Hello First of all I changed today the WLM54AGP23 card to CM9 and same result. I tried also WLM200NX and with this card nor clients in 2.4GHz nor in 5GHz connects but this test was only rudimentary... Will have to see again in labor ... BUT CM9 and WLM54AGP23 has the same issue so it's definitvly a driver or hostapd problem ...
On 11/15/2013 08:21 PM, Nick Kossifidis wrote: > 2013/11/15 Beat Meier <mbe...@swiss-wireless.com.ar>: >> Hello again >> >> Have done a lot of tests and there is no chance. WLM54AGP23 in the 2.4GHz >> band does not work right... >> What I see is that if hostapd using ath5k and WLM54AGP23 clients with >> madwifi and WLM54AGP23 will not connect in outdoor environment >> I have enabled debugfs but don't see anything. How can you see if a cliente >> tries to connect? >> At the end I will put the output of kernel.log >> >> Using any mode of any (disable/high/low) does not change anything. Not for >> the only connected client signal nor will connect any other client. >> >> In the labor I have no problems apart from lower signal put outdoor it does >> not work... >> My clients signal on the outdoor have the following signals on madwifi... >> >> X: -80 # this is the only which connects to ath5k driver too >> Y: -76 >> Z: -68 >> Nor Y nor Z connect to ath5k driver (y,z use madwifi) >> > Try running hostapd in foreground with verbose debugging on, this > sounds more like a hostapd bug than a driver bug, also the signal i > think is not relevant here, maybe X is the first that tries to connect > or something. Try to debug this further please, if you have only Y or > only Z do they connect ? I have other client which will connect which don't have madwifi... I need to install a system with ath5k instetad of madwifi as client and see ... I don't think this is a hostapd problem because i have no associacion in debug level of driver or am I wrong that I must see the driver first befor hostapd? I have the messages moving STA XX to state Y and the only one is the connected one... I thought ath5k drive will put this because I enabled debuging of driver but must have a closer look... Other strange behaviout is that I cannot enable ath5k debugging in debugfs... I have to instert the parameter loading the module i.e. cd /sys/kernel/debug/ieee80211/phy0/ath5k echo 0xffffffff > debug does not work.. debug is still on 0 BTW: I'm using kernel-3.8.13 > >> More strange the tests below... I really don't know were to go... >> Any help with this cards??? >> >> Here my output tests of the labor >> >> ########################################################### >> >> CM9 Tetst Channel 1 >> ******************** >> >> on master (CM9/19dbm): madwifi -34 with wlanconfig >> on sta (CM9/19dbm): ath5k -30 with wlanconfig # set txpower betwenn 15-19db >> same signal on master >> > What do you mean ath5k with wlanconfig ? wlanconfig is a madwifi > related tool, to play with ath5k you should use iw. Sorry... Cut & paste error from old tests.. of course iw... >> on master (CM9/19dbm): madwifi -29 with wlanconfig >> on sta (CM9/19dbm): madwifi -32 with wlanconfig # set txpower to 19 >> >> on master (CM9/19dbm): ath5k -31 with wlanconfig >> on sta (CM9/19dbm): madwifi -36 with wlanconfig # set txpower to 19 >> >> echo "sens-high" > ani on master >> not anymore connection possible from client with madwifi >> not even after setting ani-on and restart interface!! >> >> >> on master (CM9/19dbm): ath5k -39 / 3.9Mbps >> on sta (CM9/19dbm): ath5k -33 / 3.1Mbps >> >> Without trafic signal on client drops to -77!! >> Setting all 4 ani states on master and restart interface does not change >> signal nor on client nor on master >> The same on client (while on master ani off) >> > I noticed that in all your tests above the signal strength is too > much. If a signal is too strong (-40 and above, depends on the card) > it may result distortion on the receive path and result lower > throughput. Signal strength is measured at the "entry" of the phy > chain, before any processing is being done, you can have great signal > but no traffic at all -e.g. because your intermediate frequency is > jammed or gets a lot of noise-. Also giving a high tx power might also > result distortion on the tx path. > > To make sure that's not the case please make sure you have a tx power > around 15db to begin with and that the signal strength on your > receiver is -60 to -77 and retest. I had never problems in the past nor with madwifi nor with ubuiquity... I do tests with signal o -15 in the labor no problem with madwifi ... Will do test as you noted nevertheless!! >> WLM54AGP23 >> ************** >> Channel 1-6 does not work with ath5k!!!! >> ping time about 700-3000ms >> trafic about 100kbps >> I have changed both cards the new one the same effect so no card is >> defect!!! >> On outdor this channel works... More than strange to me!!! >> > Maybe because the signal strength is too much indoors (see above) ;-) > >> Channel 7-11 works >> on master (WLM54AGP23/23dbm): ath5k -36 / 3.9Mbps >> on sta (WLM54AGP23/23dbm): ath5k -26 with trafic goese to -32 / 3.1Mbps >> scanning shows ap with signal of -29 >> without trafic signal -26 on client after 5sec drops to -45 go -27 etc. >> >> Test with madwifi >> on master (WLM54AGP23/23dbm): madwifi -28 / 3.9Mbps >> on sta (WLM54AGP23/23dbm): madwifi -23 with trafic goese to -32 / 3.1Mbps >> > Again your signal strength is too much. Get a normal signal strength > of -60 to -77 and see how it goes. Just this seems to be the problem outdoor. Clients which have such a signal will not connect... What can I do apart from running hostapd in debug mode? > _______________________________________________ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel
