From owner-freebsd-net@FreeBSD.ORG Sun Sep 17 19:57:38 2006 Return-Path: X-Original-To: freebsd-net@freebsd.org Delivered-To: freebsd-net@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id B8A7316A40F for ; Sun, 17 Sep 2006 19:57:38 +0000 (UTC) (envelope-from h.nieser@xs4all.nl) Received: from smtp-vbr9.xs4all.nl (smtp-vbr9.xs4all.nl [194.109.24.29]) by mx1.FreeBSD.org (Postfix) with ESMTP id 40DE743D60 for ; Sun, 17 Sep 2006 19:57:38 +0000 (GMT) (envelope-from h.nieser@xs4all.nl) Received: from [192.168.1.64] (aphax.nl [82.92.29.227]) by smtp-vbr9.xs4all.nl (8.13.6/8.13.6) with ESMTP id k8HJvZRQ056979 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Sun, 17 Sep 2006 21:57:36 +0200 (CEST) (envelope-from h.nieser@xs4all.nl) Message-ID: <450DA861.7030507@xs4all.nl> Date: Sun, 17 Sep 2006 21:56:17 +0200 From: Hans Nieser User-Agent: Thunderbird 1.5.0.5 (X11/20060910) MIME-Version: 1.0 To: =?ISO-8859-1?Q?Sten_Daniel_S=F8rsdal?= References: <44F4CF17.8060203@xs4all.nl> <44F70983.40705@errno.com> <44F71402.1060405@xs4all.nl> <44F7558A.4030907@errno.com> <450B48C2.7030501@xs4all.nl> <450C21B4.5030703@errno.com> <450D9FE2.5030902@wm-access.no> In-Reply-To: <450D9FE2.5030902@wm-access.no> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit X-Virus-Scanned: by XS4ALL Virus Scanner Cc: freebsd-net@freebsd.org Subject: Re: iwi discarding oversized packets while mtu=1500 for src/dst X-BeenThere: freebsd-net@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Networking and TCP/IP with FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 17 Sep 2006 19:57:38 -0000 Sten Daniel Sørsdal wrote: > Sam Leffler wrote: >> Hans Nieser wrote: >>> Sam Leffler wrote: >>>> Hans Nieser wrote: >>>> >>>>> root@aphax-laptop:~# uname -a >>>>> FreeBSD aphax-laptop.lan 6.1-RELEASE FreeBSD 6.1-RELEASE #0: Thu May 11 >>>>> 07:17:09 CEST 2006 >>>>> root@aphax-laptop.nieser.local:/usr/obj/usr/src/sys/APHAX-LAPTOP i386 >>>> Are you running the iwi driver that came with 6.1-release? If so it has >>>> numerous problems that have been fixed in 6-STABLE and HEAD. I'm not >>>> sure how best to update your system except by going to 6-STABLE via a >>>> src upgrade. >>> Today I have upgraded my laptop to 6-STABLE, but unfortunately the >>> problems remain. I don't even know wether iwi is to blame, because I can't >>> figure out wether perhaps xl on my server really is sending out too large >>> packets. >>> >>> That is, so far I've been assuming that an mtu of 1500 may not mean that >>> there's exactly 1500 bytes going over the wire (overhead of protocols to >>> which the mtu doesn't apply or something, I dunno, I'm no networking >>> expert obviously :), because xl is definitely sending out packets of 1518 >>> bytes. Which iwi on my laptop doesn't like, but the NIC in my desktop >>> machine (which runs Linux) has no problem with. Maybe someone can tell me >>> which of the machines is in error here, at least I'd know what to blame :( >>> >>> The fact that my Linux box doesn't discard these packages coming from my >>> server made me suspicious of iwi initially, but maybe iwi is doing nothing >>> wrong and my Linux box is simply willing to accept these oversized packets. >>> >> I can't speak to linux but an mtu of 1500 will cause a 1518 byte packet >> to be discarded. I don't recall what your problem was but if the iwi >> driver is receiving the frame and passing it up only to be discarded by >> the 802.3 layer then you've received a frame that's too large and you >> should look at the sender side for why it's being generated. If you >> don't want to do that you can probably just up the mtu on iwi and let >> the frame through. >> > > This does sound awfully lot like the driver doesn't strip the 4 byte > checksum of a basic 802.[23] ethernet frame. My initial thought was that > it was a vlan frame but the frame protocol indicates it's an IP packet. > Perhaps it's a frame that slips through the AP's checks on it's way out > to the wireless? Hmm, I'm not exactly sure what all this means but I suppose I could spent some time capturing packets on both sides and see if anything happens in between, and maybe read a book or two on networking/tcp-ip =] I should maybe also mention that my desktop Linux box is connected over wifi as well. Perhaps if the AP is at fault, it would explain some weird packetloss problems I've been having when playing online games (and maybe in other situations as well, but I guess this kind of thing is most noticable with online games).