From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from lists.gentoo.org ([140.105.134.102] helo=robin.gentoo.org) by nuthatch.gentoo.org with esmtp (Exim 4.62) (envelope-from ) id 1Hf25B-0003D5-R3 for garchives@archives.gentoo.org; Fri, 20 Apr 2007 23:00:02 +0000 Received: from robin.gentoo.org (localhost [127.0.0.1]) by robin.gentoo.org (8.14.0/8.14.0) with SMTP id l3KMxS1j024126; Fri, 20 Apr 2007 22:59:28 GMT Received: from an-out-0708.google.com (an-out-0708.google.com [209.85.132.243]) by robin.gentoo.org (8.14.0/8.14.0) with ESMTP id l3KMxRMO024117 for ; Fri, 20 Apr 2007 22:59:28 GMT Received: by an-out-0708.google.com with SMTP id b8so1060278ana for ; Fri, 20 Apr 2007 15:59:27 -0700 (PDT) Received: by 10.100.139.9 with SMTP id m9mr2029077and.1177109966802; Fri, 20 Apr 2007 15:59:26 -0700 (PDT) Received: by 10.100.138.1 with HTTP; Fri, 20 Apr 2007 15:59:26 -0700 (PDT) Message-ID: Date: Sat, 21 Apr 2007 00:59:26 +0200 From: "Camille Huot" Sender: camille@huot.name To: gentoo-bsd@lists.gentoo.org Subject: Re: [gentoo-bsd] TCP (FTP) problem - some sites - while emerging In-Reply-To: Precedence: bulk List-Post: List-Help: List-Unsubscribe: List-Subscribe: List-Id: Gentoo Linux mail X-BeenThere: gentoo-bsd@gentoo.org Reply-to: gentoo-bsd@lists.gentoo.org MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_Part_141091_24411224.1177109966706" References: <46266DDB.9050701@skyrush.com> <46267F77.5050405@skyrush.com> <46275F9D.6060906@skyrush.com> <4628C3B4.6070809@gentoo.org> X-Google-Sender-Auth: b2354822c05fc504 X-Archives-Salt: 1b249a51-2a42-4ab7-b302-a421bd4d9f0a X-Archives-Hash: 71a98c7ff8fd021ac3fb16995a447040 ------=_Part_141091_24411224.1177109966706 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Content-Disposition: inline 2007/4/20, Nathan Smith : > > The symptoms (broken TCP traffic to some sites) > depend entirely upon the sites you are connecting to and how your > connections are routed. We saw that affected users are sending TCP data with bad checksum (this could apply to udp and/or icmp too? anyone to check?). Then, some router on the Internet can decide to drop such bad packet. I noticed Linux doesn't drop but Checkpoint does. Another point, note that the bad checksum is always good_checksum+0100 """ [bad tcp cksum 3de5 (->3ce5)!] [bad tcp cksum 322d (->312d)!] [bad tcp cksum 25ad (->24ad)!] """ I guess this is the checksum algorithm that is incompatible with -O2. Even if I won't check the source code :p However, using tcpdump I can tell that some machines apparently do not > experience this problem, so it may be processor-specific (which makes > sense given that it is an optimization problem). I use a Pentium M. > What are other people using that have been affected by the problem? Mine is pentium-4 -- Camille Huot ------=_Part_141091_24411224.1177109966706 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Content-Disposition: inline 2007/4/20, Nathan Smith <ndansmith@gmail.com>:
The symptoms (broken TCP traffic to some sites)
depend entirely upon the sites you are connecting to and how your
connections are routed.

We saw that affected users are sending TCP data with bad checksum (this could apply to udp and/or icmp too? anyone to check?).

Then, some router on the Internet can decide to drop such bad packet. I noticed Linux doesn't drop but Checkpoint does.

Another point, note that the bad checksum is always good_checksum+0100
"""
[bad tcp cksum 3de5 (->3ce5)!]
[bad tcp cksum 322d (->312d)!]
[bad tcp cksum 25ad (->24ad)!]
"""

I guess this is the checksum algorithm that is incompatible with -O2. Even if I won't check the source code :p

However, using tcpdump I can tell that some machines apparently do not
experience this problem, so it may be processor-specific (which makes
sense given that it is an optimization problem).  I use a Pentium M.
What are other people using that have been affected by the problem?

Mine is pentium-4 

--
Camille Huot ------=_Part_141091_24411224.1177109966706-- -- gentoo-bsd@gentoo.org mailing list