2012-03-20 19:47:59

by Ben Greear

Subject: avg-signal is weird in ath9k in 3.3.0

I'm testing out some code to get a stations's avg-signal through ethtool-stats,
and both it and /proc/net/wireless show ranges bouncing all over
(from around -48 to -110) on an interface running heavy traffic.

Are there any known bugs in 3.3.0 related to this?

Thanks,
Ben

--
Ben Greear <[email protected]>
Candela Technologies Inc http://www.candelatech.com

2012-03-20 22:22:59

by Felix Fietkau

Subject: Re: avg-signal is weird in ath9k in 3.3.0

On 2012-03-20 8:47 PM, Ben Greear wrote:
> I'm testing out some code to get a stations's avg-signal through ethtool-stats,
> and both it and /proc/net/wireless show ranges bouncing all over
> (from around -48 to -110) on an interface running heavy traffic.
>
> Are there any known bugs in 3.3.0 related to this?
As far as I know, this bug has always been there (RSSI values for A-MPDU
subframes are invalid, except for the last subframe).
I've sent fixes for this, which will show up in 3.4

- Felix

2012-03-21 02:12:09

by Ben Greear

Subject: Re: avg-signal is weird in ath9k in 3.3.0

On 03/20/2012 03:22 PM, Felix Fietkau wrote:
> On 2012-03-20 8:47 PM, Ben Greear wrote:
>> I'm testing out some code to get a stations's avg-signal through ethtool-stats,
>> and both it and /proc/net/wireless show ranges bouncing all over
>> (from around -48 to -110) on an interface running heavy traffic.
>>
>> Are there any known bugs in 3.3.0 related to this?
> As far as I know, this bug has always been there (RSSI values for A-MPDU
> subframes are invalid, except for the last subframe).
> I've sent fixes for this, which will show up in 3.4

I found the two patches and they apply cleanly to 3.3.0, and it
does appear to fix the problem.

Thanks,
Ben

--
Ben Greear <[email protected]>
Candela Technologies Inc http://www.candelatech.com

2012-03-20 20:47:02

by Ben Greear

Subject: Re: avg-signal is weird in ath9k in 3.3.0

On 03/20/2012 12:47 PM, Ben Greear wrote:
> I'm testing out some code to get a stations's avg-signal through ethtool-stats,
> and both it and /proc/net/wireless show ranges bouncing all over
> (from around -48 to -110) on an interface running heavy traffic.
>
> Are there any known bugs in 3.3.0 related to this?

Same with iw station dump:

[root@lec2010-ath9k-1 lanforge]# iw dev sta1 station dump
Station 00:88:99:88:99:01 (on sta1)
inactive time: 2 ms
rx bytes: 3646300587
rx packets: 10768068
tx bytes: 3444260988
tx packets: 10706688
tx retries: 187471399
tx failed: 2
signal: -47 dBm
signal avg: -64 dBm
tx bitrate: 130.0 MBit/s MCS 15
[root@lec2010-ath9k-1 lanforge]# iw dev sta1 station dump
Station 00:88:99:88:99:01 (on sta1)
inactive time: 1 ms
rx bytes: 3655139651
rx packets: 10775853
tx bytes: 3452330988
tx packets: 10714184
tx retries: 187604280
tx failed: 2
signal: 16 dBm
signal avg: -92 dBm
tx bitrate: 130.0 MBit/s MCS 15

Thanks,
Ben

>
> Thanks,
> Ben
>


--
Ben Greear <[email protected]>
Candela Technologies Inc http://www.candelatech.com