[iwn] Cedric's stuff is in -HEAD

[hiren panchasara]

On Tue, Nov 12, 2013 at 12:19 PM, Adrian Chadd <adrian at freebsd.org> wrote:
> On 12 November 2013 11:13, hiren panchasara <hiren.panchasara at gmail.com> wrote:
>> On Tue, Nov 12, 2013 at 10:45 AM, Adrian Chadd <adrian at freebsd.org> wrote:
>>> Run with wlandebug +assoc +state +debug +output +scan +crypto
>>
>> http://people.freebsd.org/~hiren/20131112_iwnfail_atwork.txt
>
> What's this showing? What was going on at this time?

I keep pings going on another window just to see when it stops.

It was doing alright (pings going through) and suddenly "stops
working" around following lines in logs:

Nov 12 10:59:51 flymockour-l7 kernel: wlan0: ieee80211_bg_scan: active
scan, ticks 7165936 duration 150
Nov 12 10:59:51 flymockour-l7 kernel: wlan0: [6c:f3:7f:4d:88:47] send
QoS null data frame on channel 1, pwr mgt ena

and then it tries scanning and what not but never recovers.

Funny (read annoying) thing is, its not failing in the same way every
time it disconnects. I do not see the same pattern.

And by "stops working" I see, it still has ip and ssid but "status: no
carrier" in ifconfig.

>
>> Thanks a bunch Adrian!
>
> so, I have a feeling that for "best" iwn(4) behaviour, we should
> likely extend net80211 to not do its own scan stuff (by putting the
> VAP into power save mode whilst doing scanning) but just to fire off a
> scan request. I _think_ iwn(4) will do it itself.
>
> We should however still fix the scan code to not be a racy, unpredictable thing.

I am not sure if I follow. Any more clues you can throw based on the logs?

Thanks again,
Hiren