Date: Sun, 3 Nov 2002 19:46:54 -0800 (PST) From: Julian Elischer <julian@elischer.org> To: Luigi Rizzo <rizzo@icir.org> Cc: net@freebsd.org Subject: Re: fragmentation issues Message-ID: <Pine.BSF.4.21.0211031942070.60008-100000@InterJet.elischer.org> In-Reply-To: <20021103193923.A34675@xorpc.icir.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, 3 Nov 2002, Luigi Rizzo wrote: > Hi, > a colleague pointed out the following problem: the various > forms of encapsulation of IP traffic might result in IP > datagrams which are larger than the IP maximum datagram > length (64k). > > The problem arises with the various IP-in-IP encapsulations > (gif, maybe faith), with IPSEC, and with some multicast > processing. > yes I have: tcp/udp in ip in ppp in udp in esp in ip in my applications.. I ended up having to set the MTU of the tunnel interfaces down to about 1400 bytes so that the final packet could get into the 1500 byte ethernet frame neeed to get to the router to the Internet. (mtu discovery then reduced the path mtu and everything is cool) > I guess the correct way to solve the problem is to do > some form of fragmentation _before_ adding the new > header to the packet. This requires the replication of > some code which is in ip_output(). > > So i was wondering if there objections if in the long > term i pull out the fragmentation code from ip_output > and put it into a separate procedure that can be used > wherever is required without having to replicate the > code (and maintain it). > > comments ? > seems reasonable. Well see how practical it is when you try do it :-) > cheers > luigi > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-net" in the body of the message > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-net" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0211031942070.60008-100000>