_connect_mode =* *<0|1>*0 = connect to wired
interface automatically
1 = do not connect to wired interface automatically*prefer_wired =* *
<True|False>*True = Switch to wired interface if a link is detected, even
if already connected to wireless
False = Do not switch to wired interface automatically



More information about the PLUG mailing list