Date: Sun, 23 Oct 2011 05:22:08 +0900 From: rozhuk.im@gmail.com To: <Rozhuk.IM@gmail.com>, <freebsd-net@freebsd.org> Subject: RE: QinQ support: implement details - need help! Message-ID: <4ea325f3.4693cc0a.45c2.6fc5@mx.google.com> In-Reply-To: <4e988844.1025cc0a.1e88.ffffb4b9@mx.google.com> References: <4e988844.1025cc0a.1e88.ffffb4b9@mx.google.com>
index | next in thread | previous in thread | raw e-mail
http://www.freebsd.org/cgi/query-pr.cgi?pr=161908 All done. IEEE 802.1Q + IEEE 802.1p IEEE 802.1ad (IEEE 802.1QinQ) - if two ng_vlan node + ethernet_type for VLAN encapsulation is tunable, default is: 0x8100 (33024) + PCP (Priority Code Point) and CFI (Canonical Format Indicator) for VLAN encapsulation is tunable per VID + VLAN filter can be deleted by VID + tunable encapsulation: on - do 802.1Q encapsulation, off - set M_VLANTAG and ether_vtag * improved encapsulation/decapsulation code * "vlan" changed to "vid" in "addfilter" and "gettable" messages * many other changes -- Rozhuk Ivan > -----Original Message----- > From: rozhuk.im@gmail.com [mailto:rozhuk.im@gmail.com] > Sent: Saturday, October 15, 2011 4:07 AM > To: freebsd-net@freebsd.org > Cc: Rozhuk.IM@gmail.com > Subject: QinQ support: implement details - need help! > > > ... > IEEE 802.1ad (802.1QinQ) specifies architecture and bridge protocols to > provide separate instances of the MAC services to multiple independent > users > of a Bridged Local Area Network in a manner that does not require > cooperation among the users, and requires a minimum of cooperation > between > the users and the provider of the MAC service. > > The idea is to provide, for example, the possibility for customers to > run > their own VLANs inside service provider's provided VLAN. This way the > service provider can just configure one VLAN for the customer and > customer > can then treat that VLAN as if it was a trunk. > ... > http://en.wikipedia.org/wiki/802.1ad > > > > "Customer" VLAN - ether_type: 0x8100 > Stored in: > struct pkthdr { > ... > union { > u_int16_t vt_vtag; /* Ethernet 802.1p+q vlan tag */ > u_int16_t vt_nrecs; /* # of IGMPv3 records in this chain > */ > } PH_vt; > SLIST_HEAD(packet_tags, m_tag) tags; /* list of packet tags */ > }; > #define ether_vtag PH_vt.vt_vtag > > > > "Service Provider" VLAN - ether_type: 0x88a8/0x8100/0x9100 > How I can store it in packet? > How to store tag for QinQinQ? > > > > > > > VLAN tags store implementation (IMHO) > > > Ethernet packet: > | Dst_MAC | Src_MAC | 802.1Q_Tag2 | 802.1Q_Tag1 | 802.1Q_Tag0 | > Ether_type/size | Payload | > > 802.1Q Tag0 = "Customer" VLAN - ether_type: 0x8100, stored in > pkthdr.PH_vt.vt_vtag (now) (IEEE 802.1Q) > 802.1Q Tag1 = "Service Provider" / MetroTag... (IEEE 802.1ad) > 802.1Q Tag2 = SomeTag... (QinQinQ) > > > I found part of old code in /usr/src/sys/dev/mxge/if_mxge.c: > ... > /* save the tag */ > #ifdef MXGE_NEW_VLAN_API > m->m_pkthdr.ether_vtag = ntohs(evl->evl_tag); > #else > { > struct m_tag *mtag; > mtag = m_tag_alloc(MTAG_VLAN, MTAG_VLAN_TAG, sizeof(u_int), > M_NOWAIT); > if (mtag == NULL) > return; > VLAN_TAG_VALUE(mtag) = ntohs(evl->evl_tag); > m_tag_prepend(m, mtag); > } > > #endif > m->m_flags |= M_VLANTAG; > ... > > > 1. Compact scheme: > #define MTAG_VLAN_T0 1035328035 /* m_tag_cookie */ > #define MTAG_VLAN_T1 (MTAG_VLAN_T0 + 1) > #define MTAG_VLAN_T2 (MTAG_VLAN_T1 + 1) > #define MTAG_VLAN_T3 (MTAG_VLAN_T2 + 1) > > Store vt_vtag in m_tag_id > ng_vlan / if_vlan will use for tag/untag one of MTAG_VLAN_Tx > m_tag_cookie > and tunable ether_type for vlan encapsulation: > lower <-> ng_vlan(MTAG_VLAN_T0) <-> ng_vlan(MTAG_VLAN_T1) <-> > ng_vlan(MTAG_VLAN_T2) <-> upper > > > 2. OpenBSD compatible > #ifdef <OPENBSD> > #define MTAG_VLAN MTAG_ABI_COMPAT /* cookie */ > #define MTAG_VLAN_TAG0 1035328035 > #else > #define MTAG_VLAN 1035328035 /* m_tag_cookie */ > #define MTAG_VLAN_TAG0 0 /* tag of VLAN interface */ > #endif > > #define MTAG_VLAN_TAG1 (MTAG_VLAN_TAG0 + 1) > #define MTAG_VLAN_TAG2 (MTAG_VLAN_TAG1 + 1) > #define MTAG_VLAN_TAG3 (MTAG_VLAN_TAG2 + 1) > > struct mv_tag { > struct m_tag tag; > u_int16_t vt_vtag; /* Ethernet 802.1p+q vlan > tag */ > }; > > ng_vlan / if_vlan will use for tag/untag one of MTAG_VLAN_TAGx m_tag_id > and > tunable ether_type for vlan encapsulation: > lower <-> ng_vlan(MTAG_VLAN_TAG0) <-> ng_vlan(MTAG_VLAN_TAG1) <-> > ng_vlan(MTAG_VLAN_TAG2) <-> upper > > > 3. Extended > Same as 2, but > > struct mv_tag { > struct m_tag tag; > u_int16_t vt_vtag; /* Ethernet 802.1p+q vlan > tag */ > u_int16_t ether_type; /* Ethernet type for TAG */ > }; > > > > Major question is: > were store 802.1Q Tag0 = "Customer" VLAN (ether_type: 0x8100, IEEE > 802.1Q): > in pkthdr.PH_vt.vt_vtag or in struct m_tag? > > > ng_vlan modifications (I can make): > + tunable ether_type for vlan encapsulation > + tunable on/off encapsulation (to prevent network adapter > encapsulation) > + tunable m_tag identifier for VLAN tag > ...??? > > > > Any comments before I start? > > > > PS: Trick: > kern.ipc.max_linkhdr should be increased via sysctl: > 20 - 1 VLAN tag (.Q) > 24 - 2 VLAN tags (QinQ) > 28 - 3 VLAN tags (QinQinQ) > 32 - 4 VLAN tags (...) > > > > -- > Rozhuk Ivan > >help
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4ea325f3.4693cc0a.45c2.6fc5>