Index: head/sys/net/bridgestp.c =================================================================== --- head/sys/net/bridgestp.c (revision 234486) +++ head/sys/net/bridgestp.c (revision 234487) @@ -1,2264 +1,2250 @@ /* $NetBSD: bridgestp.c,v 1.5 2003/11/28 08:56:48 keihan Exp $ */ /* * Copyright (c) 2000 Jason L. Wright (jason@thought.net) * Copyright (c) 2006 Andrew Thompson (thompsa@FreeBSD.org) * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE * DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * OpenBSD: bridgestp.c,v 1.5 2001/03/22 03:48:29 jason Exp */ /* * Implementation of the spanning tree protocol as defined in * ISO/IEC 802.1D-2004, June 9, 2004. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef BRIDGESTP_DEBUG #define DPRINTF(fmt, arg...) printf("bstp: " fmt, ##arg) #else #define DPRINTF(fmt, arg...) (void)0 #endif #define PV2ADDR(pv, eaddr) do { \ eaddr[0] = pv >> 40; \ eaddr[1] = pv >> 32; \ eaddr[2] = pv >> 24; \ eaddr[3] = pv >> 16; \ eaddr[4] = pv >> 8; \ eaddr[5] = pv >> 0; \ } while (0) #define INFO_BETTER 1 #define INFO_SAME 0 #define INFO_WORSE -1 const uint8_t bstp_etheraddr[] = { 0x01, 0x80, 0xc2, 0x00, 0x00, 0x00 }; LIST_HEAD(, bstp_state) bstp_list; static struct mtx bstp_list_mtx; static void bstp_transmit(struct bstp_state *, struct bstp_port *); static void bstp_transmit_bpdu(struct bstp_state *, struct bstp_port *); static void bstp_transmit_tcn(struct bstp_state *, struct bstp_port *); static void bstp_decode_bpdu(struct bstp_port *, struct bstp_cbpdu *, struct bstp_config_unit *); static void bstp_send_bpdu(struct bstp_state *, struct bstp_port *, struct bstp_cbpdu *); static int bstp_pdu_flags(struct bstp_port *); static void bstp_received_stp(struct bstp_state *, struct bstp_port *, struct mbuf **, struct bstp_tbpdu *); static void bstp_received_rstp(struct bstp_state *, struct bstp_port *, struct mbuf **, struct bstp_tbpdu *); static void bstp_received_tcn(struct bstp_state *, struct bstp_port *, struct bstp_tcn_unit *); static void bstp_received_bpdu(struct bstp_state *, struct bstp_port *, struct bstp_config_unit *); static int bstp_pdu_rcvtype(struct bstp_port *, struct bstp_config_unit *); static int bstp_pdu_bettersame(struct bstp_port *, int); static int bstp_info_cmp(struct bstp_pri_vector *, struct bstp_pri_vector *); static int bstp_info_superior(struct bstp_pri_vector *, struct bstp_pri_vector *); static void bstp_assign_roles(struct bstp_state *); static void bstp_update_roles(struct bstp_state *, struct bstp_port *); static void bstp_update_state(struct bstp_state *, struct bstp_port *); static void bstp_update_tc(struct bstp_port *); static void bstp_update_info(struct bstp_port *); static void bstp_set_other_tcprop(struct bstp_port *); static void bstp_set_all_reroot(struct bstp_state *); static void bstp_set_all_sync(struct bstp_state *); static void bstp_set_port_state(struct bstp_port *, int); static void bstp_set_port_role(struct bstp_port *, int); static void bstp_set_port_proto(struct bstp_port *, int); static void bstp_set_port_tc(struct bstp_port *, int); static void bstp_set_timer_tc(struct bstp_port *); static void bstp_set_timer_msgage(struct bstp_port *); static int bstp_rerooted(struct bstp_state *, struct bstp_port *); static uint32_t bstp_calc_path_cost(struct bstp_port *); static void bstp_notify_state(void *, int); static void bstp_notify_rtage(void *, int); static void bstp_ifupdstatus(struct bstp_state *, struct bstp_port *); static void bstp_enable_port(struct bstp_state *, struct bstp_port *); static void bstp_disable_port(struct bstp_state *, struct bstp_port *); static void bstp_tick(void *); static void bstp_timer_start(struct bstp_timer *, uint16_t); static void bstp_timer_stop(struct bstp_timer *); static void bstp_timer_latch(struct bstp_timer *); static int bstp_timer_dectest(struct bstp_timer *); static void bstp_hello_timer_expiry(struct bstp_state *, struct bstp_port *); static void bstp_message_age_expiry(struct bstp_state *, struct bstp_port *); static void bstp_migrate_delay_expiry(struct bstp_state *, struct bstp_port *); static void bstp_edge_delay_expiry(struct bstp_state *, struct bstp_port *); static int bstp_addr_cmp(const uint8_t *, const uint8_t *); static int bstp_same_bridgeid(uint64_t, uint64_t); static void bstp_reinit(struct bstp_state *); static void bstp_transmit(struct bstp_state *bs, struct bstp_port *bp) { if (bs->bs_running == 0) return; /* * a PDU can only be sent if we have tx quota left and the * hello timer is running. */ if (bp->bp_hello_timer.active == 0) { /* Test if it needs to be reset */ bstp_hello_timer_expiry(bs, bp); return; } if (bp->bp_txcount > bs->bs_txholdcount) /* Ran out of karma */ return; if (bp->bp_protover == BSTP_PROTO_RSTP) { bstp_transmit_bpdu(bs, bp); bp->bp_tc_ack = 0; } else { /* STP */ switch (bp->bp_role) { case BSTP_ROLE_DESIGNATED: bstp_transmit_bpdu(bs, bp); bp->bp_tc_ack = 0; break; case BSTP_ROLE_ROOT: bstp_transmit_tcn(bs, bp); break; } } bstp_timer_start(&bp->bp_hello_timer, bp->bp_desg_htime); bp->bp_flags &= ~BSTP_PORT_NEWINFO; } static void bstp_transmit_bpdu(struct bstp_state *bs, struct bstp_port *bp) { struct bstp_cbpdu bpdu; BSTP_LOCK_ASSERT(bs); bpdu.cbu_rootpri = htons(bp->bp_desg_pv.pv_root_id >> 48); PV2ADDR(bp->bp_desg_pv.pv_root_id, bpdu.cbu_rootaddr); bpdu.cbu_rootpathcost = htonl(bp->bp_desg_pv.pv_cost); bpdu.cbu_bridgepri = htons(bp->bp_desg_pv.pv_dbridge_id >> 48); PV2ADDR(bp->bp_desg_pv.pv_dbridge_id, bpdu.cbu_bridgeaddr); bpdu.cbu_portid = htons(bp->bp_port_id); bpdu.cbu_messageage = htons(bp->bp_desg_msg_age); bpdu.cbu_maxage = htons(bp->bp_desg_max_age); bpdu.cbu_hellotime = htons(bp->bp_desg_htime); bpdu.cbu_forwarddelay = htons(bp->bp_desg_fdelay); bpdu.cbu_flags = bstp_pdu_flags(bp); switch (bp->bp_protover) { case BSTP_PROTO_STP: bpdu.cbu_bpdutype = BSTP_MSGTYPE_CFG; break; case BSTP_PROTO_RSTP: bpdu.cbu_bpdutype = BSTP_MSGTYPE_RSTP; break; } bstp_send_bpdu(bs, bp, &bpdu); } static void bstp_transmit_tcn(struct bstp_state *bs, struct bstp_port *bp) { struct bstp_tbpdu bpdu; struct ifnet *ifp = bp->bp_ifp; struct ether_header *eh; struct mbuf *m; KASSERT(bp == bs->bs_root_port, ("%s: bad root port\n", __func__)); if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) return; MGETHDR(m, M_DONTWAIT, MT_DATA); if (m == NULL) return; m->m_pkthdr.rcvif = ifp; m->m_pkthdr.len = sizeof(*eh) + sizeof(bpdu); m->m_len = m->m_pkthdr.len; eh = mtod(m, struct ether_header *); memcpy(eh->ether_shost, IF_LLADDR(ifp), ETHER_ADDR_LEN); memcpy(eh->ether_dhost, bstp_etheraddr, ETHER_ADDR_LEN); eh->ether_type = htons(sizeof(bpdu)); bpdu.tbu_ssap = bpdu.tbu_dsap = LLC_8021D_LSAP; bpdu.tbu_ctl = LLC_UI; bpdu.tbu_protoid = 0; bpdu.tbu_protover = 0; bpdu.tbu_bpdutype = BSTP_MSGTYPE_TCN; memcpy(mtod(m, caddr_t) + sizeof(*eh), &bpdu, sizeof(bpdu)); bp->bp_txcount++; ifp->if_transmit(ifp, m); } static void bstp_decode_bpdu(struct bstp_port *bp, struct bstp_cbpdu *cpdu, struct bstp_config_unit *cu) { int flags; cu->cu_pv.pv_root_id = (((uint64_t)ntohs(cpdu->cbu_rootpri)) << 48) | (((uint64_t)cpdu->cbu_rootaddr[0]) << 40) | (((uint64_t)cpdu->cbu_rootaddr[1]) << 32) | (((uint64_t)cpdu->cbu_rootaddr[2]) << 24) | (((uint64_t)cpdu->cbu_rootaddr[3]) << 16) | (((uint64_t)cpdu->cbu_rootaddr[4]) << 8) | (((uint64_t)cpdu->cbu_rootaddr[5]) << 0); cu->cu_pv.pv_dbridge_id = (((uint64_t)ntohs(cpdu->cbu_bridgepri)) << 48) | (((uint64_t)cpdu->cbu_bridgeaddr[0]) << 40) | (((uint64_t)cpdu->cbu_bridgeaddr[1]) << 32) | (((uint64_t)cpdu->cbu_bridgeaddr[2]) << 24) | (((uint64_t)cpdu->cbu_bridgeaddr[3]) << 16) | (((uint64_t)cpdu->cbu_bridgeaddr[4]) << 8) | (((uint64_t)cpdu->cbu_bridgeaddr[5]) << 0); cu->cu_pv.pv_cost = ntohl(cpdu->cbu_rootpathcost); cu->cu_message_age = ntohs(cpdu->cbu_messageage); cu->cu_max_age = ntohs(cpdu->cbu_maxage); cu->cu_hello_time = ntohs(cpdu->cbu_hellotime); cu->cu_forward_delay = ntohs(cpdu->cbu_forwarddelay); cu->cu_pv.pv_dport_id = ntohs(cpdu->cbu_portid); cu->cu_pv.pv_port_id = bp->bp_port_id; cu->cu_message_type = cpdu->cbu_bpdutype; /* Strip off unused flags in STP mode */ flags = cpdu->cbu_flags; switch (cpdu->cbu_protover) { case BSTP_PROTO_STP: flags &= BSTP_PDU_STPMASK; /* A STP BPDU explicitly conveys a Designated Port */ cu->cu_role = BSTP_ROLE_DESIGNATED; break; case BSTP_PROTO_RSTP: flags &= BSTP_PDU_RSTPMASK; break; } cu->cu_topology_change_ack = (flags & BSTP_PDU_F_TCA) ? 1 : 0; cu->cu_proposal = (flags & BSTP_PDU_F_P) ? 1 : 0; cu->cu_agree = (flags & BSTP_PDU_F_A) ? 1 : 0; cu->cu_learning = (flags & BSTP_PDU_F_L) ? 1 : 0; cu->cu_forwarding = (flags & BSTP_PDU_F_F) ? 1 : 0; cu->cu_topology_change = (flags & BSTP_PDU_F_TC) ? 1 : 0; switch ((flags & BSTP_PDU_PRMASK) >> BSTP_PDU_PRSHIFT) { case BSTP_PDU_F_ROOT: cu->cu_role = BSTP_ROLE_ROOT; break; case BSTP_PDU_F_ALT: cu->cu_role = BSTP_ROLE_ALTERNATE; break; case BSTP_PDU_F_DESG: cu->cu_role = BSTP_ROLE_DESIGNATED; break; } } static void bstp_send_bpdu(struct bstp_state *bs, struct bstp_port *bp, struct bstp_cbpdu *bpdu) { struct ifnet *ifp; struct mbuf *m; struct ether_header *eh; BSTP_LOCK_ASSERT(bs); ifp = bp->bp_ifp; if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) return; MGETHDR(m, M_DONTWAIT, MT_DATA); if (m == NULL) return; eh = mtod(m, struct ether_header *); bpdu->cbu_ssap = bpdu->cbu_dsap = LLC_8021D_LSAP; bpdu->cbu_ctl = LLC_UI; bpdu->cbu_protoid = htons(BSTP_PROTO_ID); memcpy(eh->ether_shost, IF_LLADDR(ifp), ETHER_ADDR_LEN); memcpy(eh->ether_dhost, bstp_etheraddr, ETHER_ADDR_LEN); switch (bpdu->cbu_bpdutype) { case BSTP_MSGTYPE_CFG: bpdu->cbu_protover = BSTP_PROTO_STP; m->m_pkthdr.len = sizeof(*eh) + BSTP_BPDU_STP_LEN; eh->ether_type = htons(BSTP_BPDU_STP_LEN); memcpy(mtod(m, caddr_t) + sizeof(*eh), bpdu, BSTP_BPDU_STP_LEN); break; case BSTP_MSGTYPE_RSTP: bpdu->cbu_protover = BSTP_PROTO_RSTP; bpdu->cbu_versionlen = htons(0); m->m_pkthdr.len = sizeof(*eh) + BSTP_BPDU_RSTP_LEN; eh->ether_type = htons(BSTP_BPDU_RSTP_LEN); memcpy(mtod(m, caddr_t) + sizeof(*eh), bpdu, BSTP_BPDU_RSTP_LEN); break; default: panic("not implemented"); } m->m_pkthdr.rcvif = ifp; m->m_len = m->m_pkthdr.len; bp->bp_txcount++; ifp->if_transmit(ifp, m); } static int bstp_pdu_flags(struct bstp_port *bp) { int flags = 0; if (bp->bp_proposing && bp->bp_state != BSTP_IFSTATE_FORWARDING) flags |= BSTP_PDU_F_P; if (bp->bp_agree) flags |= BSTP_PDU_F_A; if (bp->bp_tc_timer.active) flags |= BSTP_PDU_F_TC; if (bp->bp_tc_ack) flags |= BSTP_PDU_F_TCA; switch (bp->bp_state) { case BSTP_IFSTATE_LEARNING: flags |= BSTP_PDU_F_L; break; case BSTP_IFSTATE_FORWARDING: flags |= (BSTP_PDU_F_L | BSTP_PDU_F_F); break; } switch (bp->bp_role) { case BSTP_ROLE_ROOT: flags |= (BSTP_PDU_F_ROOT << BSTP_PDU_PRSHIFT); break; case BSTP_ROLE_ALTERNATE: case BSTP_ROLE_BACKUP: /* fall through */ flags |= (BSTP_PDU_F_ALT << BSTP_PDU_PRSHIFT); break; case BSTP_ROLE_DESIGNATED: flags |= (BSTP_PDU_F_DESG << BSTP_PDU_PRSHIFT); break; } /* Strip off unused flags in either mode */ switch (bp->bp_protover) { case BSTP_PROTO_STP: flags &= BSTP_PDU_STPMASK; break; case BSTP_PROTO_RSTP: flags &= BSTP_PDU_RSTPMASK; break; } return (flags); } void bstp_input(struct bstp_port *bp, struct ifnet *ifp, struct mbuf *m) { struct bstp_state *bs = bp->bp_bs; struct ether_header *eh; struct bstp_tbpdu tpdu; uint16_t len; if (bp->bp_active == 0) { m_freem(m); return; } BSTP_LOCK(bs); eh = mtod(m, struct ether_header *); len = ntohs(eh->ether_type); if (len < sizeof(tpdu)) goto out; m_adj(m, ETHER_HDR_LEN); if (m->m_pkthdr.len > len) m_adj(m, len - m->m_pkthdr.len); if (m->m_len < sizeof(tpdu) && (m = m_pullup(m, sizeof(tpdu))) == NULL) goto out; memcpy(&tpdu, mtod(m, caddr_t), sizeof(tpdu)); /* basic packet checks */ if (tpdu.tbu_dsap != LLC_8021D_LSAP || tpdu.tbu_ssap != LLC_8021D_LSAP || tpdu.tbu_ctl != LLC_UI) goto out; if (tpdu.tbu_protoid != BSTP_PROTO_ID) goto out; /* * We can treat later versions of the PDU as the same as the maximum * version we implement. All additional parameters/flags are ignored. */ if (tpdu.tbu_protover > BSTP_PROTO_MAX) tpdu.tbu_protover = BSTP_PROTO_MAX; if (tpdu.tbu_protover != bp->bp_protover) { /* * Wait for the migration delay timer to expire before changing * protocol version to avoid flip-flops. */ if (bp->bp_flags & BSTP_PORT_CANMIGRATE) bstp_set_port_proto(bp, tpdu.tbu_protover); else goto out; } /* Clear operedge upon receiving a PDU on the port */ bp->bp_operedge = 0; bstp_timer_start(&bp->bp_edge_delay_timer, BSTP_DEFAULT_MIGRATE_DELAY); switch (tpdu.tbu_protover) { case BSTP_PROTO_STP: bstp_received_stp(bs, bp, &m, &tpdu); break; case BSTP_PROTO_RSTP: bstp_received_rstp(bs, bp, &m, &tpdu); break; } out: BSTP_UNLOCK(bs); if (m) m_freem(m); } static void bstp_received_stp(struct bstp_state *bs, struct bstp_port *bp, struct mbuf **mp, struct bstp_tbpdu *tpdu) { struct bstp_cbpdu cpdu; struct bstp_config_unit *cu = &bp->bp_msg_cu; struct bstp_tcn_unit tu; switch (tpdu->tbu_bpdutype) { case BSTP_MSGTYPE_TCN: tu.tu_message_type = tpdu->tbu_bpdutype; bstp_received_tcn(bs, bp, &tu); break; case BSTP_MSGTYPE_CFG: if ((*mp)->m_len < BSTP_BPDU_STP_LEN && (*mp = m_pullup(*mp, BSTP_BPDU_STP_LEN)) == NULL) return; memcpy(&cpdu, mtod(*mp, caddr_t), BSTP_BPDU_STP_LEN); bstp_decode_bpdu(bp, &cpdu, cu); bstp_received_bpdu(bs, bp, cu); break; } } static void bstp_received_rstp(struct bstp_state *bs, struct bstp_port *bp, struct mbuf **mp, struct bstp_tbpdu *tpdu) { struct bstp_cbpdu cpdu; struct bstp_config_unit *cu = &bp->bp_msg_cu; if (tpdu->tbu_bpdutype != BSTP_MSGTYPE_RSTP) return; if ((*mp)->m_len < BSTP_BPDU_RSTP_LEN && (*mp = m_pullup(*mp, BSTP_BPDU_RSTP_LEN)) == NULL) return; memcpy(&cpdu, mtod(*mp, caddr_t), BSTP_BPDU_RSTP_LEN); bstp_decode_bpdu(bp, &cpdu, cu); bstp_received_bpdu(bs, bp, cu); } static void bstp_received_tcn(struct bstp_state *bs, struct bstp_port *bp, struct bstp_tcn_unit *tcn) { bp->bp_rcvdtcn = 1; bstp_update_tc(bp); } static void bstp_received_bpdu(struct bstp_state *bs, struct bstp_port *bp, struct bstp_config_unit *cu) { int type; BSTP_LOCK_ASSERT(bs); /* We need to have transitioned to INFO_MINE before proceeding */ switch (bp->bp_infois) { case BSTP_INFO_DISABLED: case BSTP_INFO_AGED: return; } type = bstp_pdu_rcvtype(bp, cu); switch (type) { case BSTP_PDU_SUPERIOR: bs->bs_allsynced = 0; bp->bp_agreed = 0; bp->bp_proposing = 0; if (cu->cu_proposal && cu->cu_forwarding == 0) bp->bp_proposed = 1; if (cu->cu_topology_change) bp->bp_rcvdtc = 1; if (cu->cu_topology_change_ack) bp->bp_rcvdtca = 1; if (bp->bp_agree && !bstp_pdu_bettersame(bp, BSTP_INFO_RECEIVED)) bp->bp_agree = 0; /* copy the received priority and timers to the port */ bp->bp_port_pv = cu->cu_pv; bp->bp_port_msg_age = cu->cu_message_age; bp->bp_port_max_age = cu->cu_max_age; bp->bp_port_fdelay = cu->cu_forward_delay; bp->bp_port_htime = (cu->cu_hello_time > BSTP_MIN_HELLO_TIME ? cu->cu_hello_time : BSTP_MIN_HELLO_TIME); /* set expiry for the new info */ bstp_set_timer_msgage(bp); bp->bp_infois = BSTP_INFO_RECEIVED; bstp_assign_roles(bs); break; case BSTP_PDU_REPEATED: if (cu->cu_proposal && cu->cu_forwarding == 0) bp->bp_proposed = 1; if (cu->cu_topology_change) bp->bp_rcvdtc = 1; if (cu->cu_topology_change_ack) bp->bp_rcvdtca = 1; /* rearm the age timer */ bstp_set_timer_msgage(bp); break; case BSTP_PDU_INFERIOR: if (cu->cu_learning) { bp->bp_agreed = 1; bp->bp_proposing = 0; } break; case BSTP_PDU_INFERIORALT: /* * only point to point links are allowed fast * transitions to forwarding. */ if (cu->cu_agree && bp->bp_ptp_link) { bp->bp_agreed = 1; bp->bp_proposing = 0; } else bp->bp_agreed = 0; if (cu->cu_topology_change) bp->bp_rcvdtc = 1; if (cu->cu_topology_change_ack) bp->bp_rcvdtca = 1; break; case BSTP_PDU_OTHER: return; /* do nothing */ } /* update the state machines with the new data */ bstp_update_state(bs, bp); } static int bstp_pdu_rcvtype(struct bstp_port *bp, struct bstp_config_unit *cu) { int type; /* default return type */ type = BSTP_PDU_OTHER; switch (cu->cu_role) { case BSTP_ROLE_DESIGNATED: if (bstp_info_superior(&bp->bp_port_pv, &cu->cu_pv)) /* bpdu priority is superior */ type = BSTP_PDU_SUPERIOR; else if (bstp_info_cmp(&bp->bp_port_pv, &cu->cu_pv) == INFO_SAME) { if (bp->bp_port_msg_age != cu->cu_message_age || bp->bp_port_max_age != cu->cu_max_age || bp->bp_port_fdelay != cu->cu_forward_delay || bp->bp_port_htime != cu->cu_hello_time) /* bpdu priority is equal and timers differ */ type = BSTP_PDU_SUPERIOR; else /* bpdu is equal */ type = BSTP_PDU_REPEATED; } else /* bpdu priority is worse */ type = BSTP_PDU_INFERIOR; break; case BSTP_ROLE_ROOT: case BSTP_ROLE_ALTERNATE: case BSTP_ROLE_BACKUP: if (bstp_info_cmp(&bp->bp_port_pv, &cu->cu_pv) <= INFO_SAME) /* * not a designated port and priority is the same or * worse */ type = BSTP_PDU_INFERIORALT; break; } return (type); } static int bstp_pdu_bettersame(struct bstp_port *bp, int newinfo) { if (newinfo == BSTP_INFO_RECEIVED && bp->bp_infois == BSTP_INFO_RECEIVED && bstp_info_cmp(&bp->bp_port_pv, &bp->bp_msg_cu.cu_pv) >= INFO_SAME) return (1); if (newinfo == BSTP_INFO_MINE && bp->bp_infois == BSTP_INFO_MINE && bstp_info_cmp(&bp->bp_port_pv, &bp->bp_desg_pv) >= INFO_SAME) return (1); return (0); } static int bstp_info_cmp(struct bstp_pri_vector *pv, struct bstp_pri_vector *cpv) { if (cpv->pv_root_id < pv->pv_root_id) return (INFO_BETTER); if (cpv->pv_root_id > pv->pv_root_id) return (INFO_WORSE); if (cpv->pv_cost < pv->pv_cost) return (INFO_BETTER); if (cpv->pv_cost > pv->pv_cost) return (INFO_WORSE); if (cpv->pv_dbridge_id < pv->pv_dbridge_id) return (INFO_BETTER); if (cpv->pv_dbridge_id > pv->pv_dbridge_id) return (INFO_WORSE); if (cpv->pv_dport_id < pv->pv_dport_id) return (INFO_BETTER); if (cpv->pv_dport_id > pv->pv_dport_id) return (INFO_WORSE); return (INFO_SAME); } /* * This message priority vector is superior to the port priority vector and * will replace it if, and only if, the message priority vector is better than * the port priority vector, or the message has been transmitted from the same * designated bridge and designated port as the port priority vector. */ static int bstp_info_superior(struct bstp_pri_vector *pv, struct bstp_pri_vector *cpv) { if (bstp_info_cmp(pv, cpv) == INFO_BETTER || (bstp_same_bridgeid(pv->pv_dbridge_id, cpv->pv_dbridge_id) && (cpv->pv_dport_id & 0xfff) == (pv->pv_dport_id & 0xfff))) return (1); return (0); } static void bstp_assign_roles(struct bstp_state *bs) { struct bstp_port *bp, *rbp = NULL; struct bstp_pri_vector pv; /* default to our priority vector */ bs->bs_root_pv = bs->bs_bridge_pv; bs->bs_root_msg_age = 0; bs->bs_root_max_age = bs->bs_bridge_max_age; bs->bs_root_fdelay = bs->bs_bridge_fdelay; bs->bs_root_htime = bs->bs_bridge_htime; bs->bs_root_port = NULL; /* check if any recieved info supersedes us */ LIST_FOREACH(bp, &bs->bs_bplist, bp_next) { if (bp->bp_infois != BSTP_INFO_RECEIVED) continue; pv = bp->bp_port_pv; pv.pv_cost += bp->bp_path_cost; /* * The root priority vector is the best of the set comprising * the bridge priority vector plus all root path priority * vectors whose bridge address is not equal to us. */ if (bstp_same_bridgeid(pv.pv_dbridge_id, bs->bs_bridge_pv.pv_dbridge_id) == 0 && bstp_info_cmp(&bs->bs_root_pv, &pv) == INFO_BETTER) { /* the port vector replaces the root */ bs->bs_root_pv = pv; bs->bs_root_msg_age = bp->bp_port_msg_age + BSTP_MESSAGE_AGE_INCR; bs->bs_root_max_age = bp->bp_port_max_age; bs->bs_root_fdelay = bp->bp_port_fdelay; bs->bs_root_htime = bp->bp_port_htime; rbp = bp; } } LIST_FOREACH(bp, &bs->bs_bplist, bp_next) { /* calculate the port designated vector */ bp->bp_desg_pv.pv_root_id = bs->bs_root_pv.pv_root_id; bp->bp_desg_pv.pv_cost = bs->bs_root_pv.pv_cost; bp->bp_desg_pv.pv_dbridge_id = bs->bs_bridge_pv.pv_dbridge_id; bp->bp_desg_pv.pv_dport_id = bp->bp_port_id; bp->bp_desg_pv.pv_port_id = bp->bp_port_id; /* calculate designated times */ bp->bp_desg_msg_age = bs->bs_root_msg_age; bp->bp_desg_max_age = bs->bs_root_max_age; bp->bp_desg_fdelay = bs->bs_root_fdelay; bp->bp_desg_htime = bs->bs_bridge_htime; switch (bp->bp_infois) { case BSTP_INFO_DISABLED: bstp_set_port_role(bp, BSTP_ROLE_DISABLED); break; case BSTP_INFO_AGED: bstp_set_port_role(bp, BSTP_ROLE_DESIGNATED); bstp_update_info(bp); break; case BSTP_INFO_MINE: bstp_set_port_role(bp, BSTP_ROLE_DESIGNATED); /* update the port info if stale */ if (bstp_info_cmp(&bp->bp_port_pv, &bp->bp_desg_pv) != INFO_SAME || (rbp != NULL && (bp->bp_port_msg_age != rbp->bp_port_msg_age || bp->bp_port_max_age != rbp->bp_port_max_age || bp->bp_port_fdelay != rbp->bp_port_fdelay || bp->bp_port_htime != rbp->bp_port_htime))) bstp_update_info(bp); break; case BSTP_INFO_RECEIVED: if (bp == rbp) { /* * root priority is derived from this * port, make it the root port. */ bstp_set_port_role(bp, BSTP_ROLE_ROOT); bs->bs_root_port = bp; } else if (bstp_info_cmp(&bp->bp_port_pv, &bp->bp_desg_pv) == INFO_BETTER) { /* * the port priority is lower than the root * port. */ bstp_set_port_role(bp, BSTP_ROLE_DESIGNATED); bstp_update_info(bp); } else { if (bstp_same_bridgeid( bp->bp_port_pv.pv_dbridge_id, bs->bs_bridge_pv.pv_dbridge_id)) { /* * the designated bridge refers to * another port on this bridge. */ bstp_set_port_role(bp, BSTP_ROLE_BACKUP); } else { /* * the port is an inferior path to the * root bridge. */ bstp_set_port_role(bp, BSTP_ROLE_ALTERNATE); } } break; } } } static void bstp_update_state(struct bstp_state *bs, struct bstp_port *bp) { struct bstp_port *bp2; int synced; BSTP_LOCK_ASSERT(bs); /* check if all the ports have syncronised again */ if (!bs->bs_allsynced) { synced = 1; LIST_FOREACH(bp2, &bs->bs_bplist, bp_next) { if (!(bp2->bp_synced || bp2->bp_role == BSTP_ROLE_ROOT)) { synced = 0; break; } } bs->bs_allsynced = synced; } bstp_update_roles(bs, bp); bstp_update_tc(bp); } static void bstp_update_roles(struct bstp_state *bs, struct bstp_port *bp) { switch (bp->bp_role) { case BSTP_ROLE_DISABLED: /* Clear any flags if set */ if (bp->bp_sync || !bp->bp_synced || bp->bp_reroot) { bp->bp_sync = 0; bp->bp_synced = 1; bp->bp_reroot = 0; } break; case BSTP_ROLE_ALTERNATE: case BSTP_ROLE_BACKUP: if ((bs->bs_allsynced && !bp->bp_agree) || (bp->bp_proposed && bp->bp_agree)) { bp->bp_proposed = 0; bp->bp_agree = 1; bp->bp_flags |= BSTP_PORT_NEWINFO; DPRINTF("%s -> ALTERNATE_AGREED\n", bp->bp_ifp->if_xname); } if (bp->bp_proposed && !bp->bp_agree) { bstp_set_all_sync(bs); bp->bp_proposed = 0; DPRINTF("%s -> ALTERNATE_PROPOSED\n", bp->bp_ifp->if_xname); } /* Clear any flags if set */ if (bp->bp_sync || !bp->bp_synced || bp->bp_reroot) { bp->bp_sync = 0; bp->bp_synced = 1; bp->bp_reroot = 0; DPRINTF("%s -> ALTERNATE_PORT\n", bp->bp_ifp->if_xname); } break; case BSTP_ROLE_ROOT: if (bp->bp_state != BSTP_IFSTATE_FORWARDING && !bp->bp_reroot) { bstp_set_all_reroot(bs); DPRINTF("%s -> ROOT_REROOT\n", bp->bp_ifp->if_xname); } if ((bs->bs_allsynced && !bp->bp_agree) || (bp->bp_proposed && bp->bp_agree)) { bp->bp_proposed = 0; bp->bp_sync = 0; bp->bp_agree = 1; bp->bp_flags |= BSTP_PORT_NEWINFO; DPRINTF("%s -> ROOT_AGREED\n", bp->bp_ifp->if_xname); } if (bp->bp_proposed && !bp->bp_agree) { bstp_set_all_sync(bs); bp->bp_proposed = 0; DPRINTF("%s -> ROOT_PROPOSED\n", bp->bp_ifp->if_xname); } if (bp->bp_state != BSTP_IFSTATE_FORWARDING && (bp->bp_forward_delay_timer.active == 0 || (bstp_rerooted(bs, bp) && bp->bp_recent_backup_timer.active == 0 && bp->bp_protover == BSTP_PROTO_RSTP))) { switch (bp->bp_state) { case BSTP_IFSTATE_DISCARDING: bstp_set_port_state(bp, BSTP_IFSTATE_LEARNING); break; case BSTP_IFSTATE_LEARNING: bstp_set_port_state(bp, BSTP_IFSTATE_FORWARDING); break; } } if (bp->bp_state == BSTP_IFSTATE_FORWARDING && bp->bp_reroot) { bp->bp_reroot = 0; DPRINTF("%s -> ROOT_REROOTED\n", bp->bp_ifp->if_xname); } break; case BSTP_ROLE_DESIGNATED: if (bp->bp_recent_root_timer.active == 0 && bp->bp_reroot) { bp->bp_reroot = 0; DPRINTF("%s -> DESIGNATED_RETIRED\n", bp->bp_ifp->if_xname); } if ((bp->bp_state == BSTP_IFSTATE_DISCARDING && !bp->bp_synced) || (bp->bp_agreed && !bp->bp_synced) || (bp->bp_operedge && !bp->bp_synced) || (bp->bp_sync && bp->bp_synced)) { bstp_timer_stop(&bp->bp_recent_root_timer); bp->bp_synced = 1; bp->bp_sync = 0; DPRINTF("%s -> DESIGNATED_SYNCED\n", bp->bp_ifp->if_xname); } if (bp->bp_state != BSTP_IFSTATE_FORWARDING && !bp->bp_agreed && !bp->bp_proposing && !bp->bp_operedge) { bp->bp_proposing = 1; bp->bp_flags |= BSTP_PORT_NEWINFO; bstp_timer_start(&bp->bp_edge_delay_timer, (bp->bp_ptp_link ? BSTP_DEFAULT_MIGRATE_DELAY : bp->bp_desg_max_age)); DPRINTF("%s -> DESIGNATED_PROPOSE\n", bp->bp_ifp->if_xname); } if (bp->bp_state != BSTP_IFSTATE_FORWARDING && (bp->bp_forward_delay_timer.active == 0 || bp->bp_agreed || bp->bp_operedge) && (bp->bp_recent_root_timer.active == 0 || !bp->bp_reroot) && !bp->bp_sync) { if (bp->bp_agreed) DPRINTF("%s -> AGREED\n", bp->bp_ifp->if_xname); /* * If agreed|operedge then go straight to forwarding, * otherwise follow discard -> learn -> forward. */ if (bp->bp_agreed || bp->bp_operedge || bp->bp_state == BSTP_IFSTATE_LEARNING) { bstp_set_port_state(bp, BSTP_IFSTATE_FORWARDING); bp->bp_agreed = bp->bp_protover; } else if (bp->bp_state == BSTP_IFSTATE_DISCARDING) bstp_set_port_state(bp, BSTP_IFSTATE_LEARNING); } if (((bp->bp_sync && !bp->bp_synced) || (bp->bp_reroot && bp->bp_recent_root_timer.active) || (bp->bp_flags & BSTP_PORT_DISPUTED)) && !bp->bp_operedge && bp->bp_state != BSTP_IFSTATE_DISCARDING) { bstp_set_port_state(bp, BSTP_IFSTATE_DISCARDING); bp->bp_flags &= ~BSTP_PORT_DISPUTED; bstp_timer_start(&bp->bp_forward_delay_timer, bp->bp_protover == BSTP_PROTO_RSTP ? bp->bp_desg_htime : bp->bp_desg_fdelay); DPRINTF("%s -> DESIGNATED_DISCARD\n", bp->bp_ifp->if_xname); } break; } if (bp->bp_flags & BSTP_PORT_NEWINFO) bstp_transmit(bs, bp); } static void bstp_update_tc(struct bstp_port *bp) { switch (bp->bp_tcstate) { case BSTP_TCSTATE_ACTIVE: if ((bp->bp_role != BSTP_ROLE_DESIGNATED && bp->bp_role != BSTP_ROLE_ROOT) || bp->bp_operedge) bstp_set_port_tc(bp, BSTP_TCSTATE_LEARNING); if (bp->bp_rcvdtcn) bstp_set_port_tc(bp, BSTP_TCSTATE_TCN); if (bp->bp_rcvdtc) bstp_set_port_tc(bp, BSTP_TCSTATE_TC); if (bp->bp_tc_prop && !bp->bp_operedge) bstp_set_port_tc(bp, BSTP_TCSTATE_PROPAG); if (bp->bp_rcvdtca) bstp_set_port_tc(bp, BSTP_TCSTATE_ACK); break; case BSTP_TCSTATE_INACTIVE: if ((bp->bp_state == BSTP_IFSTATE_LEARNING || bp->bp_state == BSTP_IFSTATE_FORWARDING) && bp->bp_fdbflush == 0) bstp_set_port_tc(bp, BSTP_TCSTATE_LEARNING); break; case BSTP_TCSTATE_LEARNING: if (bp->bp_rcvdtc || bp->bp_rcvdtcn || bp->bp_rcvdtca || bp->bp_tc_prop) bstp_set_port_tc(bp, BSTP_TCSTATE_LEARNING); else if (bp->bp_role != BSTP_ROLE_DESIGNATED && bp->bp_role != BSTP_ROLE_ROOT && bp->bp_state == BSTP_IFSTATE_DISCARDING) bstp_set_port_tc(bp, BSTP_TCSTATE_INACTIVE); if ((bp->bp_role == BSTP_ROLE_DESIGNATED || bp->bp_role == BSTP_ROLE_ROOT) && bp->bp_state == BSTP_IFSTATE_FORWARDING && !bp->bp_operedge) bstp_set_port_tc(bp, BSTP_TCSTATE_DETECTED); break; /* these are transient states and go straight back to ACTIVE */ case BSTP_TCSTATE_DETECTED: case BSTP_TCSTATE_TCN: case BSTP_TCSTATE_TC: case BSTP_TCSTATE_PROPAG: case BSTP_TCSTATE_ACK: DPRINTF("Invalid TC state for %s\n", bp->bp_ifp->if_xname); break; } } static void bstp_update_info(struct bstp_port *bp) { struct bstp_state *bs = bp->bp_bs; bp->bp_proposing = 0; bp->bp_proposed = 0; if (bp->bp_agreed && !bstp_pdu_bettersame(bp, BSTP_INFO_MINE)) bp->bp_agreed = 0; if (bp->bp_synced && !bp->bp_agreed) { bp->bp_synced = 0; bs->bs_allsynced = 0; } /* copy the designated pv to the port */ bp->bp_port_pv = bp->bp_desg_pv; bp->bp_port_msg_age = bp->bp_desg_msg_age; bp->bp_port_max_age = bp->bp_desg_max_age; bp->bp_port_fdelay = bp->bp_desg_fdelay; bp->bp_port_htime = bp->bp_desg_htime; bp->bp_infois = BSTP_INFO_MINE; /* Set transmit flag but do not immediately send */ bp->bp_flags |= BSTP_PORT_NEWINFO; } /* set tcprop on every port other than the caller */ static void bstp_set_other_tcprop(struct bstp_port *bp) { struct bstp_state *bs = bp->bp_bs; struct bstp_port *bp2; BSTP_LOCK_ASSERT(bs); LIST_FOREACH(bp2, &bs->bs_bplist, bp_next) { if (bp2 == bp) continue; bp2->bp_tc_prop = 1; } } static void bstp_set_all_reroot(struct bstp_state *bs) { struct bstp_port *bp; BSTP_LOCK_ASSERT(bs); LIST_FOREACH(bp, &bs->bs_bplist, bp_next) bp->bp_reroot = 1; } static void bstp_set_all_sync(struct bstp_state *bs) { struct bstp_port *bp; BSTP_LOCK_ASSERT(bs); LIST_FOREACH(bp, &bs->bs_bplist, bp_next) { bp->bp_sync = 1; bp->bp_synced = 0; /* Not explicit in spec */ } bs->bs_allsynced = 0; } static void bstp_set_port_state(struct bstp_port *bp, int state) { if (bp->bp_state == state) return; bp->bp_state = state; switch (bp->bp_state) { case BSTP_IFSTATE_DISCARDING: DPRINTF("state changed to DISCARDING on %s\n", bp->bp_ifp->if_xname); break; case BSTP_IFSTATE_LEARNING: DPRINTF("state changed to LEARNING on %s\n", bp->bp_ifp->if_xname); bstp_timer_start(&bp->bp_forward_delay_timer, bp->bp_protover == BSTP_PROTO_RSTP ? bp->bp_desg_htime : bp->bp_desg_fdelay); break; case BSTP_IFSTATE_FORWARDING: DPRINTF("state changed to FORWARDING on %s\n", bp->bp_ifp->if_xname); bstp_timer_stop(&bp->bp_forward_delay_timer); /* Record that we enabled forwarding */ bp->bp_forward_transitions++; break; } /* notify the parent bridge */ taskqueue_enqueue(taskqueue_swi, &bp->bp_statetask); } static void bstp_set_port_role(struct bstp_port *bp, int role) { struct bstp_state *bs = bp->bp_bs; if (bp->bp_role == role) return; /* perform pre-change tasks */ switch (bp->bp_role) { case BSTP_ROLE_DISABLED: bstp_timer_start(&bp->bp_forward_delay_timer, bp->bp_desg_max_age); break; case BSTP_ROLE_BACKUP: bstp_timer_start(&bp->bp_recent_backup_timer, bp->bp_desg_htime * 2); /* fall through */ case BSTP_ROLE_ALTERNATE: bstp_timer_start(&bp->bp_forward_delay_timer, bp->bp_desg_fdelay); bp->bp_sync = 0; bp->bp_synced = 1; bp->bp_reroot = 0; break; case BSTP_ROLE_ROOT: bstp_timer_start(&bp->bp_recent_root_timer, BSTP_DEFAULT_FORWARD_DELAY); break; } bp->bp_role = role; /* clear values not carried between roles */ bp->bp_proposing = 0; bs->bs_allsynced = 0; /* initialise the new role */ switch (bp->bp_role) { case BSTP_ROLE_DISABLED: case BSTP_ROLE_ALTERNATE: case BSTP_ROLE_BACKUP: DPRINTF("%s role -> ALT/BACK/DISABLED\n", bp->bp_ifp->if_xname); bstp_set_port_state(bp, BSTP_IFSTATE_DISCARDING); bstp_timer_stop(&bp->bp_recent_root_timer); bstp_timer_latch(&bp->bp_forward_delay_timer); bp->bp_sync = 0; bp->bp_synced = 1; bp->bp_reroot = 0; break; case BSTP_ROLE_ROOT: DPRINTF("%s role -> ROOT\n", bp->bp_ifp->if_xname); bstp_set_port_state(bp, BSTP_IFSTATE_DISCARDING); bstp_timer_latch(&bp->bp_recent_root_timer); bp->bp_proposing = 0; break; case BSTP_ROLE_DESIGNATED: DPRINTF("%s role -> DESIGNATED\n", bp->bp_ifp->if_xname); bstp_timer_start(&bp->bp_hello_timer, bp->bp_desg_htime); bp->bp_agree = 0; break; } /* let the TC state know that the role changed */ bstp_update_tc(bp); } static void bstp_set_port_proto(struct bstp_port *bp, int proto) { struct bstp_state *bs = bp->bp_bs; /* supported protocol versions */ switch (proto) { case BSTP_PROTO_STP: /* we can downgrade protocols only */ bstp_timer_stop(&bp->bp_migrate_delay_timer); /* clear unsupported features */ bp->bp_operedge = 0; /* STP compat mode only uses 16 bits of the 32 */ if (bp->bp_path_cost > 65535) bp->bp_path_cost = 65535; break; case BSTP_PROTO_RSTP: bstp_timer_start(&bp->bp_migrate_delay_timer, bs->bs_migration_delay); break; default: DPRINTF("Unsupported STP version %d\n", proto); return; } bp->bp_protover = proto; bp->bp_flags &= ~BSTP_PORT_CANMIGRATE; } static void bstp_set_port_tc(struct bstp_port *bp, int state) { struct bstp_state *bs = bp->bp_bs; bp->bp_tcstate = state; /* initialise the new state */ switch (bp->bp_tcstate) { case BSTP_TCSTATE_ACTIVE: DPRINTF("%s -> TC_ACTIVE\n", bp->bp_ifp->if_xname); /* nothing to do */ break; case BSTP_TCSTATE_INACTIVE: bstp_timer_stop(&bp->bp_tc_timer); /* flush routes on the parent bridge */ bp->bp_fdbflush = 1; taskqueue_enqueue(taskqueue_swi, &bp->bp_rtagetask); bp->bp_tc_ack = 0; DPRINTF("%s -> TC_INACTIVE\n", bp->bp_ifp->if_xname); break; case BSTP_TCSTATE_LEARNING: bp->bp_rcvdtc = 0; bp->bp_rcvdtcn = 0; bp->bp_rcvdtca = 0; bp->bp_tc_prop = 0; DPRINTF("%s -> TC_LEARNING\n", bp->bp_ifp->if_xname); break; case BSTP_TCSTATE_DETECTED: bstp_set_timer_tc(bp); bstp_set_other_tcprop(bp); /* send out notification */ bp->bp_flags |= BSTP_PORT_NEWINFO; bstp_transmit(bs, bp); getmicrotime(&bs->bs_last_tc_time); DPRINTF("%s -> TC_DETECTED\n", bp->bp_ifp->if_xname); bp->bp_tcstate = BSTP_TCSTATE_ACTIVE; /* UCT */ break; case BSTP_TCSTATE_TCN: bstp_set_timer_tc(bp); DPRINTF("%s -> TC_TCN\n", bp->bp_ifp->if_xname); /* fall through */ case BSTP_TCSTATE_TC: bp->bp_rcvdtc = 0; bp->bp_rcvdtcn = 0; if (bp->bp_role == BSTP_ROLE_DESIGNATED) bp->bp_tc_ack = 1; bstp_set_other_tcprop(bp); DPRINTF("%s -> TC_TC\n", bp->bp_ifp->if_xname); bp->bp_tcstate = BSTP_TCSTATE_ACTIVE; /* UCT */ break; case BSTP_TCSTATE_PROPAG: /* flush routes on the parent bridge */ bp->bp_fdbflush = 1; taskqueue_enqueue(taskqueue_swi, &bp->bp_rtagetask); bp->bp_tc_prop = 0; bstp_set_timer_tc(bp); DPRINTF("%s -> TC_PROPAG\n", bp->bp_ifp->if_xname); bp->bp_tcstate = BSTP_TCSTATE_ACTIVE; /* UCT */ break; case BSTP_TCSTATE_ACK: bstp_timer_stop(&bp->bp_tc_timer); bp->bp_rcvdtca = 0; DPRINTF("%s -> TC_ACK\n", bp->bp_ifp->if_xname); bp->bp_tcstate = BSTP_TCSTATE_ACTIVE; /* UCT */ break; } } static void bstp_set_timer_tc(struct bstp_port *bp) { struct bstp_state *bs = bp->bp_bs; if (bp->bp_tc_timer.active) return; switch (bp->bp_protover) { case BSTP_PROTO_RSTP: bstp_timer_start(&bp->bp_tc_timer, bp->bp_desg_htime + BSTP_TICK_VAL); bp->bp_flags |= BSTP_PORT_NEWINFO; break; case BSTP_PROTO_STP: bstp_timer_start(&bp->bp_tc_timer, bs->bs_root_max_age + bs->bs_root_fdelay); break; } } static void bstp_set_timer_msgage(struct bstp_port *bp) { if (bp->bp_port_msg_age + BSTP_MESSAGE_AGE_INCR <= bp->bp_port_max_age) { bstp_timer_start(&bp->bp_message_age_timer, bp->bp_port_htime * 3); } else /* expires immediately */ bstp_timer_start(&bp->bp_message_age_timer, 0); } static int bstp_rerooted(struct bstp_state *bs, struct bstp_port *bp) { struct bstp_port *bp2; int rr_set = 0; LIST_FOREACH(bp2, &bs->bs_bplist, bp_next) { if (bp2 == bp) continue; if (bp2->bp_recent_root_timer.active) { rr_set = 1; break; } } return (!rr_set); } int bstp_set_htime(struct bstp_state *bs, int t) { /* convert seconds to ticks */ t *= BSTP_TICK_VAL; /* value can only be changed in leagacy stp mode */ if (bs->bs_protover != BSTP_PROTO_STP) return (EPERM); if (t < BSTP_MIN_HELLO_TIME || t > BSTP_MAX_HELLO_TIME) return (EINVAL); BSTP_LOCK(bs); bs->bs_bridge_htime = t; bstp_reinit(bs); BSTP_UNLOCK(bs); return (0); } int bstp_set_fdelay(struct bstp_state *bs, int t) { /* convert seconds to ticks */ t *= BSTP_TICK_VAL; if (t < BSTP_MIN_FORWARD_DELAY || t > BSTP_MAX_FORWARD_DELAY) return (EINVAL); BSTP_LOCK(bs); bs->bs_bridge_fdelay = t; bstp_reinit(bs); BSTP_UNLOCK(bs); return (0); } int bstp_set_maxage(struct bstp_state *bs, int t) { /* convert seconds to ticks */ t *= BSTP_TICK_VAL; if (t < BSTP_MIN_MAX_AGE || t > BSTP_MAX_MAX_AGE) return (EINVAL); BSTP_LOCK(bs); bs->bs_bridge_max_age = t; bstp_reinit(bs); BSTP_UNLOCK(bs); return (0); } int bstp_set_holdcount(struct bstp_state *bs, int count) { struct bstp_port *bp; if (count < BSTP_MIN_HOLD_COUNT || count > BSTP_MAX_HOLD_COUNT) return (EINVAL); BSTP_LOCK(bs); bs->bs_txholdcount = count; LIST_FOREACH(bp, &bs->bs_bplist, bp_next) bp->bp_txcount = 0; BSTP_UNLOCK(bs); return (0); } int bstp_set_protocol(struct bstp_state *bs, int proto) { struct bstp_port *bp; switch (proto) { /* Supported protocol versions */ case BSTP_PROTO_STP: case BSTP_PROTO_RSTP: break; default: return (EINVAL); } BSTP_LOCK(bs); bs->bs_protover = proto; bs->bs_bridge_htime = BSTP_DEFAULT_HELLO_TIME; LIST_FOREACH(bp, &bs->bs_bplist, bp_next) { /* reinit state */ bp->bp_infois = BSTP_INFO_DISABLED; bp->bp_txcount = 0; bstp_set_port_proto(bp, bs->bs_protover); bstp_set_port_role(bp, BSTP_ROLE_DISABLED); bstp_set_port_tc(bp, BSTP_TCSTATE_INACTIVE); bstp_timer_stop(&bp->bp_recent_backup_timer); } bstp_reinit(bs); BSTP_UNLOCK(bs); return (0); } int bstp_set_priority(struct bstp_state *bs, int pri) { if (pri < 0 || pri > BSTP_MAX_PRIORITY) return (EINVAL); /* Limit to steps of 4096 */ pri -= pri % 4096; BSTP_LOCK(bs); bs->bs_bridge_priority = pri; bstp_reinit(bs); BSTP_UNLOCK(bs); return (0); } int bstp_set_port_priority(struct bstp_port *bp, int pri) { struct bstp_state *bs = bp->bp_bs; if (pri < 0 || pri > BSTP_MAX_PORT_PRIORITY) return (EINVAL); /* Limit to steps of 16 */ pri -= pri % 16; BSTP_LOCK(bs); bp->bp_priority = pri; bstp_reinit(bs); BSTP_UNLOCK(bs); return (0); } int bstp_set_path_cost(struct bstp_port *bp, uint32_t path_cost) { struct bstp_state *bs = bp->bp_bs; if (path_cost > BSTP_MAX_PATH_COST) return (EINVAL); /* STP compat mode only uses 16 bits of the 32 */ if (bp->bp_protover == BSTP_PROTO_STP && path_cost > 65535) path_cost = 65535; BSTP_LOCK(bs); if (path_cost == 0) { /* use auto */ bp->bp_flags &= ~BSTP_PORT_ADMCOST; bp->bp_path_cost = bstp_calc_path_cost(bp); } else { bp->bp_path_cost = path_cost; bp->bp_flags |= BSTP_PORT_ADMCOST; } bstp_reinit(bs); BSTP_UNLOCK(bs); return (0); } int bstp_set_edge(struct bstp_port *bp, int set) { struct bstp_state *bs = bp->bp_bs; BSTP_LOCK(bs); if ((bp->bp_operedge = set) == 0) bp->bp_flags &= ~BSTP_PORT_ADMEDGE; else bp->bp_flags |= BSTP_PORT_ADMEDGE; BSTP_UNLOCK(bs); return (0); } int bstp_set_autoedge(struct bstp_port *bp, int set) { struct bstp_state *bs = bp->bp_bs; BSTP_LOCK(bs); if (set) { bp->bp_flags |= BSTP_PORT_AUTOEDGE; /* we may be able to transition straight to edge */ if (bp->bp_edge_delay_timer.active == 0) bstp_edge_delay_expiry(bs, bp); } else bp->bp_flags &= ~BSTP_PORT_AUTOEDGE; BSTP_UNLOCK(bs); return (0); } int bstp_set_ptp(struct bstp_port *bp, int set) { struct bstp_state *bs = bp->bp_bs; BSTP_LOCK(bs); bp->bp_ptp_link = set; BSTP_UNLOCK(bs); return (0); } int bstp_set_autoptp(struct bstp_port *bp, int set) { struct bstp_state *bs = bp->bp_bs; BSTP_LOCK(bs); if (set) { bp->bp_flags |= BSTP_PORT_AUTOPTP; if (bp->bp_role != BSTP_ROLE_DISABLED) bstp_ifupdstatus(bs, bp); } else bp->bp_flags &= ~BSTP_PORT_AUTOPTP; BSTP_UNLOCK(bs); return (0); } /* * Calculate the path cost according to the link speed. */ static uint32_t bstp_calc_path_cost(struct bstp_port *bp) { struct ifnet *ifp = bp->bp_ifp; uint32_t path_cost; /* If the priority has been manually set then retain the value */ if (bp->bp_flags & BSTP_PORT_ADMCOST) return bp->bp_path_cost; if (ifp->if_link_state == LINK_STATE_DOWN) { /* Recalc when the link comes up again */ bp->bp_flags |= BSTP_PORT_PNDCOST; return (BSTP_DEFAULT_PATH_COST); } if (ifp->if_baudrate < 1000) return (BSTP_DEFAULT_PATH_COST); /* formula from section 17.14, IEEE Std 802.1D-2004 */ path_cost = 20000000000ULL / (ifp->if_baudrate / 1000); if (path_cost > BSTP_MAX_PATH_COST) path_cost = BSTP_MAX_PATH_COST; /* STP compat mode only uses 16 bits of the 32 */ if (bp->bp_protover == BSTP_PROTO_STP && path_cost > 65535) path_cost = 65535; return (path_cost); } /* * Notify the bridge that a port state has changed, we need to do this from a * taskqueue to avoid a LOR. */ static void bstp_notify_state(void *arg, int pending) { struct bstp_port *bp = (struct bstp_port *)arg; struct bstp_state *bs = bp->bp_bs; if (bp->bp_active == 1 && bs->bs_state_cb != NULL) (*bs->bs_state_cb)(bp->bp_ifp, bp->bp_state); } /* * Flush the routes on the bridge port, we need to do this from a * taskqueue to avoid a LOR. */ static void bstp_notify_rtage(void *arg, int pending) { struct bstp_port *bp = (struct bstp_port *)arg; struct bstp_state *bs = bp->bp_bs; int age = 0; BSTP_LOCK(bs); switch (bp->bp_protover) { case BSTP_PROTO_STP: /* convert to seconds */ age = bp->bp_desg_fdelay / BSTP_TICK_VAL; break; case BSTP_PROTO_RSTP: age = 0; break; } BSTP_UNLOCK(bs); if (bp->bp_active == 1 && bs->bs_rtage_cb != NULL) (*bs->bs_rtage_cb)(bp->bp_ifp, age); /* flush is complete */ BSTP_LOCK(bs); bp->bp_fdbflush = 0; BSTP_UNLOCK(bs); } void -bstp_linkstate(struct ifnet *ifp, int state) +bstp_linkstate(struct bstp_port *bp) { - struct bstp_state *bs; - struct bstp_port *bp; + struct bstp_state *bs = bp->bp_bs; - /* search for the stp port */ - mtx_lock(&bstp_list_mtx); - LIST_FOREACH(bs, &bstp_list, bs_list) { - BSTP_LOCK(bs); - LIST_FOREACH(bp, &bs->bs_bplist, bp_next) { - if (bp->bp_ifp == ifp) { - bstp_ifupdstatus(bs, bp); - bstp_update_state(bs, bp); - /* it only exists once so return */ - BSTP_UNLOCK(bs); - mtx_unlock(&bstp_list_mtx); - return; - } - } - BSTP_UNLOCK(bs); + BSTP_LOCK(bs); + if (bp->bp_active) { + bstp_ifupdstatus(bs, bp); + bstp_update_state(bs, bp); } - mtx_unlock(&bstp_list_mtx); + BSTP_UNLOCK(bs); } static void bstp_ifupdstatus(struct bstp_state *bs, struct bstp_port *bp) { struct ifnet *ifp = bp->bp_ifp; struct ifmediareq ifmr; int error = 0; BSTP_LOCK_ASSERT(bs); bzero((char *)&ifmr, sizeof(ifmr)); error = (*ifp->if_ioctl)(ifp, SIOCGIFMEDIA, (caddr_t)&ifmr); if ((error == 0) && (ifp->if_flags & IFF_UP)) { if (ifmr.ifm_status & IFM_ACTIVE) { /* A full-duplex link is assumed to be point to point */ if (bp->bp_flags & BSTP_PORT_AUTOPTP) { bp->bp_ptp_link = ifmr.ifm_active & IFM_FDX ? 1 : 0; } /* Calc the cost if the link was down previously */ if (bp->bp_flags & BSTP_PORT_PNDCOST) { bp->bp_path_cost = bstp_calc_path_cost(bp); bp->bp_flags &= ~BSTP_PORT_PNDCOST; } if (bp->bp_role == BSTP_ROLE_DISABLED) bstp_enable_port(bs, bp); } else { if (bp->bp_role != BSTP_ROLE_DISABLED) { bstp_disable_port(bs, bp); if ((bp->bp_flags & BSTP_PORT_ADMEDGE) && bp->bp_protover == BSTP_PROTO_RSTP) bp->bp_operedge = 1; } } return; } if (bp->bp_infois != BSTP_INFO_DISABLED) bstp_disable_port(bs, bp); } static void bstp_enable_port(struct bstp_state *bs, struct bstp_port *bp) { bp->bp_infois = BSTP_INFO_AGED; bstp_assign_roles(bs); } static void bstp_disable_port(struct bstp_state *bs, struct bstp_port *bp) { bp->bp_infois = BSTP_INFO_DISABLED; bstp_assign_roles(bs); } static void bstp_tick(void *arg) { struct bstp_state *bs = arg; struct bstp_port *bp; BSTP_LOCK_ASSERT(bs); if (bs->bs_running == 0) return; CURVNET_SET(bs->bs_vnet); /* poll link events on interfaces that do not support linkstate */ if (bstp_timer_dectest(&bs->bs_link_timer)) { LIST_FOREACH(bp, &bs->bs_bplist, bp_next) { if (!(bp->bp_ifp->if_capabilities & IFCAP_LINKSTATE)) bstp_ifupdstatus(bs, bp); } bstp_timer_start(&bs->bs_link_timer, BSTP_LINK_TIMER); } LIST_FOREACH(bp, &bs->bs_bplist, bp_next) { /* no events need to happen for these */ bstp_timer_dectest(&bp->bp_tc_timer); bstp_timer_dectest(&bp->bp_recent_root_timer); bstp_timer_dectest(&bp->bp_forward_delay_timer); bstp_timer_dectest(&bp->bp_recent_backup_timer); if (bstp_timer_dectest(&bp->bp_hello_timer)) bstp_hello_timer_expiry(bs, bp); if (bstp_timer_dectest(&bp->bp_message_age_timer)) bstp_message_age_expiry(bs, bp); if (bstp_timer_dectest(&bp->bp_migrate_delay_timer)) bstp_migrate_delay_expiry(bs, bp); if (bstp_timer_dectest(&bp->bp_edge_delay_timer)) bstp_edge_delay_expiry(bs, bp); /* update the various state machines for the port */ bstp_update_state(bs, bp); if (bp->bp_txcount > 0) bp->bp_txcount--; } CURVNET_RESTORE(); callout_reset(&bs->bs_bstpcallout, hz, bstp_tick, bs); } static void bstp_timer_start(struct bstp_timer *t, uint16_t v) { t->value = v; t->active = 1; t->latched = 0; } static void bstp_timer_stop(struct bstp_timer *t) { t->value = 0; t->active = 0; t->latched = 0; } static void bstp_timer_latch(struct bstp_timer *t) { t->latched = 1; t->active = 1; } static int bstp_timer_dectest(struct bstp_timer *t) { if (t->active == 0 || t->latched) return (0); t->value -= BSTP_TICK_VAL; if (t->value <= 0) { bstp_timer_stop(t); return (1); } return (0); } static void bstp_hello_timer_expiry(struct bstp_state *bs, struct bstp_port *bp) { if ((bp->bp_flags & BSTP_PORT_NEWINFO) || bp->bp_role == BSTP_ROLE_DESIGNATED || (bp->bp_role == BSTP_ROLE_ROOT && bp->bp_tc_timer.active == 1)) { bstp_timer_start(&bp->bp_hello_timer, bp->bp_desg_htime); bp->bp_flags |= BSTP_PORT_NEWINFO; bstp_transmit(bs, bp); } } static void bstp_message_age_expiry(struct bstp_state *bs, struct bstp_port *bp) { if (bp->bp_infois == BSTP_INFO_RECEIVED) { bp->bp_infois = BSTP_INFO_AGED; bstp_assign_roles(bs); DPRINTF("aged info on %s\n", bp->bp_ifp->if_xname); } } static void bstp_migrate_delay_expiry(struct bstp_state *bs, struct bstp_port *bp) { bp->bp_flags |= BSTP_PORT_CANMIGRATE; } static void bstp_edge_delay_expiry(struct bstp_state *bs, struct bstp_port *bp) { if ((bp->bp_flags & BSTP_PORT_AUTOEDGE) && bp->bp_protover == BSTP_PROTO_RSTP && bp->bp_proposing && bp->bp_role == BSTP_ROLE_DESIGNATED) { bp->bp_operedge = 1; DPRINTF("%s -> edge port\n", bp->bp_ifp->if_xname); } } static int bstp_addr_cmp(const uint8_t *a, const uint8_t *b) { int i, d; for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) { d = ((int)a[i]) - ((int)b[i]); } return (d); } /* * compare the bridge address component of the bridgeid */ static int bstp_same_bridgeid(uint64_t id1, uint64_t id2) { u_char addr1[ETHER_ADDR_LEN]; u_char addr2[ETHER_ADDR_LEN]; PV2ADDR(id1, addr1); PV2ADDR(id2, addr2); if (bstp_addr_cmp(addr1, addr2) == 0) return (1); return (0); } void bstp_reinit(struct bstp_state *bs) { struct bstp_port *bp; struct ifnet *ifp, *mif; u_char *e_addr; void *bridgeptr; static const u_char llzero[ETHER_ADDR_LEN]; /* 00:00:00:00:00:00 */ BSTP_LOCK_ASSERT(bs); if (LIST_EMPTY(&bs->bs_bplist)) goto disablestp; mif = NULL; bridgeptr = LIST_FIRST(&bs->bs_bplist)->bp_ifp->if_bridge; KASSERT(bridgeptr != NULL, ("Invalid bridge pointer")); /* * Search through the Ethernet adapters and find the one with the * lowest value. Make sure the adapter which we take the MAC address * from is part of this bridge, so we can have more than one independent * bridges in the same STP domain. */ IFNET_RLOCK_NOSLEEP(); TAILQ_FOREACH(ifp, &V_ifnet, if_link) { if (ifp->if_type != IFT_ETHER) continue; /* Not Ethernet */ if (ifp->if_bridge != bridgeptr) continue; /* Not part of our bridge */ if (bstp_addr_cmp(IF_LLADDR(ifp), llzero) == 0) continue; /* No mac address set */ if (mif == NULL) { mif = ifp; continue; } if (bstp_addr_cmp(IF_LLADDR(ifp), IF_LLADDR(mif)) < 0) { mif = ifp; continue; } } IFNET_RUNLOCK_NOSLEEP(); if (mif == NULL) goto disablestp; e_addr = IF_LLADDR(mif); bs->bs_bridge_pv.pv_dbridge_id = (((uint64_t)bs->bs_bridge_priority) << 48) | (((uint64_t)e_addr[0]) << 40) | (((uint64_t)e_addr[1]) << 32) | (((uint64_t)e_addr[2]) << 24) | (((uint64_t)e_addr[3]) << 16) | (((uint64_t)e_addr[4]) << 8) | (((uint64_t)e_addr[5])); bs->bs_bridge_pv.pv_root_id = bs->bs_bridge_pv.pv_dbridge_id; bs->bs_bridge_pv.pv_cost = 0; bs->bs_bridge_pv.pv_dport_id = 0; bs->bs_bridge_pv.pv_port_id = 0; if (bs->bs_running && callout_pending(&bs->bs_bstpcallout) == 0) callout_reset(&bs->bs_bstpcallout, hz, bstp_tick, bs); LIST_FOREACH(bp, &bs->bs_bplist, bp_next) { bp->bp_port_id = (bp->bp_priority << 8) | (bp->bp_ifp->if_index & 0xfff); bstp_ifupdstatus(bs, bp); } bstp_assign_roles(bs); bstp_timer_start(&bs->bs_link_timer, BSTP_LINK_TIMER); return; disablestp: /* Set the bridge and root id (lower bits) to zero */ bs->bs_bridge_pv.pv_dbridge_id = ((uint64_t)bs->bs_bridge_priority) << 48; bs->bs_bridge_pv.pv_root_id = bs->bs_bridge_pv.pv_dbridge_id; bs->bs_root_pv = bs->bs_bridge_pv; /* Disable any remaining ports, they will have no MAC address */ LIST_FOREACH(bp, &bs->bs_bplist, bp_next) { bp->bp_infois = BSTP_INFO_DISABLED; bstp_set_port_role(bp, BSTP_ROLE_DISABLED); } callout_stop(&bs->bs_bstpcallout); } static int bstp_modevent(module_t mod, int type, void *data) { switch (type) { case MOD_LOAD: mtx_init(&bstp_list_mtx, "bridgestp list", NULL, MTX_DEF); LIST_INIT(&bstp_list); - bstp_linkstate_p = bstp_linkstate; break; case MOD_UNLOAD: - bstp_linkstate_p = NULL; mtx_destroy(&bstp_list_mtx); break; default: return (EOPNOTSUPP); } return (0); } static moduledata_t bstp_mod = { "bridgestp", bstp_modevent, 0 }; DECLARE_MODULE(bridgestp, bstp_mod, SI_SUB_PSEUDO, SI_ORDER_ANY); MODULE_VERSION(bridgestp, 1); void bstp_attach(struct bstp_state *bs, struct bstp_cb_ops *cb) { BSTP_LOCK_INIT(bs); callout_init_mtx(&bs->bs_bstpcallout, &bs->bs_mtx, 0); LIST_INIT(&bs->bs_bplist); bs->bs_bridge_max_age = BSTP_DEFAULT_MAX_AGE; bs->bs_bridge_htime = BSTP_DEFAULT_HELLO_TIME; bs->bs_bridge_fdelay = BSTP_DEFAULT_FORWARD_DELAY; bs->bs_bridge_priority = BSTP_DEFAULT_BRIDGE_PRIORITY; bs->bs_hold_time = BSTP_DEFAULT_HOLD_TIME; bs->bs_migration_delay = BSTP_DEFAULT_MIGRATE_DELAY; bs->bs_txholdcount = BSTP_DEFAULT_HOLD_COUNT; bs->bs_protover = BSTP_PROTO_RSTP; bs->bs_state_cb = cb->bcb_state; bs->bs_rtage_cb = cb->bcb_rtage; bs->bs_vnet = curvnet; getmicrotime(&bs->bs_last_tc_time); mtx_lock(&bstp_list_mtx); LIST_INSERT_HEAD(&bstp_list, bs, bs_list); mtx_unlock(&bstp_list_mtx); } void bstp_detach(struct bstp_state *bs) { KASSERT(LIST_EMPTY(&bs->bs_bplist), ("bstp still active")); mtx_lock(&bstp_list_mtx); LIST_REMOVE(bs, bs_list); mtx_unlock(&bstp_list_mtx); callout_drain(&bs->bs_bstpcallout); BSTP_LOCK_DESTROY(bs); } void bstp_init(struct bstp_state *bs) { BSTP_LOCK(bs); callout_reset(&bs->bs_bstpcallout, hz, bstp_tick, bs); bs->bs_running = 1; bstp_reinit(bs); BSTP_UNLOCK(bs); } void bstp_stop(struct bstp_state *bs) { struct bstp_port *bp; BSTP_LOCK(bs); LIST_FOREACH(bp, &bs->bs_bplist, bp_next) bstp_set_port_state(bp, BSTP_IFSTATE_DISCARDING); bs->bs_running = 0; callout_stop(&bs->bs_bstpcallout); BSTP_UNLOCK(bs); } int bstp_create(struct bstp_state *bs, struct bstp_port *bp, struct ifnet *ifp) { bzero(bp, sizeof(struct bstp_port)); BSTP_LOCK(bs); bp->bp_ifp = ifp; bp->bp_bs = bs; bp->bp_priority = BSTP_DEFAULT_PORT_PRIORITY; TASK_INIT(&bp->bp_statetask, 0, bstp_notify_state, bp); TASK_INIT(&bp->bp_rtagetask, 0, bstp_notify_rtage, bp); /* Init state */ bp->bp_infois = BSTP_INFO_DISABLED; bp->bp_flags = BSTP_PORT_AUTOEDGE|BSTP_PORT_AUTOPTP; bstp_set_port_state(bp, BSTP_IFSTATE_DISCARDING); bstp_set_port_proto(bp, bs->bs_protover); bstp_set_port_role(bp, BSTP_ROLE_DISABLED); bstp_set_port_tc(bp, BSTP_TCSTATE_INACTIVE); bp->bp_path_cost = bstp_calc_path_cost(bp); BSTP_UNLOCK(bs); return (0); } int bstp_enable(struct bstp_port *bp) { struct bstp_state *bs = bp->bp_bs; struct ifnet *ifp = bp->bp_ifp; KASSERT(bp->bp_active == 0, ("already a bstp member")); switch (ifp->if_type) { case IFT_ETHER: /* These can do spanning tree. */ break; default: /* Nothing else can. */ return (EINVAL); } BSTP_LOCK(bs); LIST_INSERT_HEAD(&bs->bs_bplist, bp, bp_next); bp->bp_active = 1; bp->bp_flags |= BSTP_PORT_NEWINFO; bstp_reinit(bs); bstp_update_roles(bs, bp); BSTP_UNLOCK(bs); return (0); } void bstp_disable(struct bstp_port *bp) { struct bstp_state *bs = bp->bp_bs; KASSERT(bp->bp_active == 1, ("not a bstp member")); BSTP_LOCK(bs); bstp_disable_port(bs, bp); LIST_REMOVE(bp, bp_next); bp->bp_active = 0; bstp_reinit(bs); BSTP_UNLOCK(bs); } /* * The bstp_port structure is about to be freed by the parent bridge. */ void bstp_destroy(struct bstp_port *bp) { KASSERT(bp->bp_active == 0, ("port is still attached")); taskqueue_drain(taskqueue_swi, &bp->bp_statetask); taskqueue_drain(taskqueue_swi, &bp->bp_rtagetask); } Index: head/sys/net/bridgestp.h =================================================================== --- head/sys/net/bridgestp.h (revision 234486) +++ head/sys/net/bridgestp.h (revision 234487) @@ -1,397 +1,395 @@ /* $NetBSD: if_bridgevar.h,v 1.4 2003/07/08 07:13:50 itojun Exp $ */ /* * Copyright 2001 Wasabi Systems, Inc. * All rights reserved. * * Written by Jason R. Thorpe for Wasabi Systems, Inc. