--- zzzz-none-000/linux-3.10.107/drivers/infiniband/hw/mlx4/mad.c 2017-06-27 09:49:32.000000000 +0000 +++ scorpion-7490-727/linux-3.10.107/drivers/infiniband/hw/mlx4/mad.c 2021-02-04 17:41:59.000000000 +0000 @@ -64,14 +64,6 @@ #define GUID_TBL_BLK_NUM_ENTRIES 8 #define GUID_TBL_BLK_SIZE (GUID_TBL_ENTRY_SIZE * GUID_TBL_BLK_NUM_ENTRIES) -/* Counters should be saturate once they reach their maximum value */ -#define ASSIGN_32BIT_COUNTER(counter, value) do {\ - if ((value) > U32_MAX) \ - counter = cpu_to_be32(U32_MAX); \ - else \ - counter = cpu_to_be32(value); \ -} while (0) - struct mlx4_mad_rcv_buf { struct ib_grh grh; u8 payload[256]; @@ -111,8 +103,9 @@ } int mlx4_MAD_IFC(struct mlx4_ib_dev *dev, int mad_ifc_flags, - int port, struct ib_wc *in_wc, struct ib_grh *in_grh, - void *in_mad, void *response_mad) + int port, const struct ib_wc *in_wc, + const struct ib_grh *in_grh, + const void *in_mad, void *response_mad) { struct mlx4_cmd_mailbox *inmailbox, *outmailbox; void *inbox; @@ -220,7 +213,7 @@ * Snoop SM MADs for port info, GUID info, and P_Key table sets, so we can * synthesize LID change, Client-Rereg, GID change, and P_Key change events. */ -static void smp_snoop(struct ib_device *ibdev, u8 port_num, struct ib_mad *mad, +static void smp_snoop(struct ib_device *ibdev, u8 port_num, const struct ib_mad *mad, u16 prev_lid) { struct ib_port_info *pinfo; @@ -356,7 +349,7 @@ } } -static void forward_trap(struct mlx4_ib_dev *dev, u8 port_num, struct ib_mad *mad) +static void forward_trap(struct mlx4_ib_dev *dev, u8 port_num, const struct ib_mad *mad) { int qpn = mad->mad_hdr.mgmt_class != IB_MGMT_CLASS_SUBN_LID_ROUTED; struct ib_mad_send_buf *send_buf; @@ -366,7 +359,8 @@ if (agent) { send_buf = ib_create_send_mad(agent, qpn, 0, 0, IB_MGMT_MAD_HDR, - IB_MGMT_MAD_DATA, GFP_ATOMIC); + IB_MGMT_MAD_DATA, GFP_ATOMIC, + IB_MGMT_BASE_VERSION); if (IS_ERR(send_buf)) return; /* @@ -463,7 +457,8 @@ struct ib_grh *grh, struct ib_mad *mad) { struct ib_sge list; - struct ib_send_wr wr, *bad_wr; + struct ib_ud_wr wr; + struct ib_send_wr *bad_wr; struct mlx4_ib_demux_pv_ctx *tun_ctx; struct mlx4_ib_demux_pv_qp *tun_qp; struct mlx4_rcv_tunnel_mad *tun_mad; @@ -475,6 +470,7 @@ int ret = 0; u16 tun_pkey_ix; u16 cached_pkey; + u8 is_eth = dev->dev->caps.port_type[port] == MLX4_PORT_TYPE_ETH; if (dest_qpt > IB_QPT_GSI) return -EINVAL; @@ -485,10 +481,6 @@ if (!tun_ctx || tun_ctx->state != DEMUX_PV_STATE_ACTIVE) return -EAGAIN; - /* QP0 forwarding only for Dom0 */ - if (!dest_qpt && (mlx4_master_func_num(dev->dev) != slave)) - return -EINVAL; - if (!dest_qpt) tun_qp = &tun_ctx->qp[0]; else @@ -517,6 +509,10 @@ * The driver will set the force loopback bit in post_send */ memset(&attr, 0, sizeof attr); attr.port_num = port; + if (is_eth) { + memcpy(&attr.grh.dgid.raw[0], &grh->dgid.raw[0], 16); + attr.ah_flags = IB_AH_GRH; + } ah = ib_create_ah(tun_ctx->pd, &attr); if (IS_ERR(ah)) return -ENOMEM; @@ -530,7 +526,7 @@ tun_tx_ix = (++tun_qp->tx_ix_head) & (MLX4_NUM_TUNNEL_BUFS - 1); spin_unlock(&tun_qp->tx_lock); if (ret) - goto out; + goto end; tun_mad = (struct mlx4_rcv_tunnel_mad *) (tun_qp->tx_ring[tun_tx_ix].buf.addr); if (tun_qp->tx_ring[tun_tx_ix].ah) @@ -548,11 +544,36 @@ /* adjust tunnel data */ tun_mad->hdr.pkey_index = cpu_to_be16(tun_pkey_ix); - tun_mad->hdr.sl_vid = cpu_to_be16(((u16)(wc->sl)) << 12); - tun_mad->hdr.slid_mac_47_32 = cpu_to_be16(wc->slid); tun_mad->hdr.flags_src_qp = cpu_to_be32(wc->src_qp & 0xFFFFFF); tun_mad->hdr.g_ml_path = (grh && (wc->wc_flags & IB_WC_GRH)) ? 0x80 : 0; + if (is_eth) { + u16 vlan = 0; + if (mlx4_get_slave_default_vlan(dev->dev, port, slave, &vlan, + NULL)) { + /* VST mode */ + if (vlan != wc->vlan_id) + /* Packet vlan is not the VST-assigned vlan. + * Drop the packet. + */ + goto out; + else + /* Remove the vlan tag before forwarding + * the packet to the VF. + */ + vlan = 0xffff; + } else { + vlan = wc->vlan_id; + } + + tun_mad->hdr.sl_vid = cpu_to_be16(vlan); + memcpy((char *)&tun_mad->hdr.mac_31_0, &(wc->smac[0]), 4); + memcpy((char *)&tun_mad->hdr.slid_mac_47_32, &(wc->smac[4]), 2); + } else { + tun_mad->hdr.sl_vid = cpu_to_be16(((u16)(wc->sl)) << 12); + tun_mad->hdr.slid_mac_47_32 = cpu_to_be16(wc->slid); + } + ib_dma_sync_single_for_device(&dev->ib_dev, tun_qp->tx_ring[tun_tx_ix].buf.map, sizeof (struct mlx4_rcv_tunnel_mad), @@ -560,23 +581,29 @@ list.addr = tun_qp->tx_ring[tun_tx_ix].buf.map; list.length = sizeof (struct mlx4_rcv_tunnel_mad); - list.lkey = tun_ctx->mr->lkey; + list.lkey = tun_ctx->pd->local_dma_lkey; - wr.wr.ud.ah = ah; - wr.wr.ud.port_num = port; - wr.wr.ud.remote_qkey = IB_QP_SET_QKEY; - wr.wr.ud.remote_qpn = dqpn; - wr.next = NULL; - wr.wr_id = ((u64) tun_tx_ix) | MLX4_TUN_SET_WRID_QPN(dest_qpt); - wr.sg_list = &list; - wr.num_sge = 1; - wr.opcode = IB_WR_SEND; - wr.send_flags = IB_SEND_SIGNALED; + wr.ah = ah; + wr.port_num = port; + wr.remote_qkey = IB_QP_SET_QKEY; + wr.remote_qpn = dqpn; + wr.wr.next = NULL; + wr.wr.wr_id = ((u64) tun_tx_ix) | MLX4_TUN_SET_WRID_QPN(dest_qpt); + wr.wr.sg_list = &list; + wr.wr.num_sge = 1; + wr.wr.opcode = IB_WR_SEND; + wr.wr.send_flags = IB_SEND_SIGNALED; - ret = ib_post_send(src_qp, &wr, &bad_wr); -out: - if (ret) - ib_destroy_ah(ah); + ret = ib_post_send(src_qp, &wr.wr, &bad_wr); + if (!ret) + return 0; + out: + spin_lock(&tun_qp->tx_lock); + tun_qp->tx_ix_tail++; + spin_unlock(&tun_qp->tx_lock); + tun_qp->tx_ring[tun_tx_ix].ah = NULL; +end: + ib_destroy_ah(ah); return ret; } @@ -588,6 +615,41 @@ int err; int slave; u8 *slave_id; + int is_eth = 0; + + if (rdma_port_get_link_layer(ibdev, port) == IB_LINK_LAYER_INFINIBAND) + is_eth = 0; + else + is_eth = 1; + + if (is_eth) { + if (!(wc->wc_flags & IB_WC_GRH)) { + mlx4_ib_warn(ibdev, "RoCE grh not present.