> ether_demux currently assumes that all vlan-tagged packets that it > sees have had the vlan stripped out and the M_VLAN tag is set, so it > never checks the ether type for a vlan. However ng_ether_rcv_upper > currently does not guarantee that this is the case(and there may be > other code paths where this is also true). Does anybody have any > strong feelings as to where the fix should go? Making ether_demux > handle it is guaranteed to catch all cases but it does add a bit more > overhead to check for a vlan tag at each stage.
In what cases vlan-tagged packet can be received by ng_ether_rcv_upper ? Another side of vlan and netgraph implementation problem is in: PR 152141 http://lists.freebsd.org/pipermail/freebsd-net/2011-February/027964.html Tagget packet -> ether_input --> (M_VLANTAG set) --> ng_ether.lower --> ng_bridge --> ng_ether.lower --> ether_output_frame --> ifp->if_transmit Untagged packet may be transmitted. ng_ether.lower and ether_output_frame does not check: is M_VLANTAG handled by iface driver IMHO ether_output_frame should do this check. /* * If underlying interface can not do VLAN tag insertion itself * then attach a packet tag that holds it. */ if ((m->m_flags & M_VLANTAG) && (ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) { m = ether_vlanencap(m, m->m_pkthdr.ether_vtag); if (m == NULL) { ifp->if_oerrors++; return (ENOBUFS); } m->m_flags &= ~M_VLANTAG; } (from if_bridge.c) -- Rozhuk Ivan _______________________________________________ [email protected] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-net To unsubscribe, send any mail to "[email protected]"
