Re: Wireless network problem with Access Point selection using ifup

[Mark Hounschell <dmarkh@cfl.rr.com>]

On 12/10/2012 06:29 AM, Johannes Berg wrote:
> On Mon, 2012-12-10 at 06:10 -0500, Mark Hounschell wrote:
>
>> The driver is use has was identified in the original email. So I guess this
>> is the right place to report this issue?
>
> Yeah, I guess Larry should take a look.
>
>> In each of the wpa log files I See
>> in the beginning, a section like this:
>>
>> nl80211: Received scan results (4 BSSes)
>> wlan0: BSS: Start scan result update 1
>> wlan0: BSS: Add new id 0 BSSID c8:d7:19:7a:ab:1e SSID 'hounschell-1'
>> wlan0: BSS: Add new id 1 BSSID 20:aa:4b:b2:dd:79 SSID 'hounschell-1'
>> wlan0: BSS: Add new id 2 BSSID 00:18:f8:ff:72:d5 SSID 'hounschell-1'
>> wlan0: BSS: Add new id 3 BSSID 00:1a:70:6d:99:50 SSID 'MACDONALD'
>>
>> Every time the first one listed is the one actually selected and each time
>> is different and usually NOT the closest.
>
> That's expected, the order is random by the order in which it was found
> in scanning etc. There's no inherent order here.
>
>> It is a Desktop in a fixed location. The AP is within 15ft of the ASUS USB
>> adapter and in direct line of sight.  There may be more than one problem
>> here. I think one of the problems is right here. Above 20:aa:4b:b2:dd:79 is
>> the closest AP. Why is it not first in the list?
>
> See above.
>
>> Why, when it selects one
>> of the others and it can't figure out it selected the wrong one and
>> reselect the right one seems like a different problem. After all when it
>> does select the correct one, all is good but the actual bandwidth.
>
> Well so I suspect the driver is also not reporting useful signal
> strength so that wpa_s doesn't know which one to select. And then when
> the connection fails, the driver is refusing scanning so it can't even
> reconnect to a better AP ...
>
> IOW, all driver problems, and I'm not familiar enough with that driver
> to tell you what's going on.
>

Should I file and bug report and/or contact the author?

Mark