Re: wpa_supplicant: lagg0 not working anymore (if_iwm/if_iwlwifi)

From: FreeBSD User <freebsd_at_walstatt-de.de>
Date: Mon, 10 Jul 2023 19:39:05 UTC
Am Mon, 10 Jul 2023 00:46:45 +0000 (UTC)
"Bjoern A. Zeeb" <bzeeb-lists@lists.zabbadoz.net> schrieb:

> On Sun, 9 Jul 2023, FreeBSD User wrote:
> 
> > Hello,
> >
> > running 13-STABLE (13.2-STABLE #29 stable/13-n255792-d295e418ae7e: Sun Jul  9 10:08:25 CEST
> > 2023 amd64) on a Lenovo T560 with Intel i8260 WiFi chipset, combining the LAN (if_em) and
> > WiFi (if_iwm) via if_lagg for convenient switching between LAN and WiFi APs of our
> > departments, worked fine until recently.
> >
> > Althoug running wlan0 with several debug options, I can't fathom what is going wrong, there
> > seems to be an authentication problem.  

Hello,

> 
> If lagg stopped working with iwm (what's your last good version for
> that?) then this is likely an issue outside of wireless as "legacy"
> wireless (iwm or net80211) haven't changed much if at all in the last
> months.  Networking on the other hand had a reasonable amount of changes.
> 
> Would be interesting for someone to use two ethernet and see if it works
> there?  Probably asking on freebsd-net as a start would be good.
> 
> The logs below are hard to read given they are wrapped but it seems
> there's little information on lagg in there?
> 
> Can you explain the actual problem you encounter a bit more?

I use a "classical" straight forward failover configuration between NIC (em0) and WiFi (iwm0
or, now, iwlwifi0).

Last time I changed something on that config lies in the past around a year for now when
trying if_iwlwifi on the T560 Laptop with i8260 WiFi chipset, which didn't work well with
iwlwifi that time. I do mention this particular fact since I update the software (13-STABLE)
from time to time to be prepared for lab-work when needed, so a couple of weeks ago I realized
that WiFi wasn't working anymore as expected (still the mentioned failover/lagg config).

As mentioned initially, configuring each interface, WiFi (no matter whether iwm0 or iwlwifi0)
works as expected. With the failover config it fails.

I found the culprit!

In my former config, the WiFi was given the MAC address of the NIC via
"config_args_wlan0="ether MAC-ADDRESS ...." or ifconfig_wlan0="ether MAC-ADDRESS-OF-NIC ...."
(latter one seems correct, since it seems omitting the ether part assigns the NICs MAC address
automatically to the WiFi, so I thing using config_args_wlan0="" is the worng place to do
so...).

Now I simply swapped this relation by giving the NIC the MAC-of-the-WiFi via "ether ..." - and
surprise, surprise, it works!

I do not know whether this is intentionally, the opposite around worked for a while since, as
mentioned, I didn't change the config for almost a year and used the Notebook in a campaign
with both WiFi and NIC via lagg/failover (my only metric here ...).

The problem is with if_iwm and if_iwlwifi the same way, so I guess it might be something
strange with if_lagg handling the ether?

Kind regards,

Oliver
> 
> Does switching from ethernet to wifi work?
> Does switching from wifi to ethernet work?
> 
> In each case what happens to packets, networking, lagg, wpa, wifi state?
> 
> What do the link states of the underlying interfaces say (in ifconfig)?
> 
> /bz
> 



-- 
O. Hartmann