From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stephen Hemminger Subject: Re: [PATCH 3/4] llc: use a device based hash table to speed up multicast delivery Date: Thu, 3 Dec 2009 16:37:28 -0800 Message-ID: <20091203163728.44a9c0f1@nehalam> References: <1259879498-27860-1-git-send-email-opurdila@ixiacom.com> <1259879498-27860-4-git-send-email-opurdila@ixiacom.com> <20091203152513.643c471a@nehalam> <200912040153.11395.opurdila@ixiacom.com> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org, Arnaldo Carvalho de Melo To: Octavian Purdila Return-path: Received: from mail.vyatta.com ([76.74.103.46]:46668 "EHLO mail.vyatta.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754560AbZLDAhm (ORCPT ); Thu, 3 Dec 2009 19:37:42 -0500 In-Reply-To: <200912040153.11395.opurdila@ixiacom.com> Sender: netdev-owner@vger.kernel.org List-ID: On Fri, 4 Dec 2009 01:53:11 +0200 Octavian Purdila wrote: > On Friday 04 December 2009 01:25:13 you wrote: > > On Fri, 4 Dec 2009 00:31:37 +0200 > > > > Octavian Purdila wrote: > > > This patch adds a per SAP device based hash table to solve the > > > multicast delivery scalability issues for the case where the are a > > > large number of interfaces and a large number of sockets (bound to the > > > same SAP) are used. > > > > Rather than adding hash table and rwlock, why not hash list RCU > > and a single spin lock > > > > I have a partial version with RCU and single spinlock, but then I ran into a > (Eric's I think) patch which moved the UDP lock per bucket. And since RCU > can't help on the write side (in this instance each time we bound or delete > the socket) it was not clear to me what is the best approach. The lock is held for such a brief period on connection setup that a single spinlock is probably ok.