From owner-freebsd-net@FreeBSD.ORG Tue Feb 11 19:49:55 2014 Return-Path: Delivered-To: freebsd-net@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) (using TLSv1 with cipher ADH-AES256-SHA (256/256 bits)) (No client certificate requested) by hub.freebsd.org (Postfix) with ESMTPS id 51B22282; Tue, 11 Feb 2014 19:49:55 +0000 (UTC) Received: from bigwig.baldwin.cx (bigwig.baldwin.cx [IPv6:2001:470:1f11:75::1]) (using TLSv1 with cipher ADH-CAMELLIA256-SHA (256/256 bits)) (No client certificate requested) by mx1.freebsd.org (Postfix) with ESMTPS id 2A82419B1; Tue, 11 Feb 2014 19:49:55 +0000 (UTC) Received: from jhbbsd.localnet (unknown [209.249.190.124]) by bigwig.baldwin.cx (Postfix) with ESMTPSA id 1D93BB9C7; Tue, 11 Feb 2014 14:49:54 -0500 (EST) From: John Baldwin To: freebsd-net@freebsd.org Subject: Re: ixgbe/NFS m_defrag() instrumentation Date: Tue, 11 Feb 2014 13:58:54 -0500 User-Agent: KMail/1.13.5 (FreeBSD/8.4-CBSD-20130906; KDE/4.5.5; amd64; ; ) References: <21233.52147.912022.488615@hergotha.csail.mit.edu> In-Reply-To: <21233.52147.912022.488615@hergotha.csail.mit.edu> MIME-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Message-Id: <201402111358.54764.jhb@freebsd.org> X-Greylist: Sender succeeded SMTP AUTH, not delayed by milter-greylist-4.2.7 (bigwig.baldwin.cx); Tue, 11 Feb 2014 14:49:54 -0500 (EST) Cc: rmacklem@freebsd.org, Garrett Wollman X-BeenThere: freebsd-net@freebsd.org X-Mailman-Version: 2.1.17 Precedence: list List-Id: Networking and TCP/IP with FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 11 Feb 2014 19:49:55 -0000 On Wednesday, February 05, 2014 12:27:15 am Garrett Wollman wrote: > I instrumented calls to m_defrag() in ixgbe. As expected, it gets > called *a lot* when NFS is running with the default read size of 64k. > A simple benchmark (single-threaded sequential read of a 128 GB file > which I didn't even run to completion) tells the tale: > > $ sysctl dev.ix.0.mbuf_defrag_attempted > dev.ix.0.mbuf_defrag_attempted: 1737994 > > (There's already a similar counter for m_defrag() failures, which made > it easy to add this counter. Unfortunately, there is no analogous > instrumentation in cxgbe so I couldn't do likewise for that NIC.) Is this due to the other thread Rick has about TSO interacting with NFS? Do these go away if you disable TSO on the interface? -- John Baldwin