Date: Tue, 28 Aug 2007 21:37:18 +0100 From: "Bruce M. Simpson" <bms@FreeBSD.org> To: "Christian S.J. Peron" <csjp@FreeBSD.org> Cc: FreeBSD Current <current@freebsd.org>, Andrew Thompson <thompsa@FreeBSD.org> Subject: Re: multicast packets from bpf Message-ID: <46D4877E.700@FreeBSD.org> In-Reply-To: <20070828135929.GA2305@sub.vaned.net> References: <20070828040026.GB42201@heff.fud.org.nz> <46D3C9F3.2010802@FreeBSD.org> <20070828135929.GA2305@sub.vaned.net>
next in thread | previous in thread | raw e-mail | index | archive | help
Christian S.J. Peron wrote: > I think that tap(4) is a bit different since the only kind of frames it > handles are Ethernet. As Andrew points out the tapwrite check probably isn't needed now. > This is not the case for bpf(4). I wonder if it > makes sense to add this check into ether_output()? IIRC bpf will call > the network interface's output routine, in the Ethernet/bridge case it > should be ether_output(). > This approach avoids touching the device-independent paths, and, providing the check resides in the AF_UNSPEC case (as ARP resolution should do the right thing) is reasonably neat. But it doesn't handle the case where there are link-layer netgraph nodes between bpf and if_bridge, something which the first change would deal with. regards BMS
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?46D4877E.700>