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed for the NetBSD Project by * Wasabi Systems, Inc. * 4. The name of Wasabi Systems, Inc. may not be used to endorse * or promote products derived from this software without specific prior * written permission. * * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. */ /* * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net) * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by Jason L. Wright * 4. The name of the author may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE * DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * OpenBSD: if_bridge.h,v 1.14 2001/03/22 03:48:29 jason Exp * * $FreeBSD$ */ /* * Data structure and control definitions for STP interfaces. */ #include #include /* STP port states */ #define BSTP_IFSTATE_DISABLED 0 #define BSTP_IFSTATE_LISTENING 1 #define BSTP_IFSTATE_LEARNING 2 #define BSTP_IFSTATE_FORWARDING 3 #define BSTP_IFSTATE_BLOCKING 4 #define BSTP_IFSTATE_DISCARDING 5 #define BSTP_TCSTATE_ACTIVE 1 #define BSTP_TCSTATE_DETECTED 2 #define BSTP_TCSTATE_INACTIVE 3 #define BSTP_TCSTATE_LEARNING 4 #define BSTP_TCSTATE_PROPAG 5 #define BSTP_TCSTATE_ACK 6 #define BSTP_TCSTATE_TC 7 #define BSTP_TCSTATE_TCN 8 #define BSTP_ROLE_DISABLED 0 #define BSTP_ROLE_ROOT 1 #define BSTP_ROLE_DESIGNATED 2 #define BSTP_ROLE_ALTERNATE 3 #define BSTP_ROLE_BACKUP 4 #ifdef _KERNEL /* STP port flags */ #define BSTP_PORT_CANMIGRATE 0x0001 #define BSTP_PORT_NEWINFO 0x0002 #define BSTP_PORT_DISPUTED 0x0004 #define BSTP_PORT_ADMCOST 0x0008 #define BSTP_PORT_AUTOEDGE 0x0010 #define BSTP_PORT_AUTOPTP 0x0020 #define BSTP_PORT_ADMEDGE 0x0040 #define BSTP_PORT_PNDCOST 0x0080 /* BPDU priority */ #define BSTP_PDU_SUPERIOR 1 #define BSTP_PDU_REPEATED 2 #define BSTP_PDU_INFERIOR 3 #define BSTP_PDU_INFERIORALT 4 #define BSTP_PDU_OTHER 5 /* BPDU flags */ #define BSTP_PDU_PRMASK 0x0c /* Port Role */ #define BSTP_PDU_PRSHIFT 2 /* Port Role offset */ #define BSTP_PDU_F_UNKN 0x00 /* Unknown port (00) */ #define BSTP_PDU_F_ALT 0x01 /* Alt/Backup port (01) */ #define BSTP_PDU_F_ROOT 0x02 /* Root port (10) */ #define BSTP_PDU_F_DESG 0x03 /* Designated port (11) */ #define BSTP_PDU_STPMASK 0x81 /* strip unused STP flags */ #define BSTP_PDU_RSTPMASK 0x7f /* strip unused RSTP flags */ #define BSTP_PDU_F_TC 0x01 /* Topology change */ #define BSTP_PDU_F_P 0x02 /* Proposal flag */ #define BSTP_PDU_F_L 0x10 /* Learning flag */ #define BSTP_PDU_F_F 0x20 /* Forwarding flag */ #define BSTP_PDU_F_A 0x40 /* Agreement flag */ #define BSTP_PDU_F_TCA 0x80 /* Topology change ack */ /* * Spanning tree defaults. */ #define BSTP_DEFAULT_MAX_AGE (20 * 256) #define BSTP_DEFAULT_HELLO_TIME (2 * 256) #define BSTP_DEFAULT_FORWARD_DELAY (15 * 256) #define BSTP_DEFAULT_HOLD_TIME (1 * 256) #define BSTP_DEFAULT_MIGRATE_DELAY (3 * 256) #define BSTP_DEFAULT_HOLD_COUNT 6 #define BSTP_DEFAULT_BRIDGE_PRIORITY 0x8000 #define BSTP_DEFAULT_PORT_PRIORITY 0x80 #define BSTP_DEFAULT_PATH_COST 55 #define BSTP_MIN_HELLO_TIME (1 * 256) #define BSTP_MIN_MAX_AGE (6 * 256) #define BSTP_MIN_FORWARD_DELAY (4 * 256) #define BSTP_MIN_HOLD_COUNT 1 #define BSTP_MAX_HELLO_TIME (2 * 256) #define BSTP_MAX_MAX_AGE (40 * 256) #define BSTP_MAX_FORWARD_DELAY (30 * 256) #define BSTP_MAX_HOLD_COUNT 10 #define BSTP_MAX_PRIORITY 61440 #define BSTP_MAX_PORT_PRIORITY 240 #define BSTP_MAX_PATH_COST 200000000 /* BPDU message types */ #define BSTP_MSGTYPE_CFG 0x00 /* Configuration */ #define BSTP_MSGTYPE_RSTP 0x02 /* Rapid STP */ #define BSTP_MSGTYPE_TCN 0x80 /* Topology chg notification */ /* Protocol versions */ #define BSTP_PROTO_ID 0x00 #define BSTP_PROTO_STP 0x00 #define BSTP_PROTO_RSTP 0x02 #define BSTP_PROTO_MAX BSTP_PROTO_RSTP #define BSTP_INFO_RECEIVED 1 #define BSTP_INFO_MINE 2 #define BSTP_INFO_AGED 3 #define BSTP_INFO_DISABLED 4 #define BSTP_MESSAGE_AGE_INCR (1 * 256) /* in 256ths of a second */ #define BSTP_TICK_VAL (1 * 256) /* in 256ths of a second */ #define BSTP_LINK_TIMER (BSTP_TICK_VAL * 15) /* * Driver callbacks for STP state changes */ typedef void (*bstp_state_cb_t)(struct ifnet *, int); typedef void (*bstp_rtage_cb_t)(struct ifnet *, int); struct bstp_cb_ops { bstp_state_cb_t bcb_state; bstp_rtage_cb_t bcb_rtage; }; /* * Because BPDU's do not make nicely aligned structures, two different * declarations are used: bstp_?bpdu (wire representation, packed) and * bstp_*_unit (internal, nicely aligned version). */ /* configuration bridge protocol data unit */ struct bstp_cbpdu { uint8_t cbu_dsap; /* LLC: destination sap */ uint8_t cbu_ssap; /* LLC: source sap */ uint8_t cbu_ctl; /* LLC: control */ uint16_t cbu_protoid; /* protocol id */ uint8_t cbu_protover; /* protocol version */ uint8_t cbu_bpdutype; /* message type */ uint8_t cbu_flags; /* flags (below) */ /* root id */ uint16_t cbu_rootpri; /* root priority */ uint8_t cbu_rootaddr[6]; /* root address */ uint32_t cbu_rootpathcost; /* root path cost */ /* bridge id */ uint16_t cbu_bridgepri; /* bridge priority */ uint8_t cbu_bridgeaddr[6]; /* bridge address */ uint16_t cbu_portid; /* port id */ uint16_t cbu_messageage; /* current message age */ uint16_t cbu_maxage; /* maximum age */ uint16_t cbu_hellotime; /* hello time */ uint16_t cbu_forwarddelay; /* forwarding delay */ uint8_t cbu_versionlen; /* version 1 length */ } __packed; #define BSTP_BPDU_STP_LEN (3 + 35) /* LLC + STP pdu */ #define BSTP_BPDU_RSTP_LEN (3 + 36) /* LLC + RSTP pdu */ /* topology change notification bridge protocol data unit */ struct bstp_tbpdu { uint8_t tbu_dsap; /* LLC: destination sap */ uint8_t tbu_ssap; /* LLC: source sap */ uint8_t tbu_ctl; /* LLC: control */ uint16_t tbu_protoid; /* protocol id */ uint8_t tbu_protover; /* protocol version */ uint8_t tbu_bpdutype; /* message type */ } __packed; /* * Timekeeping structure used in spanning tree code. */ struct bstp_timer { int active; int latched; int value; }; struct bstp_pri_vector { uint64_t pv_root_id; uint32_t pv_cost; uint64_t pv_dbridge_id; uint16_t pv_dport_id; uint16_t pv_port_id; }; struct bstp_config_unit { struct bstp_pri_vector cu_pv; uint16_t cu_message_age; uint16_t cu_max_age; uint16_t cu_forward_delay; uint16_t cu_hello_time; uint8_t cu_message_type; uint8_t cu_topology_change_ack; uint8_t cu_topology_change; uint8_t cu_proposal; uint8_t cu_agree; uint8_t cu_learning; uint8_t cu_forwarding; uint8_t cu_role; }; struct bstp_tcn_unit { uint8_t tu_message_type; }; struct bstp_port { LIST_ENTRY(bstp_port) bp_next; struct ifnet *bp_ifp; /* parent if */ struct bstp_state *bp_bs; uint8_t bp_active; uint8_t bp_protover; uint32_t bp_flags; uint32_t bp_path_cost; uint16_t bp_port_msg_age; uint16_t bp_port_max_age; uint16_t bp_port_fdelay; uint16_t bp_port_htime; uint16_t bp_desg_msg_age; uint16_t bp_desg_max_age; uint16_t bp_desg_fdelay; uint16_t bp_desg_htime; struct bstp_timer bp_edge_delay_timer; struct bstp_timer bp_forward_delay_timer; struct bstp_timer bp_hello_timer; struct bstp_timer bp_message_age_timer; struct bstp_timer bp_migrate_delay_timer; struct bstp_timer bp_recent_backup_timer; struct bstp_timer bp_recent_root_timer; struct bstp_timer bp_tc_timer; struct bstp_config_unit bp_msg_cu; struct bstp_pri_vector bp_desg_pv; struct bstp_pri_vector bp_port_pv; uint16_t bp_port_id; uint8_t bp_state; uint8_t bp_tcstate; uint8_t bp_role; uint8_t bp_infois; uint8_t bp_tc_ack; uint8_t bp_tc_prop; uint8_t bp_fdbflush; uint8_t bp_priority; uint8_t bp_ptp_link; uint8_t bp_agree; uint8_t bp_agreed; uint8_t bp_sync; uint8_t bp_synced; uint8_t bp_proposing; uint8_t bp_proposed; uint8_t bp_operedge; uint8_t bp_reroot; uint8_t bp_rcvdtc; uint8_t bp_rcvdtca; uint8_t bp_rcvdtcn; uint32_t bp_forward_transitions; uint8_t bp_txcount; struct task bp_statetask; struct task bp_rtagetask; }; /* * Software state for each bridge STP. */ struct bstp_state { LIST_ENTRY(bstp_state) bs_list; uint8_t bs_running; struct mtx bs_mtx; struct bstp_pri_vector bs_bridge_pv; struct bstp_pri_vector bs_root_pv; struct bstp_port *bs_root_port; uint8_t bs_protover; uint16_t bs_migration_delay; uint16_t bs_edge_delay; uint16_t bs_bridge_max_age; uint16_t bs_bridge_fdelay; uint16_t bs_bridge_htime; uint16_t bs_root_msg_age; uint16_t bs_root_max_age; uint16_t bs_root_fdelay; uint16_t bs_root_htime; uint16_t bs_hold_time; uint16_t bs_bridge_priority; uint8_t bs_txholdcount; uint8_t bs_allsynced; struct callout bs_bstpcallout; /* STP callout */ struct bstp_timer bs_link_timer; struct timeval bs_last_tc_time; LIST_HEAD(, bstp_port) bs_bplist; bstp_state_cb_t bs_state_cb; bstp_rtage_cb_t bs_rtage_cb; struct vnet *bs_vnet; }; #define BSTP_LOCK_INIT(_bs) mtx_init(&(_bs)->bs_mtx, "bstp", NULL, MTX_DEF) #define BSTP_LOCK_DESTROY(_bs) mtx_destroy(&(_bs)->bs_mtx) #define BSTP_LOCK(_bs) mtx_lock(&(_bs)->bs_mtx) #define BSTP_UNLOCK(_bs) mtx_unlock(&(_bs)->bs_mtx) #define BSTP_LOCK_ASSERT(_bs) mtx_assert(&(_bs)->bs_mtx, MA_OWNED) extern const uint8_t bstp_etheraddr[]; -extern void (*bstp_linkstate_p)(struct ifnet *ifp, int state); - void bstp_attach(struct bstp_state *, struct bstp_cb_ops *); void bstp_detach(struct bstp_state *); void bstp_init(struct bstp_state *); void bstp_stop(struct bstp_state *); int bstp_create(struct bstp_state *, struct bstp_port *, struct ifnet *); int bstp_enable(struct bstp_port *); void bstp_disable(struct bstp_port *); void bstp_destroy(struct bstp_port *); -void bstp_linkstate(struct ifnet *, int); +void bstp_linkstate(struct bstp_port *); int bstp_set_htime(struct bstp_state *, int); int bstp_set_fdelay(struct bstp_state *, int); int bstp_set_maxage(struct bstp_state *, int); int bstp_set_holdcount(struct bstp_state *, int); int bstp_set_protocol(struct bstp_state *, int); int bstp_set_priority(struct bstp_state *, int); int bstp_set_port_priority(struct bstp_port *, int); int bstp_set_path_cost(struct bstp_port *, uint32_t); int bstp_set_edge(struct bstp_port *, int); int bstp_set_autoedge(struct bstp_port *, int); int bstp_set_ptp(struct bstp_port *, int); int bstp_set_autoptp(struct bstp_port *, int); void bstp_input(struct bstp_port *, struct ifnet *, struct mbuf *); #endif /* _KERNEL */ Index: head/sys/net/if.c =================================================================== --- head/sys/net/if.c (revision 234486) +++ head/sys/net/if.c (revision 234487) @@ -1,3413 +1,3413 @@ /*- * Copyright (c) 1980, 1986, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)if.c 8.5 (Berkeley) 1/9/95 * $FreeBSD$ */ #include "opt_compat.h" #include "opt_inet6.h" #include "opt_inet.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #if defined(INET) || defined(INET6) /*XXX*/ #include #include #include #ifdef INET6 #include #include #endif #endif #ifdef INET #include #endif #include #ifdef COMPAT_FREEBSD32 #include #include #endif struct ifindex_entry { struct ifnet *ife_ifnet; }; SYSCTL_NODE(_net, PF_LINK, link, CTLFLAG_RW, 0, "Link layers"); SYSCTL_NODE(_net_link, 0, generic, CTLFLAG_RW, 0, "Generic link-management"); TUNABLE_INT("net.link.ifqmaxlen", &ifqmaxlen); SYSCTL_INT(_net_link, OID_AUTO, ifqmaxlen, CTLFLAG_RDTUN, &ifqmaxlen, 0, "max send queue size"); /* Log link state change events */ static int log_link_state_change = 1; SYSCTL_INT(_net_link, OID_AUTO, log_link_state_change, CTLFLAG_RW, &log_link_state_change, 0, "log interface link state change events"); /* Interface description */ static unsigned int ifdescr_maxlen = 1024; SYSCTL_UINT(_net, OID_AUTO, ifdescr_maxlen, CTLFLAG_RW, &ifdescr_maxlen, 0, "administrative maximum length for interface description"); static MALLOC_DEFINE(M_IFDESCR, "ifdescr", "ifnet descriptions"); /* global sx for non-critical path ifdescr */ static struct sx ifdescr_sx; SX_SYSINIT(ifdescr_sx, &ifdescr_sx, "ifnet descr"); -void (*bstp_linkstate_p)(struct ifnet *ifp, int state); +void (*bridge_linkstate_p)(struct ifnet *ifp); void (*ng_ether_link_state_p)(struct ifnet *ifp, int state); void (*lagg_linkstate_p)(struct ifnet *ifp, int state); /* These are external hooks for CARP. */ void (*carp_linkstate_p)(struct ifnet *ifp); void (*carp_demote_adj_p)(int, char *); #if defined(INET) || defined(INET6) int (*carp_forus_p)(struct ifnet *ifp, u_char *dhost); int (*carp_output_p)(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa); int (*carp_ioctl_p)(struct ifreq *, u_long, struct thread *); int (*carp_attach_p)(struct ifaddr *, int); void (*carp_detach_p)(struct ifaddr *); #endif #ifdef INET int (*carp_iamatch_p)(struct ifaddr *, uint8_t **); #endif #ifdef INET6 struct ifaddr *(*carp_iamatch6_p)(struct ifnet *ifp, struct in6_addr *taddr6); caddr_t (*carp_macmatch6_p)(struct ifnet *ifp, struct mbuf *m, const struct in6_addr *taddr); #endif struct mbuf *(*tbr_dequeue_ptr)(struct ifaltq *, int) = NULL; /* * XXX: Style; these should be sorted alphabetically, and unprototyped * static functions should be prototyped. Currently they are sorted by * declaration order. */ static void if_attachdomain(void *); static void if_attachdomain1(struct ifnet *); static int ifconf(u_long, caddr_t); static void if_freemulti(struct ifmultiaddr *); static void if_init(void *); static void if_grow(void); static void if_route(struct ifnet *, int flag, int fam); static int if_setflag(struct ifnet *, int, int, int *, int); static int if_transmit(struct ifnet *ifp, struct mbuf *m); static void if_unroute(struct ifnet *, int flag, int fam); static void link_rtrequest(int, struct rtentry *, struct rt_addrinfo *); static int if_rtdel(struct radix_node *, void *); static int ifhwioctl(u_long, struct ifnet *, caddr_t, struct thread *); static int if_delmulti_locked(struct ifnet *, struct ifmultiaddr *, int); static void do_link_state_change(void *, int); static int if_getgroup(struct ifgroupreq *, struct ifnet *); static int if_getgroupmembers(struct ifgroupreq *); static void if_delgroups(struct ifnet *); static void if_attach_internal(struct ifnet *, int); static void if_detach_internal(struct ifnet *, int); #ifdef INET6 /* * XXX: declare here to avoid to include many inet6 related files.. * should be more generalized? */ extern void nd6_setmtu(struct ifnet *); #endif VNET_DEFINE(int, if_index); int ifqmaxlen = IFQ_MAXLEN; VNET_DEFINE(struct ifnethead, ifnet); /* depend on static init XXX */ VNET_DEFINE(struct ifgrouphead, ifg_head); static VNET_DEFINE(int, if_indexlim) = 8; /* Table of ifnet by index. */ VNET_DEFINE(struct ifindex_entry *, ifindex_table); #define V_if_indexlim VNET(if_indexlim) #define V_ifindex_table VNET(ifindex_table) /* * The global network interface list (V_ifnet) and related state (such as * if_index, if_indexlim, and ifindex_table) are protected by an sxlock and * an rwlock. Either may be acquired shared to stablize the list, but both * must be acquired writable to modify the list. This model allows us to * both stablize the interface list during interrupt thread processing, but * also to stablize it over long-running ioctls, without introducing priority * inversions and deadlocks. */ struct rwlock ifnet_rwlock; struct sx ifnet_sxlock; /* * The allocation of network interfaces is a rather non-atomic affair; we * need to select an index before we are ready to expose the interface for * use, so will use this pointer value to indicate reservation. */ #define IFNET_HOLD (void *)(uintptr_t)(-1) static if_com_alloc_t *if_com_alloc[256]; static if_com_free_t *if_com_free[256]; static MALLOC_DEFINE(M_IFNET, "ifnet", "interface internals"); MALLOC_DEFINE(M_IFADDR, "ifaddr", "interface address"); MALLOC_DEFINE(M_IFMADDR, "ether_multi", "link-level multicast address"); struct ifnet * ifnet_byindex_locked(u_short idx) { if (idx > V_if_index) return (NULL); if (V_ifindex_table[idx].ife_ifnet == IFNET_HOLD) return (NULL); return (V_ifindex_table[idx].ife_ifnet); } struct ifnet * ifnet_byindex(u_short idx) { struct ifnet *ifp; IFNET_RLOCK_NOSLEEP(); ifp = ifnet_byindex_locked(idx); IFNET_RUNLOCK_NOSLEEP(); return (ifp); } struct ifnet * ifnet_byindex_ref(u_short idx) { struct ifnet *ifp; IFNET_RLOCK_NOSLEEP(); ifp = ifnet_byindex_locked(idx); if (ifp == NULL || (ifp->if_flags & IFF_DYING)) { IFNET_RUNLOCK_NOSLEEP(); return (NULL); } if_ref(ifp); IFNET_RUNLOCK_NOSLEEP(); return (ifp); } /* * Allocate an ifindex array entry; return 0 on success or an error on * failure. */ static int ifindex_alloc_locked(u_short *idxp) { u_short idx; IFNET_WLOCK_ASSERT(); retry: /* * Try to find an empty slot below V_if_index. If we fail, take the * next slot. */ for (idx = 1; idx <= V_if_index; idx++) { if (V_ifindex_table[idx].ife_ifnet == NULL) break; } /* Catch if_index overflow. */ if (idx < 1) return (ENOSPC); if (idx >= V_if_indexlim) { if_grow(); goto retry; } if (idx > V_if_index) V_if_index = idx; *idxp = idx; return (0); } static void ifindex_free_locked(u_short idx) { IFNET_WLOCK_ASSERT(); V_ifindex_table[idx].ife_ifnet = NULL; while (V_if_index > 0 && V_ifindex_table[V_if_index].ife_ifnet == NULL) V_if_index--; } static void ifindex_free(u_short idx) { IFNET_WLOCK(); ifindex_free_locked(idx); IFNET_WUNLOCK(); } static void ifnet_setbyindex_locked(u_short idx, struct ifnet *ifp) { IFNET_WLOCK_ASSERT(); V_ifindex_table[idx].ife_ifnet = ifp; } static void ifnet_setbyindex(u_short idx, struct ifnet *ifp) { IFNET_WLOCK(); ifnet_setbyindex_locked(idx, ifp); IFNET_WUNLOCK(); } struct ifaddr * ifaddr_byindex(u_short idx) { struct ifaddr *ifa; IFNET_RLOCK_NOSLEEP(); ifa = ifnet_byindex_locked(idx)->if_addr; if (ifa != NULL) ifa_ref(ifa); IFNET_RUNLOCK_NOSLEEP(); return (ifa); } /* * Network interface utility routines. * * Routines with ifa_ifwith* names take sockaddr *'s as * parameters. */ static void vnet_if_init(const void *unused __unused) { TAILQ_INIT(&V_ifnet); TAILQ_INIT(&V_ifg_head); IFNET_WLOCK(); if_grow(); /* create initial table */ IFNET_WUNLOCK(); vnet_if_clone_init(); } VNET_SYSINIT(vnet_if_init, SI_SUB_INIT_IF, SI_ORDER_SECOND, vnet_if_init, NULL); /* ARGSUSED*/ static void if_init(void *dummy __unused) { IFNET_LOCK_INIT(); if_clone_init(); } SYSINIT(interfaces, SI_SUB_INIT_IF, SI_ORDER_FIRST, if_init, NULL); #ifdef VIMAGE static void vnet_if_uninit(const void *unused __unused) { VNET_ASSERT(TAILQ_EMPTY(&V_ifnet), ("%s:%d tailq &V_ifnet=%p " "not empty", __func__, __LINE__, &V_ifnet)); VNET_ASSERT(TAILQ_EMPTY(&V_ifg_head), ("%s:%d tailq &V_ifg_head=%p " "not empty", __func__, __LINE__, &V_ifg_head)); free((caddr_t)V_ifindex_table, M_IFNET); } VNET_SYSUNINIT(vnet_if_uninit, SI_SUB_INIT_IF, SI_ORDER_FIRST, vnet_if_uninit, NULL); #endif static void if_grow(void) { int oldlim; u_int n; struct ifindex_entry *e; IFNET_WLOCK_ASSERT(); oldlim = V_if_indexlim; IFNET_WUNLOCK(); n = (oldlim << 1) * sizeof(*e); e = malloc(n, M_IFNET, M_WAITOK | M_ZERO); IFNET_WLOCK(); if (V_if_indexlim != oldlim) { free(e, M_IFNET); return; } if (V_ifindex_table != NULL) { memcpy((caddr_t)e, (caddr_t)V_ifindex_table, n/2); free((caddr_t)V_ifindex_table, M_IFNET); } V_if_indexlim <<= 1; V_ifindex_table = e; } /* * Allocate a struct ifnet and an index for an interface. A layer 2 * common structure will also be allocated if an allocation routine is * registered for the passed type. */ struct ifnet * if_alloc(u_char type) { struct ifnet *ifp; u_short idx; ifp = malloc(sizeof(struct ifnet), M_IFNET, M_WAITOK|M_ZERO); IFNET_WLOCK(); if (ifindex_alloc_locked(&idx) != 0) { IFNET_WUNLOCK(); free(ifp, M_IFNET); return (NULL); } ifnet_setbyindex_locked(idx, IFNET_HOLD); IFNET_WUNLOCK(); ifp->if_index = idx; ifp->if_type = type; ifp->if_alloctype = type; if (if_com_alloc[type] != NULL) { ifp->if_l2com = if_com_alloc[type](type, ifp); if (ifp->if_l2com == NULL) { free(ifp, M_IFNET); ifindex_free(idx); return (NULL); } } IF_ADDR_LOCK_INIT(ifp); TASK_INIT(&ifp->if_linktask, 0, do_link_state_change, ifp); ifp->if_afdata_initialized = 0; IF_AFDATA_LOCK_INIT(ifp); TAILQ_INIT(&ifp->if_addrhead); TAILQ_INIT(&ifp->if_multiaddrs); TAILQ_INIT(&ifp->if_groups); #ifdef MAC mac_ifnet_init(ifp); #endif ifq_init(&ifp->if_snd, ifp); refcount_init(&ifp->if_refcount, 1); /* Index reference. */ ifnet_setbyindex(ifp->if_index, ifp); return (ifp); } /* * Do the actual work of freeing a struct ifnet, and layer 2 common * structure. This call is made when the last reference to an * interface is released. */ static void if_free_internal(struct ifnet *ifp) { KASSERT((ifp->if_flags & IFF_DYING), ("if_free_internal: interface not dying")); if (if_com_free[ifp->if_alloctype] != NULL) if_com_free[ifp->if_alloctype](ifp->if_l2com, ifp->if_alloctype); #ifdef MAC mac_ifnet_destroy(ifp); #endif /* MAC */ if (ifp->if_description != NULL) free(ifp->if_description, M_IFDESCR); IF_AFDATA_DESTROY(ifp); IF_ADDR_LOCK_DESTROY(ifp); ifq_delete(&ifp->if_snd); free(ifp, M_IFNET); } /* * Deregister an interface and free the associated storage. */ void if_free(struct ifnet *ifp) { ifp->if_flags |= IFF_DYING; /* XXX: Locking */ IFNET_WLOCK(); KASSERT(ifp == ifnet_byindex_locked(ifp->if_index), ("%s: freeing unallocated ifnet", ifp->if_xname)); ifindex_free_locked(ifp->if_index); IFNET_WUNLOCK(); if (!refcount_release(&ifp->if_refcount)) return; if_free_internal(ifp); } /* * Interfaces to keep an ifnet type-stable despite the possibility of the * driver calling if_free(). If there are additional references, we defer * freeing the underlying data structure. */ void if_ref(struct ifnet *ifp) { /* We don't assert the ifnet list lock here, but arguably should. */ refcount_acquire(&ifp->if_refcount); } void if_rele(struct ifnet *ifp) { if (!refcount_release(&ifp->if_refcount)) return; if_free_internal(ifp); } void ifq_init(struct ifaltq *ifq, struct ifnet *ifp) { mtx_init(&ifq->ifq_mtx, ifp->if_xname, "if send queue", MTX_DEF); if (ifq->ifq_maxlen == 0) ifq->ifq_maxlen = ifqmaxlen; ifq->altq_type = 0; ifq->altq_disc = NULL; ifq->altq_flags &= ALTQF_CANTCHANGE; ifq->altq_tbr = NULL; ifq->altq_ifp = ifp; } void ifq_delete(struct ifaltq *ifq) { mtx_destroy(&ifq->ifq_mtx); } /* * Perform generic interface initalization tasks and attach the interface * to the list of "active" interfaces. If vmove flag is set on entry * to if_attach_internal(), perform only a limited subset of initialization * tasks, given that we are moving from one vnet to another an ifnet which * has already been fully initialized. * * XXX: * - The decision to return void and thus require this function to * succeed is questionable. * - We should probably do more sanity checking. For instance we don't * do anything to insure if_xname is unique or non-empty. */ void if_attach(struct ifnet *ifp) { if_attach_internal(ifp, 0); } static void if_attach_internal(struct ifnet *ifp, int vmove) { unsigned socksize, ifasize; int namelen, masklen; struct sockaddr_dl *sdl; struct ifaddr *ifa; if (ifp->if_index == 0 || ifp != ifnet_byindex(ifp->if_index)) panic ("%s: BUG: if_attach called without if_alloc'd input()\n", ifp->if_xname); #ifdef VIMAGE ifp->if_vnet = curvnet; if (ifp->if_home_vnet == NULL) ifp->if_home_vnet = curvnet; #endif if_addgroup(ifp, IFG_ALL); getmicrotime(&ifp->if_lastchange); ifp->if_data.ifi_epoch = time_uptime; ifp->if_data.ifi_datalen = sizeof(struct if_data); KASSERT((ifp->if_transmit == NULL && ifp->if_qflush == NULL) || (ifp->if_transmit != NULL && ifp->if_qflush != NULL), ("transmit and qflush must both either be set or both be NULL")); if (ifp->if_transmit == NULL) { ifp->if_transmit = if_transmit; ifp->if_qflush = if_qflush; } if (!vmove) { #ifdef MAC mac_ifnet_create(ifp); #endif /* * Create a Link Level name for this device. */ namelen = strlen(ifp->if_xname); /* * Always save enough space for any possiable name so we * can do a rename in place later. */ masklen = offsetof(struct sockaddr_dl, sdl_data[0]) + IFNAMSIZ; socksize = masklen + ifp->if_addrlen; if (socksize < sizeof(*sdl)) socksize = sizeof(*sdl); socksize = roundup2(socksize, sizeof(long)); ifasize = sizeof(*ifa) + 2 * socksize; ifa = malloc(ifasize, M_IFADDR, M_WAITOK | M_ZERO); ifa_init(ifa); sdl = (struct sockaddr_dl *)(ifa + 1); sdl->sdl_len = socksize; sdl->sdl_family = AF_LINK; bcopy(ifp->if_xname, sdl->sdl_data, namelen); sdl->sdl_nlen = namelen; sdl->sdl_index = ifp->if_index; sdl->sdl_type = ifp->if_type; ifp->if_addr = ifa; ifa->ifa_ifp = ifp; ifa->ifa_rtrequest = link_rtrequest; ifa->ifa_addr = (struct sockaddr *)sdl; sdl = (struct sockaddr_dl *)(socksize + (caddr_t)sdl); ifa->ifa_netmask = (struct sockaddr *)sdl; sdl->sdl_len = masklen; while (namelen != 0) sdl->sdl_data[--namelen] = 0xff; TAILQ_INSERT_HEAD(&ifp->if_addrhead, ifa, ifa_link); /* Reliably crash if used uninitialized. */ ifp->if_broadcastaddr = NULL; } #ifdef VIMAGE else { /* * Update the interface index in the link layer address * of the interface. */ for (ifa = ifp->if_addr; ifa != NULL; ifa = TAILQ_NEXT(ifa, ifa_link)) { if (ifa->ifa_addr->sa_family == AF_LINK) { sdl = (struct sockaddr_dl *)ifa->ifa_addr; sdl->sdl_index = ifp->if_index; } } } #endif IFNET_WLOCK(); TAILQ_INSERT_TAIL(&V_ifnet, ifp, if_link); #ifdef VIMAGE curvnet->vnet_ifcnt++; #endif IFNET_WUNLOCK(); if (domain_init_status >= 2) if_attachdomain1(ifp); EVENTHANDLER_INVOKE(ifnet_arrival_event, ifp); if (IS_DEFAULT_VNET(curvnet)) devctl_notify("IFNET", ifp->if_xname, "ATTACH", NULL); /* Announce the interface. */ rt_ifannouncemsg(ifp, IFAN_ARRIVAL); } static void if_attachdomain(void *dummy) { struct ifnet *ifp; int s; s = splnet(); TAILQ_FOREACH(ifp, &V_ifnet, if_link) if_attachdomain1(ifp); splx(s); } SYSINIT(domainifattach, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_SECOND, if_attachdomain, NULL); static void if_attachdomain1(struct ifnet *ifp) { struct domain *dp; int s; s = splnet(); /* * Since dp->dom_ifattach calls malloc() with M_WAITOK, we * cannot lock ifp->if_afdata initialization, entirely. */ if (IF_AFDATA_TRYLOCK(ifp) == 0) { splx(s); return; } if (ifp->if_afdata_initialized >= domain_init_status) { IF_AFDATA_UNLOCK(ifp); splx(s); printf("if_attachdomain called more than once on %s\n", ifp->if_xname); return; } ifp->if_afdata_initialized = domain_init_status; IF_AFDATA_UNLOCK(ifp); /* address family dependent data region */ bzero(ifp->if_afdata, sizeof(ifp->if_afdata)); for (dp = domains; dp; dp = dp->dom_next) { if (dp->dom_ifattach) ifp->if_afdata[dp->dom_family] = (*dp->dom_ifattach)(ifp); } splx(s); } /* * Remove any unicast or broadcast network addresses from an interface. */ void if_purgeaddrs(struct ifnet *ifp) { struct ifaddr *ifa, *next; TAILQ_FOREACH_SAFE(ifa, &ifp->if_addrhead, ifa_link, next) { if (ifa->ifa_addr->sa_family == AF_LINK) continue; #ifdef INET /* XXX: Ugly!! ad hoc just for INET */ if (ifa->ifa_addr->sa_family == AF_INET) { struct ifaliasreq ifr; bzero(&ifr, sizeof(ifr)); ifr.ifra_addr = *ifa->ifa_addr; if (ifa->ifa_dstaddr) ifr.ifra_broadaddr = *ifa->ifa_dstaddr; if (in_control(NULL, SIOCDIFADDR, (caddr_t)&ifr, ifp, NULL) == 0) continue; } #endif /* INET */ #ifdef INET6 if (ifa->ifa_addr->sa_family == AF_INET6) { in6_purgeaddr(ifa); /* ifp_addrhead is already updated */ continue; } #endif /* INET6 */ TAILQ_REMOVE(&ifp->if_addrhead, ifa, ifa_link); ifa_free(ifa); } } /* * Remove any multicast network addresses from an interface when an ifnet * is going away. */ static void if_purgemaddrs(struct ifnet *ifp) { struct ifmultiaddr *ifma; struct ifmultiaddr *next; IF_ADDR_WLOCK(ifp); TAILQ_FOREACH_SAFE(ifma, &ifp->if_multiaddrs, ifma_link, next) if_delmulti_locked(ifp, ifma, 1); IF_ADDR_WUNLOCK(ifp); } /* * Detach an interface, removing it from the list of "active" interfaces. * If vmove flag is set on entry to if_detach_internal(), perform only a * limited subset of cleanup tasks, given that we are moving an ifnet from * one vnet to another, where it must be fully operational. * * XXXRW: There are some significant questions about event ordering, and * how to prevent things from starting to use the interface during detach. */ void if_detach(struct ifnet *ifp) { if_detach_internal(ifp, 0); } static void if_detach_internal(struct ifnet *ifp, int vmove) { struct ifaddr *ifa; struct radix_node_head *rnh; int i, j; struct domain *dp; struct ifnet *iter; int found = 0; IFNET_WLOCK(); TAILQ_FOREACH(iter, &V_ifnet, if_link) if (iter == ifp) { TAILQ_REMOVE(&V_ifnet, ifp, if_link); found = 1; break; } #ifdef VIMAGE if (found) curvnet->vnet_ifcnt--; #endif IFNET_WUNLOCK(); if (!found) { if (vmove) panic("%s: ifp=%p not on the ifnet tailq %p", __func__, ifp, &V_ifnet); else return; /* XXX this should panic as well? */ } /* * Remove/wait for pending events. */ taskqueue_drain(taskqueue_swi, &ifp->if_linktask); /* * Remove routes and flush queues. */ if_down(ifp); #ifdef ALTQ if (ALTQ_IS_ENABLED(&ifp->if_snd)) altq_disable(&ifp->if_snd); if (ALTQ_IS_ATTACHED(&ifp->if_snd)) altq_detach(&ifp->if_snd); #endif if_purgeaddrs(ifp); #ifdef INET in_ifdetach(ifp); #endif #ifdef INET6 /* * Remove all IPv6 kernel structs related to ifp. This should