From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: Re: [PATCH 2/2] netfilter: get rid of atomic ops in fast path Date: Sun, 20 Mar 2011 20:06:03 +0100 Message-ID: <1300647963.2831.291.camel@edumazet-laptop> References: <1300468038.2888.160.camel@edumazet-laptop> <4D86121C.7050503@trash.net> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: David Miller , Netfilter Development Mailinglist , netdev , Jan Engelhardt To: Patrick McHardy Return-path: In-Reply-To: <4D86121C.7050503@trash.net> Sender: netfilter-devel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org Le dimanche 20 mars 2011 =C3=A0 15:41 +0100, Patrick McHardy a =C3=A9cr= it : > Am 18.03.2011 18:07, schrieb Eric Dumazet: > > We currently use a percpu spinlock to 'protect' rule bytes/packets > > counters, after various attempts to use RCU instead. > >=20 > > Lately we added a seqlock so that get_counters() can run without > > blocking BH or 'writers'. But we really use the seqcount in it. > >=20 > > Spinlock itself is only locked by the current/owner cpu, so we can > > remove it completely. > >=20 > > This cleanups api, using correct 'writer' vs 'reader' semantic. > >=20 > > At replace time, the get_counters() call makes sure all cpus are do= ne > > using the old table. >=20 > I think this will have to wait until net-next opens up again since > its not a bugfix. Sure :) -- To unsubscribe from this list: send the line "unsubscribe netfilter-dev= el" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html