for wpa+ral, you should definitely run -current or 4.8.
On 2010-09-18, Todd Carson <t...@daybefore.net> wrote:
> On Sat, Sep 18, 2010 at 01:01:43PM +0100, Joe Martel wrote:
>> > If you do 'ifconfig ral0 down; ifconfig ral0 up' on the hostap
>> > box it might temporarily fix things
>>
>> Yes it does, thanks.
>> Have added a cron job to run this every 24hrs.
>
> I have a similar problem against which ifconfig down; up isn't effective.
> My card is an RT2561, and when I turn on ifconfig debug on both the client
> and AP, I see that the AP responds to the association request and sends
> the first packet of the 4-way handshake. The client receives the
> association response, fails to receive the handshake initiation, and
> receives a deauth after timing out.
>
> AP side:
> ral0: received assoc_req from xx:xx:xx:xx:xx:xx rssi 105 mode 11g
> ral0: sending assoc_resp to xx:xx:xx:xx:xx:xx on channel 6 mode 11g
> ral0: sending msg 1/4 of the 4-way handshake to xx:xx:xx:xx:xx:xx
> last message repeated 2 times
> ral0: station xx:xx:xx:xx:xx:xx deauthenticate (reason 15)
> ral0: sending deauth to xx:xx:xx:xx:xx:xx on channel 6 mode 11g
>
> client side:
> athn0: sending auth to yy:yy:yy:yy:yy:yy on channel 6 mode 11g
> athn0: received auth from yy:yy:yy:yy:yy:yy rssi 47 mode 11g
> athn0: sending assoc_req to yy:yy:yy:yy:yy:yy on channel 6 mode 11g
> athn0: received assoc_resp from yy:yy:yy:yy:yy:yy rssi 46 mode 11g
> athn0: associated with yy:yy:yy:yy:yy:yy (...)
> athn0: received deauth from yy:yy:yy:yy:yy:yy rssi 46 mode 11g
>
> Turning off WPA2 doesn't allow clients to use the network; it just fails
> differently, though I unfortunately don't have log captures of that at the
> moment. The only reliable way I've found to resurrect the AP is a cold
> power cycle.
>
> Is this also a known issue? Is there any more targeted tracing I could
> try compiling into my kernels besides RAL_DEBUG, ATHN_DEBUG,
> IEEE80211_DEBUG? I'm wondering if this has something to do with crypto
> parameters on the card not being reset in all cases, but I don't know
> enough about either the hardware or the 802.11 protocol to have any idea
> whether that makes sense.
