Date: Thu, 1 Mar 2007 15:53:13 -0500 From: Andre Guibert de Bruet <andy@siliconlandmark.com> To: Andre Oppermann <andre@freebsd.org> Cc: freebsd-net@freebsd.org, freebsd-current@freebsd.org, rwatson@freebsd.org, Andrew Gallatin <gallatin@cs.duke.edu>, kmacy@freebsd.org Subject: Re: Large TCP send socket buffer optimizations Message-ID: <DABA390F-20B5-4602-9DAE-F0C39B37B56A@siliconlandmark.com> In-Reply-To: <45E7020D.2000605@freebsd.org> References: <45E6E2E8.5060408@freebsd.org> <17894.61337.772803.325091@grasshopper.cs.duke.edu> <45E7020D.2000605@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mar 1, 2007, at 11:40 AM, Andre Oppermann wrote: > Andrew Gallatin wrote: >> Andre Oppermann writes: >> > This patch solves the problem by maintaining an offset pointer >> in the socket >> > buffer to give tcp_output() the closest mbuf right away >> avoiding the traversal >> > from the beginning. >> > > With this patch we should be able to compete nicely for the >> Internet land >> > speed record again. >> Great news. I will test soon. >> Have you found the bug which makes a freebsd sender spew >> duplicate acks? This will have a negative affect on >> any benchmarks.. > > No, not yet. And my current test environment doesn't lend itself > to reproduce > the bug easily. I'm very likely to need further input from you and > will send > you some patches for instrumentation in the near future. Please CC me on the patches. Did you get a chance to check out the pcap that I linked to? Andy /* Andre Guibert de Bruet * 6f43 6564 7020 656f 2e74 4220 7469 6a20 */ /* Code poet / Sysadmin * 636f 656b 2e79 5320 7379 6461 696d 2e6e */ /* GSM: +1 734 846 8758 * 5520 494e 2058 6c73 7565 6874 002e 0000 */ /* WWW: siliconlandmark.com * C/C++, Java, Perl, PHP, SQL, XHTML, XML */
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?DABA390F-20B5-4602-9DAE-F0C39B37B56A>