Red Hat Bugzilla – Bug 179982
2.6.15-1.1830_FC4 changes wlan0 to eth1? hostap broken?
Last modified: 2007-11-30 17:11:23 EST
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8) Gecko/20051201 Fedora/1.5-1.1.fc4.nr Firefox/1.5
Description of problem:
I noticed that wlan0 no longer works and it seems to be replaced with eth1. The driver identifies the card as a Prism I card instead of a Prism 2.5 card. It seems to be using the orinoco driver instead of the hostap one, even though they *both* load up.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Just insert the card.
This bug is biting me too, but I think its a udev bug, not a kernel bug, because
even if I go back to booting the previous FC4 kernel, I continue to have the
I've got a Linksys WPC11v3 PCMCIA wifi card that uses the orinoco driver. Since
the udev update to 0.71 as soon as I try to bring up the interface for the wifi
card, I start getting an un-ending stream of errors like this:
cannot change name of wlan0 to devWXYZ: File exists
where WXYZ seems to be a random number that changes each time the error appears
(and it appears many times per second, but gradually slows down over time).
This is really really bad.
Please can the submittor of this bug change the 'component' to udev and the
priority/severity to high?
It appears the 1831 update fixed the problem (unless some other component was
also updated). I have my wlan0 back! Thanks!
Its not fixed for me. Only downgrading udev avoids this.
'rpm -q udev' shows me running 'udev-071-0.FC4.2'
I wonder what else has changed. Under 1830 I tried installing hostap-drivers
and it didn't fix the problem, leading me to belive that it wasn't some sort of
kernel issue. I didn't know udev could assign the interface names.
FWIW -- I had a converse problem. Orinoco_cs which was eth1 got renamed to
wifi0/wlan0, and ifcfg-eth1 ceased to work as is.
I had to:
1) change the interface names eth1 -> wlan0
2) add "MODE=Managed", because otherwise the driver (hostap?) defaulted to
"Master" (running in AP mode)
So, the fix 1) is easy but should be automatic; 2) is a high-priority problem.
Or maybe the problem is misasscociating orinoco_cs cards to use hostap?
The problem is the hostap and orinoco drivers both claim support for the same
hardware. I'm not sure what the best way forward is here. John, any ideas?
I don't have a really good answer for this...do both drivers work equally
well? It looks like there may be hw covered by orinoco that isn't covered by
hostap? But, none of that is PCI...
Maybe notting has suggestions? I suppose we could remove configuration for
one of the drivers...hmmm...
The problem is that given a particular modalias, udev will load both drivers, so
it's "whichever matches first wins". We could selectively blacklist one of the
drivers, but I don't know enough to say which we should.
There was a recent patch upstream which touches close to this issue. Test
kernels w/ that patch available here:
Please give those kernels a try and post the results here...thanks!
I'm not sure who was the target of comment #9..
I didn't need to test the Linville kernel, because the recent 2.6.15-1.1833_FC4
kernel seemed to fix my orinoco_cs -> hostap misassociation problem by default.
Closed due to lack of response. Please reopen when the requested information
Urmm, this is very badly broken in FC5 as well. Can someone please reopen this?