Date: Tue, 23 Nov 1999 11:47:14 -0800 (PST) From: Julian Elischer <julian@whistle.com> To: Vern Nichols <nichols@roundview.com> Cc: freebsd-net@FreeBSD.ORG Subject: Re: netgraph'd enet nics anyone? Message-ID: <Pine.BSF.4.20.9911231126300.3171-100000@home.elischer.org> In-Reply-To: <Pine.BSF.4.10.9911230959350.20815-100000@ns.sys.roundview.com>
next in thread | previous in thread | raw e-mail | index | archive | help
> I'm currently playing around with and learning what I can about netgraph > to see if I want to use it in a new project. I thought I'd play around by > trying to use it on the ethernet card (rl0). There is already the following: if_ethersubr.c when compiled with options NETGRAPH already generates a netgraph node for every etherenet NIC. You can decide if you want to pre-empt ALL packets, or just get those that are unrecognised by present protocols. If you have rl0 and compile if_ethersubr.c with options NETGRAPH you already have a node called rl0: There is an 'Iface' type that generated a system interface that shows up in ifconfig. This feeds data in to the bottom of ip abd effectively gives us a netgraph attachment to the botom of IP, without us needing to touch IP. There is a 'ksocket' type that pretends to be a socket. This allows us to catch packets at the top of udp or tcp that are destined to certain addresses. Once again, this is the equivalent of having a netgraph attachment at the top of tcp/udp, except without having to actually touch tcp or udp. On Tue, 23 Nov 1999, Vern Nichols wrote: > > > I'm wondering if anyone has done that and if I'd also have to make a TCP > netgraph node to make it work the way I'd expect. I think I'd be looking > at something like: > > SocketNode->TCPNode(does not seem to exist)->IFaceNode(IP)->NICNode(enet) > > Or am I off base on how netgraph can/should be used? I see some > ngether_*() routines in if_ethersubr.c so I assume someone is already > looking into a netgraph enet node. > > I'm currenlty working on 3.2 with the netgraph 3.0 patches applied. I > see netgraph appears to be integrated into 4.0. Is it worth moving up > to 4.0 before I start getting too deep in netgraph? It's already applied to 3.3-STABLEB as well. basically you can already do all that you say. though our aim is not to replace the present 'fast path from socket to NIC' but to allow people to make diversions if needed. e.g. [tcp socket] [netgraph socket] | | | | | [arbitrary processig node] | | | | | | | [ksocket node] | | | | | /------------+ | / [--tcp/udp/ip--] | \ | \ | \ | \ | \ | \ | | | | | [iface node] | | | | | [arbitrary processing nodes] | | | / | / | / | / | / |norm /netgraph [NIC (with netgraph option] > > Thanks for any insite. > > Vern A. Nichols Roundview, Inc. > nichols@roundview.com 10700 Old Co Rd. 15, Suite 303 > 612.591.7777 ext. 14 Plymouth, MN 55441 > julian 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.4.20.9911231126300.3171-100000>