\n"); + return -EINVAL; + } + if (mad->mad_hdr.mgmt_class != IB_MGMT_CLASS_CM) { + mlx4_ib_warn(ibdev, "RoCE mgmt class is not CM\n"); + return -EINVAL; + } + if (mlx4_get_slave_from_roce_gid(dev->dev, port, grh->dgid.raw, &slave)) { + mlx4_ib_warn(ibdev, "failed matching grh\n"); + return -ENOENT; + } + if (slave >= dev->dev->caps.sqp_demux) { + mlx4_ib_warn(ibdev, "slave id: %d is bigger than allowed:%d\n", + slave, dev->dev->caps.sqp_demux); + return -ENOENT; + } + + if (mlx4_ib_demux_cm_handler(ibdev, port, NULL, mad)) + return 0; + + err = mlx4_ib_send_to_slave(dev, slave, port, wc->qp->qp_type, wc, grh, mad); + if (err) + pr_debug("failed sending to slave %d via tunnel qp (%d)\n", + slave, err); + return 0; + } /* Initially assume that this mad is for us */ slave = mlx4_master_func_num(dev->dev); @@ -610,6 +672,21 @@ } /* Class-specific handling */ switch (mad->mad_hdr.mgmt_class) { + case IB_MGMT_CLASS_SUBN_LID_ROUTED: + case IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE: + /* 255 indicates the dom0 */ + if (slave != 255 && slave != mlx4_master_func_num(dev->dev)) { + if (!mlx4_vf_smi_enabled(dev->dev, slave, port)) + return -EPERM; + /* for a VF. drop unsolicited MADs */ + if (!(mad->mad_hdr.method & IB_MGMT_METHOD_RESP)) { + mlx4_ib_warn(ibdev, "demux QP0. rejecting unsolicited mad for slave %d class 0x%x, method 0x%x\n", + slave, mad->mad_hdr.mgmt_class, + mad->mad_hdr.method); + return -EINVAL; + } + } + break; case IB_MGMT_CLASS_SUBN_ADM: if (mlx4_ib_demux_sa_handler(ibdev, port, slave, (struct ib_sa_mad *) mad)) @@ -646,8 +723,8 @@ } static int ib_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, - struct ib_wc *in_wc, struct ib_grh *in_grh, - struct ib_mad *in_mad, struct ib_mad *out_mad) + const struct ib_wc *in_wc, const struct ib_grh *in_grh, + const struct ib_mad *in_mad, struct ib_mad *out_mad) { u16 slid, prev_lid = 0; int err; @@ -749,34 +826,39 @@ } static int iboe_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, - struct ib_wc *in_wc, struct ib_grh *in_grh, - struct ib_mad *in_mad, struct ib_mad *out_mad) + const struct ib_wc *in_wc, const struct ib_grh *in_grh, + const struct ib_mad *in_mad, struct ib_mad *out_mad) { - struct mlx4_cmd_mailbox *mailbox; + struct mlx4_counter counter_stats; struct mlx4_ib_dev *dev = to_mdev(ibdev); - int err; - u32 inmod = dev->counters[port_num - 1] & 0xffff; - u8 mode; + struct counter_index *tmp_counter; + int err = IB_MAD_RESULT_FAILURE, stats_avail = 0; if (in_mad->mad_hdr.mgmt_class != IB_MGMT_CLASS_PERF_MGMT) return -EINVAL; - mailbox = mlx4_alloc_cmd_mailbox(dev->dev); - if (IS_ERR(mailbox)) - return IB_MAD_RESULT_FAILURE; - - err = mlx4_cmd_box(dev->dev, 0, mailbox->dma, inmod, 0, - MLX4_CMD_QUERY_IF_STAT, MLX4_CMD_TIME_CLASS_C, - MLX4_CMD_WRAPPED); - if (err) - err = IB_MAD_RESULT_FAILURE; - else { + memset(&counter_stats, 0, sizeof(counter_stats)); + mutex_lock(&dev->counters_table[port_num - 1].mutex); + list_for_each_entry(tmp_counter, + &dev->counters_table[port_num - 1].counters_list, + list) { + err = mlx4_get_counter_stats(dev->dev, + tmp_counter->index, + &counter_stats, 0); + if (err) { + err = IB_MAD_RESULT_FAILURE; + stats_avail = 