From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Adam Langley" Subject: Re: [PATCH] TCP: Add seq and ack numbers to TCP_INFO Date: Wed, 9 Jul 2008 11:02:54 -0700 Message-ID: <396556a20807091102n5bb061daw91f41512faa8474@mail.gmail.com> References: <396556a20805301217k293e5718h6bbf02bfe069003@europa> <20080709104934.390cd81b@speedy> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Cc: davem@davemloft.net, netdev@vger.kernel.org To: "Stephen Hemminger" Return-path: Received: from rv-out-0506.google.com ([209.85.198.239]:1068 "EHLO rv-out-0506.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751562AbYGISCz (ORCPT ); Wed, 9 Jul 2008 14:02:55 -0400 Received: by rv-out-0506.google.com with SMTP id k40so3649018rvb.1 for ; Wed, 09 Jul 2008 11:02:54 -0700 (PDT) In-Reply-To: <20080709104934.390cd81b@speedy> Content-Disposition: inline Sender: netdev-owner@vger.kernel.org List-ID: On Wed, Jul 9, 2008 at 10:49 AM, Stephen Hemminger wrote: > Of what possible use is this to an application? I worry that applications would > come to expect or use certain characteristics of the network and TCP which > would be broken by later changes. Well, when you consider what else is in tcp_info (window scaling, SACK info, MTU, RTT, reorder detection), the sequence numbers seem positively pedestrian. There are a bunch of crazy reasons why userspace might be interested in the seq/ack numbers: handing connections between hosts, checking the backlog, analyzing the RNG strength of remote hosts etc. All experimental crap to be sure, but it's very handy in those experiments. I used a very similar patch when doing some of the very early basic on backlogs that Jerry Chu has taken up now. And sequence numbers are pretty fundamental to TCP. Certainly much more so than much of the other stuff that we're currently happy to export with sufficient warnings. AGL -- Adam Langley agl@imperialviolet.org http://www.imperialviolet.org