From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: Re: [PATCH v2] rfs: Receive Flow Steering Date: Tue, 06 Apr 2010 17:10:52 +0200 Message-ID: <1270566652.2081.38.camel@edumazet-laptop> References: <1270559096.2081.35.camel@edumazet-laptop> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: davem@davemloft.net, netdev@vger.kernel.org To: Tom Herbert Return-path: Received: from mail-bw0-f209.google.com ([209.85.218.209]:55781 "EHLO mail-bw0-f209.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753717Ab0DFPK7 (ORCPT ); Tue, 6 Apr 2010 11:10:59 -0400 Received: by bwz1 with SMTP id 1so3710054bwz.21 for ; Tue, 06 Apr 2010 08:10:56 -0700 (PDT) In-Reply-To: Sender: netdev-owner@vger.kernel.org List-ID: Le mardi 06 avril 2010 =C3=A0 07:25 -0700, Tom Herbert a =C3=A9crit : > > 2) inet_rps_save_rxhash(sk, skb->rxhash); > > > > It should have a check to make sure some part of the stack d= oesnt feed > > many different rxhash for a given socket (Make sure we dont pollute= flow > > table with pseudo random values) > > > If packets for a connection are always received on the same device, i= s > it reasonable to assume the rxhash is constant for that connection? >=20 > I suppose it's possible that packets for a same sockets are being > constantly received on two different devices that are giving differen= t > rxhashes. This would already be bad in that OOO is probably happenin= g > anyway. I don't know if thrashing the sock_flow_table is going to > aggravate this scenario much. >=20 > Are there any other degenerative cases you're worried about? No, I was only considering this mostly as a debugging aid, before RPS i= s stable, because mismatches could be unnoticed and performance not optimal. =20