Date: Wed, 7 Dec 2011 21:33:05 +0530
From: Vasanthakumar Thiagarajan <vthiagar@qca.qualcomm.com>
To: Johannes Berg <johannes@sipsolutions.net>
CC: <linville@tuxdriver.com>, <linux-wireless@vger.kernel.org>
Subject: Re: [PATCH] cfg80211: Fix race in bss timeout
Message-ID: <20111207160303.GE12498@chvasanth-lnx> (sfid-20111207_170331_359041_361D0CB5)
References: <1323270687-5989-1-git-send-email-vthiagar@qca.qualcomm.com>
 <1323271070.3404.40.camel@jlt3.sipsolutions.net>
 <20111207153052.GB12498@chvasanth-lnx>
 <1323272248.3404.43.camel@jlt3.sipsolutions.net>
 <20111207155325.GC12498@chvasanth-lnx>
 <1323273375.3404.44.camel@jlt3.sipsolutions.net>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
In-Reply-To: <1323273375.3404.44.camel@jlt3.sipsolutions.net>
Sender: linux-wireless-owner@vger.kernel.org

On Wed, Dec 07, 2011 at 04:56:15PM +0100, Johannes Berg wrote:
> On Wed, 2011-12-07 at 21:23 +0530, Vasanthakumar Thiagarajan wrote:
> > On Wed, Dec 07, 2011 at 04:37:28PM +0100, Johannes Berg wrote:
> > > > 
> > > > These warnings are added in cfg80211_roamed().
> > > 
> > > But this can be called directly by the driver with a NULL BSS.
> > > 
> > > Ohh. I see what you did, you didn't allow drivers calling this now. I
> > > think you should export this function still since otherwise the race
> > > windows might get tiny, but isn't actually completely closed (the first
> > > get_bss() might find it, the next a millisecond later not)
> > 
> > I can understand the millisecond delay between cfg80211_get_bss() in driver and
> > __cfg80211_roamed() as there can be delay in event_work getting
> > scheduled. I don't understand the delay between driver and
> > cfg80211_roamed() as the later one is direct call. Or this can be
> > done later if experienced the delay?.
> 
> The computer takes time to execute the functions, so time passes, right?

Ok. As driver adds new bss entry if it is not available it is always
safe to use the bss from driver to avoid any such race. Thanks!

Vasanth