From owner-svn-src-all@FreeBSD.ORG Sat Mar 15 17:33:27 2014 Return-Path: Delivered-To: svn-src-all@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [8.8.178.115]) (using TLSv1 with cipher ADH-AES256-SHA (256/256 bits)) (No client certificate requested) by hub.freebsd.org (Postfix) with ESMTPS id E4FE3492; Sat, 15 Mar 2014 17:33:27 +0000 (UTC) Received: from cyrus.watson.org (cyrus.watson.org [198.74.231.69]) by mx1.freebsd.org (Postfix) with ESMTP id B8EFDCA; Sat, 15 Mar 2014 17:33:27 +0000 (UTC) Received: from [10.108.26.30] (unknown [46.233.116.87]) by cyrus.watson.org (Postfix) with ESMTPSA id 2FA9646B0C; Sat, 15 Mar 2014 13:33:24 -0400 (EDT) Content-Type: text/plain; charset=iso-8859-1 Mime-Version: 1.0 (Mac OS X Mail 7.2 \(1874\)) Subject: Re: svn commit: r263198 - in head/sys: amd64/conf conf net netinet netinet6 sys From: "Robert N. M. Watson" In-Reply-To: Date: Sat, 15 Mar 2014 17:33:22 +0000 Content-Transfer-Encoding: quoted-printable Message-Id: <33030492-A1A3-4A00-B2B3-66B9B2DFB461@FreeBSD.org> References: <201403150057.s2F0vofg081606@svn.freebsd.org> To: Adrian Chadd X-Mailer: Apple Mail (2.1874) Cc: "svn-src-head@freebsd.org" , "svn-src-all@freebsd.org" , "src-committers@freebsd.org" X-BeenThere: svn-src-all@freebsd.org X-Mailman-Version: 2.1.17 Precedence: list List-Id: "SVN commit messages for the entire src tree \(except for " user" and " projects" \)" List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 15 Mar 2014 17:33:28 -0000 On 15 Mar 2014, at 17:29, Adrian Chadd wrote: >> I'd characterise this work as "early" in that it would benefit from >> performance optimisation, device-driver work, and feature enhancement = (e.g., >> not just stuff like load rebalancing, but also statistics to detect = RSS >> configuration problems leading to work being directed to the wrong = CPU, >> etc). In any case, hopefully this is a useful starting point for = people >> doing work in this space. >=20 > Yup. I have a bunch of diagnostics and statistics to add to ensure > things are staying on the right CPU. >=20 > I'll spin them up on netperf soon and dive into UDP testing/fixing. Excellent -- netstat support for monitoring queue and RSS behaviour is = definitely the next step to run in, here. I'd like to see us extend the = kernel's device-driver independent data structures and statistics to = allow interfaces to expose additional hardware queue = information/statistics as well as higher-level netisr, pcbgroup, and RSS = information. Robert=