[Prism54-users] Getting stuck when out of range

[Claas Hilbrecht]

--Am Montag, 31. Mai 2004 20:27 +0200 Jens Maurer <Jens.Maurer@gmx.net> 
schrieb:

> When we can't reach our connection partner any more, we can't
> send the packets, and sending stalls.  We tell the upper networking
> layers "my queue is full, don't send more stuff".

I don't think in my case the connection is really "dropped" because I'm out 
of range. Both stations do not move around and an automated ifdown/ifup 
reenables the connection instantly. And the connection only drops if I use 
the connection with maximum transfer speed. Since 26.05.2004 the link is 
stable but I didn't sent any files over the link so far. Only some mails 
and jabber chat activity. What is also interessting is that both stations 
use the nearly the same hardware. And the AP station nevers fails so far 
(the wlan card is up for about 3 weeks now). Only my client wlan that 
connects to the AP fail. The following scipt is used to detect a wlan 
outage:

while [ 1 ]
do
  chk=`iwconfig eth2 | grep ESSID:\"xxx\"`
  if [ -z "$chk" ]
  then
    logger wlanchk QQQ: Das Funkuhrnetz ist ausgefallen, versuche RESET
    ifconfig eth2 down
    ifconfig eth2 up
    sleep 2
    iwpriv eth2 s_profile 3
    iwpriv eth2 s_frameburst 5000
  fi
  sleep 5
done

> The question is whether we just forget to re-enable the upper
> networking layers, or whether the firmware is unable to
> re-establish the association with the base station, and
> tell us when it did that.

I can reproduce the problem in a few minutes. I'm willing to test patches 
too. WEP seems to be involved too. If I disable WEP the connection seems to 
be more "stable". With WEP it takes about 5-15 minutes to get an outage, 
without WEP it takes about 30-60 minutes.

-- 
 Claas Hilbrecht
 http://www.jucs-kramkiste.de