Thanks, Jouni. I see you're correct regarding the countermeasures not
being triggered in that log.

There are actually two different scenarios that may or may not be
related: the MIC failures when multiple stations are associated within a
tight window, and the periodic MIC failures when stations are connected
for long periods of time (which may increase in frequency with an
increase in clients).

The AP log from comment #7 is an example of the second scenario, and
looking at last night's run, we're coming up on 15.5 hours with three
MIC failures and no countermeasures triggered.

Here's a snippet from the AP log from the first scenario (tight window
association):

May 26 16:00:15 Severity:5      Associated with station 1c:4b:d6:a3:75:60
May 26 16:00:15 Severity:5      Installed unicast CCMP key for supplicant 
1c:4b:d6:a3:75:60
May 26 16:00:16 Severity:5      Associated with station 1c:4b:d6:55:e8:01
May 26 16:00:16 Severity:5      Installed unicast CCMP key for supplicant 
1c:4b:d6:55:e8:01
May 26 16:00:16 Severity:1      MIC integrity error (reported from STA) 
src_addr=1c:4b:d6:55:e8:01
May 26 16:00:16 Severity:5      Disassociated with station 1c:4b:d6:55:e8:01
May 26 16:00:17 Severity:5      Rotated TKIP group key.
May 26 16:00:20 Severity:5      Deauthenticating with station 1c:4b:d6:55:e8:01 
(reserved 2).
May 26 16:00:20 Severity:5      Disassociated with station 1c:4b:d6:55:e8:01
May 26 16:00:23 Severity:5      Associated with station 1c:4b:d6:55:ef:bf
May 26 16:00:23 Severity:5      Installed unicast CCMP key for supplicant 
1c:4b:d6:55:ef:bf
May 26 16:00:23 Severity:1      MIC integrity error (reported from STA) 
src_addr=1c:4b:d6:55:ef:bf
May 26 16:00:23 Severity:1      Start TKIP countermeasures mode.
May 26 16:00:23 Severity:5      Deauthenticating with station 1c:4b:d6:55:ef:bf 
(reserved 14).
May 26 16:00:23 Severity:5      Deauthenticating with station 1c:4b:d6:a3:75:60 
(reserved 14).
May 26 16:00:23 Severity:5      Disassociated with station 1c:4b:d6:55:ef:bf
May 26 16:00:23 Severity:5      Disassociated with station 1c:4b:d6:a3:75:60
May 26 16:00:23 Severity:5      Rotated TKIP group key.
May 26 16:01:23 Severity:1      End TKIP countermeasures mode.

I think especially if you look at 1c:4b:d6:55:ef:bf, you'll see the CCMP
pairwise key for the station, but the group key is clearly still TKIP,
and as you noted the TKIP group key seems to be rotated whenever a
station comes and goes. But the station can still screw up the group key
rotation and trigger the TKIP countermeasures.  We're seeing this
behavior on our Cisco APs as well, but the Apple AirPort is handy for
the test.

I believe all the failures are similar and that the variance we've seen
has to do only with the number of stations. The population suffers from
random disassociation/reassociation (either the no probe response or the
MIC failure), which then triggers a group rekey, which increases the
odds of another random disassociation/reassociation in another member of
the population.  Someone with better math than me could probably work
out how this plays out as the population increases, but we've seen it's
pretty painful in groups of 30.

I'll jump on the wpa_supplicant debug verbosity right now.

-- 
ath9k module causing MIC challenge failures
https://bugs.launchpad.net/bugs/580753
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.

-- 
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to