I had gotten that, but the issue is that this particular syslog excerpt
has so many things happening in it it's hard to figure out what's wrong
from what's okay: for instance, at some point wpasupplicant gets killed
with signal 15, and there's at least one instance of either a
suspend/resume cycle or a case where networking/wifi was disabled then
re-enabled. It muddies the waters a little.

I think I may have gotten to some good info; but it's nothing that
clearly points towards a wpasupplicant issue. There's two cases here:

1) You'll pretty often roam from a BSS to another due to differring
signal levels. I think we *may* want to consider whether roaming for
this little difference is really worth it (didn't really check the code
to see what kind of threshold is set).

Oct 11 17:51:44 panhandle wpa_supplicant[772]: Considering within-ESS 
reassociation
Oct 11 17:51:44 panhandle wpa_supplicant[772]: Current BSS: 00:0b:86:d5:1e:91 
level=175
Oct 11 17:51:44 panhandle wpa_supplicant[772]: Selected BSS: 00:0b:86:d5:16:51 
level=177
Oct 11 17:51:44 panhandle wpa_supplicant[772]: Trying to associate with 
00:0b:86:d5:16:51 (SSID='wpa.mcgill.ca' freq=5745 MHz)

That's part of why you shortly after that see a "deauth by local choice
(reason=3)": reason 3 is WLAN_REASON_DEAUTH_LEAVING; due to roaming or
any such cases.

2) Not all wpa.mcgill.ca APs seem to properly respond to the auth
requests; seemingly due to overloading. Consider the following instance:

Oct 11 17:47:58 panhandle kernel: [55574.563224] wlan0: authenticate with 
00:0b:86:d5:16:41 (try 1)
Oct 11 17:47:58 panhandle kernel: [55574.566014] wlan0: 00:0b:86:d5:16:41 
denied authentication (status 17)
Oct 11 17:48:02 panhandle NetworkManager[568]: <warn> Activation 
(wlan0/wireless): association took too long.
Oct 11 17:48:02 panhandle NetworkManager[568]: <info> (wlan0): device state 
change: 5 -> 6 (reason 0)
Oct 11 17:48:02 panhandle wpa_supplicant[772]: dbus: Unregister network object 
'/fi/w1/wpa_supplicant1/Interfaces/1/Networks/0'
Oct 11 17:48:02 panhandle wpa_supplicant[772]: No keys have been configured - 
skip key clearing
Oct 11 17:48:02 panhandle wpa_supplicant[772]: State: ASSOCIATING -> 
DISCONNECTED

Status 17 is "WLAN_STATUS_AP_UNABLE_TO_HANDLE_NEW_STA"; basically the AP
responds that it can't deal with the new association; more or less
meaning it's already to capacity with clients and can't handle more.
(See http://lists.shmoo.com/pipermail/hostap/2004-May/006769.html)

FWIW, I'm using the ieee_802_11_defs.h file from the wpasupplicant code
to match status and reason codes, if you're interesting in knowing what
the numbers mean ;)
(http://hostap.epitest.fi/gitweb/gitweb.cgi?p=hostap.git;a=blob;f=src/common/ieee802_11_defs.h).
You can similarly get the NM reason/status codes here:
http://cgit.freedesktop.org/NetworkManager/NetworkManager/tree/include/NetworkManager.h.

The best I can suggest at this point is to contact McGill's network
administrators; discuss the matter with them (pointing out this bug
report), and see if there's more information that can be retrieved from
comparing your system's logs with logs from the APs and auth system.
I'll keep this bug report open in case there really is something wrong
with wpasupplicant, but for now nothing clearly indicates it's the case.

Furthermore, you may wish to test this all with the latest release of
Ubuntu: Oneiric Ocelot (11.10), since there were some changes in NM and
wpasupplicant which might help (and it's just a good idea to give it a
shot anyway).

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/872578

Title:
  The infamous "deauthenticating by local choice (reason=3)"

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/wpasupplicant/+bug/872578/+subscriptions

-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to