From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: Re: rps: some comments Date: Fri, 08 Jan 2010 07:27:31 +0100 Message-ID: <4B46D053.3000604@gmail.com> References: <65634d660911201528k5a07135el471b65fff9dd7c9d@mail.gmail.com> <20091120154046.67252d23@nehalam> <65634d660912171304p751e1698mbc9de50dade4317d@mail.gmail.com> <65634d661001051732qd64e79dt37e6247f8b0dc863@mail.gmail.com> <4B44258C.2050302@gmail.com> <4B44D89B.8070006@gmail.com> <65634d661001061454v389d311fjb245de21e0ab8092@mail.gmail.com> <4B45A623.7070507@gmail.com> <4B461D11.6050603@gmail.com> <65634d661001071607k63c33a0q4f402942d64ea77e@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: David Miller , Linux Netdev List To: Tom Herbert Return-path: Received: from gw1.cosmosbay.com ([212.99.114.194]:43904 "EHLO gw1.cosmosbay.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750982Ab0AHG1j (ORCPT ); Fri, 8 Jan 2010 01:27:39 -0500 In-Reply-To: <65634d661001071607k63c33a0q4f402942d64ea77e@mail.gmail.com> Sender: netdev-owner@vger.kernel.org List-ID: Le 08/01/2010 01:07, Tom Herbert a =E9crit : > I kind of like the way it is right now, which is to enqueue even on > the local CPU. The advantage of that is that all the packets receive= d > in the napi run are dispatched to all the participating CPUs before > any are processed, this is good for average and worst case latency on > a packet burst. I see... this makes sense, thanks. Do you know if some non multi queue NIC would be able to provide us a suitable hash to fill skb->rxhash in NIC driver itself ?