be done * before removing routing entries below, since IPv6 interface direct * routes are expected to be removed by the IPv6-specific kernel API. * Otherwise, the kernel will detect some inconsistency and bark it. */ in6_ifdetach(ifp); #endif if_purgemaddrs(ifp); if (!vmove) { /* * Prevent further calls into the device driver via ifnet. */ if_dead(ifp); /* * Remove link ifaddr pointer and maybe decrement if_index. * Clean up all addresses. */ ifp->if_addr = NULL; /* We can now free link ifaddr. */ if (!TAILQ_EMPTY(&ifp->if_addrhead)) { ifa = TAILQ_FIRST(&ifp->if_addrhead); TAILQ_REMOVE(&ifp->if_addrhead, ifa, ifa_link); ifa_free(ifa); } } /* * Delete all remaining routes using this interface * Unfortuneatly the only way to do this is to slog through * the entire routing table looking for routes which point * to this interface...oh well... */ for (i = 1; i <= AF_MAX; i++) { for (j = 0; j < rt_numfibs; j++) { rnh = rt_tables_get_rnh(j, i); if (rnh == NULL) continue; RADIX_NODE_HEAD_LOCK(rnh); (void) rnh->rnh_walktree(rnh, if_rtdel, ifp); RADIX_NODE_HEAD_UNLOCK(rnh); } } /* Announce that the interface is gone. */ rt_ifannouncemsg(ifp, IFAN_DEPARTURE); EVENTHANDLER_INVOKE(ifnet_departure_event, ifp); if (IS_DEFAULT_VNET(curvnet)) devctl_notify("IFNET", ifp->if_xname, "DETACH", NULL); if_delgroups(ifp); /* * We cannot hold the lock over dom_ifdetach calls as they might * sleep, for example trying to drain a callout, thus open up the * theoretical race with re-attaching. */ IF_AFDATA_LOCK(ifp); i = ifp->if_afdata_initialized; ifp->if_afdata_initialized = 0; IF_AFDATA_UNLOCK(ifp); for (dp = domains; i > 0 && dp; dp = dp->dom_next) { if (dp->dom_ifdetach && ifp->if_afdata[dp->dom_family]) (*dp->dom_ifdetach)(ifp, ifp->if_afdata[dp->dom_family]); } } #ifdef VIMAGE /* * if_vmove() performs a limited version of if_detach() in current * vnet and if_attach()es the ifnet to the vnet specified as 2nd arg. * An attempt is made to shrink if_index in current vnet, find an * unused if_index in target vnet and calls if_grow() if necessary, * and finally find an unused if_xname for the target vnet. */ void if_vmove(struct ifnet *ifp, struct vnet *new_vnet) { u_short idx; /* * Detach from current vnet, but preserve LLADDR info, do not * mark as dead etc. so that the ifnet can be reattached later. */ if_detach_internal(ifp, 1); /* * Unlink the ifnet from ifindex_table[] in current vnet, and shrink * the if_index for that vnet if possible. * * NOTE: IFNET_WLOCK/IFNET_WUNLOCK() are assumed to be unvirtualized, * or we'd lock on one vnet and unlock on another. */ IFNET_WLOCK(); ifindex_free_locked(ifp->if_index); IFNET_WUNLOCK(); /* * Perform interface-specific reassignment tasks, if provided by * the driver. */ if (ifp->if_reassign != NULL) ifp->if_reassign(ifp, new_vnet, NULL); /* * Switch to the context of the target vnet. */ CURVNET_SET_QUIET(new_vnet); IFNET_WLOCK(); if (ifindex_alloc_locked(&idx) != 0) { IFNET_WUNLOCK(); panic("if_index overflow"); } ifp->if_index = idx; ifnet_setbyindex_locked(ifp->if_index, ifp); IFNET_WUNLOCK(); if_attach_internal(ifp, 1); CURVNET_RESTORE(); } /* * Move an ifnet to or from another child prison/vnet, specified by the jail id. */ static int if_vmove_loan(struct thread *td, struct ifnet *ifp, char *ifname, int jid) { struct prison *pr; struct ifnet *difp; /* Try to find the prison within our visibility. */ sx_slock(&allprison_lock); pr = prison_find_child(td->td_ucred->cr_prison, jid); sx_sunlock(&allprison_lock); if (pr == NULL) return (ENXIO); prison_hold_locked(pr); mtx_unlock(&pr->pr_mtx); /* Do not try to move the iface from and to the same prison. */ if (pr->pr_vnet == ifp->if_vnet) { prison_free(pr); return (EEXIST); } /* Make sure the named iface does not exists in the dst. prison/vnet. */ /* XXX Lock interfaces to avoid races. */ CURVNET_SET_QUIET(pr->pr_vnet); difp = ifunit(ifname); CURVNET_RESTORE(); if (difp != NULL) { prison_free(pr); return (EEXIST); } /* Move the interface into the child jail/vnet. */ if_vmove(ifp, pr->pr_vnet); /* Report the new if_xname back to the userland. */ sprintf(ifname, "%s", ifp->if_xname); prison_free(pr); return (0); } static int if_vmove_reclaim(struct thread *td, char *ifname, int jid) { struct prison *pr; struct vnet *vnet_dst; struct ifnet *ifp; /* Try to find the prison within our visibility. */ sx_slock(&allprison_lock); pr = prison_find_child(td->td_ucred->cr_prison, jid); sx_sunlock(&allprison_lock); if (pr == NULL) return (ENXIO); prison_hold_locked(pr); mtx_unlock(&pr->pr_mtx); /* Make sure the named iface exists in the source prison/vnet. */ CURVNET_SET(pr->pr_vnet); ifp = ifunit(ifname); /* XXX Lock to avoid races. */ if (ifp == NULL) { CURVNET_RESTORE(); prison_free(pr); return (ENXIO); } /* Do not try to move the iface from and to the same prison. */ vnet_dst = TD_TO_VNET(td); if (vnet_dst == ifp->if_vnet) { CURVNET_RESTORE(); prison_free(pr); return (EEXIST); } /* Get interface back from child jail/vnet. */ if_vmove(ifp, vnet_dst); CURVNET_RESTORE(); /* Report the new if_xname back to the userland. */ sprintf(ifname, "%s", ifp->if_xname); prison_free(pr); return (0); } #endif /* VIMAGE */ /* * Add a group to an interface */ int if_addgroup(struct ifnet *ifp, const char *groupname) { struct ifg_list *ifgl; struct ifg_group *ifg = NULL; struct ifg_member *ifgm; if (groupname[0] && groupname[strlen(groupname) - 1] >= '0' && groupname[strlen(groupname) - 1] <= '9') return (EINVAL); IFNET_WLOCK(); TAILQ_FOREACH(ifgl, &ifp->if_groups, ifgl_next) if (!strcmp(ifgl->ifgl_group->ifg_group, groupname)) { IFNET_WUNLOCK(); return (EEXIST); } if ((ifgl = (struct ifg_list *)malloc(sizeof(struct ifg_list), M_TEMP, M_NOWAIT)) == NULL) { IFNET_WUNLOCK(); return (ENOMEM); } if ((ifgm = (struct ifg_member *)malloc(sizeof(struct ifg_member), M_TEMP, M_NOWAIT)) == NULL) { free(ifgl, M_TEMP); IFNET_WUNLOCK(); return (ENOMEM); } TAILQ_FOREACH(ifg, &V_ifg_head, ifg_next) if (!strcmp(ifg->ifg_group, groupname)) break; if (ifg == NULL) { if ((ifg = (struct ifg_group *)malloc(sizeof(struct ifg_group), M_TEMP, M_NOWAIT)) == NULL) { free(ifgl, M_TEMP); free(ifgm, M_TEMP); IFNET_WUNLOCK(); return (ENOMEM); } strlcpy(ifg->ifg_group, groupname, sizeof(ifg->ifg_group)); ifg->ifg_refcnt = 0; TAILQ_INIT(&ifg->ifg_members); EVENTHANDLER_INVOKE(group_attach_event, ifg); TAILQ_INSERT_TAIL(&V_ifg_head, ifg, ifg_next); } ifg->ifg_refcnt++; ifgl->ifgl_group = ifg; ifgm->ifgm_ifp = ifp; IF_ADDR_WLOCK(ifp); TAILQ_INSERT_TAIL(&ifg->ifg_members, ifgm, ifgm_next); TAILQ_INSERT_TAIL(&ifp->if_groups, ifgl, ifgl_next); IF_ADDR_WUNLOCK(ifp); IFNET_WUNLOCK(); EVENTHANDLER_INVOKE(group_change_event, groupname); return (0); } /* * Remove a group from an interface */ int if_delgroup(struct ifnet *ifp, const char *groupname) { struct ifg_list *ifgl; struct ifg_member *ifgm; IFNET_WLOCK(); TAILQ_FOREACH(ifgl, &ifp->if_groups, ifgl_next) if (!strcmp(ifgl->ifgl_group->ifg_group, groupname)) break; if (ifgl == NULL) { IFNET_WUNLOCK(); return (ENOENT); } IF_ADDR_WLOCK(ifp); TAILQ_REMOVE(&ifp->if_groups, ifgl, ifgl_next); IF_ADDR_WUNLOCK(ifp); TAILQ_FOREACH(ifgm, &ifgl->ifgl_group->ifg_members, ifgm_next) if (ifgm->ifgm_ifp == ifp) break; if (ifgm != NULL) { TAILQ_REMOVE(&ifgl->ifgl_group->ifg_members, ifgm, ifgm_next); free(ifgm, M_TEMP); } if (--ifgl->ifgl_group->ifg_refcnt == 0) { TAILQ_REMOVE(&V_ifg_head, ifgl->ifgl_group, ifg_next); EVENTHANDLER_INVOKE(group_detach_event, ifgl->ifgl_group); free(ifgl->ifgl_group, M_TEMP); } IFNET_WUNLOCK(); free(ifgl, M_TEMP); EVENTHANDLER_INVOKE(group_change_event, groupname); return (0); } /* * Remove an interface from all groups */ static void if_delgroups(struct ifnet *ifp) { struct ifg_list *ifgl; struct ifg_member *ifgm; char groupname[IFNAMSIZ]; IFNET_WLOCK(); while (!TAILQ_EMPTY(&ifp->if_groups)) { ifgl = TAILQ_FIRST(&ifp->if_groups); strlcpy(groupname, ifgl->ifgl_group->ifg_group, IFNAMSIZ); IF_ADDR_WLOCK(ifp); TAILQ_REMOVE(&ifp->if_groups, ifgl, ifgl_next); IF_ADDR_WUNLOCK(ifp); TAILQ_FOREACH(ifgm, &ifgl->ifgl_group->ifg_members, ifgm_next) if (ifgm->ifgm_ifp == ifp) break; if (ifgm != NULL) { TAILQ_REMOVE(&ifgl->ifgl_group->ifg_members, ifgm, ifgm_next); free(ifgm, M_TEMP); } if (--ifgl->ifgl_group->ifg_refcnt == 0) { TAILQ_REMOVE(&V_ifg_head, ifgl->ifgl_group, ifg_next); EVENTHANDLER_INVOKE(group_detach_event, ifgl->ifgl_group); free(ifgl->ifgl_group, M_TEMP); } IFNET_WUNLOCK(); free(ifgl, M_TEMP); EVENTHANDLER_INVOKE(group_change_event, groupname); IFNET_WLOCK(); } IFNET_WUNLOCK(); } /* * Stores all groups from an interface in memory pointed * to by data */ static int if_getgroup(struct ifgroupreq *data, struct ifnet *ifp) { int len, error; struct ifg_list *ifgl; struct ifg_req ifgrq, *ifgp; struct ifgroupreq *ifgr = data; if (ifgr->ifgr_len == 0) { IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ifgl, &ifp->if_groups, ifgl_next) ifgr->ifgr_len += sizeof(struct ifg_req); IF_ADDR_RUNLOCK(ifp); return (0); } len = ifgr->ifgr_len; ifgp = ifgr->ifgr_groups; /* XXX: wire */ IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ifgl, &ifp->if_groups, ifgl_next) { if (len < sizeof(ifgrq)) { IF_ADDR_RUNLOCK(ifp); return (EINVAL); } bzero(&ifgrq, sizeof ifgrq); strlcpy(ifgrq.ifgrq_group, ifgl->ifgl_group->ifg_group, sizeof(ifgrq.ifgrq_group)); if ((error = copyout(&ifgrq, ifgp, sizeof(struct ifg_req)))) { IF_ADDR_RUNLOCK(ifp); return (error); } len -= sizeof(ifgrq); ifgp++; } IF_ADDR_RUNLOCK(ifp); return (0); } /* * Stores all members of a group in memory pointed to by data */ static int if_getgroupmembers(struct ifgroupreq *data) { struct ifgroupreq *ifgr = data; struct ifg_group *ifg; struct ifg_member *ifgm; struct ifg_req ifgrq, *ifgp; int len, error; IFNET_RLOCK(); TAILQ_FOREACH(ifg, &V_ifg_head, ifg_next) if (!strcmp(ifg->ifg_group, ifgr->ifgr_name)) break; if (ifg == NULL) { IFNET_RUNLOCK(); return (ENOENT); } if (ifgr->ifgr_len == 0) { TAILQ_FOREACH(ifgm, &ifg->ifg_members, ifgm_next) ifgr->ifgr_len += sizeof(ifgrq); IFNET_RUNLOCK(); return (0); } len = ifgr->ifgr_len; ifgp = ifgr->ifgr_groups; TAILQ_FOREACH(ifgm, &ifg->ifg_members, ifgm_next) { if (len < sizeof(ifgrq)) { IFNET_RUNLOCK(); return (EINVAL); } bzero(&ifgrq, sizeof ifgrq); strlcpy(ifgrq.ifgrq_member, ifgm->ifgm_ifp->if_xname, sizeof(ifgrq.ifgrq_member)); if ((error = copyout(&ifgrq, ifgp, sizeof(struct ifg_req)))) { IFNET_RUNLOCK(); return (error); } len -= sizeof(ifgrq); ifgp++; } IFNET_RUNLOCK(); return (0); } /* * Delete Routes for a Network Interface * * Called for each routing entry via the rnh->rnh_walktree() call above * to delete all route entries referencing a detaching network interface. * * Arguments: * rn pointer to node in the routing table * arg argument passed to rnh->rnh_walktree() - detaching interface * * Returns: * 0 successful * errno failed - reason indicated * */ static int if_rtdel(struct radix_node *rn, void *arg) { struct rtentry *rt = (struct rtentry *)rn; struct ifnet *ifp = arg; int err; if (rt->rt_ifp == ifp) { /* * Protect (sorta) against walktree recursion problems * with cloned routes */ if ((rt->rt_flags & RTF_UP) == 0) return (0); err = rtrequest_fib(RTM_DELETE, rt_key(rt), rt->rt_gateway, rt_mask(rt), rt->rt_flags|RTF_RNH_LOCKED, (struct rtentry **) NULL, rt->rt_fibnum); if (err) { log(LOG_WARNING, "if_rtdel: error %d\n", err); } } return (0); } /* * Wrapper functions for struct ifnet address list locking macros. These are * used by kernel modules to avoid encoding programming interface or binary * interface assumptions that may be violated when kernel-internal locking * approaches change. */ void if_addr_rlock(struct ifnet *ifp) { IF_ADDR_RLOCK(ifp); } void if_addr_runlock(struct ifnet *ifp) { IF_ADDR_RUNLOCK(ifp); } void if_maddr_rlock(struct ifnet *ifp) { IF_ADDR_RLOCK(ifp); } void if_maddr_runlock(struct ifnet *ifp) { IF_ADDR_RUNLOCK(ifp); } /* * Initialization, destruction and refcounting functions for ifaddrs. */ void ifa_init(struct ifaddr *ifa) { mtx_init(&ifa->ifa_mtx, "ifaddr", NULL, MTX_DEF); refcount_init(&ifa->ifa_refcnt, 1); ifa->if_data.ifi_datalen = sizeof(ifa->if_data); } void ifa_ref(struct ifaddr *ifa) { refcount_acquire(&ifa->ifa_refcnt); } void ifa_free(struct ifaddr *ifa) { if (refcount_release(&ifa->ifa_refcnt)) { mtx_destroy(&ifa->ifa_mtx); free(ifa, M_IFADDR); } } int ifa_add_loopback_route(struct ifaddr *ifa, struct sockaddr *ia) { int error = 0; struct rtentry *rt = NULL; struct rt_addrinfo info; static struct sockaddr_dl null_sdl = {sizeof(null_sdl), AF_LINK}; bzero(&info, sizeof(info)); info.rti_ifp = V_loif; info.rti_flags = ifa->ifa_flags | RTF_HOST | RTF_STATIC; info.rti_info[RTAX_DST] = ia; info.rti_info[RTAX_GATEWAY] = (struct sockaddr *)&null_sdl; error = rtrequest1_fib(RTM_ADD, &info, &rt, 0); if (error == 0 && rt != NULL) { RT_LOCK(rt); ((struct sockaddr_dl *)rt->rt_gateway)->sdl_type = ifa->ifa_ifp->if_type; ((struct sockaddr_dl *)rt->rt_gateway)->sdl_index = ifa->ifa_ifp->if_index; RT_REMREF(rt); RT_UNLOCK(rt); } else if (error != 0) log(LOG_DEBUG, "%s: insertion failed: %u\n", __func__, error); return (error); } int ifa_del_loopback_route(struct ifaddr *ifa, struct sockaddr *ia) { int error = 0; struct rt_addrinfo info; struct sockaddr_dl null_sdl; bzero(&null_sdl, sizeof(null_sdl)); null_sdl.sdl_len = sizeof(null_sdl); null_sdl.sdl_family = AF_LINK; null_sdl.sdl_type = ifa->ifa_ifp->if_type; null_sdl.sdl_index = ifa->ifa_ifp->if_index; bzero(&info, sizeof(info)); info.rti_flags = ifa->ifa_flags | RTF_HOST | RTF_STATIC; info.rti_info[RTAX_DST] = ia; info.rti_info[RTAX_GATEWAY] = (struct sockaddr *)&null_sdl; error = rtrequest1_fib(RTM_DELETE, &info, NULL, 0); if (error != 0) log(LOG_DEBUG, "%s: deletion failed: %u\n", __func__, error); return (error); } /* * XXX: Because sockaddr_dl has deeper structure than the sockaddr * structs used to represent other address families, it is necessary * to perform a different comparison. */ #define sa_equal(a1, a2) \ (bcmp((a1), (a2), ((a1))->sa_len) == 0) #define sa_dl_equal(a1, a2) \ ((((struct sockaddr_dl *)(a1))->sdl_len == \ ((struct sockaddr_dl *)(a2))->sdl_len) && \ (bcmp(LLADDR((struct sockaddr_dl *)(a1)), \ LLADDR((struct sockaddr_dl *)(a2)), \ ((struct sockaddr_dl *)(a1))->sdl_alen) == 0)) /* * Locate an interface based on a complete address. */ /*ARGSUSED*/ static struct ifaddr * ifa_ifwithaddr_internal(struct sockaddr *addr, int getref) { struct ifnet *ifp; struct ifaddr *ifa; IFNET_RLOCK_NOSLEEP(); TAILQ_FOREACH(ifp, &V_ifnet, if_link) { IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != addr->sa_family) continue; if (sa_equal(addr, ifa->ifa_addr)) { if (getref) ifa_ref(ifa); IF_ADDR_RUNLOCK(ifp); goto done; } /* IP6 doesn't have broadcast */ if ((ifp->if_flags & IFF_BROADCAST) && ifa->ifa_broadaddr && ifa->ifa_broadaddr->sa_len != 0 && sa_equal(ifa->ifa_broadaddr, addr)) { if (getref) ifa_ref(ifa); IF_ADDR_RUNLOCK(ifp); goto done; } } IF_ADDR_RUNLOCK(ifp); } ifa = NULL; done: IFNET_RUNLOCK_NOSLEEP(); return (ifa); } struct ifaddr * ifa_ifwithaddr(struct sockaddr *addr) { return (ifa_ifwithaddr_internal(addr, 1)); } int ifa_ifwithaddr_check(struct sockaddr *addr) { return (ifa_ifwithaddr_internal(addr, 0) != NULL); } /* * Locate an interface based on the broadcast address. */ /* ARGSUSED */ struct ifaddr * ifa_ifwithbroadaddr(struct sockaddr *addr) { struct ifnet *ifp; struct ifaddr *ifa; IFNET_RLOCK_NOSLEEP(); TAILQ_FOREACH(ifp, &V_ifnet, if_link) { IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != addr->sa_family) continue; if ((ifp->if_flags & IFF_BROADCAST) && ifa->ifa_broadaddr && ifa->ifa_broadaddr->sa_len != 0 && sa_equal(ifa->ifa_broadaddr, addr)) { ifa_ref(ifa); IF_ADDR_RUNLOCK(ifp); goto done; } } IF_ADDR_RUNLOCK(ifp); } ifa = NULL; done: IFNET_RUNLOCK_NOSLEEP(); return (ifa); } /* * Locate the point to point interface with a given destination address. */ /*ARGSUSED*/ struct ifaddr * ifa_ifwithdstaddr(struct sockaddr *addr) { struct ifnet *ifp; struct ifaddr *ifa; IFNET_RLOCK_NOSLEEP(); TAILQ_FOREACH(ifp, &V_ifnet, if_link) { if ((ifp->if_flags & IFF_POINTOPOINT) == 0) continue; IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != addr->sa_family) continue; if (ifa->ifa_dstaddr != NULL && sa_equal(addr, ifa->ifa_dstaddr)) { ifa_ref(ifa); IF_ADDR_RUNLOCK(ifp); goto done; } } IF_ADDR_RUNLOCK(ifp); } ifa = NULL; done: IFNET_RUNLOCK_NOSLEEP(); return (ifa); } /* * Find an interface on a specific network. If many, choice * is most specific found. */ struct ifaddr * ifa_ifwithnet(struct sockaddr *addr, int ignore_ptp) { struct ifnet *ifp; struct ifaddr *ifa; struct ifaddr *ifa_maybe = NULL; u_int af = addr->sa_family; char *addr_data = addr->sa_data, *cplim; /* * AF_LINK addresses can be looked up directly by their index number, * so do that if we can. */ if (af == AF_LINK) { struct sockaddr_dl *sdl = (struct sockaddr_dl *)addr; if (sdl->sdl_index && sdl->sdl_index <= V_if_index) return (ifaddr_byindex(sdl->sdl_index)); } /* * Scan though each interface, looking for ones that have addresses * in this address family. Maintain a reference on ifa_maybe once * we find one, as we release the IF_ADDR_RLOCK() that kept it stable * when we move onto the next interface. */ IFNET_RLOCK_NOSLEEP(); TAILQ_FOREACH(ifp, &V_ifnet, if_link) { IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { char *cp, *cp2, *cp3; if (ifa->ifa_addr->sa_family != af) next: continue; if (af == AF_INET && ifp->if_flags & IFF_POINTOPOINT && !ignore_ptp) { /* * This is a bit broken as it doesn't * take into account that the remote end may * be a single node in the network we are * looking for. * The trouble is that we don't know the * netmask for the remote end. */ if (ifa->ifa_dstaddr != NULL && sa_equal(addr, ifa->ifa_dstaddr)) { ifa_ref(ifa); IF_ADDR_RUNLOCK(ifp); goto done; } } else { /* * if we have a special address handler, * then use it instead of the generic one. */ if (ifa->ifa_claim_addr) { if ((*ifa->ifa_claim_addr)(ifa, addr)) { ifa_ref(ifa); IF_ADDR_RUNLOCK(ifp); goto done; } continue; } /* * Scan all the bits in the ifa's address. * If a bit dissagrees with what we are * looking for, mask it with the netmask * to see if it really matters. * (A byte at a time) */ if (ifa->ifa_netmask == 0) continue; cp = addr_data; cp2 = ifa->ifa_addr->sa_data; cp3 = ifa->ifa_netmask->sa_data; cplim = ifa->ifa_netmask->sa_len + (char *)ifa->ifa_netmask; while (cp3 < cplim) if ((*cp++ ^ *cp2++) & *cp3++) goto next; /* next address! */ /* * If the netmask of what we just found * is more specific than what we had before * (if we had one) then remember the new one * before continuing to search * for an even better one. */ if (ifa_maybe == NULL || rn_refines((caddr_t)ifa->ifa_netmask, (caddr_t)ifa_maybe->ifa_netmask)) { if (ifa_maybe != NULL) ifa_free(ifa_maybe); ifa_maybe = ifa; ifa_ref(ifa_maybe); } } } IF_ADDR_RUNLOCK(ifp); } ifa = ifa_maybe; ifa_maybe = NULL; done: IFNET_RUNLOCK_NOSLEEP(); if (ifa_maybe != NULL) ifa_free(ifa_maybe); return (ifa); } /* * Find an interface address specific to an interface best matching * a given address. */ struct ifaddr * ifaof_ifpforaddr(struct sockaddr *addr, struct ifnet *ifp) { struct ifaddr *ifa; char *cp, *cp2, *cp3; char *cplim; struct ifaddr *ifa_maybe = NULL; u_int af = addr->sa_family; if (af >= AF_MAX) return (NULL); IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family != af) continue; if (ifa_maybe == NULL) ifa_maybe = ifa; if (ifa->ifa_netmask == 0) { if (sa_equal(addr, ifa->ifa_addr) || (ifa->ifa_dstaddr && sa_equal(addr, ifa->ifa_dstaddr))) goto done; continue; } if (ifp->if_flags & IFF_POINTOPOINT) { if (sa_equal(addr, ifa->ifa_dstaddr)) goto done; } else { cp = addr->sa_data; cp2 = ifa->ifa_addr->sa_data; cp3 = ifa->ifa_netmask->sa_data; cplim = ifa->ifa_netmask->sa_len + (char *)ifa->ifa_netmask; for (; cp3 < cplim; cp3++) if ((*cp++ ^ *cp2++) & *cp3) break; if (cp3 == cplim) goto done; } } ifa = ifa_maybe; done: if (ifa != NULL) ifa_ref(ifa); IF_ADDR_RUNLOCK(ifp); return (ifa); } #include /* * Default action when installing a route with a Link Level gateway. * Lookup an appropriate real ifa to point to. * This should be moved to /sys/net/link.c eventually. */ static void link_rtrequest(int cmd, struct rtentry *rt, struct rt_addrinfo *info) { struct ifaddr *ifa, *oifa; struct sockaddr *dst; struct ifnet *ifp; RT_LOCK_ASSERT(rt); if (cmd != RTM_ADD || ((ifa = rt->rt_ifa) == 0) || ((ifp = ifa->ifa_ifp) == 0) || ((dst = rt_key(rt)) == 0)) return; ifa = ifaof_ifpforaddr(dst, ifp); if (ifa) { oifa = rt->rt_ifa; rt->rt_ifa = ifa; ifa_free(oifa); if (ifa->ifa_rtrequest && ifa->ifa_rtrequest != link_rtrequest) ifa->ifa_rtrequest(cmd, rt, info); } } /* * Mark an interface down and notify protocols of * the transition. * NOTE: must be called at splnet or eqivalent. */ static void if_unroute(struct ifnet *ifp, int flag, int fam) { struct ifaddr *ifa; KASSERT(flag == IFF_UP, ("if_unroute: flag != IFF_UP")); ifp->if_flags &= ~flag; getmicrotime(&ifp->if_lastchange); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) if (fam == PF_UNSPEC || (fam == ifa->ifa_addr->sa_family)) pfctlinput(PRC_IFDOWN, ifa->ifa_addr); ifp->if_qflush(ifp); if (ifp->if_carp) (*carp_linkstate_p)(ifp); rt_ifmsg(ifp); } /* * Mark an interface up and notify protocols of * the transition. * NOTE: must be called at splnet or eqivalent. */ static void if_route(struct ifnet *ifp, int flag, int fam) { struct ifaddr *ifa; KASSERT(flag == IFF_UP, ("if_route: flag != IFF_UP")); ifp->if_flags |= flag; getmicrotime(&ifp->if_lastchange); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) if (fam == PF_UNSPEC || (fam == ifa->ifa_addr->sa_family)) pfctlinput(PRC_IFUP, ifa->ifa_addr); if (ifp->if_carp) (*carp_linkstate_p)(ifp); rt_ifmsg(ifp); #ifdef INET6 in6_if_up(ifp); #endif } void (*vlan_link_state_p)(struct ifnet *); /* XXX: private from if_vlan */ void (*vlan_trunk_cap_p)(struct ifnet *); /* XXX: private from if_vlan */ struct ifnet *(*vlan_trunkdev_p)(struct ifnet *); struct ifnet *(*vlan_devat_p)(struct ifnet *, uint16_t); int (*vlan_tag_p)(struct ifnet *, uint16_t *); int (*vlan_setcookie_p)(struct ifnet *, void *); void *(*vlan_cookie_p)(struct ifnet *); /* * Handle a change in the interface link state. To avoid LORs * between driver lock and upper layer locks, as well as possible * recursions, we post event to taskqueue, and all job * is done in static do_link_state_change(). */ void if_link_state_change(struct ifnet *ifp, int link_state) { /* Return if state hasn't changed. */ if (ifp->if_link_state == link_state) return; ifp->if_link_state = link_state; taskqueue_enqueue(taskqueue_swi, &ifp->if_linktask); } static void do_link_state_change(void *arg, int pending) { struct ifnet *ifp = (struct ifnet *)arg; int link_state = ifp->if_link_state; CURVNET_SET(ifp->if_vnet); /* Notify that the link state has changed. */ rt_ifmsg(ifp); if (ifp->if_vlantrunk != NULL) (*vlan_link_state_p)(ifp); if ((ifp->if_type == IFT_ETHER || ifp->if_type == IFT_L2VLAN) && IFP2AC(ifp)->ac_netgraph != NULL) (*ng_ether_link_state_p)(ifp, link_state); if (ifp->if_carp) (*carp_linkstate_p)(ifp); if (ifp->if_bridge) - (*bstp_linkstate_p)(ifp, link_state); + (*bridge_linkstate_p)(ifp); if (ifp->if_lagg) (*lagg_linkstate_p)(ifp, link_state); if (IS_DEFAULT_VNET(curvnet)) devctl_notify("IFNET", ifp->if_xname, (link_state == LINK_STATE_UP) ? "LINK_UP" : "LINK_DOWN", NULL); if (pending > 1) if_printf(ifp, "%d link states coalesced\n", pending); if (log_link_state_change) log(LOG_NOTICE, "%s: link state changed to %s\n", ifp->if_xname, (link_state == LINK_STATE_UP) ? "UP" : "DOWN" ); EVENTHANDLER_INVOKE(ifnet_link_event, ifp, ifp->if_link_state); CURVNET_RESTORE(); } /* * Mark an interface down and notify protocols of * the transition. * NOTE: must be called at splnet or eqivalent. */ void if_down(struct ifnet *ifp) { if_unroute(ifp, IFF_UP, AF_UNSPEC); } /* * Mark an interface up and notify protocols of * the transition. * NOTE: must be called at splnet or eqivalent. */ void if_up(struct ifnet *ifp) { if_route(ifp, IFF_UP, AF_UNSPEC); } /* * Flush an interface queue. */ void if_qflush(struct ifnet *ifp) { struct mbuf *m, *n; struct ifaltq *ifq; ifq = &ifp->if_snd; IFQ_LOCK(ifq); #ifdef ALTQ if (ALTQ_IS_ENABLED(ifq)) ALTQ_PURGE(ifq); #endif n = ifq->ifq_head; while ((m = n) != 0) { n = m->m_act; m_freem(m); } ifq->ifq_head = 0; ifq->ifq_tail = 0; ifq->ifq_len = 0; IFQ_UNLOCK(ifq); } /* * Map interface name to interface structure pointer, with or without * returning a reference. */ struct ifnet * ifunit_ref(const char *name) { struct ifnet *ifp; IFNET_RLOCK_NOSLEEP(); TAILQ_FOREACH(ifp, &V_ifnet, if_link) { if (strncmp(name, ifp->if_xname, IFNAMSIZ) == 0 && !