From owner-freebsd-hackers Wed May 28 19:49:52 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id TAA19523 for hackers-outgoing; Wed, 28 May 1997 19:49:52 -0700 (PDT) Received: from friley01.res.iastate.edu (friley01.res.iastate.edu [129.186.78.1]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA19518 for ; Wed, 28 May 1997 19:49:50 -0700 (PDT) Received: from friley01.res.iastate.edu (localhost [127.0.0.1]) by friley01.res.iastate.edu (8.8.5/8.8.5) with ESMTP id VAA01968 for ; Wed, 28 May 1997 21:49:40 -0500 (CDT) Message-Id: <199705290249.VAA01968@friley01.res.iastate.edu> X-Mailer: exmh version 2.0gamma 1/27/96 To: hackers@freebsd.org Subject: TCP/IP bug? Unnecessary fragmentation... Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Wed, 28 May 1997 21:49:39 -0500 From: Chris Csanady Sender: owner-hackers@freebsd.org X-Loop: FreeBSD.org Precedence: bulk I'm not really sure how this occurs, but it does not seem quite right. If one does a write of 125 bytes, it seems to get fragmented into a 100, and 25 byte packet, and sent seperately. At first I thought this might be some weird artifact of TCP_NODELAY, but this does not seem to make a difference. Regardless of the way the data fits in mbufs/clusters, I think that it someone does a write syscall, and it fits within the mtu/mss/window, it should be sent as one piece. Is there any reason why it is not done this way? --Chris Csanady