From owner-freebsd-net Mon May 29 14:39:27 2000 Delivered-To: freebsd-net@freebsd.org Received: from dustdevil.waterspout.com (standpipe.waterspout.com [208.13.60.152]) by hub.freebsd.org (Postfix) with ESMTP id 9010837B63D for ; Mon, 29 May 2000 14:39:23 -0700 (PDT) (envelope-from csg@dustdevil.waterspout.com) Received: (from csg@localhost) by dustdevil.waterspout.com (8.9.3/8.9.3) id QAA15998; Mon, 29 May 2000 16:39:17 -0500 (EST) (envelope-from csg) Date: Mon, 29 May 2000 16:39:17 -0500 From: "C. Stephen Gunn" To: Victor Ponomarev Cc: Net Subject: Re: VLAN improvement needed... Message-ID: <20000529163917.A15841@waterspout.com> References: <392E754B.BCF8AC96@unet.ru> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 1.0.1i In-Reply-To: <392E754B.BCF8AC96@unet.ru>; from vick@unet.ru on Fri, May 26, 2000 at 04:59:56PM +0400 Sender: owner-freebsd-net@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Fri, May 26, 2000 at 04:59:56PM +0400, Victor Ponomarev wrote: > Now VLAN support in stable is bad. I'm not sure what you mean by now. if_vlan has always had some shortcomings. I don't think anything has changed recently. > When host send a packet about 1514 byte switch trunk port add 4 byte and > router trunk port substitute vlan tag to another and send it back to > switch. The latter remove vlan header and send packet to appropriate > vlan ports. > > Currently FreeBSD router simply drop large packet on it's interface. > That's very bad... Actually several ethernet cards drop these frames as giants. There are also a few (ti in particular) that do not. > The existing solution on these problem for Intel card may be found at > http://www.euitt.upm.es/~pjlobo/ I've reviewed this patch carefully before. I still don't see any changes the the FXP driver that allow reception of tagged frames. I do have some modifications around that allow larger frame sizes on 3Com 3c905B-TX cards. I'll dug that patch up, and put it online: http://www.physics.purdue.edu/~csg/xl-vlan.patch > But there's another problem with small ip packet. When BSD router strip > ethernet header for payload < 46 it strip padding bytes also. But when > it reinserted data with another vlan header it don't add padding bytes > and we have runts packets on interface. Do you think this is related to VLANs because of the encapsulation? Do you have a method to duplicate this behavior? - Steve To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-net" in the body of the message