(ifp->if_flags & IFF_DYING)) break; } if (ifp != NULL) if_ref(ifp); IFNET_RUNLOCK_NOSLEEP(); return (ifp); } struct ifnet * ifunit(const char *name) { struct ifnet *ifp; IFNET_RLOCK_NOSLEEP(); TAILQ_FOREACH(ifp, &V_ifnet, if_link) { if (strncmp(name, ifp->if_xname, IFNAMSIZ) == 0) break; } IFNET_RUNLOCK_NOSLEEP(); return (ifp); } /* * Hardware specific interface ioctls. */ static int ifhwioctl(u_long cmd, struct ifnet *ifp, caddr_t data, struct thread *td) { struct ifreq *ifr; struct ifstat *ifs; int error = 0; int new_flags, temp_flags; size_t namelen, onamelen; size_t descrlen; char *descrbuf, *odescrbuf; char new_name[IFNAMSIZ]; struct ifaddr *ifa; struct sockaddr_dl *sdl; ifr = (struct ifreq *)data; switch (cmd) { case SIOCGIFINDEX: ifr->ifr_index = ifp->if_index; break; case SIOCGIFFLAGS: temp_flags = ifp->if_flags | ifp->if_drv_flags; ifr->ifr_flags = temp_flags & 0xffff; ifr->ifr_flagshigh = temp_flags >> 16; break; case SIOCGIFCAP: ifr->ifr_reqcap = ifp->if_capabilities; ifr->ifr_curcap = ifp->if_capenable; break; #ifdef MAC case SIOCGIFMAC: error = mac_ifnet_ioctl_get(td->td_ucred, ifr, ifp); break; #endif case SIOCGIFMETRIC: ifr->ifr_metric = ifp->if_metric; break; case SIOCGIFMTU: ifr->ifr_mtu = ifp->if_mtu; break; case SIOCGIFPHYS: ifr->ifr_phys = ifp->if_physical; break; case SIOCGIFDESCR: error = 0; sx_slock(&ifdescr_sx); if (ifp->if_description == NULL) error = ENOMSG; else { /* space for terminating nul */ descrlen = strlen(ifp->if_description) + 1; if (ifr->ifr_buffer.length < descrlen) ifr->ifr_buffer.buffer = NULL; else error = copyout(ifp->if_description, ifr->ifr_buffer.buffer, descrlen); ifr->ifr_buffer.length = descrlen; } sx_sunlock(&ifdescr_sx); break; case SIOCSIFDESCR: error = priv_check(td, PRIV_NET_SETIFDESCR); if (error) return (error); /* * Copy only (length-1) bytes to make sure that * if_description is always nul terminated. The * length parameter is supposed to count the * terminating nul in. */ if (ifr->ifr_buffer.length > ifdescr_maxlen) return (ENAMETOOLONG); else if (ifr->ifr_buffer.length == 0) descrbuf = NULL; else { descrbuf = malloc(ifr->ifr_buffer.length, M_IFDESCR, M_WAITOK | M_ZERO); error = copyin(ifr->ifr_buffer.buffer, descrbuf, ifr->ifr_buffer.length - 1); if (error) { free(descrbuf, M_IFDESCR); break; } } sx_xlock(&ifdescr_sx); odescrbuf = ifp->if_description; ifp->if_description = descrbuf; sx_xunlock(&ifdescr_sx); getmicrotime(&ifp->if_lastchange); free(odescrbuf, M_IFDESCR); break; case SIOCGIFFIB: ifr->ifr_fib = ifp->if_fib; break; case SIOCSIFFIB: error = priv_check(td, PRIV_NET_SETIFFIB); if (error) return (error); if (ifr->ifr_fib >= rt_numfibs) return (EINVAL); ifp->if_fib = ifr->ifr_fib; break; case SIOCSIFFLAGS: error = priv_check(td, PRIV_NET_SETIFFLAGS); if (error) return (error); /* * Currently, no driver owned flags pass the IFF_CANTCHANGE * check, so we don't need special handling here yet. */ new_flags = (ifr->ifr_flags & 0xffff) | (ifr->ifr_flagshigh << 16); if (ifp->if_flags & IFF_SMART) { /* Smart drivers twiddle their own routes */ } else if (ifp->if_flags & IFF_UP && (new_flags & IFF_UP) == 0) { int s = splimp(); if_down(ifp); splx(s); } else if (new_flags & IFF_UP && (ifp->if_flags & IFF_UP) == 0) { int s = splimp(); if_up(ifp); splx(s); } /* See if permanently promiscuous mode bit is about to flip */ if ((ifp->if_flags ^ new_flags) & IFF_PPROMISC) { if (new_flags & IFF_PPROMISC) ifp->if_flags |= IFF_PROMISC; else if (ifp->if_pcount == 0) ifp->if_flags &= ~IFF_PROMISC; log(LOG_INFO, "%s: permanently promiscuous mode %s\n", ifp->if_xname, (new_flags & IFF_PPROMISC) ? "enabled" : "disabled"); } ifp->if_flags = (ifp->if_flags & IFF_CANTCHANGE) | (new_flags &~ IFF_CANTCHANGE); if (ifp->if_ioctl) { (void) (*ifp->if_ioctl)(ifp, cmd, data); } getmicrotime(&ifp->if_lastchange); break; case SIOCSIFCAP: error = priv_check(td, PRIV_NET_SETIFCAP); if (error) return (error); if (ifp->if_ioctl == NULL) return (EOPNOTSUPP); if (ifr->ifr_reqcap & ~ifp->if_capabilities) return (EINVAL); error = (*ifp->if_ioctl)(ifp, cmd, data); if (error == 0) getmicrotime(&ifp->if_lastchange); break; #ifdef MAC case SIOCSIFMAC: error = mac_ifnet_ioctl_set(td->td_ucred, ifr, ifp); break; #endif case SIOCSIFNAME: error = priv_check(td, PRIV_NET_SETIFNAME); if (error) return (error); error = copyinstr(ifr->ifr_data, new_name, IFNAMSIZ, NULL); if (error != 0) return (error); if (new_name[0] == '\0') return (EINVAL); if (ifunit(new_name) != NULL) return (EEXIST); /* * XXX: Locking. Nothing else seems to lock if_flags, * and there are numerous other races with the * ifunit() checks not being atomic with namespace * changes (renames, vmoves, if_attach, etc). */ ifp->if_flags |= IFF_RENAMING; /* Announce the departure of the interface. */ rt_ifannouncemsg(ifp, IFAN_DEPARTURE); EVENTHANDLER_INVOKE(ifnet_departure_event, ifp); log(LOG_INFO, "%s: changing name to '%s'\n", ifp->if_xname, new_name); strlcpy(ifp->if_xname, new_name, sizeof(ifp->if_xname)); ifa = ifp->if_addr; IFA_LOCK(ifa); sdl = (struct sockaddr_dl *)ifa->ifa_addr; namelen = strlen(new_name); onamelen = sdl->sdl_nlen; /* * Move the address if needed. This is safe because we * allocate space for a name of length IFNAMSIZ when we * create this in if_attach(). */ if (namelen != onamelen) { bcopy(sdl->sdl_data + onamelen, sdl->sdl_data + namelen, sdl->sdl_alen); } bcopy(new_name, sdl->sdl_data, namelen); sdl->sdl_nlen = namelen; sdl = (struct sockaddr_dl *)ifa->ifa_netmask; bzero(sdl->sdl_data, onamelen); while (namelen != 0) sdl->sdl_data[--namelen] = 0xff; IFA_UNLOCK(ifa); EVENTHANDLER_INVOKE(ifnet_arrival_event, ifp); /* Announce the return of the interface. */ rt_ifannouncemsg(ifp, IFAN_ARRIVAL); ifp->if_flags &= ~IFF_RENAMING; break; #ifdef VIMAGE case SIOCSIFVNET: error = priv_check(td, PRIV_NET_SETIFVNET); if (error) return (error); error = if_vmove_loan(td, ifp, ifr->ifr_name, ifr->ifr_jid); break; #endif case SIOCSIFMETRIC: error = priv_check(td, PRIV_NET_SETIFMETRIC); if (error) return (error); ifp->if_metric = ifr->ifr_metric; getmicrotime(&ifp->if_lastchange); break; case SIOCSIFPHYS: error = priv_check(td, PRIV_NET_SETIFPHYS); if (error) return (error); if (ifp->if_ioctl == NULL) return (EOPNOTSUPP); error = (*ifp->if_ioctl)(ifp, cmd, data); if (error == 0) getmicrotime(&ifp->if_lastchange); break; case SIOCSIFMTU: { u_long oldmtu = ifp->if_mtu; error = priv_check(td, PRIV_NET_SETIFMTU); if (error) return (error); if (ifr->ifr_mtu < IF_MINMTU || ifr->ifr_mtu > IF_MAXMTU) return (EINVAL); if (ifp->if_ioctl == NULL) return (EOPNOTSUPP); error = (*ifp->if_ioctl)(ifp, cmd, data); if (error == 0) { getmicrotime(&ifp->if_lastchange); rt_ifmsg(ifp); } /* * If the link MTU changed, do network layer specific procedure. */ if (ifp->if_mtu != oldmtu) { #ifdef INET6 nd6_setmtu(ifp); #endif } break; } case SIOCADDMULTI: case SIOCDELMULTI: if (cmd == SIOCADDMULTI) error = priv_check(td, PRIV_NET_ADDMULTI); else error = priv_check(td, PRIV_NET_DELMULTI); if (error) return (error); /* Don't allow group membership on non-multicast interfaces. */ if ((ifp->if_flags & IFF_MULTICAST) == 0) return (EOPNOTSUPP); /* Don't let users screw up protocols' entries. */ if (ifr->ifr_addr.sa_family != AF_LINK) return (EINVAL); if (cmd == SIOCADDMULTI) { struct ifmultiaddr *ifma; /* * Userland is only permitted to join groups once * via the if_addmulti() KPI, because it cannot hold * struct ifmultiaddr * between calls. It may also * lose a race while we check if the membership * already exists. */ IF_ADDR_RLOCK(ifp); ifma = if_findmulti(ifp, &ifr->ifr_addr); IF_ADDR_RUNLOCK(ifp); if (ifma != NULL) error = EADDRINUSE; else error = if_addmulti(ifp, &ifr->ifr_addr, &ifma); } else { error = if_delmulti(ifp, &ifr->ifr_addr); } if (error == 0) getmicrotime(&ifp->if_lastchange); break; case SIOCSIFPHYADDR: case SIOCDIFPHYADDR: #ifdef INET6 case SIOCSIFPHYADDR_IN6: #endif case SIOCSLIFPHYADDR: case SIOCSIFMEDIA: case SIOCSIFGENERIC: error = priv_check(td, PRIV_NET_HWIOCTL); if (error) return (error); if (ifp->if_ioctl == NULL) return (EOPNOTSUPP); error = (*ifp->if_ioctl)(ifp, cmd, data); if (error == 0) getmicrotime(&ifp->if_lastchange); break; case SIOCGIFSTATUS: ifs = (struct ifstat *)data; ifs->ascii[0] = '\0'; case SIOCGIFPSRCADDR: case SIOCGIFPDSTADDR: case SIOCGLIFPHYADDR: case SIOCGIFMEDIA: case SIOCGIFGENERIC: if (ifp->if_ioctl == NULL) return (EOPNOTSUPP); error = (*ifp->if_ioctl)(ifp, cmd, data); break; case SIOCSIFLLADDR: error = priv_check(td, PRIV_NET_SETLLADDR); if (error) return (error); error = if_setlladdr(ifp, ifr->ifr_addr.sa_data, ifr->ifr_addr.sa_len); EVENTHANDLER_INVOKE(iflladdr_event, ifp); break; case SIOCAIFGROUP: { struct ifgroupreq *ifgr = (struct ifgroupreq *)ifr; error = priv_check(td, PRIV_NET_ADDIFGROUP); if (error) return (error); if ((error = if_addgroup(ifp, ifgr->ifgr_group))) return (error); break; } case SIOCGIFGROUP: if ((error = if_getgroup((struct ifgroupreq *)ifr, ifp))) return (error); break; case SIOCDIFGROUP: { struct ifgroupreq *ifgr = (struct ifgroupreq *)ifr; error = priv_check(td, PRIV_NET_DELIFGROUP); if (error) return (error); if ((error = if_delgroup(ifp, ifgr->ifgr_group))) return (error); break; } default: error = ENOIOCTL; break; } return (error); } #ifdef COMPAT_FREEBSD32 struct ifconf32 { int32_t ifc_len; union { uint32_t ifcu_buf; uint32_t ifcu_req; } ifc_ifcu; }; #define SIOCGIFCONF32 _IOWR('i', 36, struct ifconf32) #endif /* * Interface ioctls. */ int ifioctl(struct socket *so, u_long cmd, caddr_t data, struct thread *td) { struct ifnet *ifp; struct ifreq *ifr; int error; int oif_flags; CURVNET_SET(so->so_vnet); switch (cmd) { case SIOCGIFCONF: case OSIOCGIFCONF: error = ifconf(cmd, data); CURVNET_RESTORE(); return (error); #ifdef COMPAT_FREEBSD32 case SIOCGIFCONF32: { struct ifconf32 *ifc32; struct ifconf ifc; ifc32 = (struct ifconf32 *)data; ifc.ifc_len = ifc32->ifc_len; ifc.ifc_buf = PTRIN(ifc32->ifc_buf); error = ifconf(SIOCGIFCONF, (void *)&ifc); CURVNET_RESTORE(); if (error == 0) ifc32->ifc_len = ifc.ifc_len; return (error); } #endif } ifr = (struct ifreq *)data; switch (cmd) { #ifdef VIMAGE case SIOCSIFRVNET: error = priv_check(td, PRIV_NET_SETIFVNET); if (error == 0) error = if_vmove_reclaim(td, ifr->ifr_name, ifr->ifr_jid); CURVNET_RESTORE(); return (error); #endif case SIOCIFCREATE: case SIOCIFCREATE2: error = priv_check(td, PRIV_NET_IFCREATE); if (error == 0) error = if_clone_create(ifr->ifr_name, sizeof(ifr->ifr_name), cmd == SIOCIFCREATE2 ? ifr->ifr_data : NULL); CURVNET_RESTORE(); return (error); case SIOCIFDESTROY: error = priv_check(td, PRIV_NET_IFDESTROY); if (error == 0) error = if_clone_destroy(ifr->ifr_name); CURVNET_RESTORE(); return (error); case SIOCIFGCLONERS: error = if_clone_list((struct if_clonereq *)data); CURVNET_RESTORE(); return (error); case SIOCGIFGMEMB: error = if_getgroupmembers((struct ifgroupreq *)data); CURVNET_RESTORE(); return (error); #if defined(INET) || defined(INET6) case SIOCSVH: case SIOCGVH: if (carp_ioctl_p == NULL) error = EPROTONOSUPPORT; else error = (*carp_ioctl_p)(ifr, cmd, td); CURVNET_RESTORE(); return (error); #endif } ifp = ifunit_ref(ifr->ifr_name); if (ifp == NULL) { CURVNET_RESTORE(); return (ENXIO); } error = ifhwioctl(cmd, ifp, data, td); if (error != ENOIOCTL) { if_rele(ifp); CURVNET_RESTORE(); return (error); } oif_flags = ifp->if_flags; if (so->so_proto == NULL) { if_rele(ifp); CURVNET_RESTORE(); return (EOPNOTSUPP); } #ifndef COMPAT_43 error = ((*so->so_proto->pr_usrreqs->pru_control)(so, cmd, data, ifp, td)); if (error == EOPNOTSUPP && ifp != NULL && ifp->if_ioctl != NULL) error = (*ifp->if_ioctl)(ifp, cmd, data); #else { u_long ocmd = cmd; switch (cmd) { case SIOCSIFDSTADDR: case SIOCSIFADDR: case SIOCSIFBRDADDR: case SIOCSIFNETMASK: #if BYTE_ORDER != BIG_ENDIAN if (ifr->ifr_addr.sa_family == 0 && ifr->ifr_addr.sa_len < 16) { ifr->ifr_addr.sa_family = ifr->ifr_addr.sa_len; ifr->ifr_addr.sa_len = 16; } #else if (ifr->ifr_addr.sa_len == 0) ifr->ifr_addr.sa_len = 16; #endif break; case OSIOCGIFADDR: cmd = SIOCGIFADDR; break; case OSIOCGIFDSTADDR: cmd = SIOCGIFDSTADDR; break; case OSIOCGIFBRDADDR: cmd = SIOCGIFBRDADDR; break; case OSIOCGIFNETMASK: cmd = SIOCGIFNETMASK; } error = ((*so->so_proto->pr_usrreqs->pru_control)(so, cmd, data, ifp, td)); if (error == EOPNOTSUPP && ifp != NULL && ifp->if_ioctl != NULL) error = (*ifp->if_ioctl)(ifp, cmd, data); switch (ocmd) { case OSIOCGIFADDR: case OSIOCGIFDSTADDR: case OSIOCGIFBRDADDR: case OSIOCGIFNETMASK: *(u_short *)&ifr->ifr_addr = ifr->ifr_addr.sa_family; } } #endif /* COMPAT_43 */ if ((oif_flags ^ ifp->if_flags) & IFF_UP) { #ifdef INET6 if (ifp->if_flags & IFF_UP) { int s = splimp(); in6_if_up(ifp); splx(s); } #endif } if_rele(ifp); CURVNET_RESTORE(); return (error); } /* * The code common to handling reference counted flags, * e.g., in ifpromisc() and if_allmulti(). * The "pflag" argument can specify a permanent mode flag to check, * such as IFF_PPROMISC for promiscuous mode; should be 0 if none. * * Only to be used on stack-owned flags, not driver-owned flags. */ static int if_setflag(struct ifnet *ifp, int flag, int pflag, int *refcount, int onswitch) { struct ifreq ifr; int error; int oldflags, oldcount; /* Sanity checks to catch programming errors */ KASSERT((flag & (IFF_DRV_OACTIVE|IFF_DRV_RUNNING)) == 0, ("%s: setting driver-owned flag %d", __func__, flag)); if (onswitch) KASSERT(*refcount >= 0, ("%s: increment negative refcount %d for flag %d", __func__, *refcount, flag)); else KASSERT(*refcount > 0, ("%s: decrement non-positive refcount %d for flag %d", __func__, *refcount, flag)); /* In case this mode is permanent, just touch refcount */ if (ifp->if_flags & pflag) { *refcount += onswitch ? 1 : -1; return (0); } /* Save ifnet parameters for if_ioctl() may fail */ oldcount = *refcount; oldflags = ifp->if_flags; /* * See if we aren't the only and touching refcount is enough. * Actually toggle interface flag if we are the first or last. */ if (onswitch) { if ((*refcount)++) return (0); ifp->if_flags |= flag; } else { if (--(*refcount)) return (0); ifp->if_flags &= ~flag; } /* Call down the driver since we've changed interface flags */ if (ifp->if_ioctl == NULL) { error = EOPNOTSUPP; goto recover; } ifr.ifr_flags = ifp->if_flags & 0xffff; ifr.ifr_flagshigh = ifp->if_flags >> 16; error = (*ifp->if_ioctl)(ifp, SIOCSIFFLAGS, (caddr_t)&ifr); if (error) goto recover; /* Notify userland that interface flags have changed */ rt_ifmsg(ifp); return (0); recover: /* Recover after driver error */ *refcount = oldcount; ifp->if_flags = oldflags; return (error); } /* * Set/clear promiscuous mode on interface ifp based on the truth value * of pswitch. The calls are reference counted so that only the first * "on" request actually has an effect, as does the final "off" request. * Results are undefined if the "off" and "on" requests are not matched. */ int ifpromisc(struct ifnet *ifp, int pswitch) { int error; int oldflags = ifp->if_flags; error = if_setflag(ifp, IFF_PROMISC, IFF_PPROMISC, &ifp->if_pcount, pswitch); /* If promiscuous mode status has changed, log a message */ if (error == 0 && ((ifp->if_flags ^ oldflags) & IFF_PROMISC)) log(LOG_INFO, "%s: promiscuous mode %s\n", ifp->if_xname, (ifp->if_flags & IFF_PROMISC) ? "enabled" : "disabled"); return (error); } /* * Return interface configuration * of system. List may be used * in later ioctl's (above) to get * other information. */ /*ARGSUSED*/ static int ifconf(u_long cmd, caddr_t data) { struct ifconf *ifc = (struct ifconf *)data; struct ifnet *ifp; struct ifaddr *ifa; struct ifreq ifr; struct sbuf *sb; int error, full = 0, valid_len, max_len; /* Limit initial buffer size to MAXPHYS to avoid DoS from userspace. */ max_len = MAXPHYS - 1; /* Prevent hostile input from being able to crash the system */ if (ifc->ifc_len <= 0) return (EINVAL); again: if (ifc->ifc_len <= max_len) { max_len = ifc->ifc_len; full = 1; } sb = sbuf_new(NULL, NULL, max_len + 1, SBUF_FIXEDLEN); max_len = 0; valid_len = 0; IFNET_RLOCK(); TAILQ_FOREACH(ifp, &V_ifnet, if_link) { int addrs; /* * Zero the ifr_name buffer to make sure we don't * disclose the contents of the stack. */ memset(ifr.ifr_name, 0, sizeof(ifr.ifr_name)); if (strlcpy(ifr.ifr_name, ifp->if_xname, sizeof(ifr.ifr_name)) >= sizeof(ifr.ifr_name)) { sbuf_delete(sb); IFNET_RUNLOCK(); return (ENAMETOOLONG); } addrs = 0; IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { struct sockaddr *sa = ifa->ifa_addr; if (prison_if(curthread->td_ucred, sa) != 0) continue; addrs++; #ifdef COMPAT_43 if (cmd == OSIOCGIFCONF) { struct osockaddr *osa = (struct osockaddr *)&ifr.ifr_addr; ifr.ifr_addr = *sa; osa->sa_family = sa->sa_family; sbuf_bcat(sb, &ifr, sizeof(ifr)); max_len += sizeof(ifr); } else #endif if (sa->sa_len <= sizeof(*sa)) { ifr.ifr_addr = *sa; sbuf_bcat(sb, &ifr, sizeof(ifr)); max_len += sizeof(ifr); } else { sbuf_bcat(sb, &ifr, offsetof(struct ifreq, ifr_addr)); max_len += offsetof(struct ifreq, ifr_addr); sbuf_bcat(sb, sa, sa->sa_len); max_len += sa->sa_len; } if (sbuf_error(sb) == 0) valid_len = sbuf_len(sb); } IF_ADDR_RUNLOCK(ifp); if (addrs == 0) { bzero((caddr_t)&ifr.ifr_addr, sizeof(ifr.ifr_addr)); sbuf_bcat(sb, &ifr, sizeof(ifr)); max_len += sizeof(ifr); if (sbuf_error(sb) == 0) valid_len = sbuf_len(sb); } } IFNET_RUNLOCK(); /* * If we didn't allocate enough space (uncommon), try again. If * we have already allocated as much space as we are allowed, * return what we've got. */ if (valid_len != max_len && !full) { sbuf_delete(sb); goto again; } ifc->ifc_len = valid_len; sbuf_finish(sb); error = copyout(sbuf_data(sb), ifc->ifc_req, ifc->ifc_len); sbuf_delete(sb); return (error); } /* * Just like ifpromisc(), but for all-multicast-reception mode. */ int if_allmulti(struct ifnet *ifp, int onswitch) { return (if_setflag(ifp, IFF_ALLMULTI, 0, &ifp->if_amcount, onswitch)); } struct ifmultiaddr * if_findmulti(struct ifnet *ifp, struct sockaddr *sa) { struct ifmultiaddr *ifma; IF_ADDR_LOCK_ASSERT(ifp); TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) { if (sa->sa_family == AF_LINK) { if (sa_dl_equal(ifma->ifma_addr, sa)) break; } else { if (sa_equal(ifma->ifma_addr, sa)) break; } } return ifma; } /* * Allocate a new ifmultiaddr and initialize based on passed arguments. We * make copies of passed sockaddrs. The ifmultiaddr will not be added to * the ifnet multicast address list here, so the caller must do that and * other setup work (such as notifying the device driver). The reference * count is initialized to 1. */ static struct ifmultiaddr * if_allocmulti(struct ifnet *ifp, struct sockaddr *sa, struct sockaddr *llsa, int mflags) { struct ifmultiaddr *ifma; struct sockaddr *dupsa; ifma = malloc(sizeof *ifma, M_IFMADDR, mflags | M_ZERO); if (ifma == NULL) return (NULL); dupsa = malloc(sa->sa_len, M_IFMADDR, mflags); if (dupsa == NULL) { free(ifma, M_IFMADDR); return (NULL); } bcopy(sa, dupsa, sa->sa_len); ifma->ifma_addr = dupsa; ifma->ifma_ifp = ifp; ifma->ifma_refcount = 1; ifma->ifma_protospec = NULL; if (llsa == NULL) { ifma->ifma_lladdr = NULL; return (ifma); } dupsa = malloc(llsa->sa_len, M_IFMADDR, mflags); if (dupsa == NULL) { free(ifma->ifma_addr, M_IFMADDR); free(ifma, M_IFMADDR); return (NULL); } bcopy(llsa, dupsa, llsa->sa_len); ifma->ifma_lladdr = dupsa; return (ifma); } /* * if_freemulti: free ifmultiaddr structure and possibly attached related * addresses. The caller is responsible for implementing reference * counting, notifying the driver, handling routing messages, and releasing * any dependent link layer state. */ static void if_freemulti(struct ifmultiaddr *ifma) { KASSERT(ifma->ifma_refcount == 0, ("if_freemulti: refcount %d", ifma->ifma_refcount)); KASSERT(ifma->ifma_protospec == NULL, ("if_freemulti: protospec not NULL")); if (ifma->ifma_lladdr != NULL) free(ifma->ifma_lladdr, M_IFMADDR); free(ifma->ifma_addr, M_IFMADDR); free(ifma, M_IFMADDR); } /* * Register an additional multicast address with a network interface. * * - If the address is already present, bump the reference count on the * address and return. * - If the address is not link-layer, look up a link layer address. * - Allocate address structures for one or both addresses, and attach to the * multicast address list on the interface. If automatically adding a link * layer address, the protocol address will own a reference to the link * layer address, to be freed when it is freed. * - Notify the network device driver of an addition to the multicast address * list. * * 'sa' points to caller-owned memory with the desired multicast address. * * 'retifma' will be used to return a pointer to the resulting multicast * address reference, if desired. */ int if_addmulti(struct ifnet *ifp, struct sockaddr *sa, struct ifmultiaddr **retifma) { struct ifmultiaddr *ifma, *ll_ifma; struct sockaddr *llsa; int error; /* * If the address is already present, return a new reference to it; * otherwise, allocate storage and set up a new address. */ IF_ADDR_WLOCK(ifp); ifma = if_findmulti(ifp, sa); if (ifma != NULL) { ifma->ifma_refcount++; if (retifma != NULL) *retifma = ifma; IF_ADDR_WUNLOCK(ifp); return (0); } /* * The address isn't already present; resolve the protocol address * into a link layer address, and then look that up, bump its * refcount or allocate an ifma for that also. If 'llsa' was * returned, we will need to free it later. */ llsa = NULL; ll_ifma = NULL; if (ifp->if_resolvemulti != NULL) { error = ifp->if_resolvemulti(ifp, &llsa, sa); if (error) goto unlock_out; } /* * Allocate the new address. Don't hook it up yet, as we may also * need to allocate a link layer multicast address. */ ifma = if_allocmulti(ifp, sa, llsa, M_NOWAIT); if (ifma == NULL) { error = ENOMEM; goto free_llsa_out; } /* * If a link layer address is found, we'll need to see if it's * already present in the address list, or allocate is as well. * When this block finishes, the link layer address will be on the * list. */ if (llsa != NULL) { ll_ifma = if_findmulti(ifp, llsa); if (ll_ifma == NULL) { ll_ifma = if_allocmulti(ifp, llsa, NULL, M_NOWAIT); if (ll_ifma == NULL) { --ifma->ifma_refcount; if_freemulti(ifma); error = ENOMEM; goto free_llsa_out; } TAILQ_INSERT_HEAD(&ifp->if_multiaddrs, ll_ifma, ifma_link); } else ll_ifma->ifma_refcount++; ifma->ifma_llifma = ll_ifma; } /* * We now have a new multicast address, ifma, and possibly a new or * referenced link layer address. Add the primary address to the * ifnet address list. */ TAILQ_INSERT_HEAD(&ifp->if_multiaddrs, ifma, ifma_link); if (retifma != NULL) *retifma = ifma; /* * Must generate the message while holding the lock so that 'ifma' * pointer is still valid. */ rt_newmaddrmsg(RTM_NEWMADDR, ifma); IF_ADDR_WUNLOCK(ifp); /* * We are certain we have added something, so call down to the * interface to let them know about it. */ if (ifp->if_ioctl != NULL) { (void) (*ifp->if_ioctl)(ifp, SIOCADDMULTI, 0); } if (llsa != NULL) free(llsa, M_IFMADDR); return (0); free_llsa_out: if (llsa != NULL) free(llsa, M_IFMADDR); unlock_out: IF_ADDR_WUNLOCK(ifp); return (error); } /* * Delete a multicast group membership by network-layer group address. * * Returns ENOENT if the entry could not be found. If ifp no longer * exists, results are undefined. This entry point should only be used * from subsystems which do appropriate locking to hold ifp for the * duration of the call. * Network-layer protocol domains must use if_delmulti_ifma(). */ int if_delmulti(struct ifnet *ifp, struct sockaddr *sa) { struct ifmultiaddr *ifma; int lastref; #ifdef INVARIANTS struct ifnet *oifp; IFNET_RLOCK_NOSLEEP(); TAILQ_FOREACH(oifp, &V_ifnet, if_link) if (ifp == oifp) break; if (ifp != oifp) ifp = NULL; IFNET_RUNLOCK_NOSLEEP(); KASSERT(ifp != NULL, ("%s: ifnet went away", __func__)); #endif if (ifp == NULL) return (ENOENT); IF_ADDR_WLOCK(ifp); lastref = 0; ifma = if_findmulti(ifp, sa); if (ifma != NULL) lastref = if_delmulti_locked(ifp, ifma, 0); IF_ADDR_WUNLOCK(ifp); if (ifma == NULL) return (ENOENT); if (lastref && ifp->if_ioctl != NULL) { (void)(*ifp->if_ioctl)(ifp, SIOCDELMULTI, 0); } return (0); } /* * Delete all multicast group membership for an interface. * Should be used to quickly flush all multicast filters. */ void if_delallmulti(struct ifnet *ifp) { struct ifmultiaddr *ifma; struct ifmultiaddr *next; IF_ADDR_WLOCK(ifp); TAILQ_FOREACH_SAFE(ifma, &ifp->if_multiaddrs, ifma_link, next) if_delmulti_locked(ifp, ifma, 0); IF_ADDR_WUNLOCK(ifp); } /* * Delete a multicast group membership by group membership pointer. * Network-layer protocol domains must use this routine. * * It is safe to call this routine if the ifp disappeared. */ void if_delmulti_ifma(struct ifmultiaddr *ifma) { struct ifnet *ifp; int lastref; ifp = ifma->ifma_ifp; #ifdef DIAGNOSTIC if (ifp == NULL) { printf("%s: ifma_ifp seems to be detached\n", __func__); } else { struct ifnet *oifp; IFNET_RLOCK_NOSLEEP(); TAILQ_FOREACH(oifp, &V_ifnet, if_link) if (ifp == oifp) break; if (ifp != oifp) { printf("%s: ifnet %p disappeared\n", __func__, ifp); ifp = NULL; } IFNET_RUNLOCK_NOSLEEP(); } #endif /* * If and only if the ifnet instance exists: Acquire the address lock. */ if (ifp != NULL) IF_ADDR_WLOCK(ifp); lastref = if_delmulti_locked(ifp, ifma, 0); if (ifp != NULL) { /* * If and only if the ifnet instance exists: * Release the address lock. * If the group was left: update the hardware hash filter. */ IF_ADDR_WUNLOCK(ifp); if (lastref && ifp->if_ioctl != NULL) { (void)(*ifp->if_ioctl)(ifp, SIOCDELMULTI, 0); } } } /* * Perform deletion of network-layer and/or link-layer multicast address. * * Return 0 if the reference count was decremented. * Return 1 if the final reference was released, indicating that the * hardware hash filter should be reprogrammed. */ static int if_delmulti_locked(struct ifnet *ifp, struct ifmultiaddr *ifma, int detaching) { struct ifmultiaddr *ll_ifma; if (ifp != NULL && ifma->ifma_ifp != NULL) { KASSERT(ifma->ifma_ifp == ifp, ("%s: inconsistent ifp %p", __func__, ifp)); IF_ADDR_WLOCK_ASSERT(ifp); } ifp = ifma->ifma_ifp; /* * If the ifnet is detaching, null out references to ifnet, * so that upper protocol layers will notice, and not attempt * to obtain locks for an ifnet which no longer exists. The * routing socket announcement must happen before the ifnet * instance is detached from the system. */ if (detaching) { #ifdef DIAGNOSTIC printf("%s: detaching ifnet instance %p\n", __func__, ifp); #endif /* * ifp may already be nulled out if we are being reentered * to delete the ll_ifma. */ if (ifp != NULL) { rt_newmaddrmsg(RTM_DELMADDR, ifma); ifma->ifma_ifp = NULL; } } if (--ifma->ifma_refcount > 0) return 0; /* * If this ifma is a network-layer ifma, a link-layer ifma may * have been associated with it. Release it first if so. */ ll_ifma = ifma->ifma_llifma; if (ll_ifma != NULL) { KASSERT(ifma->ifma_lladdr != NULL, ("%s: llifma w/o lladdr", __func__)); if (detaching) ll_ifma->ifma_ifp = NULL; /* XXX */ if (--ll_ifma->ifma_refcount == 0) { if (ifp != NULL) { TAILQ_REMOVE(&ifp->if_multiaddrs, ll_ifma, ifma_link); } if_freemulti(ll_ifma); } } if (ifp != NULL) TAILQ_REMOVE(&ifp->if_multiaddrs, ifma, ifma_link); if_freemulti(ifma); /* * The last reference to this instance of struct ifmultiaddr * was released; the hardware should be notified of this change. */ return 1; } /* * Set the link layer address on an interface. * * At this time we only support certain types of interfaces, * and we don't allow the length of the address to change. */ int if_setlladdr(struct ifnet *ifp, const u_char *lladdr, int len) { struct sockaddr_dl *sdl; struct ifaddr *ifa; struct ifreq ifr; IF_ADDR_RLOCK(ifp); ifa = ifp->if_addr; if (ifa == NULL) { IF_ADDR_RUNLOCK(ifp); return (EINVAL); } ifa_ref(ifa); IF_ADDR_RUNLOCK(ifp); sdl = (struct sockaddr_dl *)ifa->ifa_addr; if (sdl == NULL) { ifa_free(ifa); return (EINVAL); } if (len != sdl->sdl_alen) { /* don't allow length to change */ ifa_free(ifa); return (EINVAL); } switch (ifp->if_type) { case IFT_ETHER: case IFT_FDDI: case IFT_XETHER: case IFT_ISO88025: case IFT_L2VLAN: case IFT_BRIDGE: case IFT_ARCNET: case IFT_IEEE8023ADLAG: case IFT_IEEE80211: bcopy(lladdr, LLADDR(sdl), len); ifa_free(ifa); break; default: ifa_free(ifa); return (ENODEV); } /* * If the interface is already up, we need * to re-init it in order to reprogram its * address filter. */ if ((ifp->if_flags & IFF_UP) != 0) { if (ifp->if_ioctl) { ifp->if_flags &= ~IFF_UP; ifr.ifr_flags = ifp->if_flags & 0xffff; ifr.ifr_flagshigh = ifp->if_flags >> 16; (*ifp->if_ioctl)(ifp, SIOCSIFFLAGS, (caddr_t)&ifr); ifp->if_flags |= IFF_UP; ifr.ifr_flags = ifp->if_flags & 0xffff; ifr.ifr_flagshigh = ifp->if_flags >> 16; (*ifp->if_ioctl)(ifp, SIOCSIFFLAGS, (caddr_t)&ifr); } #ifdef INET /* * Also send gratuitous ARPs to notify other nodes about * the address change. */ TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { if (ifa->ifa_addr->sa_family == AF_INET) arp_ifinit(ifp, ifa); } #endif } return (0); } /* * The name argument must be a pointer to storage which will last as * long as the interface does. For physical devices, the result of * device_get_name(dev) is a good choice and for pseudo-devices a * static string works well. */ void if_initname(struct ifnet *ifp, const char *name, int unit) { ifp->if_dname = name; ifp->if_dunit = unit; if (unit != IF_DUNIT_NONE) snprintf(ifp->if_xname, IFNAMSIZ, "%s%d", name, unit); else strlcpy(ifp->if_xname, name, IFNAMSIZ); } int if_printf(struct ifnet *ifp, const char * fmt, ...) { va_list ap; int retval; retval = printf("%s: ", ifp->if_xname); va_start(ap, fmt); retval += vprintf(fmt, ap); va_end(ap); return (retval); } void if_start(struct ifnet *ifp) { (*(ifp)->if_start)(ifp); } /* * Backwards compatibility interface for drivers * that have not implemented it */ static int if_transmit(struct ifnet *ifp, struct mbuf *m) { int error; IFQ_HANDOFF(ifp, m, error); return (error); } int if_handoff(struct ifqueue *ifq, struct mbuf *m, struct ifnet *ifp, int adjust) { int active = 0; IF_LOCK(ifq); if (_IF_QFULL(ifq)) { _IF_DROP(ifq); IF_UNLOCK(ifq); m_freem(m); return (0); } if (ifp != NULL) { ifp->if_obytes += m->m_pkthdr.len + adjust; if (m->m_flags & (M_BCAST|M_MCAST)) ifp->if_omcasts++; active = ifp->if_drv_flags & IFF_DRV_OACTIVE; } _IF_ENQUEUE(ifq, m); IF_UNLOCK(ifq); if (ifp != NULL && !active) (*(ifp)->if_start)(ifp); return (1); } void if_register_com_alloc(u_char type, if_com_alloc_t *a, if_com_free_t *f) { KASSERT(if_com_alloc[type] == NULL, ("if_register_com_alloc: %d already registered", type)); KASSERT(if_com_free[type] == NULL, ("if_register_com_alloc: %d free already registered", type)); if_com_alloc[type] = a; if_com_free[type] = f; } void if_deregister_com_alloc(u_char type) { KASSERT(if_com_alloc[type] != NULL, ("if_deregister_com_alloc: %d not registered", type)); KASSERT(if_com_free[type] != NULL, ("if_deregister_com_alloc: %d free not registered", type)); if_com_alloc[type] = NULL; if_com_free[type] = NULL; } Index: head/sys/net/if_bridge.c =================================================================== --- head/sys/net/if_bridge.c (revision 234486) +++ head/sys/net/if_bridge.c (revision 234487) @@ -1,3477 +1,3518 @@ /* $NetBSD: if_bridge.c,v 1.31 2005/06/01 19:45:34 jdc Exp $ */ /* * Copyright 2001 Wasabi Systems, Inc. * All rights reserved. * * Written by Jason R. Thorpe for Wasabi Systems, Inc. