Red Hat Bugzilla – Bug 135629
wireless nic and builtin nic both claim eth0
Last modified: 2007-11-30 17:10:51 EST
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.2)
Description of problem:
I am using a thinkpad T21 with builtin nic (Ethernet Pro 100 according
to lspci), which claims eth0 using the e100 driver. I also have a
pcmcia US Robotics wireless card that uses the orinoco_cs driver.
If I insert the wireless card before enabling the wired ethernet, the
wireless driver claims eth0, but fails to function. If I then remove
the wireless card and run "ifconfig", the mac address of the wired
eth0 is displayed, but without any of the usual statistics (RX
packets, TX packets, etc...). If I then enable the card with
"ifconfig eth0 up" and insert the wireless card, the wireless card
claims eth1 (as it should) and works correctly.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
2.insert wireless card
Actual Results: wireless card fails to function, claiming a device
name already in use by builtin card (eth0)
Expected Results: device claims device name eth1 and works correctly
/etc/modprobe.conf has a line:
alias eth0 e100
I have similar problems with a T30 laptop. The difference is that my wireless is built-in
(Prism 2.5) so I cannot "pop out" the card !!!
This problem is critical because it causes networking to fail, or become funky, with little
recourse to fixing it. (for example, when wired networking moves to eth1, software
locked using FlexLM no longer recognizes the MAC address on eth0).
I would like to provide some additional clues:
1. Per newsgroups, it seems to be affected by *availability of nearby wireless networks*.
I have struggled most of this afternoon trying to get the wired connection back onto eth0.
Finally, i pulled the plug on my wireless router and it *immediately* went back to normal
behavior. I rebooted a couple of times and it solidly remained on eth0.
2. Regression test - I plugged the wireless base stn. back in and rebooted. kudzu did
not do anything out of the ordinary.
3. I see that there are also submitted bugs about the Prism2.5 card showing up as
"Ethernet" preventing proper choice of a driver and proper initialization of wireless
settings. I have that too.
4. system-config-network also insists on displaying "eth2" with the Intel EtherExpress/
100 driver. I never put it there, and eth2 doesn't show up in kudzu -p but somebody
wrote it into modprobe.conf.
system-config-network should be binding interfaces to NICs based upon their MAC
address. This doesn't look like a kernel bug.
You should execute system-config-network and bind the interface name
to the MAC address!