Big delays between an AR5416 AP and its clients

Gustau Pérez gperez at entel.upc.edu
Sat Dec 3 17:47:17 UTC 2011


On 03/12/2011 17:22, Adrian Chadd wrote:
> 2011/12/4 Gustau Pérez<gperez at entel.upc.edu>:
>
>>    I see nothing particular in my dmesg with wlandebug enabled. I see the
>> AR5416 complaining about a beacon off-channel for a MAC I don't know about.
>> I also see the iphone's MAC going power save with a message like this:
>>
>>              power save mode on, 2 sta's in ps mode
> Well, it _is_ doing power saving, which I expected. This doesn't at
> all correlate with the periodicity of the issue?

    No it doesn't. The period of the issue was 0.1sec, because it 
happens after 10 pings and the time between pings was 0.01 seconds. 
During the tests the power save mode message only appeared 4 times, and 
they did during a second and then never appeared again. As I was doing 5 
times 6000 pings and the problem repeats every 10 pings, I would have 
expected the message about 5*600 times. So I would say they don't 
correlate at all.

    During normal operation the power save mode on also appears. Other 
stations also trigger that message.

>
>>   However, I see these the following entries in my dmesg when using r227364:
>>
>>         Dec  3 16:46:27 harkserver kernel: Timeout while waiting for nf
>>         to load: AR_PHY_AGC_CONTROL=0x41d1a
>>         Dec  3 16:46:27 harkserver kernel: Timeout while waiting for nf
>>         to load: AR_PHY_AGC_CONTROL=0x41d1a
>>
>>         Dec  3 16:18:27 harkserver kernel: ar5416PerCalibrationN: NF
>>
>>         calibration didn't finish; delaying CCA
>>         Dec  3 16:18:27 harkserver kernel: ar5416PerCalibrationN: NF
>>
>>         calibration didn't finish; delaying CCA
>>
>>   I don't see those messages with a previous revision because they appear to
>> be part of r227364. Those problems appear from time to time, could they and
>> the latency increase be related?
> That could be related. If the device thinks the air is constantly busy
> (or it's spending a lot of time trying to TX) then it may fail NF
> calibration.
> I just wonder what I've changed which may be causing the NIC to behave this way.
>
> What's interesting (!) is that I have seen similar reports with ath9k
> in Linux. I wonder if part of the 11n work (that ath9k has also
> borrowed) has the same broken behaviour on the AR5416.

    Well, if they did have that issue, maybe we could also borrow their 
solution. Did they fix the issue. Do you have any link I could read and 
try while you finish unpacking yourself :) ?

> Can you please file all of this in the PR? I'll have finished moving
> by December 15 so I can start trying to replicate this issue and help
> you debug it.
>
>
    You can count on this. I'll fill the PR with the collected info. It 
may get quite long, so maybe I'll fill it tomorrow.

    Thanks,

    Gus


More information about the freebsd-wireless mailing list