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed for the NetBSD Project by * Wasabi Systems, Inc. * 4. The name of Wasabi Systems, Inc. may not be used to endorse * or promote products derived from this software without specific prior * written permission. * * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL WASABI SYSTEMS, INC * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. */ /* * Copyright (c) 1999, 2000 Jason L. Wright (jason@thought.net) * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE * DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * OpenBSD: if_bridge.c,v 1.60 2001/06/15 03:38:33 itojun Exp */ /* * Network interface bridge support. * * TODO: * * - Currently only supports Ethernet-like interfaces (Ethernet, * 802.11, VLANs on Ethernet, etc.) Figure out a nice way * to bridge other types of interfaces (FDDI-FDDI, and maybe * consider heterogenous bridges). */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #include #include #include #include #include #include #include #include /* for net/if.h */ #include #include /* string functions */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* for struct arpcom */ #include #include #include #include #ifdef INET6 #include #include #endif #if defined(INET) || defined(INET6) #include #endif #include #include /* for struct arpcom */ #include #include #include #include #include #include #include /* * Size of the route hash table. Must be a power of two. */ #ifndef BRIDGE_RTHASH_SIZE #define BRIDGE_RTHASH_SIZE 1024 #endif #define BRIDGE_RTHASH_MASK (BRIDGE_RTHASH_SIZE - 1) /* * Default maximum number of addresses to cache. */ #ifndef BRIDGE_RTABLE_MAX #define BRIDGE_RTABLE_MAX 2000 #endif /* * Timeout (in seconds) for entries learned dynamically. */ #ifndef BRIDGE_RTABLE_TIMEOUT #define BRIDGE_RTABLE_TIMEOUT (20 * 60) /* same as ARP */ #endif /* * Number of seconds between walks of the route list. */ #ifndef BRIDGE_RTABLE_PRUNE_PERIOD #define BRIDGE_RTABLE_PRUNE_PERIOD (5 * 60) #endif /* * List of capabilities to possibly mask on the member interface. */ #define BRIDGE_IFCAPS_MASK (IFCAP_TOE|IFCAP_TSO|IFCAP_TXCSUM) /* * List of capabilities to strip */ #define BRIDGE_IFCAPS_STRIP IFCAP_LRO /* * Bridge interface list entry. */ struct bridge_iflist { LIST_ENTRY(bridge_iflist) bif_next; struct ifnet *bif_ifp; /* member if */ struct bstp_port bif_stp; /* STP state */ uint32_t bif_flags; /* member if flags */ int bif_savedcaps; /* saved capabilities */ uint32_t bif_addrmax; /* max # of addresses */ uint32_t bif_addrcnt; /* cur. # of addresses */ uint32_t bif_addrexceeded;/* # of address violations */ }; /* * Bridge route node. */ struct bridge_rtnode { LIST_ENTRY(bridge_rtnode) brt_hash; /* hash table linkage */ LIST_ENTRY(bridge_rtnode) brt_list; /* list linkage */ struct bridge_iflist *brt_dst; /* destination if */ unsigned long brt_expire; /* expiration time */ uint8_t brt_flags; /* address flags */ uint8_t brt_addr[ETHER_ADDR_LEN]; uint16_t brt_vlan; /* vlan id */ }; #define brt_ifp brt_dst->bif_ifp /* * Software state for each bridge. */ struct bridge_softc { struct ifnet *sc_ifp; /* make this an interface */ LIST_ENTRY(bridge_softc) sc_list; struct mtx sc_mtx; struct cv sc_cv; uint32_t sc_brtmax; /* max # of addresses */ uint32_t sc_brtcnt; /* cur. # of addresses */ uint32_t sc_brttimeout; /* rt timeout in seconds */ struct callout sc_brcallout; /* bridge callout */ uint32_t sc_iflist_ref; /* refcount for sc_iflist */ uint32_t sc_iflist_xcnt; /* refcount for sc_iflist */ LIST_HEAD(, bridge_iflist) sc_iflist; /* member interface list */ LIST_HEAD(, bridge_rtnode) *sc_rthash; /* our forwarding table */ LIST_HEAD(, bridge_rtnode) sc_rtlist; /* list version of above */ uint32_t sc_rthash_key; /* key for hash */ LIST_HEAD(, bridge_iflist) sc_spanlist; /* span ports list */ struct bstp_state sc_stp; /* STP state */ uint32_t sc_brtexceeded; /* # of cache drops */ struct ifnet *sc_ifaddr; /* member mac copied from */ u_char sc_defaddr[6]; /* Default MAC address */ }; static struct mtx bridge_list_mtx; eventhandler_tag bridge_detach_cookie = NULL; int bridge_rtable_prune_period = BRIDGE_RTABLE_PRUNE_PERIOD; uma_zone_t bridge_rtnode_zone; static int bridge_clone_create(struct if_clone *, int, caddr_t); static void bridge_clone_destroy(struct ifnet *); static int bridge_ioctl(struct ifnet *, u_long, caddr_t); static void bridge_mutecaps(struct bridge_softc *); static void bridge_set_ifcap(struct bridge_softc *, struct bridge_iflist *, int); static void bridge_ifdetach(void *arg __unused, struct ifnet *); static void bridge_init(void *); static void bridge_dummynet(struct mbuf *, struct ifnet *); static void bridge_stop(struct ifnet *, int); static void bridge_start(struct ifnet *); static struct mbuf *bridge_input(struct ifnet *, struct mbuf *); static int bridge_output(struct ifnet *, struct mbuf *, struct sockaddr *, struct rtentry *); static void bridge_enqueue(struct bridge_softc *, struct ifnet *, struct mbuf *); static void bridge_rtdelete(struct bridge_softc *, struct ifnet *ifp, int); static void bridge_forward(struct bridge_softc *, struct bridge_iflist *, struct mbuf *m); static void bridge_timer(void *); static void bridge_broadcast(struct bridge_softc *, struct ifnet *, struct mbuf *, int); static void bridge_span(struct bridge_softc *, struct mbuf *); static int bridge_rtupdate(struct bridge_softc *, const uint8_t *, uint16_t, struct bridge_iflist *, int, uint8_t); static struct ifnet *bridge_rtlookup(struct bridge_softc *, const uint8_t *, uint16_t); static void bridge_rttrim(struct bridge_softc *); static void bridge_rtage(struct bridge_softc *); static void bridge_rtflush(struct bridge_softc *, int); static int bridge_rtdaddr(struct bridge_softc *, const uint8_t *, uint16_t); static int bridge_rtable_init(struct bridge_softc *); static void bridge_rtable_fini(struct bridge_softc *); static int bridge_rtnode_addr_cmp(const uint8_t *, const uint8_t *); static struct bridge_rtnode *bridge_rtnode_lookup(struct bridge_softc *, const uint8_t *, uint16_t); static int bridge_rtnode_insert(struct bridge_softc *, struct bridge_rtnode *); static void bridge_rtnode_destroy(struct bridge_softc *, struct bridge_rtnode *); static void bridge_rtable_expire(struct ifnet *, int); static void bridge_state_change(struct ifnet *, int); static struct bridge_iflist *bridge_lookup_member(struct bridge_softc *, const char *name); static struct bridge_iflist *bridge_lookup_member_if(struct bridge_softc *, struct ifnet *ifp); static void bridge_delete_member(struct bridge_softc *, struct bridge_iflist *, int); static void bridge_delete_span(struct bridge_softc *, struct bridge_iflist *); static int bridge_ioctl_add(struct bridge_softc *, void *); static int bridge_ioctl_del(struct bridge_softc *, void *); static int bridge_ioctl_gifflags(struct bridge_softc *, void *); static int bridge_ioctl_sifflags(struct bridge_softc *, void *); static int bridge_ioctl_scache(struct bridge_softc *, void *); static int bridge_ioctl_gcache(struct bridge_softc *, void *); static int bridge_ioctl_gifs(struct bridge_softc *, void *); static int bridge_ioctl_rts(struct bridge_softc *, void *); static int bridge_ioctl_saddr(struct bridge_softc *, void *); static int bridge_ioctl_sto(struct bridge_softc *, void *); static int bridge_ioctl_gto(struct bridge_softc *, void *); static int bridge_ioctl_daddr(struct bridge_softc *, void *); static int bridge_ioctl_flush(struct bridge_softc *, void *); static int bridge_ioctl_gpri(struct bridge_softc *, void *); static int bridge_ioctl_spri(struct bridge_softc *, void *); static int bridge_ioctl_ght(struct bridge_softc *, void *); static int bridge_ioctl_sht(struct bridge_softc *, void *); static int bridge_ioctl_gfd(struct bridge_softc *, void *); static int bridge_ioctl_sfd(struct bridge_softc *, void *); static int bridge_ioctl_gma(struct bridge_softc *, void *); static int bridge_ioctl_sma(struct bridge_softc *, void *); static int bridge_ioctl_sifprio(struct bridge_softc *, void *); static int bridge_ioctl_sifcost(struct bridge_softc *, void *); static int bridge_ioctl_sifmaxaddr(struct bridge_softc *, void *); static int bridge_ioctl_addspan(struct bridge_softc *, void *); static int bridge_ioctl_delspan(struct bridge_softc *, void *); static int bridge_ioctl_gbparam(struct bridge_softc *, void *); static int bridge_ioctl_grte(struct bridge_softc *, void *); static int bridge_ioctl_gifsstp(struct bridge_softc *, void *); static int bridge_ioctl_sproto(struct bridge_softc *, void *); static int bridge_ioctl_stxhc(struct bridge_softc *, void *); static int bridge_pfil(struct mbuf **, struct ifnet *, struct ifnet *, int); static int bridge_ip_checkbasic(struct mbuf **mp); #ifdef INET6 static int bridge_ip6_checkbasic(struct mbuf **mp); #endif /* INET6 */ static int bridge_fragment(struct ifnet *, struct mbuf *, struct ether_header *, int, struct llc *); +static void bridge_linkstate(struct ifnet *ifp); +extern void (*bridge_linkstate_p)(struct ifnet *ifp); + /* The default bridge vlan is 1 (IEEE 802.1Q-2003 Table 9-2) */ #define VLANTAGOF(_m) \ (_m->m_flags & M_VLANTAG) ? EVL_VLANOFTAG(_m->m_pkthdr.ether_vtag) : 1 static struct bstp_cb_ops bridge_ops = { .bcb_state = bridge_state_change, .bcb_rtage = bridge_rtable_expire }; SYSCTL_DECL(_net_link); static SYSCTL_NODE(_net_link, IFT_BRIDGE, bridge, CTLFLAG_RW, 0, "Bridge"); static int pfil_onlyip = 1; /* only pass IP[46] packets when pfil is enabled */ static int pfil_bridge = 1; /* run pfil hooks on the bridge interface */ static int pfil_member = 1; /* run pfil hooks on the member interface */ static int pfil_ipfw = 0; /* layer2 filter with ipfw */ static int pfil_ipfw_arp = 0; /* layer2 filter with ipfw */ static int pfil_local_phys = 0; /* run pfil hooks on the physical interface for locally destined packets */ static int log_stp = 0; /* log STP state changes */ static int bridge_inherit_mac = 0; /* share MAC with first bridge member */ TUNABLE_INT("net.link.bridge.pfil_onlyip", &pfil_onlyip); SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_onlyip, CTLFLAG_RW, &pfil_onlyip, 0, "Only pass IP packets when pfil is enabled"); TUNABLE_INT("net.link.bridge.ipfw_arp", &pfil_ipfw_arp); SYSCTL_INT(_net_link_bridge, OID_AUTO, ipfw_arp, CTLFLAG_RW, &pfil_ipfw_arp, 0, "Filter ARP packets through IPFW layer2"); TUNABLE_INT("net.link.bridge.pfil_bridge", &pfil_bridge); SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_bridge, CTLFLAG_RW, &pfil_bridge, 0, "Packet filter on the bridge interface"); TUNABLE_INT("net.link.bridge.pfil_member", &pfil_member); SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_member, CTLFLAG_RW, &pfil_member, 0, "Packet filter on the member interface"); TUNABLE_INT("net.link.bridge.pfil_local_phys", &pfil_local_phys); SYSCTL_INT(_net_link_bridge, OID_AUTO, pfil_local_phys, CTLFLAG_RW, &pfil_local_phys, 0, "Packet filter on the physical interface for locally destined packets"); TUNABLE_INT("net.link.bridge.log_stp", &log_stp); SYSCTL_INT(_net_link_bridge, OID_AUTO, log_stp, CTLFLAG_RW, &log_stp, 0, "Log STP state changes"); TUNABLE_INT("net.link.bridge.inherit_mac", &bridge_inherit_mac); SYSCTL_INT(_net_link_bridge, OID_AUTO, inherit_mac, CTLFLAG_RW, &bridge_inherit_mac, 0, "Inherit MAC address from the first bridge member"); struct bridge_control { int (*bc_func)(struct bridge_softc *, void *); int bc_argsize; int bc_flags; }; #define BC_F_COPYIN 0x01 /* copy arguments in */ #define BC_F_COPYOUT 0x02 /* copy arguments out */ #define BC_F_SUSER 0x04 /* do super-user check */ const struct bridge_control bridge_control_table[] = { { bridge_ioctl_add, sizeof(struct ifbreq), BC_F_COPYIN|BC_F_SUSER }, { bridge_ioctl_del, sizeof(struct ifbreq), BC_F_COPYIN|BC_F_SUSER }, { bridge_ioctl_gifflags, sizeof(struct ifbreq), BC_F_COPYIN|BC_F_COPYOUT }, { bridge_ioctl_sifflags, sizeof(struct ifbreq), BC_F_COPYIN|BC_F_SUSER }, { bridge_ioctl_scache, sizeof(struct ifbrparam), BC_F_COPYIN|BC_F_SUSER }, { bridge_ioctl_gcache, sizeof(struct ifbrparam), BC_F_COPYOUT }, { bridge_ioctl_gifs, sizeof(struct ifbifconf), BC_F_COPYIN|BC_F_COPYOUT }, { bridge_ioctl_rts, sizeof(struct ifbaconf), BC_F_COPYIN|BC_F_COPYOUT }, { bridge_ioctl_saddr, sizeof(struct ifbareq), BC_F_COPYIN|BC_F_SUSER }, { bridge_ioctl_sto, sizeof(struct ifbrparam), BC_F_COPYIN|BC_F_SUSER }, { bridge_ioctl_gto, sizeof(struct ifbrparam), BC_F_COPYOUT }, { bridge_ioctl_daddr, sizeof(struct ifbareq), BC_F_COPYIN|BC_F_SUSER }, { bridge_ioctl_flush, sizeof(struct ifbreq), BC_F_COPYIN|BC_F_SUSER }, { bridge_ioctl_gpri, sizeof(struct ifbrparam), BC_F_COPYOUT }, { bridge_ioctl_spri, sizeof(struct ifbrparam), BC_F_COPYIN|BC_F_SUSER }, { bridge_ioctl_ght, sizeof(struct ifbrparam), BC_F_COPYOUT }, { bridge_ioctl_sht, sizeof(struct ifbrparam), BC_F_COPYIN|BC_F_SUSER }, { bridge_ioctl_gfd, sizeof(struct ifbrparam), BC_F_COPYOUT }, { bridge_ioctl_sfd, sizeof(struct ifbrparam), BC_F_COPYIN|BC_F_SUSER }, { bridge_ioctl_gma, sizeof(struct ifbrparam), BC_F_COPYOUT }, { bridge_ioctl_sma, sizeof(struct ifbrparam), BC_F_COPYIN|BC_F_SUSER }, { bridge_ioctl_sifprio, sizeof(struct ifbreq), BC_F_COPYIN|BC_F_SUSER }, { bridge_ioctl_sifcost, sizeof(struct ifbreq), BC_F_COPYIN|BC_F_SUSER }, { bridge_ioctl_addspan, sizeof(struct ifbreq), BC_F_COPYIN|BC_F_SUSER }, { bridge_ioctl_delspan, sizeof(struct ifbreq), BC_F_COPYIN|BC_F_SUSER }, { bridge_ioctl_gbparam, sizeof(struct ifbropreq), BC_F_COPYOUT }, { bridge_ioctl_grte, sizeof(struct ifbrparam), BC_F_COPYOUT }, { bridge_ioctl_gifsstp, sizeof(struct ifbpstpconf), BC_F_COPYIN|BC_F_COPYOUT }, { bridge_ioctl_sproto, sizeof(struct ifbrparam), BC_F_COPYIN|BC_F_SUSER }, { bridge_ioctl_stxhc, sizeof(struct ifbrparam), BC_F_COPYIN|BC_F_SUSER }, { bridge_ioctl_sifmaxaddr, sizeof(struct ifbreq), BC_F_COPYIN|BC_F_SUSER }, }; const int bridge_control_table_size = sizeof(bridge_control_table) / sizeof(bridge_control_table[0]); LIST_HEAD(, bridge_softc) bridge_list; IFC_SIMPLE_DECLARE(bridge, 0); static int bridge_modevent(module_t mod, int type, void *data) { switch (type) { case MOD_LOAD: mtx_init(&bridge_list_mtx, "if_bridge list", NULL, MTX_DEF); if_clone_attach(&bridge_cloner); bridge_rtnode_zone = uma_zcreate("bridge_rtnode", sizeof(struct bridge_rtnode), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0); LIST_INIT(&bridge_list); bridge_input_p = bridge_input; bridge_output_p = bridge_output; bridge_dn_p = bridge_dummynet; + bridge_linkstate_p = bridge_linkstate; bridge_detach_cookie = EVENTHANDLER_REGISTER( ifnet_departure_event, bridge_ifdetach, NULL, EVENTHANDLER_PRI_ANY); break; case MOD_UNLOAD: EVENTHANDLER_DEREGISTER(ifnet_departure_event, bridge_detach_cookie); if_clone_detach(&bridge_cloner); uma_zdestroy(bridge_rtnode_zone); bridge_input_p = NULL; bridge_output_p = NULL; bridge_dn_p = NULL; + bridge_linkstate_p = NULL; mtx_destroy(&bridge_list_mtx); break; default: return (EOPNOTSUPP); } return (0); } static moduledata_t bridge_mod = { "if_bridge", bridge_modevent, 0 }; DECLARE_MODULE(if_bridge, bridge_mod, SI_SUB_PSEUDO, SI_ORDER_ANY); MODULE_DEPEND(if_bridge, bridgestp, 1, 1, 1); /* * handler for net.link.bridge.pfil_ipfw */ static int sysctl_pfil_ipfw(SYSCTL_HANDLER_ARGS) { int enable = pfil_ipfw; int error; error = sysctl_handle_int(oidp, &enable, 0, req); enable = (enable) ? 1 : 0; if (enable != pfil_ipfw) { pfil_ipfw = enable; /* * Disable pfil so that ipfw doesnt run twice, if the user * really wants both then they can re-enable pfil_bridge and/or * pfil_member. Also allow non-ip packets as ipfw can filter by * layer2 type. */ if (pfil_ipfw) { pfil_onlyip = 0; pfil_bridge = 0; pfil_member = 0; } } return (error); } SYSCTL_PROC(_net_link_bridge, OID_AUTO, ipfw, CTLTYPE_INT|CTLFLAG_RW, &pfil_ipfw, 0, &sysctl_pfil_ipfw, "I", "Layer2 filter with IPFW"); /* * bridge_clone_create: * * Create a new bridge instance. */ static int bridge_clone_create(struct if_clone *ifc, int unit, caddr_t params) { struct bridge_softc *sc, *sc2; struct ifnet *bifp, *ifp; int fb, retry; unsigned long hostid; sc = malloc(sizeof(*sc), M_DEVBUF, M_WAITOK|M_ZERO); ifp = sc->sc_ifp = if_alloc(IFT_ETHER); if (ifp == NULL) { free(sc, M_DEVBUF); return (ENOSPC); } BRIDGE_LOCK_INIT(sc); sc->sc_brtmax = BRIDGE_RTABLE_MAX; sc->sc_brttimeout = BRIDGE_RTABLE_TIMEOUT; /* Initialize our routing table. */ bridge_rtable_init(sc); callout_init_mtx(&sc->sc_brcallout, &sc->sc_mtx, 0); LIST_INIT(&sc->sc_iflist); LIST_INIT(&sc->sc_spanlist); ifp->if_softc = sc; if_initname(ifp, ifc->ifc_name, unit); ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST; ifp->if_ioctl = bridge_ioctl; ifp->if_start = bridge_start; ifp->if_init = bridge_init; ifp->if_type = IFT_BRIDGE; IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen); ifp->if_snd.ifq_drv_maxlen = ifqmaxlen; IFQ_SET_READY(&ifp->if_snd); /* * Generate an ethernet address with a locally administered address. * * Since we are using random ethernet addresses for the bridge, it is * possible that we might have address collisions, so make sure that * this hardware address isn't already in use on another bridge. * The first try uses the hostid and falls back to arc4rand(). */ fb = 0; getcredhostid(curthread->td_ucred, &hostid); for (retry = 1; retry != 0;) { if (fb || hostid == 0) { arc4rand(sc->sc_defaddr, ETHER_ADDR_LEN, 1); sc->sc_defaddr[0] &= ~1;/* clear multicast bit */ sc->sc_defaddr[0] |= 2; /* set the LAA bit */ } else { sc->sc_defaddr[0] = 0x2; sc->sc_defaddr[1] = (hostid >> 24) & 0xff; sc->sc_defaddr[2] = (hostid >> 16) & 0xff; sc->sc_defaddr[3] = (hostid >> 8 ) & 0xff; sc->sc_defaddr[4] = hostid & 0xff; sc->sc_defaddr[5] = ifp->if_dunit & 0xff; } fb = 1; retry = 0; mtx_lock(&bridge_list_mtx); LIST_FOREACH(sc2, &bridge_list, sc_list) { bifp = sc2->sc_ifp; if (memcmp(sc->sc_defaddr, IF_LLADDR(bifp), ETHER_ADDR_LEN) == 0) retry = 1; } mtx_unlock(&bridge_list_mtx); } bstp_attach(&sc->sc_stp, &bridge_ops); ether_ifattach(ifp, sc->sc_defaddr); /* Now undo some of the damage... */ ifp->if_baudrate = 0; ifp->if_type = IFT_BRIDGE; mtx_lock(&bridge_list_mtx); LIST_INSERT_HEAD(&bridge_list, sc, sc_list); mtx_unlock(&bridge_list_mtx); return (0); } /* * bridge_clone_destroy: * * Destroy a bridge instance. */ static void bridge_clone_destroy(struct ifnet *ifp) { struct bridge_softc *sc = ifp->if_softc; struct bridge_iflist *bif; BRIDGE_LOCK(sc); bridge_stop(ifp, 1); ifp->if_flags &= ~IFF_UP; while ((bif = LIST_FIRST(&sc->sc_iflist)) != NULL) bridge_delete_member(sc, bif, 0); while ((bif = LIST_FIRST(&sc->sc_spanlist)) != NULL) { bridge_delete_span(sc, bif); } BRIDGE_UNLOCK(sc); callout_drain(&sc->sc_brcallout); mtx_lock(&bridge_list_mtx); LIST_REMOVE(sc, sc_list); mtx_unlock(&bridge_list_mtx); bstp_detach(&sc->sc_stp); ether_ifdetach(ifp); if_free(ifp); /* Tear down the routing table. */ bridge_rtable_fini(sc); BRIDGE_LOCK_DESTROY(sc); free(sc, M_DEVBUF); } /* * bridge_ioctl: * * Handle a control request from the operator. */ static int bridge_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data) { struct bridge_softc *sc = ifp->if_softc; struct ifreq *ifr = (struct ifreq *)data; struct bridge_iflist *bif; struct thread *td = curthread; union { struct ifbreq ifbreq; struct ifbifconf ifbifconf; struct ifbareq ifbareq; struct ifbaconf ifbaconf; struct ifbrparam ifbrparam; struct ifbropreq ifbropreq; } args; struct ifdrv *ifd = (struct ifdrv *) data; const struct bridge_control *bc; int error = 0; switch (cmd) { case SIOCADDMULTI: case SIOCDELMULTI: break; case SIOCGDRVSPEC: case SIOCSDRVSPEC: if (ifd->ifd_cmd >= bridge_control_table_size) { error = EINVAL; break; } bc = &bridge_control_table[ifd->ifd_cmd]; if (cmd == SIOCGDRVSPEC && (bc->bc_flags & BC_F_COPYOUT) == 0) { error = EINVAL; break; } else if (cmd == SIOCSDRVSPEC && (bc->bc_flags & BC_F_COPYOUT) != 0) { error = EINVAL; break; } if (bc->bc_flags & BC_F_SUSER) { error = priv_check(td, PRIV_NET_BRIDGE); if (error) break; } if (ifd->ifd_len != bc->bc_argsize || ifd->ifd_len > sizeof(args)) { error = EINVAL; break; } bzero(&args, sizeof(args)); if (bc->bc_flags & BC_F_COPYIN) { error = copyin(ifd->ifd_data, &args, ifd->ifd_len); if (error) break; } BRIDGE_LOCK(sc); error = (*bc->bc_func)(sc, &args); BRIDGE_UNLOCK(sc); if (error) break; if (bc->bc_flags & BC_F_COPYOUT) error = copyout(&args, ifd->ifd_data, ifd->ifd_len); break; case SIOCSIFFLAGS: if (!(ifp->if_flags & IFF_UP) && (ifp->if_drv_flags & IFF_DRV_RUNNING)) { /* * If interface is marked down and it is running, * then stop and disable it. */ BRIDGE_LOCK(sc); bridge_stop(ifp, 1); BRIDGE_UNLOCK(sc); } else if ((ifp->if_flags & IFF_UP) && !