Will Keaney wrote: > Bob Copeland wrote: > >>> I also have to manually set the reg domain to US every time I load the >>> ath5k module. �I tried configuring wpa_supplicant.conf to do this >>> itself, but that doesn't seem to work. �It would be nice to have a >>> module parameter for setting this at load time. >>> >>> >> I may be wrong but I think that only works in wpa_supplicant if you >> use -Dnl80211 (instead of -Dwext). >> >> FWIW in the default world roaming domain you can still connect to most >> APs. Besides iw/wpa_supplicant, pending patches to ath5k will set the >> default regdomain automatically based on the value stored in the eeprom. >> >> > Ahh, interesting. I didn't see nl80211 documented in the wpa_supplicant > example configs or man page. I'll give that a shot. > I seem to get much better scanning results with the region set properly, > than on the world roaming domain. > I don't know if it's relevant or not, but when using -Dnl80211 I see a bit of odd behavior: - wpa_supplicant is able to set the region automatically. Yay! - scanning produces quite an extensive list of APs, but most of them don't have SSIDs listed. This isn't in itself too strange, since I know that there are 2 networks nearby each with several repeaters that do not broadcast SSID. However, one of them (chmc_guest) is identified in the scanning output when using -Dwext. - All of the networks show a signal strength of 0. -Dwext shows varying signal strengths for each network. Considering that I am unable to associate with any of them, I have to wonder which driver is producing spurious output. I would also like to note that the maximum Tx Power of the card is significantly different between madwifi and ath5k. In madwifi, it maxes at 17dBm; in ath5k, it maxes at 27dBm. I don't know if it actually is using these values, or if one or both drivers are reporting Tx power incorrectly. Will
Attachment:
signature.asc
Description: OpenPGP digital signature