Subject: Re: [PATCH 001/002] [MAC80211] Retry probe request few times
From: Marcel Holtmann <marcel@holtmann.org>
To: Maxim Levitsky <maximlevitsky@gmail.com>
Cc: linux-wireless <linux-wireless@vger.kernel.org>,
	linville <linville@tuxdriver.com>,
	Reinette Chatre <reinette.chatre@intel.com>,
	Johannes Berg <johannes@sipsolutions.net>
In-Reply-To: <1249450153.6083.4.camel@maxim-laptop>
References: <1249056817.20593.1.camel@maxim-laptop>
	 <1249056896.20593.3.camel@maxim-laptop>
	 <1249438947.3094.30.camel@localhost.localdomain>
	 <1249450153.6083.4.camel@maxim-laptop>
Content-Type: text/plain
Date: Wed, 05 Aug 2009 08:45:04 -0700
Message-Id: <1249487104.3094.43.camel@localhost.localdomain>
Mime-Version: 1.0
Sender: linux-wireless-owner@vger.kernel.org

Hi Maxim,

> > > From: Maxim Levitsky <maximlevitsky@gmail.com>
> > > Date: Fri, 31 Jul 2009 18:54:12 +0300
> > > Subject: [PATCH] [MAC80211] Retry probe request few times
> > > 
> > > Retry 5 times  (chosen arbitary ), before assuming
> > > that station is out of range.
> > 
> > so today I got the disconnect :(
> > 
> > [54632.657912] No probe response from AP 00:1c:f0:xx:xx:xx after 500ms, try 1
> > [54633.154560] No probe response from AP 00:1c:f0:xx:xx:xx after 500ms, try 2
> > [54873.231210] No probe response from AP 00:1c:f0:xx:xx:xx after 500ms, try 1
> > [55113.467840] No probe response from AP 00:1c:f0:xx:xx:xx after 500ms, try 1
> > [55113.964510] No probe response from AP 00:1c:f0:xx:xx:xx after 500ms, try 2
> > [55114.464516] No probe response from AP 00:1c:f0:xx:xx:xx after 500ms, try 3
> > [55114.967868] No probe response from AP 00:1c:f0:xx:xx:xx after 500ms, try 4
> > [55115.464511] No probe response from AP 00:1c:f0:xx:xx:xx after 500ms, disconnecting.
> > 
> > Should we increase the value to 8 or something? Or just accept that
> > sometimes we get a disconnect?
> > 
> I think a disconnect or two aren't a problem, they are always handled
> automatically, and can be caused by natural events.
> 
> Even without these patches (I have WPA2, and disconnects were every 4
> five seconds), and still it was possible to use network.

this assumption only works if you are not using NetworkManager or alike
where every new connect triggers DHCP again. If we wanna survive them,
then we have to teach them to handle the lease time more intelligent
which is kinda tricky. Might be worth doing anyway. I have to play with
it a little bit.

> One disconnect in a day is really nothing to worry about (I have seen
> such here as well)
> 
> You can increase try count, probably won't hurt much (each try is 0.5
> seconds, so even 10 tries gives total of 5 seconds, before disconnect on
> a really unaccessible AP, anyway)

Or just increase the try to 1 second. I still think there is something
odd going on with the iwlwifi driver here, but so far nobody saw
anything that is obviously wrong.

Regards

Marcel