Date: Tue, 10 Jul 2001 12:59:15 -0400 (EDT) From: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> To: "Crist Clark" <crist.clark@globalstar.com> Cc: Darren Reed <avalon@coombs.anu.edu.au>, freebsd-security@FreeBSD.ORG Subject: Re: FW: Small TCP packets == very large overhead == DoS? Message-ID: <200107101659.f6AGxFa02831@khavrinen.lcs.mit.edu> In-Reply-To: <3B4B30E7.28607AEE@globalstar.com> References: <200107100938.TAA13064@caligula.anu.edu.au> <3B4B30E7.28607AEE@globalstar.com>
next in thread | previous in thread | raw e-mail | index | archive | help
[I'm not sure who said what here....] >> > I am looking for where (if anywhere) the specification comes out and says >> > that segment "size" is the same as "length." Why isn't the MSS called the MSL >> > after the RFC has gone to such pains to define "length?" Because the MSS is specifically about ``how long a packet am I prepared to reassemble'', for which the control bits are not relevant. To quote another standards body out-of-context: the standard is what it says. At the time TCP was developed, it was thought that some small machines might have very tiny reassembly (i.e., *IP layer*) buffers, which might be only able to reassamble (say) two 576-byte IP packets. The purpose of the MSS option was to inform the other side that sending longer packets would be unproductive. The developers of the 4.2BSD TCP stack misinterpreted this, and took it to mean ``this is the size I want to send'', which caused no end of confusion ten years later when Path MTU Discovery became accepted practice. -GAWollman To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-security" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200107101659.f6AGxFa02831>