Dear list, I am currently trying to tune my R52n ath9k minipci card for maximum performance in a \ high interference city environment. Mode is 802.11n in the 5 GHz band. What I see is better download compared to an R52 ath5k in 802.11a mode, but worse \ upload. I suspect either that the 802.11n modulation is less robust to noise OR that \ AMSDU aggregation is in use and is worsening resiliency to interference. See this \ monitoring screenshot: https://pasteboard.co/o8ykSMG.png The first part is 802.11a, \ the second is 802.11n, and the third was obtained by limiting the uplink to legacy \ rates by: iw dev wlan0 set bitrates legacy-5 6 9 12 18 ht-mcs-5 The problem: Unfortunately, the result of that is that only the 6 Mb/s rate is used. \ When I allow any ht-mcs-5 rate, none of the legacy rates is ever selected and the \ throughput is worse. It seems that the wandering minstrel algorithm is failing. The \ other side is a Mikrotik router and it freely mixes legacy and mcs rates to get \ optimal throughput, the Linux driver does not. 1. How would I debug this? Last time I \ did it in Madwifi, minstrel had some debug stats in procfs. 2. How to find out if \ AMSDU aggregation is in use and to disable it to see if that is the thing hurting \ performance? The kernel version is 4.9.0 from Debian. Thanks for your insights, Tomas Vondra
