Message-ID: <4DE54596.8090003@openwrt.org> (sfid-20110531_214635_078381_7E4FC85B)
Date: Tue, 31 May 2011 21:46:30 +0200
From: Felix Fietkau <nbd@openwrt.org>
MIME-Version: 1.0
To: Johannes Berg <johannes@sipsolutions.net>
CC: linux-wireless@vger.kernel.org, linville@tuxdriver.com
Subject: Re: [PATCH 2/3] mac80211: create a STA entry for a WDS interface
 after receiving a beacon
References: <1306869394-83047-1-git-send-email-nbd@openwrt.org>	 <1306869394-83047-2-git-send-email-nbd@openwrt.org> <1306870318.29989.0.camel@jlt3.sipsolutions.net>
In-Reply-To: <1306870318.29989.0.camel@jlt3.sipsolutions.net>
Content-Type: text/plain; charset=UTF-8; format=flowed
Sender: linux-wireless-owner@vger.kernel.org

On 2011-05-31 9:31 PM, Johannes Berg wrote:
> On Tue, 2011-05-31 at 21:16 +0200, Felix Fietkau wrote:
>
>>  +	rcu_read_lock();
>>  +
>>  +	sta = sta_info_get(sdata, sdata->u.wds.remote_addr);
>>  +
>>  +	if (!sta) {
>>  +		rcu_read_unlock();
>>  +		sta = sta_info_alloc(sdata, sdata->u.wds.remote_addr,
>>  +				     GFP_KERNEL);
>>  +		if (!sta)
>>  +			return;
>>  +
>>  +		new = true;
>>  +	}
>
> ...
>
>>  +	rcu_read_unlock();
>
> definitely doesn't look right.
It's a bit weird because in the case of a new sta being allocated, the 
rcu read lock is acquired by sta_info_insert_rcu, and the code inbetween 
does not need to be covered by the rcu read lock as the sta entry has 
not been added to the list yet.
It should be correct though, unless I'm missing something...

- Felix