netdev - Re: TCP IPv4 strange retransmits

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <Pine.LNX.4.64.0803042309240.15040@kivilampi-30.cs.helsinki.fi>
Date:	Tue, 4 Mar 2008 23:19:38 +0200 (EET)
From:	"Ilpo Järvinen" <ilpo.jarvinen@...sinki.fi>
To:	Arnd Hannemann <hannemann@...s.rwth-aachen.de>
cc:	Netdev <netdev@...r.kernel.org>
Subject: Re: TCP IPv4 strange retransmits

On Tue, 4 Mar 2008, Ilpo Järvinen wrote:

> In addition, the ACKs are processed in order and their effects are 
> immediate even if there's more information awaiting to be processed.

Before somebody asks or suggests (one might be tempted to think it's a 
good idea), no, we likely don't want to do it other way around unless 
somebody first proves that it won't negatively affect TCP's ACK clock,
and would benefits only some corner-case like this (and even in such 
case, one might get bitten by the tcp_max_burst). It would of course be 
possible to come up with a solution that reverse those _and_ fixes the 
ACK clock problems caused by such approach. The problems are similar to 
what LRO is causing btw, so it might not be complete waste of efforts to 
fix the ACK clock problems and reuse the solution in LRO as well.

-- 
 i.