(ifp->if_drv_flags & IFF_DRV_RUNNING)) { /* * If interface is marked up and it is stopped, then * start it. */ (*ifp->if_init)(sc); } break; case SIOCSIFMTU: if (ifr->ifr_mtu < 576) { error = EINVAL; break; } if (LIST_EMPTY(&sc->sc_iflist)) { sc->sc_ifp->if_mtu = ifr->ifr_mtu; break; } BRIDGE_LOCK(sc); LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { if (bif->bif_ifp->if_mtu != ifr->ifr_mtu) { log(LOG_NOTICE, "%s: invalid MTU: %lu(%s)" " != %d\n", sc->sc_ifp->if_xname, bif->bif_ifp->if_mtu, bif->bif_ifp->if_xname, ifr->ifr_mtu); error = EINVAL; break; } } if (!error) sc->sc_ifp->if_mtu = ifr->ifr_mtu; BRIDGE_UNLOCK(sc); break; default: /* * drop the lock as ether_ioctl() will call bridge_start() and * cause the lock to be recursed. */ error = ether_ioctl(ifp, cmd, data); break; } return (error); } /* * bridge_mutecaps: * * Clear or restore unwanted capabilities on the member interface */ static void bridge_mutecaps(struct bridge_softc *sc) { struct bridge_iflist *bif; int enabled, mask; /* Initial bitmask of capabilities to test */ mask = BRIDGE_IFCAPS_MASK; LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { /* Every member must support it or its disabled */ mask &= bif->bif_savedcaps; } LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { enabled = bif->bif_ifp->if_capenable; enabled &= ~BRIDGE_IFCAPS_STRIP; /* strip off mask bits and enable them again if allowed */ enabled &= ~BRIDGE_IFCAPS_MASK; enabled |= mask; bridge_set_ifcap(sc, bif, enabled); } } static void bridge_set_ifcap(struct bridge_softc *sc, struct bridge_iflist *bif, int set) { struct ifnet *ifp = bif->bif_ifp; struct ifreq ifr; int error; bzero(&ifr, sizeof(ifr)); ifr.ifr_reqcap = set; if (ifp->if_capenable != set) { error = (*ifp->if_ioctl)(ifp, SIOCSIFCAP, (caddr_t)&ifr); if (error) if_printf(sc->sc_ifp, "error setting interface capabilities on %s\n", ifp->if_xname); } } /* * bridge_lookup_member: * * Lookup a bridge member interface. */ static struct bridge_iflist * bridge_lookup_member(struct bridge_softc *sc, const char *name) { struct bridge_iflist *bif; struct ifnet *ifp; BRIDGE_LOCK_ASSERT(sc); LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { ifp = bif->bif_ifp; if (strcmp(ifp->if_xname, name) == 0) return (bif); } return (NULL); } /* * bridge_lookup_member_if: * * Lookup a bridge member interface by ifnet*. */ static struct bridge_iflist * bridge_lookup_member_if(struct bridge_softc *sc, struct ifnet *member_ifp) { struct bridge_iflist *bif; BRIDGE_LOCK_ASSERT(sc); LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { if (bif->bif_ifp == member_ifp) return (bif); } return (NULL); } /* * bridge_delete_member: * * Delete the specified member interface. */ static void bridge_delete_member(struct bridge_softc *sc, struct bridge_iflist *bif, int gone) { struct ifnet *ifs = bif->bif_ifp; struct ifnet *fif = NULL; BRIDGE_LOCK_ASSERT(sc); if (bif->bif_flags & IFBIF_STP) bstp_disable(&bif->bif_stp); ifs->if_bridge = NULL; BRIDGE_XLOCK(sc); LIST_REMOVE(bif, bif_next); BRIDGE_XDROP(sc); /* * If removing the interface that gave the bridge its mac address, set * the mac address of the bridge to the address of the next member, or * to its default address if no members are left. */ if (bridge_inherit_mac && sc->sc_ifaddr == ifs) { if (LIST_EMPTY(&sc->sc_iflist)) { bcopy(sc->sc_defaddr, IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN); sc->sc_ifaddr = NULL; } else { fif = LIST_FIRST(&sc->sc_iflist)->bif_ifp; bcopy(IF_LLADDR(fif), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN); sc->sc_ifaddr = fif; } EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp); } bridge_mutecaps(sc); /* recalcuate now this interface is removed */ bridge_rtdelete(sc, ifs, IFBF_FLUSHALL); KASSERT(bif->bif_addrcnt == 0, ("%s: %d bridge routes referenced", __func__, bif->bif_addrcnt)); BRIDGE_UNLOCK(sc); if (!gone) { switch (ifs->if_type) { case IFT_ETHER: case IFT_L2VLAN: /* * Take the interface out of promiscuous mode. */ (void) ifpromisc(ifs, 0); break; case IFT_GIF: break; default: #ifdef DIAGNOSTIC panic("bridge_delete_member: impossible"); #endif break; } /* reneable any interface capabilities */ bridge_set_ifcap(sc, bif, bif->bif_savedcaps); } bstp_destroy(&bif->bif_stp); /* prepare to free */ + bridge_linkstate(ifs); BRIDGE_LOCK(sc); free(bif, M_DEVBUF); } /* * bridge_delete_span: * * Delete the specified span interface. */ static void bridge_delete_span(struct bridge_softc *sc, struct bridge_iflist *bif) { BRIDGE_LOCK_ASSERT(sc); KASSERT(bif->bif_ifp->if_bridge == NULL, ("%s: not a span interface", __func__)); LIST_REMOVE(bif, bif_next); free(bif, M_DEVBUF); } static int bridge_ioctl_add(struct bridge_softc *sc, void *arg) { struct ifbreq *req = arg; struct bridge_iflist *bif = NULL; struct ifnet *ifs; int error = 0; ifs = ifunit(req->ifbr_ifsname); if (ifs == NULL) return (ENOENT); if (ifs->if_ioctl == NULL) /* must be supported */ return (EINVAL); /* If it's in the span list, it can't be a member. */ LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) if (ifs == bif->bif_ifp) return (EBUSY); if (ifs->if_bridge == sc) return (EEXIST); if (ifs->if_bridge != NULL) return (EBUSY); bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO); if (bif == NULL) return (ENOMEM); bif->bif_ifp = ifs; bif->bif_flags = IFBIF_LEARNING | IFBIF_DISCOVER; bif->bif_savedcaps = ifs->if_capenable; switch (ifs->if_type) { case IFT_ETHER: case IFT_L2VLAN: case IFT_GIF: /* permitted interface types */ break; default: error = EINVAL; goto out; } /* Allow the first Ethernet member to define the MTU */ if (LIST_EMPTY(&sc->sc_iflist)) sc->sc_ifp->if_mtu = ifs->if_mtu; else if (sc->sc_ifp->if_mtu != ifs->if_mtu) { if_printf(sc->sc_ifp, "invalid MTU: %lu(%s) != %lu\n", ifs->if_mtu, ifs->if_xname, sc->sc_ifp->if_mtu); error = EINVAL; goto out; } /* * Assign the interface's MAC address to the bridge if it's the first * member and the MAC address of the bridge has not been changed from * the default randomly generated one. */ if (bridge_inherit_mac && LIST_EMPTY(&sc->sc_iflist) && !memcmp(IF_LLADDR(sc->sc_ifp), sc->sc_defaddr, ETHER_ADDR_LEN)) { bcopy(IF_LLADDR(ifs), IF_LLADDR(sc->sc_ifp), ETHER_ADDR_LEN); sc->sc_ifaddr = ifs; EVENTHANDLER_INVOKE(iflladdr_event, sc->sc_ifp); } ifs->if_bridge = sc; bstp_create(&sc->sc_stp, &bif->bif_stp, bif->bif_ifp); /* * XXX: XLOCK HERE!?! * * NOTE: insert_***HEAD*** should be safe for the traversals. */ LIST_INSERT_HEAD(&sc->sc_iflist, bif, bif_next); /* Set interface capabilities to the intersection set of all members */ bridge_mutecaps(sc); + BRIDGE_UNLOCK(sc); + /* Update the linkstate for the bridge */ + bridge_linkstate(ifs); + /* Place the interface into promiscuous mode */ switch (ifs->if_type) { - case IFT_ETHER: - case IFT_L2VLAN: - /* - * Place the interface into promiscuous mode. - */ - BRIDGE_UNLOCK(sc); - error = ifpromisc(ifs, 1); - BRIDGE_LOCK(sc); - break; + case IFT_ETHER: + case IFT_L2VLAN: + error = ifpromisc(ifs, 1); + break; } + BRIDGE_LOCK(sc); if (error) bridge_delete_member(sc, bif, 0); out: if (error) { if (bif != NULL) free(bif, M_DEVBUF); } return (error); } static int bridge_ioctl_del(struct bridge_softc *sc, void *arg) { struct ifbreq *req = arg; struct bridge_iflist *bif; bif = bridge_lookup_member(sc, req->ifbr_ifsname); if (bif == NULL) return (ENOENT); bridge_delete_member(sc, bif, 0); return (0); } static int bridge_ioctl_gifflags(struct bridge_softc *sc, void *arg) { struct ifbreq *req = arg; struct bridge_iflist *bif; struct bstp_port *bp; bif = bridge_lookup_member(sc, req->ifbr_ifsname); if (bif == NULL) return (ENOENT); bp = &bif->bif_stp; req->ifbr_ifsflags = bif->bif_flags; req->ifbr_state = bp->bp_state; req->ifbr_priority = bp->bp_priority; req->ifbr_path_cost = bp->bp_path_cost; req->ifbr_portno = bif->bif_ifp->if_index & 0xfff; req->ifbr_proto = bp->bp_protover; req->ifbr_role = bp->bp_role; req->ifbr_stpflags = bp->bp_flags; req->ifbr_addrcnt = bif->bif_addrcnt; req->ifbr_addrmax = bif->bif_addrmax; req->ifbr_addrexceeded = bif->bif_addrexceeded; /* Copy STP state options as flags */ if (bp->bp_operedge) req->ifbr_ifsflags |= IFBIF_BSTP_EDGE; if (bp->bp_flags & BSTP_PORT_AUTOEDGE) req->ifbr_ifsflags |= IFBIF_BSTP_AUTOEDGE; if (bp->bp_ptp_link) req->ifbr_ifsflags |= IFBIF_BSTP_PTP; if (bp->bp_flags & BSTP_PORT_AUTOPTP) req->ifbr_ifsflags |= IFBIF_BSTP_AUTOPTP; if (bp->bp_flags & BSTP_PORT_ADMEDGE) req->ifbr_ifsflags |= IFBIF_BSTP_ADMEDGE; if (bp->bp_flags & BSTP_PORT_ADMCOST) req->ifbr_ifsflags |= IFBIF_BSTP_ADMCOST; return (0); } static int bridge_ioctl_sifflags(struct bridge_softc *sc, void *arg) { struct ifbreq *req = arg; struct bridge_iflist *bif; struct bstp_port *bp; int error; bif = bridge_lookup_member(sc, req->ifbr_ifsname); if (bif == NULL) return (ENOENT); bp = &bif->bif_stp; if (req->ifbr_ifsflags & IFBIF_SPAN) /* SPAN is readonly */ return (EINVAL); if (req->ifbr_ifsflags & IFBIF_STP) { if ((bif->bif_flags & IFBIF_STP) == 0) { error = bstp_enable(&bif->bif_stp); if (error) return (error); } } else { if ((bif->bif_flags & IFBIF_STP) != 0) bstp_disable(&bif->bif_stp); } /* Pass on STP flags */ bstp_set_edge(bp, req->ifbr_ifsflags & IFBIF_BSTP_EDGE ? 1 : 0); bstp_set_autoedge(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOEDGE ? 1 : 0); bstp_set_ptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_PTP ? 1 : 0); bstp_set_autoptp(bp, req->ifbr_ifsflags & IFBIF_BSTP_AUTOPTP ? 1 : 0); /* Save the bits relating to the bridge */ bif->bif_flags = req->ifbr_ifsflags & IFBIFMASK; return (0); } static int bridge_ioctl_scache(struct bridge_softc *sc, void *arg) { struct ifbrparam *param = arg; sc->sc_brtmax = param->ifbrp_csize; bridge_rttrim(sc); return (0); } static int bridge_ioctl_gcache(struct bridge_softc *sc, void *arg) { struct ifbrparam *param = arg; param->ifbrp_csize = sc->sc_brtmax; return (0); } static int bridge_ioctl_gifs(struct bridge_softc *sc, void *arg) { struct ifbifconf *bifc = arg; struct bridge_iflist *bif; struct ifbreq breq; char *buf, *outbuf; int count, buflen, len, error = 0; count = 0; LIST_FOREACH(bif, &sc->sc_iflist, bif_next) count++; LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) count++; buflen = sizeof(breq) * count; if (bifc->ifbic_len == 0) { bifc->ifbic_len = buflen; return (0); } BRIDGE_UNLOCK(sc); outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO); BRIDGE_LOCK(sc); count = 0; buf = outbuf; len = min(bifc->ifbic_len, buflen); bzero(&breq, sizeof(breq)); LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { if (len < sizeof(breq)) break; strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname, sizeof(breq.ifbr_ifsname)); /* Fill in the ifbreq structure */ error = bridge_ioctl_gifflags(sc, &breq); if (error) break; memcpy(buf, &breq, sizeof(breq)); count++; buf += sizeof(breq); len -= sizeof(breq); } LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) { if (len < sizeof(breq)) break; strlcpy(breq.ifbr_ifsname, bif->bif_ifp->if_xname, sizeof(breq.ifbr_ifsname)); breq.ifbr_ifsflags = bif->bif_flags; breq.ifbr_portno = bif->bif_ifp->if_index & 0xfff; memcpy(buf, &breq, sizeof(breq)); count++; buf += sizeof(breq); len -= sizeof(breq); } BRIDGE_UNLOCK(sc); bifc->ifbic_len = sizeof(breq) * count; error = copyout(outbuf, bifc->ifbic_req, bifc->ifbic_len); BRIDGE_LOCK(sc); free(outbuf, M_TEMP); return (error); } static int bridge_ioctl_rts(struct bridge_softc *sc, void *arg) { struct ifbaconf *bac = arg; struct bridge_rtnode *brt; struct ifbareq bareq; char *buf, *outbuf; int count, buflen, len, error = 0; if (bac->ifbac_len == 0) return (0); count = 0; LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) count++; buflen = sizeof(bareq) * count; BRIDGE_UNLOCK(sc); outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO); BRIDGE_LOCK(sc); count = 0; buf = outbuf; len = min(bac->ifbac_len, buflen); bzero(&bareq, sizeof(bareq)); LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) { if (len < sizeof(bareq)) goto out; strlcpy(bareq.ifba_ifsname, brt->brt_ifp->if_xname, sizeof(bareq.ifba_ifsname)); memcpy(bareq.ifba_dst, brt->brt_addr, sizeof(brt->brt_addr)); bareq.ifba_vlan = brt->brt_vlan; if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC && time_uptime < brt->brt_expire) bareq.ifba_expire = brt->brt_expire - time_uptime; else bareq.ifba_expire = 0; bareq.ifba_flags = brt->brt_flags; memcpy(buf, &bareq, sizeof(bareq)); count++; buf += sizeof(bareq); len -= sizeof(bareq); } out: BRIDGE_UNLOCK(sc); bac->ifbac_len = sizeof(bareq) * count; error = copyout(outbuf, bac->ifbac_req, bac->ifbac_len); BRIDGE_LOCK(sc); free(outbuf, M_TEMP); return (error); } static int bridge_ioctl_saddr(struct bridge_softc *sc, void *arg) { struct ifbareq *req = arg; struct bridge_iflist *bif; int error; bif = bridge_lookup_member(sc, req->ifba_ifsname); if (bif == NULL) return (ENOENT); error = bridge_rtupdate(sc, req->ifba_dst, req->ifba_vlan, bif, 1, req->ifba_flags); return (error); } static int bridge_ioctl_sto(struct bridge_softc *sc, void *arg) { struct ifbrparam *param = arg; sc->sc_brttimeout = param->ifbrp_ctime; return (0); } static int bridge_ioctl_gto(struct bridge_softc *sc, void *arg) { struct ifbrparam *param = arg; param->ifbrp_ctime = sc->sc_brttimeout; return (0); } static int bridge_ioctl_daddr(struct bridge_softc *sc, void *arg) { struct ifbareq *req = arg; return (bridge_rtdaddr(sc, req->ifba_dst, req->ifba_vlan)); } static int bridge_ioctl_flush(struct bridge_softc *sc, void *arg) { struct ifbreq *req = arg; bridge_rtflush(sc, req->ifbr_ifsflags); return (0); } static int bridge_ioctl_gpri(struct bridge_softc *sc, void *arg) { struct ifbrparam *param = arg; struct bstp_state *bs = &sc->sc_stp; param->ifbrp_prio = bs->bs_bridge_priority; return (0); } static int bridge_ioctl_spri(struct bridge_softc *sc, void *arg) { struct ifbrparam *param = arg; return (bstp_set_priority(&sc->sc_stp, param->ifbrp_prio)); } static int bridge_ioctl_ght(struct bridge_softc *sc, void *arg) { struct ifbrparam *param = arg; struct bstp_state *bs = &sc->sc_stp; param->ifbrp_hellotime = bs->bs_bridge_htime >> 8; return (0); } static int bridge_ioctl_sht(struct bridge_softc *sc, void *arg) { struct ifbrparam *param = arg; return (bstp_set_htime(&sc->sc_stp, param->ifbrp_hellotime)); } static int bridge_ioctl_gfd(struct bridge_softc *sc, void *arg) { struct ifbrparam *param = arg; struct bstp_state *bs = &sc->sc_stp; param->ifbrp_fwddelay = bs->bs_bridge_fdelay >> 8; return (0); } static int bridge_ioctl_sfd(struct bridge_softc *sc, void *arg) { struct ifbrparam *param = arg; return (bstp_set_fdelay(&sc->sc_stp, param->ifbrp_fwddelay)); } static int bridge_ioctl_gma(struct bridge_softc *sc, void *arg) { struct ifbrparam *param = arg; struct bstp_state *bs = &sc->sc_stp; param->ifbrp_maxage = bs->bs_bridge_max_age >> 8; return (0); } static int bridge_ioctl_sma(struct bridge_softc *sc, void *arg) { struct ifbrparam *param = arg; return (bstp_set_maxage(&sc->sc_stp, param->ifbrp_maxage)); } static int bridge_ioctl_sifprio(struct bridge_softc *sc, void *arg) { struct ifbreq *req = arg; struct bridge_iflist *bif; bif = bridge_lookup_member(sc, req->ifbr_ifsname); if (bif == NULL) return (ENOENT); return (bstp_set_port_priority(&bif->bif_stp, req->ifbr_priority)); } static int bridge_ioctl_sifcost(struct bridge_softc *sc, void *arg) { struct ifbreq *req = arg; struct bridge_iflist *bif; bif = bridge_lookup_member(sc, req->ifbr_ifsname); if (bif == NULL) return (ENOENT); return (bstp_set_path_cost(&bif->bif_stp, req->ifbr_path_cost)); } static int bridge_ioctl_sifmaxaddr(struct bridge_softc *sc, void *arg) { struct ifbreq *req = arg; struct bridge_iflist *bif; bif = bridge_lookup_member(sc, req->ifbr_ifsname); if (bif == NULL) return (ENOENT); bif->bif_addrmax = req->ifbr_addrmax; return (0); } static int bridge_ioctl_addspan(struct bridge_softc *sc, void *arg) { struct ifbreq *req = arg; struct bridge_iflist *bif = NULL; struct ifnet *ifs; ifs = ifunit(req->ifbr_ifsname); if (ifs == NULL) return (ENOENT); LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) if (ifs == bif->bif_ifp) return (EBUSY); if (ifs->if_bridge != NULL) return (EBUSY); switch (ifs->if_type) { case IFT_ETHER: case IFT_GIF: case IFT_L2VLAN: break; default: return (EINVAL); } bif = malloc(sizeof(*bif), M_DEVBUF, M_NOWAIT|M_ZERO); if (bif == NULL) return (ENOMEM); bif->bif_ifp = ifs; bif->bif_flags = IFBIF_SPAN; LIST_INSERT_HEAD(&sc->sc_spanlist, bif, bif_next); return (0); } static int bridge_ioctl_delspan(struct bridge_softc *sc, void *arg) { struct ifbreq *req = arg; struct bridge_iflist *bif; struct ifnet *ifs; ifs = ifunit(req->ifbr_ifsname); if (ifs == NULL) return (ENOENT); LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) if (ifs == bif->bif_ifp) break; if (bif == NULL) return (ENOENT); bridge_delete_span(sc, bif); return (0); } static int bridge_ioctl_gbparam(struct bridge_softc *sc, void *arg) { struct ifbropreq *req = arg; struct bstp_state *bs = &sc->sc_stp; struct bstp_port *root_port; req->ifbop_maxage = bs->bs_bridge_max_age >> 8; req->ifbop_hellotime = bs->bs_bridge_htime >> 8; req->ifbop_fwddelay = bs->bs_bridge_fdelay >> 8; root_port = bs->bs_root_port; if (root_port == NULL) req->ifbop_root_port = 0; else req->ifbop_root_port = root_port->bp_ifp->if_index; req->ifbop_holdcount = bs->bs_txholdcount; req->ifbop_priority = bs->bs_bridge_priority; req->ifbop_protocol = bs->bs_protover; req->ifbop_root_path_cost = bs->bs_root_pv.pv_cost; req->ifbop_bridgeid = bs->bs_bridge_pv.pv_dbridge_id; req->ifbop_designated_root = bs->bs_root_pv.pv_root_id; req->ifbop_designated_bridge = bs->bs_root_pv.pv_dbridge_id; req->ifbop_last_tc_time.tv_sec = bs->bs_last_tc_time.tv_sec; req->ifbop_last_tc_time.tv_usec = bs->bs_last_tc_time.tv_usec; return (0); } static int bridge_ioctl_grte(struct bridge_softc *sc, void *arg) { struct ifbrparam *param = arg; param->ifbrp_cexceeded = sc->sc_brtexceeded; return (0); } static int bridge_ioctl_gifsstp(struct bridge_softc *sc, void *arg) { struct ifbpstpconf *bifstp = arg; struct bridge_iflist *bif; struct bstp_port *bp; struct ifbpstpreq bpreq; char *buf, *outbuf; int count, buflen, len, error = 0; count = 0; LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { if ((bif->bif_flags & IFBIF_STP) != 0) count++; } buflen = sizeof(bpreq) * count; if (bifstp->ifbpstp_len == 0) { bifstp->ifbpstp_len = buflen; return (0); } BRIDGE_UNLOCK(sc); outbuf = malloc(buflen, M_TEMP, M_WAITOK | M_ZERO); BRIDGE_LOCK(sc); count = 0; buf = outbuf; len = min(bifstp->ifbpstp_len, buflen); bzero(&bpreq, sizeof(bpreq)); LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { if (len < sizeof(bpreq)) break; if ((bif->bif_flags & IFBIF_STP) == 0) continue; bp = &bif->bif_stp; bpreq.ifbp_portno = bif->bif_ifp->if_index & 0xfff; bpreq.ifbp_fwd_trans = bp->bp_forward_transitions; bpreq.ifbp_design_cost = bp->bp_desg_pv.pv_cost; bpreq.ifbp_design_port = bp->bp_desg_pv.pv_port_id; bpreq.ifbp_design_bridge = bp->bp_desg_pv.pv_dbridge_id; bpreq.ifbp_design_root = bp->bp_desg_pv.pv_root_id; memcpy(buf, &bpreq, sizeof(bpreq)); count++; buf += sizeof(bpreq); len -= sizeof(bpreq); } BRIDGE_UNLOCK(sc); bifstp->ifbpstp_len = sizeof(bpreq) * count; error = copyout(outbuf, bifstp->ifbpstp_req, bifstp->ifbpstp_len); BRIDGE_LOCK(sc); free(outbuf, M_TEMP); return (error); } static int bridge_ioctl_sproto(struct bridge_softc *sc, void *arg) { struct ifbrparam *param = arg; return (bstp_set_protocol(&sc->sc_stp, param->ifbrp_proto)); } static int bridge_ioctl_stxhc(struct bridge_softc *sc, void *arg) { struct ifbrparam *param = arg; return (bstp_set_holdcount(&sc->sc_stp, param->ifbrp_txhc)); } /* * bridge_ifdetach: * * Detach an interface from a bridge. Called when a member * interface is detaching. */ static void bridge_ifdetach(void *arg __unused, struct ifnet *ifp) { struct bridge_softc *sc = ifp->if_bridge; struct bridge_iflist *bif; /* Check if the interface is a bridge member */ if (sc != NULL) { BRIDGE_LOCK(sc); bif = bridge_lookup_member_if(sc, ifp); if (bif != NULL) bridge_delete_member(sc, bif, 1); BRIDGE_UNLOCK(sc); return; } /* Check if the interface is a span port */ mtx_lock(&bridge_list_mtx); LIST_FOREACH(sc, &bridge_list, sc_list) { BRIDGE_LOCK(sc); LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) if (ifp == bif->bif_ifp) { bridge_delete_span(sc, bif); break; } BRIDGE_UNLOCK(sc); } mtx_unlock(&bridge_list_mtx); } /* * bridge_init: * * Initialize a bridge interface. */ static void bridge_init(void *xsc) { struct bridge_softc *sc = (struct bridge_softc *)xsc; struct ifnet *ifp = sc->sc_ifp; if (ifp->if_drv_flags & IFF_DRV_RUNNING) return; BRIDGE_LOCK(sc); callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz, bridge_timer, sc); ifp->if_drv_flags |= IFF_DRV_RUNNING; bstp_init(&sc->sc_stp); /* Initialize Spanning Tree */ BRIDGE_UNLOCK(sc); } /* * bridge_stop: * * Stop the bridge interface. */ static void bridge_stop(struct ifnet *ifp, int disable) { struct bridge_softc *sc = ifp->if_softc; BRIDGE_LOCK_ASSERT(sc); if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) return; callout_stop(&sc->sc_brcallout); bstp_stop(&sc->sc_stp); bridge_rtflush(sc, IFBF_FLUSHDYN); ifp->if_drv_flags &= ~IFF_DRV_RUNNING; } /* * bridge_enqueue: * * Enqueue a packet on a bridge member interface. * */ static void bridge_enqueue(struct bridge_softc *sc, struct ifnet *dst_ifp, struct mbuf *m) { int len, err = 0; short mflags; struct mbuf *m0; len = m->m_pkthdr.len; mflags = m->m_flags; /* We may be sending a fragment so traverse the mbuf */ for (; m; m = m0) { m0 = m->m_nextpkt; m->m_nextpkt = NULL; /* * If underlying interface can not do VLAN tag insertion itself * then attach a packet tag that holds it. */ if ((m->m_flags & M_VLANTAG) && (dst_ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) { m = ether_vlanencap(m, m->m_pkthdr.ether_vtag); if (m == NULL) { if_printf(dst_ifp, "unable to prepend VLAN header\n"); dst_ifp->if_oerrors++; continue; } m->m_flags &= ~M_VLANTAG; } if (err == 0) dst_ifp->if_transmit(dst_ifp, m); } if (err == 0) { sc->sc_ifp->if_opackets++; sc->sc_ifp->if_obytes += len; if (mflags & M_MCAST) sc->sc_ifp->if_omcasts++; } } /* * bridge_dummynet: * * Receive a queued packet from dummynet and pass it on to the output * interface. * * The mbuf has the Ethernet header already attached. */ static void bridge_dummynet(struct mbuf *m, struct ifnet *ifp) { struct bridge_softc *sc; sc = ifp->if_bridge; /* * The packet didnt originate from a member interface. This should only * ever happen if a member interface is removed while packets are * queued for it. */ if (sc == NULL) { m_freem(m); return; } if (PFIL_HOOKED(&V_inet_pfil_hook) #ifdef INET6 || PFIL_HOOKED(&V_inet6_pfil_hook) #endif ) { if (bridge_pfil(&m, sc->sc_ifp, ifp, PFIL_OUT) != 0) return; if (m == NULL) return; } bridge_enqueue(sc, ifp, m); } /* * bridge_output: * * Send output from a bridge member interface. This * performs the bridging function for locally originated * packets. * * The mbuf has the Ethernet header already attached. We must * enqueue or free the mbuf before returning. */ static int bridge_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *sa, struct rtentry *rt) { struct ether_header *eh; struct ifnet *dst_if; struct bridge_softc *sc; uint16_t vlan; if (m->m_len < ETHER_HDR_LEN) { m = m_pullup(m, ETHER_HDR_LEN); if (m == NULL) return (0); } eh = mtod(m, struct ether_header *); sc = ifp->if_bridge; vlan = VLANTAGOF(m); BRIDGE_LOCK(sc); /* * If bridge is down, but the original output interface is up, * go ahead and send out that interface. Otherwise, the packet * is dropped below. */ if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { dst_if = ifp; goto sendunicast; } /* * If the packet is a multicast, or we don't know a better way to * get there, send to all interfaces. */ if (ETHER_IS_MULTICAST(eh->ether_dhost)) dst_if = NULL; else dst_if = bridge_rtlookup(sc, eh->ether_dhost, vlan); if (dst_if == NULL) { struct bridge_iflist *bif; struct mbuf *mc; int error = 0, used = 0; bridge_span(sc, m); BRIDGE_LOCK2REF(sc, error); if (error) { m_freem(m); return (0); } LIST_FOREACH(bif, &sc->sc_iflist, bif_next) { dst_if = bif->bif_ifp; if (dst_if->if_type == IFT_GIF) continue; if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) continue; /* * If this is not the original output interface, * and the interface is participating in spanning * tree, make sure the port is in a state that * allows forwarding. */ if (dst_if != ifp && (bif->bif_flags & IFBIF_STP) && bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) continue; if (LIST_NEXT(bif, bif_next) == NULL) { used = 1; mc = m; } else { mc = m_copypacket(m, M_DONTWAIT); if (mc == NULL) { sc->sc_ifp->if_oerrors++; continue; } } bridge_enqueue(sc, dst_if, mc); } if (used == 0) m_freem(m); BRIDGE_UNREF(sc); return (0); } sendunicast: /* * XXX Spanning tree consideration here? */ bridge_span(sc, m); if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) { m_freem(m); BRIDGE_UNLOCK(sc); return (0); } BRIDGE_UNLOCK(sc); bridge_enqueue(sc, dst_if, m); return (0); } /* * bridge_start: * * Start output on a bridge. * */ static void bridge_start(struct ifnet *ifp) { struct bridge_softc *sc; struct mbuf *m; struct ether_header *eh; struct ifnet *dst_if; sc = ifp->if_softc; ifp->if_drv_flags |= IFF_DRV_OACTIVE; for (;;) { IFQ_DEQUEUE(&ifp->if_snd, m); if (m == 0) break; ETHER_BPF_MTAP(ifp, m); eh = mtod(m, struct ether_header *); dst_if = NULL; BRIDGE_LOCK(sc); if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) { dst_if = bridge_rtlookup(sc, eh->ether_dhost, 1); } if (dst_if == NULL) bridge_broadcast(sc, ifp, m, 0); else { BRIDGE_UNLOCK(sc); bridge_enqueue(sc, dst_if, m); } } ifp->if_drv_flags &= ~IFF_DRV_OACTIVE; } /* * bridge_forward: * * The forwarding function of the bridge. * * NOTE: Releases the lock on return. */ static void bridge_forward(struct bridge_softc *sc, struct bridge_iflist *sbif, struct mbuf *m) { struct bridge_iflist *dbif; struct ifnet *src_if, *dst_if, *ifp; struct ether_header *eh; uint16_t vlan; uint8_t *dst; int error; src_if = m->m_pkthdr.rcvif; ifp = sc->sc_ifp; ifp->if_ipackets++; ifp->if_ibytes += m->m_pkthdr.len; vlan = VLANTAGOF(m); if ((sbif->bif_flags & IFBIF_STP) && sbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) goto drop; eh = mtod(m, struct ether_header *); dst = eh->ether_dhost; /* If the interface is learning, record the address. */ if (sbif->bif_flags & IFBIF_LEARNING) { error = bridge_rtupdate(sc, eh->ether_shost, vlan, sbif, 0, IFBAF_DYNAMIC); /* * If the interface has addresses limits then deny any source * that is not in the cache. */ if (error && sbif->bif_addrmax) goto drop; } if ((sbif->bif_flags & IFBIF_STP) != 0 && sbif->bif_stp.bp_state == BSTP_IFSTATE_LEARNING) goto drop; /* * At this point, the port either doesn't participate * in spanning tree or it is in the forwarding state. */ /* * If the packet is unicast, destined for someone on * "this" side of the bridge, drop it. */ if ((m->m_flags & (M_BCAST|M_MCAST)) == 0) { dst_if = bridge_rtlookup(sc, dst, vlan); if (src_if == dst_if) goto drop; } else { /* * Check if its a reserved multicast address, any address * listed in 802.1D section 7.12.6 may not be forwarded by the * bridge. * This is currently 01-80-C2-00-00-00 to 01-80-C2-00-00-0F */ if (dst[0] == 0x01 && dst[1] == 0x80 && dst[2] == 0xc2 && dst[3] == 0x00 && dst[4] == 0x00 && dst[5] <= 0x0f) goto drop; /* ...forward it to all interfaces. */ ifp->if_imcasts++; dst_if = NULL; } /* * If we have a destination interface which is a member of our bridge, * OR this is a unicast packet, push it through the bpf(4) machinery. * For broadcast or multicast packets, don't bother because it will * be reinjected into ether_input. We do this before we pass the packets * through the pfil(9) framework, as it is possible that pfil(9) will * drop the packet, or possibly modify it, making it difficult to debug * firewall issues on the bridge. */ if (dst_if != NULL || (m->m_flags & (M_BCAST | M_MCAST)) == 0) ETHER_BPF_MTAP(ifp, m); /* run the packet filter */ if (PFIL_HOOKED(&V_inet_pfil_hook) #ifdef INET6 || PFIL_HOOKED(&V_inet6_pfil_hook) #endif ) { BRIDGE_UNLOCK(sc); if (bridge_pfil(&m, ifp, src_if, PFIL_IN) != 0) return; if (m == NULL) return; BRIDGE_LOCK(sc); } if (dst_if == NULL) { bridge_broadcast(sc, src_if, m, 1); return; } /* * At this point, we're dealing with a unicast frame * going to a different interface. */ if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) goto drop; dbif = bridge_lookup_member_if(sc, dst_if); if (dbif == NULL) /* Not a member of the bridge (anymore?) */ goto drop; /* Private segments can not talk to each other */ if (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE) goto drop; if ((dbif->bif_flags & IFBIF_STP) && dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) goto drop; BRIDGE_UNLOCK(sc); if (PFIL_HOOKED(&V_inet_pfil_hook) #ifdef INET6 || PFIL_HOOKED(&V_inet6_pfil_hook) #endif ) { if (bridge_pfil(&m, ifp, dst_if, PFIL_OUT) != 0) return; if (m == NULL) return; } bridge_enqueue(sc, dst_if, m); return; drop: BRIDGE_UNLOCK(sc); m_freem(m); } /* * bridge_input: * * Receive input from a member interface. Queue the packet for * bridging if it is not for us. */ static struct mbuf * bridge_input(struct ifnet *ifp, struct mbuf *m) { struct bridge_softc *sc = ifp->if_bridge; struct bridge_iflist *bif, *bif2; struct ifnet *bifp; struct ether_header *eh; struct mbuf *mc, *mc2; uint16_t vlan; int error; if ((sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) return (m); bifp = sc->sc_ifp; vlan = VLANTAGOF(m); /* * Implement support for bridge monitoring. If this flag has been * set on this interface, discard the packet once we push it through * the bpf(4) machinery, but before we do, increment the byte and * packet counters associated with this interface. */ if ((bifp->if_flags & IFF_MONITOR) != 0) { m->m_pkthdr.rcvif = bifp; ETHER_BPF_MTAP(bifp, m); bifp->if_ipackets++; bifp->if_ibytes += m->m_pkthdr.len; m_freem(m); return (NULL); } BRIDGE_LOCK(sc); bif = bridge_lookup_member_if(sc, ifp); if (bif == NULL) { BRIDGE_UNLOCK(sc); return (m); } eh = mtod(m, struct ether_header *); bridge_span(sc, m); if (m->m_flags & (M_BCAST|M_MCAST)) { /* Tap off 802.1D packets; they do not get forwarded. */ if (memcmp(eh->ether_dhost, bstp_etheraddr, ETHER_ADDR_LEN) == 0) { bstp_input(&bif->bif_stp, ifp, m); /* consumes mbuf */ BRIDGE_UNLOCK(sc); return (NULL); } if ((bif->bif_flags & IFBIF_STP) && bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) { BRIDGE_UNLOCK(sc); return (m); } /* * Make a deep copy of the packet and enqueue the copy * for bridge processing; return the original packet for * local processing. */ mc = m_dup(m, M_DONTWAIT); if (mc == NULL) { BRIDGE_UNLOCK(sc); return (m); } /* Perform the bridge forwarding function with the copy. */ bridge_forward(sc, bif, mc); /* * Reinject the mbuf as arriving on the bridge so we have a * chance at claiming multicast packets. We can not loop back * here from ether_input as a bridge is never a member of a * bridge. */ KASSERT(bifp->if_bridge == NULL, ("loop created in bridge_input")); mc2 = m_dup(m, M_DONTWAIT); if (mc2 != NULL) { /* Keep the layer3 header aligned */ int i = min(mc2->m_pkthdr.len, max_protohdr); mc2 = m_copyup(mc2, i, ETHER_ALIGN); } if (mc2 != NULL) { mc2->m_pkthdr.rcvif = bifp; (*bifp->if_input)(bifp, mc2); } /* Return the original packet for local processing. */ return (m); } if ((bif->bif_flags & IFBIF_STP) && bif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) { BRIDGE_UNLOCK(sc); return (m); } #if (defined(INET) || defined(INET6)) # define OR_CARP_CHECK_WE_ARE_DST(iface) \ || ((iface)->if_carp \ && (*carp_forus_p)((iface), eh->ether_dhost)) # define OR_CARP_CHECK_WE_ARE_SRC(iface) \ || ((iface)->if_carp \ && (*carp_forus_p)((iface), eh->ether_shost)) #else # define OR_CARP_CHECK_WE_ARE_DST(iface) # define OR_CARP_CHECK_WE_ARE_SRC(iface) #endif #ifdef INET6 # define OR_PFIL_HOOKED_INET6 \ || PFIL_HOOKED(&V_inet6_pfil_hook) #else # define OR_PFIL_HOOKED_INET6 #endif #define GRAB_OUR_PACKETS(iface) \ if ((iface)->if_type == IFT_GIF) \ continue; \ /* It is destined for us. */ \ if (memcmp(IF_LLADDR((iface)), eh->ether_dhost, ETHER_ADDR_LEN) == 0 \ OR_CARP_CHECK_WE_ARE_DST((iface)) \ ) { \ if ((iface)->if_type == IFT_BRIDGE) { \ ETHER_BPF_MTAP(iface, m); \ iface->if_ipackets++; \ /* Filter on the physical interface. */ \ if (pfil_local_phys && \ (PFIL_HOOKED(&V_inet_pfil_hook) \ OR_PFIL_HOOKED_INET6)) { \ if (bridge_pfil(&m, NULL, ifp, \ PFIL_IN) != 0 || m == NULL) { \ BRIDGE_UNLOCK(sc); \ return (NULL); \ } \ } \ } \ if (bif->bif_flags & IFBIF_LEARNING) { \ error = bridge_rtupdate(sc, eh->ether_shost, \ vlan, bif, 0, IFBAF_DYNAMIC); \ if (error && bif->bif_addrmax) { \ BRIDGE_UNLOCK(sc); \ m_freem(m); \ return (NULL); \ } \ } \ m->m_pkthdr.rcvif = iface; \ BRIDGE_UNLOCK(sc); \ return (m); \ } \ \ /* We just received a packet that we sent out. */ \ if (memcmp(IF_LLADDR((iface)), eh->ether_shost, ETHER_ADDR_LEN) == 0 \ OR_CARP_CHECK_WE_ARE_SRC((iface)) \ ) { \ BRIDGE_UNLOCK(sc); \ m_freem(m); \ return (NULL); \ } /* * Unicast. Make sure it's not for the bridge. */ do { GRAB_OUR_PACKETS(bifp) } while (0); /* * Give a chance for ifp at first priority. This will help when the * packet comes through the interface like VLAN's with the same MACs * on several interfaces from the same bridge. This also will save * some CPU cycles in case the destination interface and the input * interface (eq ifp) are the same. */ do { GRAB_OUR_PACKETS(ifp) } while (0); /* Now check the all bridge members. */ LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) { GRAB_OUR_PACKETS(bif2->bif_ifp) } #undef OR_CARP_CHECK_WE_ARE_DST #undef OR_CARP_CHECK_WE_ARE_SRC #undef OR_PFIL_HOOKED_INET6 #undef GRAB_OUR_PACKETS /* Perform the bridge forwarding function. */ bridge_forward(sc, bif, m); return (NULL); } /* * bridge_broadcast: * * Send a frame to all interfaces that are members of * the bridge, except for the one on which the packet * arrived. * * NOTE: Releases the lock on return. */ static void bridge_broadcast(struct bridge_softc *sc, struct ifnet *src_if, struct mbuf *m, int runfilt) { struct bridge_iflist *dbif, *sbif; struct mbuf *mc; struct ifnet *dst_if; int error = 0, used = 0, i; sbif = bridge_lookup_member_if(sc, src_if); BRIDGE_LOCK2REF(sc, error); if (error) { m_freem(m); return; } /* Filter on the bridge interface before broadcasting */ if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook) #ifdef INET6 || PFIL_HOOKED(&V_inet6_pfil_hook) #endif )) { if (bridge_pfil(&m, sc->sc_ifp, NULL, PFIL_OUT) != 0) goto out; if (m == NULL) goto out; } LIST_FOREACH(dbif, &sc->sc_iflist, bif_next) { dst_if = dbif->bif_ifp; if (dst_if == src_if) continue; /* Private segments can not talk to each other */ if (sbif && (sbif->bif_flags & dbif->bif_flags & IFBIF_PRIVATE)) continue; if ((dbif->bif_flags & IFBIF_STP) && dbif->bif_stp.bp_state == BSTP_IFSTATE_DISCARDING) continue; if ((dbif->bif_flags & IFBIF_DISCOVER) == 0 && (m->m_flags & (M_BCAST|M_MCAST)) == 0) continue; if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) continue; if (LIST_NEXT(dbif, bif_next) == NULL) { mc = m; used = 1; } else { mc = m_dup(m, M_DONTWAIT); if (mc == NULL) { sc->sc_ifp->if_oerrors++; continue; } } /* * Filter on the output interface. Pass a NULL bridge interface * pointer so we do not redundantly filter on the bridge for * each interface we broadcast on. */ if (runfilt && (PFIL_HOOKED(&V_inet_pfil_hook) #ifdef INET6 || PFIL_HOOKED(&V_inet6_pfil_hook) #endif )) { if (used == 0) { /* Keep the layer3 header aligned */ i = min(mc->m_pkthdr.len, max_protohdr); mc = m_copyup(mc, i, ETHER_ALIGN); if (mc == NULL) { sc->sc_ifp->if_oerrors++; continue; } } if (bridge_pfil(&mc, NULL, dst_if, PFIL_OUT) != 0) continue; if (mc == NULL) continue; } bridge_enqueue(sc, dst_if, mc); } if (used == 0) m_freem(m); out: BRIDGE_UNREF(sc); } /* * bridge_span: * * Duplicate a packet out one or more interfaces that are in span mode, * the original mbuf is unmodified. */ static void bridge_span(struct bridge_softc *sc, struct mbuf *m) { struct bridge_iflist *bif; struct ifnet *dst_if; struct mbuf *mc; if (LIST_EMPTY(&sc->sc_spanlist)) return; LIST_FOREACH(bif, &sc->sc_spanlist, bif_next) { dst_if = bif->bif_ifp; if ((dst_if->if_drv_flags & IFF_DRV_RUNNING) == 0) continue; mc = m_copypacket(m, M_DONTWAIT); if (mc == NULL) { sc->sc_ifp->if_oerrors++; continue; } bridge_enqueue(sc, dst_if, mc); } } /* * bridge_rtupdate: * * Add a bridge routing entry. */ static int bridge_rtupdate(struct bridge_softc *sc, const uint8_t *dst, uint16_t vlan, struct bridge_iflist *bif, int setflags, uint8_t flags) { struct bridge_rtnode *brt; int error; BRIDGE_LOCK_ASSERT(sc); /* Check the source address is valid and not multicast. */ if (ETHER_IS_MULTICAST(dst) || (dst[0] == 0 && dst[1] == 0 && dst[2] == 0 && dst[3] == 0 && dst[4] == 0 && dst[5] == 0) != 0) return (EINVAL); /* 802.1p frames map to vlan 1 */ if (vlan == 0) vlan = 1; /* * A route for this destination might already exist. If so, * update it, otherwise create a new one. */ if ((brt = bridge_rtnode_lookup(sc, dst, vlan)) == NULL) { if (sc->sc_brtcnt >= sc->sc_brtmax) { sc->sc_brtexceeded++; return (ENOSPC); } /* Check per interface address limits (if enabled) */ if (bif->bif_addrmax && bif->bif_addrcnt >= bif->bif_addrmax) { bif->bif_addrexceeded++; return (ENOSPC); } /* * Allocate a new bridge forwarding node, and * initialize the expiration time and Ethernet * address. */ brt = uma_zalloc(bridge_rtnode_zone, M_NOWAIT | M_ZERO); if (brt == NULL) return (ENOMEM); if (bif->bif_flags & IFBIF_STICKY) brt->brt_flags = IFBAF_STICKY; else brt->brt_flags = IFBAF_DYNAMIC; memcpy(brt->brt_addr, dst, ETHER_ADDR_LEN); brt->brt_vlan = vlan; if ((error = bridge_rtnode_insert(sc, brt)) != 0) { uma_zfree(bridge_rtnode_zone, brt); return (error); } brt->brt_dst = bif; bif->bif_addrcnt++; } if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC && brt->brt_dst != bif) { brt->brt_dst->bif_addrcnt--; brt->brt_dst = bif; brt->brt_dst->bif_addrcnt++; } if ((flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) brt->brt_expire = time_uptime + sc->sc_brttimeout; if (setflags) brt->brt_flags = flags; return (0); } /* * bridge_rtlookup: * * Lookup the destination interface for an address. */ static struct ifnet * bridge_rtlookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan) { struct bridge_rtnode *brt; BRIDGE_LOCK_ASSERT(sc); if ((brt = bridge_rtnode_lookup(sc, addr, vlan)) == NULL) return (NULL); return (brt->brt_ifp); } /* * bridge_rttrim: * * Trim the routine table so that we have a number * of routing entries less than or equal to the * maximum number. */ static void bridge_rttrim(struct bridge_softc *sc) { struct bridge_rtnode *brt, *nbrt; BRIDGE_LOCK_ASSERT(sc); /* Make sure we actually need to do this. */ if (sc->sc_brtcnt <= sc->sc_brtmax) return; /* Force an aging cycle; this might trim enough addresses. */ bridge_rtage(sc); if (sc->sc_brtcnt <= sc->sc_brtmax) return; LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) { bridge_rtnode_destroy(sc, brt); if (sc->sc_brtcnt <= sc->sc_brtmax) return; } } } /* * bridge_timer: * * Aging timer for the bridge. */ static void bridge_timer(void *arg) { struct bridge_softc *sc = arg; BRIDGE_LOCK_ASSERT(sc); bridge_rtage(sc); if (sc->sc_ifp->if_drv_flags & IFF_DRV_RUNNING) callout_reset(&sc->sc_brcallout, bridge_rtable_prune_period * hz, bridge_timer, sc); } /* * bridge_rtage: * * Perform an aging cycle. */ static void bridge_rtage(struct bridge_softc *sc) { struct bridge_rtnode *brt, *nbrt; BRIDGE_LOCK_ASSERT(sc); LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { if ((brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) { if (time_uptime >= brt->brt_expire) bridge_rtnode_destroy(sc, brt); } } } /* * bridge_rtflush: * * Remove all dynamic addresses from the bridge. */ static void bridge_rtflush(struct bridge_softc *sc, int full) { struct bridge_rtnode *brt, *nbrt; BRIDGE_LOCK_ASSERT(sc); LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { if (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) bridge_rtnode_destroy(sc, brt); } } /* * bridge_rtdaddr: * * Remove an address from the table. */ static int bridge_rtdaddr(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan) { struct bridge_rtnode *brt; int found = 0; BRIDGE_LOCK_ASSERT(sc); /* * If vlan is zero then we want to delete for all vlans so the lookup * may return more than one. */ while ((brt = bridge_rtnode_lookup(sc, addr, vlan)) != NULL) { bridge_rtnode_destroy(sc, brt); found = 1; } return (found ? 0 : ENOENT); } /* * bridge_rtdelete: * * Delete routes to a speicifc member interface. */ static void bridge_rtdelete(struct bridge_softc *sc, struct ifnet *ifp, int full) { struct bridge_rtnode *brt, *nbrt; BRIDGE_LOCK_ASSERT(sc); LIST_FOREACH_SAFE(brt, &sc->sc_rtlist, brt_list, nbrt) { if (brt->brt_ifp == ifp && (full || (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC)) bridge_rtnode_destroy(sc, brt); } } /* * bridge_rtable_init: * * Initialize the route table for this bridge. */ static int bridge_rtable_init(struct bridge_softc *sc) { int i; sc->sc_rthash = malloc(sizeof(*sc->sc_rthash) * BRIDGE_RTHASH_SIZE, M_DEVBUF, M_NOWAIT); if (sc->sc_rthash == NULL) return (ENOMEM); for (i = 0; i < BRIDGE_RTHASH_SIZE; i++) LIST_INIT(&sc->sc_rthash[i]); sc->sc_rthash_key = arc4random(); LIST_INIT(&sc->sc_rtlist); return (0); } /* * bridge_rtable_fini: * * Deconstruct the route table for this bridge. */ static void bridge_rtable_fini(struct bridge_softc *sc) { KASSERT(sc->sc_brtcnt == 0, ("%s: %d bridge routes referenced", __func__, sc->sc_brtcnt)); free(sc->sc_rthash, M_DEVBUF); } /* * The following hash function is adapted from "Hash Functions" by Bob Jenkins * ("Algorithm Alley", Dr. Dobbs Journal, September 1997). */ #define mix(a, b, c) \ do { \ a -= b; a -= c; a ^= (c >> 13); \ b -= c; b -= a; b ^= (a << 8); \ c -= a; c -= b; c ^= (b >> 13); \ a -= b; a -= c; a ^= (c >> 12); \ b -= c; b -= a; b ^= (a << 16); \ c -= a; c -= b; c ^= (b >> 5); \ a -= b; a -= c; a ^= (c >> 3); \ b -= c; b -= a; b ^= (a << 10); \ c -= a; c -= b; c ^= (b >> 15); \ } while (/*CONSTCOND*/0) static __inline uint32_t bridge_rthash(struct bridge_softc *sc, const uint8_t *addr) { uint32_t a = 0x9e3779b9, b = 0x9e3779b9, c = sc->sc_rthash_key; b += addr[5] << 8; b += addr[4]; a += addr[3] << 24; a += addr[2] << 16; a += addr[1] << 8; a += addr[0]; mix(a, b, c); return (c & BRIDGE_RTHASH_MASK); } #undef mix static int bridge_rtnode_addr_cmp(const uint8_t *a, const uint8_t *b) { int i, d; for (i = 0, d = 0; i < ETHER_ADDR_LEN && d == 0; i++) { d = ((int)a[i]) - ((int)b[i]); } return (d); } /* * bridge_rtnode_lookup: * * Look up a bridge route node for the specified destination. Compare the * vlan id or if zero then just return the first match. */ static struct bridge_rtnode * bridge_rtnode_lookup(struct bridge_softc *sc, const uint8_t *addr, uint16_t vlan) { struct bridge_rtnode *brt; uint32_t hash; int dir; BRIDGE_LOCK_ASSERT(sc); hash = bridge_rthash(sc, addr); LIST_FOREACH(brt, &sc->sc_rthash[hash], brt_hash) { dir = bridge_rtnode_addr_cmp(addr, brt->brt_addr); if (dir == 0 && (brt->brt_vlan == vlan || vlan == 0)) return (brt); if (dir > 0) return (NULL); } return (NULL); } /* * bridge_rtnode_insert: * * Insert the specified bridge node into the route table. We * assume the entry is not already in the table. */ static int bridge_rtnode_insert(struct bridge_softc *sc, struct bridge_rtnode *brt) { struct bridge_rtnode *lbrt; uint32_t hash; int dir; BRIDGE_LOCK_ASSERT(sc); hash = bridge_rthash(sc, brt->brt_addr); lbrt = LIST_FIRST(&sc->sc_rthash[hash]); if (lbrt == NULL) { LIST_INSERT_HEAD(&sc->sc_rthash[hash], brt, brt_hash); goto out; } do { dir = bridge_rtnode_addr_cmp(brt->brt_addr, lbrt->brt_addr); if (dir == 0 && brt->brt_vlan == lbrt->brt_vlan) return (EEXIST); if (dir > 0) { LIST_INSERT_BEFORE(lbrt, brt, brt_hash); goto out; } if (LIST_NEXT(lbrt, brt_hash) == NULL) { LIST_INSERT_AFTER(lbrt, brt, brt_hash); goto out; } lbrt = LIST_NEXT(lbrt, brt_hash); } while (lbrt != NULL); #ifdef DIAGNOSTIC panic("bridge_rtnode_insert: impossible"); #endif out: LIST_INSERT_HEAD(&sc->sc_rtlist, brt, brt_list); sc->sc_brtcnt++; return (0); } /* * bridge_rtnode_destroy: * * Destroy a bridge rtnode. */ static void bridge_rtnode_destroy(struct bridge_softc *sc, struct bridge_rtnode *brt) { BRIDGE_LOCK_ASSERT(sc); LIST_REMOVE(brt, brt_hash); LIST_REMOVE(brt, brt_list); sc->sc_brtcnt--; brt->brt_dst->bif_addrcnt--; uma_zfree(bridge_rtnode_zone, brt); } /* * bridge_rtable_expire: * * Set the expiry time for all routes on an interface. */ static void bridge_rtable_expire(struct ifnet *ifp, int age) { struct bridge_softc *sc = ifp->if_bridge; struct bridge_rtnode *brt; BRIDGE_LOCK(sc); /* * If the age is zero then flush, otherwise set all the expiry times to * age for the interface */ if (age == 0) bridge_rtdelete(sc, ifp, IFBF_FLUSHDYN); else { LIST_FOREACH(brt, &sc->sc_rtlist, brt_list) { /* Cap the expiry time to 'age' */ if (brt->brt_ifp == ifp && brt->brt_expire > time_uptime + age && (brt->brt_flags & IFBAF_TYPEMASK) == IFBAF_DYNAMIC) brt->brt_expire = time_uptime + age; } } BRIDGE_UNLOCK(sc); } /* * bridge_state_change: * * Callback from the bridgestp code when a port changes states. */ static void bridge_state_change(struct ifnet *ifp, int state) { struct bridge_softc *sc = ifp->if_bridge; static const char *stpstates[] = { "disabled", "listening", "learning", "forwarding", "blocking", "discarding" }; if (log_stp) log(LOG_NOTICE, "%s: state changed to %s on %s\n", sc->sc_ifp->if_xname, stpstates[state], ifp->if_xname); } /* * Send bridge packets through pfil if they are one of the types pfil can deal * with, or if they are ARP or REVARP. (pfil will pass ARP and REVARP without * question.) If *bifp or *ifp are NULL then packet filtering is skipped for * that interface. */ static int bridge_pfil(struct mbuf **mp, struct ifnet *bifp, struct ifnet *ifp, int dir) { int snap, error, i, hlen; struct ether_header *eh1, eh2; struct ip_fw_args args; struct ip *ip; struct llc llc1; u_int16_t ether_type; snap = 0; error = -1; /* Default error if not error == 0 */ #if 0 /* we may return with the IP fields swapped, ensure its not shared */ KASSERT(M_WRITABLE(*mp), ("%s: modifying a shared mbuf", __func__)); #endif if (pfil_bridge == 0 && pfil_member == 0 && pfil_ipfw == 0) return (0); /* filtering is disabled */ i = min((*mp)->m_pkthdr.len, max_protohdr); if ((*mp)->m_len < i) { *mp = m_pullup(*mp, i); if (*mp == NULL) { printf("%s: m_pullup failed\n", __func__); return (-1); } } eh1 = mtod(*mp, struct ether_header *); ether_type = ntohs(eh1->ether_type); /* * Check for SNAP/LLC. */ if (ether_type < ETHERMTU) { struct llc *llc2 = (struct llc *)(eh1 + 1); if ((*mp)->m_len >= ETHER_HDR_LEN + 8 && llc2->llc_dsap == LLC_SNAP_LSAP && llc2->llc_ssap == LLC_SNAP_LSAP && llc2->llc_control == LLC_UI) { ether_type = htons(llc2->llc_un.type_snap.ether_type); snap = 1; } } /* * If we're trying to filter bridge traffic, don't look at anything * other than IP and ARP traffic. If the filter doesn't understand * IPv6, don't allow IPv6 through the bridge either. This is lame * since if we really wanted, say, an AppleTalk filter, we are hosed, * but of course we don't have an AppleTalk filter to begin with. * (Note that since pfil doesn't understand ARP it will pass *ALL* * ARP traffic.) */ switch (ether_type) { case ETHERTYPE_ARP: case ETHERTYPE_REVARP: if (pfil_ipfw_arp == 0) return (0); /* Automatically pass */ break; case ETHERTYPE_IP: #ifdef INET6 case ETHERTYPE_IPV6: #endif /* INET6 */ break; default: /* * Check to see if the user wants to pass non-ip * packets, these will not be checked by pfil(9) and * passed unconditionally so the default is to drop. */ if (pfil_onlyip) goto bad; } /* Strip off the Ethernet header and keep a copy. */ m_copydata(*mp, 0, ETHER_HDR_LEN, (caddr_t) &eh2); m_adj(*mp, ETHER_HDR_LEN); /* Strip off snap header, if present */ if (snap) { m_copydata(*mp, 0, sizeof(struct llc), (caddr_t) &llc1); m_adj(*mp, sizeof(struct llc)); } /* * Check the IP header for alignment and errors */ if (dir == PFIL_IN) { switch (ether_type) { case ETHERTYPE_IP: error = bridge_ip_checkbasic(mp); break; #ifdef INET6 case ETHERTYPE_IPV6: error = bridge_ip6_checkbasic(mp); break; #endif /* INET6 */ default: error = 0; } if (error) goto bad; } /* XXX this section is also in if_ethersubr.c */ // XXX PFIL_OUT or DIR_OUT ? if (V_ip_fw_chk_ptr && pfil_ipfw != 0 && dir == PFIL_OUT && ifp != NULL) { struct m_tag *mtag; error = -1; /* fetch the start point from existing tags, if any */ mtag = m_tag_locate(*mp, MTAG_IPFW_RULE, 0, NULL); if (mtag == NULL) { args.rule.slot = 0; } else { struct ipfw_rule_ref *r; /* XXX can we free the tag after use ? */ mtag->m_tag_id = PACKET_TAG_NONE; r = (struct ipfw_rule_ref *)(mtag + 1); /* packet already partially processed ? */ if (r->info & IPFW_ONEPASS) goto ipfwpass; args.rule = *r; } args.m = *mp; args.oif = ifp; args.next_hop = NULL; args.next_hop6 = NULL; args.eh = &eh2; args.inp = NULL; /* used by ipfw uid/gid/jail rules */ i = V_ip_fw_chk_ptr(&args); *mp = args.m; if (*mp == NULL) return (error); if (ip_dn_io_ptr && (i == IP_FW_DUMMYNET)) { /* put the Ethernet header back on */ M_PREPEND(*mp, ETHER_HDR_LEN, M_DONTWAIT); if (*mp == NULL) return (error); bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN); /* * Pass the pkt to dummynet, which consumes it. The * packet will return to us via bridge_dummynet(). */ args.oif = ifp; ip_dn_io_ptr(mp, DIR_FWD | PROTO_IFB, &args); return (error); } if (i != IP_FW_PASS) /* drop */ goto bad; } ipfwpass: error = 0; /* * Run the packet through pfil */ switch (ether_type) { case ETHERTYPE_IP: /* * before calling the firewall, swap fields the same as * IP does. here we assume the header is contiguous */ ip = mtod(*mp, struct ip *); ip->ip_len = ntohs(ip->ip_len); ip->ip_off = ntohs(ip->ip_off); /* * Run pfil on the member interface and the bridge, both can * be skipped by clearing pfil_member or pfil_bridge. * * Keep the order: * in_if -> bridge_if -> out_if */ if (pfil_bridge && dir == PFIL_OUT && bifp != NULL) error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp, dir, NULL); if (*mp == NULL || error != 0) /* filter may consume */ break; if (pfil_member && ifp != NULL) error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp, dir, NULL); if (*mp == NULL || error != 0) /* filter may consume */ break; if (pfil_bridge && dir == PFIL_IN && bifp != NULL) error = pfil_run_hooks(&V_inet_pfil_hook, mp, bifp, dir, NULL); if (*mp == NULL || error != 0) /* filter may consume */ break; /* check if we need to fragment the packet */ if (pfil_member && ifp != NULL && dir == PFIL_OUT) { i = (*mp)->m_pkthdr.len; if (i > ifp->if_mtu) { error = bridge_fragment(ifp, *mp, &eh2, snap, &llc1); return (error); } } /* Recalculate the ip checksum and restore byte ordering */ ip = mtod(*mp, struct ip *); hlen = ip->ip_hl << 2; if (hlen < sizeof(struct ip)) goto bad; if (hlen > (*mp)->m_len) { if ((*mp = m_pullup(*mp, hlen)) == 0) goto bad; ip = mtod(*mp, struct ip *); if (ip == NULL) goto bad; } ip->ip_len = htons(ip->ip_len); ip->ip_off = htons(ip->ip_off); ip->ip_sum = 0; if (hlen == sizeof(struct ip)) ip->ip_sum = in_cksum_hdr(ip); else ip->ip_sum = in_cksum(*mp, hlen); break; #ifdef INET6 case ETHERTYPE_IPV6: if (pfil_bridge && dir == PFIL_OUT && bifp != NULL) error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp, dir, NULL); if (*mp == NULL || error != 0) /* filter may consume */ break; if (pfil_member && ifp != NULL) error = pfil_run_hooks(&V_inet6_pfil_hook, mp, ifp, dir, NULL); if (*mp == NULL || error != 0) /* filter may consume */ break; if (pfil_bridge && dir == PFIL_IN && bifp != NULL) error = pfil_run_hooks(&V_inet6_pfil_hook, mp, bifp, dir, NULL); break; #endif default: error = 0; break; } if (*mp == NULL) return (error); if (error != 0) goto bad; error = -1; /* * Finally, put everything back the way it was and return */ if (snap) { M_PREPEND(*mp, sizeof(struct llc), M_DONTWAIT); if (*mp == NULL) return (error); bcopy(&llc1, mtod(*mp, caddr_t), sizeof(struct llc)); } M_PREPEND(*mp, ETHER_HDR_LEN, M_DONTWAIT); if (*mp == NULL) return (error); bcopy(&eh2, mtod(*mp, caddr_t), ETHER_HDR_LEN); return (0); bad: m_freem(*mp); *mp = NULL; return (error); } /* * Perform basic checks on header size since * pfil assumes ip_input has already processed * it for it. Cut-and-pasted from ip_input.c. * Given how simple the IPv6 version is, * does the IPv4 version really need to be * this complicated? * * XXX Should we update ipstat here, or not? * XXX Right now we update ipstat but not * XXX csum_counter. */ static int bridge_ip_checkbasic(struct mbuf **mp) { struct mbuf *m = *mp; struct ip *ip; int len, hlen; u_short sum; if (*mp == NULL) return (-1); if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { if ((m = m_copyup(m, sizeof(struct ip), (max_linkhdr + 3) & ~3)) == NULL) { /* XXXJRT new stat, please */ KMOD_IPSTAT_INC(ips_toosmall); goto bad; } } else if (__predict_false(m->m_len < sizeof (struct ip))) { if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { KMOD_IPSTAT_INC(ips_toosmall); goto bad; } } ip = mtod(m, struct ip *); if (ip == NULL) goto bad; if (ip->ip_v != IPVERSION) { KMOD_IPSTAT_INC(ips_badvers); goto bad; } hlen = ip->ip_hl << 2; if (hlen < sizeof(struct ip)) { /* minimum header length */ KMOD_IPSTAT_INC(ips_badhlen); goto bad; } if (hlen > m->m_len) { if ((m = m_pullup(m, hlen)) == 0) { KMOD_IPSTAT_INC(ips_badhlen); goto bad; } ip = mtod(m, struct ip *); if (ip == NULL) goto bad; } if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) { sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID); } else { if (hlen == sizeof(struct ip)) { sum = in_cksum_hdr(ip); } else { sum = in_cksum(m, hlen); } } if (sum) { KMOD_IPSTAT_INC(ips_badsum); goto bad; } /* Retrieve the packet length. */ len = ntohs(ip->ip_len); /* * Check for additional length bogosity */ if (len < hlen) { KMOD_IPSTAT_INC(ips_badlen); goto bad; } /* * Check that the amount of data in the buffers * is as at least much as the IP header would have us expect. * Drop packet if shorter than we expect. */ if (m->m_pkthdr.len < len) { KMOD_IPSTAT_INC(ips_tooshort); goto bad; } /* Checks out, proceed */ *mp = m; return (0); bad: *mp = m; return (-1); } #ifdef INET6 /* * Same as above, but for IPv6. * Cut-and-pasted from ip6_input.c. * XXX Should we update ip6stat, or not? */ static int bridge_ip6_checkbasic(struct mbuf **mp) { struct mbuf *m = *mp; struct ip6_hdr *ip6; /* * If the IPv6 header is not aligned, slurp it up into a new * mbuf with space for link headers, in the event we forward * it. Otherwise, if it is aligned, make sure the entire base * IPv6 header is in the first mbuf of the chain. */ if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { struct ifnet *inifp = m->m_pkthdr.rcvif; if ((m = m_copyup(m, sizeof(struct ip6_hdr), (max_linkhdr + 3) & ~3)) == NULL) { /* XXXJRT new stat, please */ V_ip6stat.ip6s_toosmall++; in6_ifstat_inc(inifp, ifs6_in_hdrerr); goto bad; } } else if (__predict_false(m->m_len < sizeof(struct ip6_hdr))) { struct ifnet *inifp = m->m_pkthdr.rcvif; if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { V_ip6stat.ip6s_toosmall++; in6_ifstat_inc(inifp, ifs6_in_hdrerr); goto bad; } } ip6 = mtod(m, struct ip6_hdr *); if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) { V_ip6stat.ip6s_badvers++; in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); goto bad; } /* Checks out, proceed */ *mp = m; return (0); bad: *mp = m; return (-1); } #endif /* INET6 */ /* * bridge_fragment: * * Return a fragmented mbuf chain. */ static int bridge_fragment(struct ifnet *ifp, struct mbuf *m, struct ether_header *eh, int snap, struct llc *llc) { struct mbuf *m0; struct ip *ip; int error = -1; if (m->m_len < sizeof(struct ip) && (m = m_pullup(m, sizeof(struct ip))) == NULL) goto out; ip = mtod(m, struct ip *); error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist, CSUM_DELAY_IP); if (error) goto out; /* walk the chain and re-add the Ethernet header */ for (m0 = m; m0; m0 = m0->m_nextpkt) { if (error == 0) { if (snap) { M_PREPEND(m0, sizeof(struct llc), M_DONTWAIT); if (m0 == NULL) { error = ENOBUFS; continue; } bcopy(llc, mtod(m0, caddr_t), sizeof(struct llc)); } M_PREPEND(m0, ETHER_HDR_LEN, M_DONTWAIT); if (m0 == NULL) { error = ENOBUFS; continue; } bcopy(eh, mtod(m0, caddr_t), ETHER_HDR_LEN); } else m_freem(m); } if (error == 0) KMOD_IPSTAT_INC(ips_fragmented); return (error); out: if (m != NULL) m_freem(m); return (error); } + +static void +bridge_linkstate(struct ifnet *ifp) +{ + struct bridge_softc *sc = ifp->if_bridge; + struct bridge_iflist *bif, *bif2; + int new_link, hasls; + + BRIDGE_LOCK(sc); + bif = bridge_lookup_member_if(sc, ifp); + if (bif == NULL) { + BRIDGE_UNLOCK(sc); + return; + } + new_link = LINK_STATE_DOWN; + hasls = 0; + /* Our link is considered up if at least one of our ports is active */ + LIST_FOREACH(bif2, &sc->sc_iflist, bif_next) { + if (bif2->bif_ifp->if_capabilities & IFCAP_LINKSTATE) + hasls++; + if (bif2->bif_ifp->if_link_state == LINK_STATE_UP) { + new_link = LINK_STATE_UP; + break; + } + } + if (!LIST_EMPTY(&sc->sc_iflist) && !hasls) { + /* If no interfaces support link-state then we default to up */ + new_link = LINK_STATE_UP; + } + if_link_state_change(sc->sc_ifp, new_link); + BRIDGE_UNLOCK(sc); + + bstp_linkstate(&bif->bif_stp); +} +