0; + break; + } + stats_avail = 1; + } + mutex_unlock(&dev->counters_table[port_num - 1].mutex); + if (stats_avail) { memset(out_mad->data, 0, sizeof out_mad->data); - mode = ((struct mlx4_counter *)mailbox->buf)->counter_mode; - switch (mode & 0xf) { + switch (counter_stats.counter_mode & 0xf) { case 0: - edit_counter(mailbox->buf, - (void *)(out_mad->data + 40)); + edit_counter(&counter_stats, + (void *)(out_mad->data + 40)); err = IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_REPLY; break; default: @@ -784,25 +866,43 @@ } } - mlx4_free_cmd_mailbox(dev->dev, mailbox); - return err; } int mlx4_ib_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, - struct ib_wc *in_wc, struct ib_grh *in_grh, - struct ib_mad *in_mad, struct ib_mad *out_mad) + const struct ib_wc *in_wc, const struct ib_grh *in_grh, + const struct ib_mad_hdr *in, size_t in_mad_size, + struct ib_mad_hdr *out, size_t *out_mad_size, + u16 *out_mad_pkey_index) { - switch (rdma_port_get_link_layer(ibdev, port_num)) { - case IB_LINK_LAYER_INFINIBAND: + struct mlx4_ib_dev *dev = to_mdev(ibdev); + const struct ib_mad *in_mad = (const struct ib_mad *)in; + struct ib_mad *out_mad = (struct ib_mad *)out; + enum rdma_link_layer link = rdma_port_get_link_layer(ibdev, port_num); + + if (WARN_ON_ONCE(in_mad_size != sizeof(*in_mad) || + *out_mad_size != sizeof(*out_mad))) + return IB_MAD_RESULT_FAILURE; + + /* iboe_process_mad() which uses the HCA flow-counters to implement IB PMA + * queries, should be called only by VFs and for that specific purpose + */ + if (link == IB_LINK_LAYER_INFINIBAND) { + if (mlx4_is_slave(dev->dev) && + in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_PERF_MGMT && + in_mad->mad_hdr.attr_id == IB_PMA_PORT_COUNTERS) + return iboe_process_mad(ibdev, mad_flags, port_num, in_wc, + in_grh, in_mad, out_mad); + return ib_process_mad(ibdev, mad_flags, port_num, in_wc, in_grh, in_mad, out_mad); - case IB_LINK_LAYER_ETHERNET: - return iboe_process_mad(ibdev, mad_flags, port_num, in_wc, - in_grh, in_mad, out_mad); - default: - return -EINVAL; } + + if (link == IB_LINK_LAYER_ETHERNET) + return iboe_process_mad(ibdev, mad_flags, port_num, in_wc, + in_grh, in_mad, out_mad); + + return -EINVAL; } static void send_handler(struct ib_mad_agent *agent, @@ -827,7 +927,7 @@ agent = ib_register_mad_agent(&dev->ib_dev, p + 1, q ? IB_QPT_GSI : IB_QPT_SMI, NULL, 0, send_handler, - NULL, NULL); + NULL, NULL, 0); if (IS_ERR(agent)) { ret = PTR_ERR(agent); goto err; @@ -980,6 +1080,27 @@ /* Generate GUID changed event */ if (changed_attr & MLX4_EQ_PORT_INFO_GID_PFX_CHANGE_MASK) { + if (mlx4_is_master(dev->dev)) { + union ib_gid gid; + int err = 0; + + if (!eqe->event.port_mgmt_change.params.port_info.gid_prefix) + err = __mlx4_ib_query_gid(&dev->ib_dev, port, 0, &gid, 1); + else + gid.global.subnet_prefix = + eqe->event.port_mgmt_change.params.port_info.gid_prefix; + if (err) { + pr_warn("Could not change QP1 subnet prefix for port %d: query_gid error (%d)\n", + port, err); + } else { + pr_debug("Changing QP1 subnet prefix for port %d. old=0x%llx. new=0x%llx\n", + port, + (u64)atomic64_read(&dev->sriov.demux[port - 1].subnet_prefix), + be64_to_cpu(gid.global.subnet_prefix)); + atomic64_set(&dev->sriov.demux[port - 1].subnet_prefix, + be64_to_cpu(gid.global.subnet_prefix)); + } + } mlx4_ib_dispatch_event(dev, port, IB_EVENT_GID_CHANGE); /*if master, notify all slaves*/ if (mlx4_is_master(dev->dev)) @@ -1051,7 +1172,7 @@ sg_list.addr = tun_qp->ring[index].map; sg_list.length = size; - sg_list.lkey = ctx->mr->lkey; + sg_list.lkey = ctx->pd->local_dma_lkey; recv_wr.next = NULL; recv_wr.sg_list = &sg_list; @@ -1088,11 +1209,13 @@ int mlx4_ib_send_to_wire(struct mlx4_ib_dev *dev, int slave, u8 port, - enum ib_qp_type dest_qpt, u16 pkey_index, u32 remote_qpn, - u32 qkey, struct ib_ah_attr *attr, struct ib_mad *mad) + enum ib_qp_type dest_qpt, u16 pkey_index, + u32 remote_qpn, u32 qkey, struct ib_ah_attr *attr, + u8 *s_mac, u16 vlan_id, struct ib_mad *mad) { struct ib_sge list; - struct ib_send_wr wr, *bad_wr; + struct ib_ud_wr wr; + struct ib_send_wr *bad_wr; struct mlx4_ib_demux_pv_ctx *sqp_ctx; struct mlx4_ib_demux_pv_qp *sqp; struct mlx4_mad_snd_buf *sqp_mad; @@ -1111,10 +1234,6 @@ if (!sqp_ctx || sqp_ctx->state != DEMUX_PV_STATE_ACTIVE) return -EAGAIN; - /* QP0 forwarding only for Dom0 */ - if (dest_qpt == IB_QPT_SMI && (mlx4_master_func_num(dev->dev) != slave)) - return -EINVAL; - if (dest_qpt == IB_QPT_SMI) { src_qpnum = 0; sqp = &sqp_ctx->qp[0]; @@ -1165,27 +1284,55 @@ list.addr = sqp->tx_ring[wire_tx_ix].buf.map; list.length = sizeof (struct mlx4_mad_snd_buf); - list.lkey = sqp_ctx->mr->lkey; + list.lkey = sqp_ctx->pd->local_dma_lkey; - wr.wr.ud.ah = ah; - wr.wr.ud.port_num = port; - wr.wr.ud.pkey_index = wire_pkey_ix; - wr.wr.ud.remote_qkey = qkey; - wr.wr.ud.remote_qpn = remote_qpn; - wr.next = NULL; - wr.wr_id = ((u64) wire_tx_ix) | MLX4_TUN_SET_WRID_QPN(src_qpnum); - wr.sg_list = &list; - wr.num_sge = 1; - wr.opcode = IB_WR_SEND; - wr.send_flags = IB_SEND_SIGNALED; + wr.ah = ah; + wr.port_num = port; + wr.pkey_index = wire_pkey_ix; + wr.remote_qkey = qkey; + wr.remote_qpn = remote_qpn; + wr.wr.next = NULL; + wr.wr.wr_id = ((u64) wire_tx_ix) | MLX4_TUN_SET_WRID_QPN(src_qpnum); + wr.wr.sg_list = &list; + wr.wr.num_sge = 1; + wr.wr.opcode = IB_WR_SEND; + wr.wr.send_flags = IB_SEND_SIGNALED; + if (s_mac) + memcpy(to_mah(ah)->av.eth.s_mac, s_mac, 6); + if (vlan_id < 0x1000) + vlan_id |= (attr->sl & 7) << 13; + to_mah(ah)->av.eth.vlan = cpu_to_be16(vlan_id); - ret = ib_post_send(send_qp, &wr, &bad_wr); + + ret = ib_post_send(send_qp, &wr.wr, &bad_wr); + if (!ret) + return 0; + + spin_lock(&sqp->tx_lock); + sqp->tx_ix_tail++; + spin_unlock(&sqp->tx_lock); + sqp->tx_ring[wire_tx_ix].ah = NULL; out: - if (ret) - ib_destroy_ah(ah); + ib_destroy_ah(ah); return ret; } +static int get_slave_base_gid_ix(struct mlx4_ib_dev *dev, int slave, int port) +{ + if (rdma_port_get_link_layer(&dev->ib_dev, port) == IB_LINK_LAYER_INFINIBAND) + return slave; + return mlx4_get_base_gid_ix(dev->dev, slave, port); +} + +static void fill_in_real_sgid_index(struct mlx4_ib_dev *dev, int slave, int port, + struct ib_ah_attr *ah_attr) +{ + if (rdma_port_get_link_layer(&dev->ib_dev, port) == IB_LINK_LAYER_INFINIBAND) + ah_attr->grh.sgid_index = slave; + else + ah_attr->grh.sgid_index += get_slave_base_gid_ix(dev, slave, port); +} + static void mlx4_ib_multiplex_mad(struct mlx4_ib_demux_pv_ctx *ctx, struct ib_wc *wc) { struct mlx4_ib_dev *dev = to_mdev(ctx->ib_dev); @@ -1196,6 +1343,8 @@ struct ib_ah_attr ah_attr; u8 *slave_id; int slave; + int port; + u16 vlan_id; /* Get slave that sent this packet */ if (wc->src_qp < dev->dev->phys_caps.base_proxy_sqpn || @@ -1211,11 +1360,6 @@ "belongs to another slave\n", wc->src_qp); return; } - if (slave != mlx4_master_func_num(dev->dev) && !(wc->src_qp & 0x2)) { - mlx4_ib_warn(ctx->ib_dev, "can't multiplex bad sqp:%d: " - "non-master trying to send QP0 packets\n", wc->src_qp); - return; - } /* Map transaction ID */ ib_dma_sync_single_for_cpu(ctx->ib_dev, tun_qp->ring[wr_ix].map, @@ -1243,6 +1387,12 @@ /* Class-specific handling */ switch (tunnel->mad.mad_hdr.mgmt_class) { + case IB_MGMT_CLASS_SUBN_LID_ROUTED: + case IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE: + if (slave != mlx4_master_func_num(dev->dev) && + !mlx4_vf_smi_enabled(dev->dev, slave, ctx->port)) + return; + break; case IB_MGMT_CLASS_SUBN_ADM: if (mlx4_ib_multiplex_sa_handler(ctx->ib_dev, ctx->port, slave, (struct ib_sa_mad *) &tunnel->mad)) @@ -1271,13 +1421,22 @@ * stadard address handle by decoding the tunnelled mlx4_ah fields */ memcpy(&ah.av, &tunnel->hdr.av, sizeof (struct mlx4_av)); ah.ibah.device = ctx->ib_dev; - mlx4_ib_query_ah(&ah.ibah, &ah_attr); - if ((ah_attr.ah_flags & IB_AH_GRH) && - (ah_attr.grh.sgid_index != slave)) { - mlx4_ib_warn(ctx->ib_dev, "slave:%d accessed invalid sgid_index:%d\n", - slave, ah_attr.grh.sgid_index); + + port = be32_to_cpu(ah.av.ib.port_pd) >> 24; + port = mlx4_slave_convert_port(dev->dev, slave, port); + if (port < 0) return; - } + ah.av.ib.port_pd = cpu_to_be32(port << 24 | (be32_to_cpu(ah.av.ib.port_pd) & 0xffffff)); + + mlx4_ib_query_ah(&ah.ibah, &ah_attr); + if (ah_attr.ah_flags & IB_AH_GRH) + fill_in_real_sgid_index(dev, slave, ctx->port, &ah_attr); + + memcpy(ah_attr.dmac, tunnel->hdr.mac, 6); + vlan_id = be16_to_cpu(tunnel->hdr.vlan); + /* if slave have default vlan use it */ + mlx4_get_slave_default_vlan(dev->dev, ctx->port, slave, + &vlan_id, &ah_attr.sl); mlx4_ib_send_to_wire(dev, slave, ctx->port, is_proxy_qp0(dev, wc->src_qp, slave) ? @@ -1285,7 +1444,7 @@ be16_to_cpu(tunnel->hdr.pkey_index), be32_to_cpu(tunnel->hdr.remote_qpn), be32_to_cpu(tunnel->hdr.qkey), - &ah_attr, &tunnel->mad); + &ah_attr, wc->smac, vlan_id, &tunnel->mad); } static int mlx4_ib_alloc_pv_bufs(struct mlx4_ib_demux_pv_ctx *ctx, @@ -1330,6 +1489,10 @@ tun_qp->ring[i].addr, rx_buf_size, DMA_FROM_DEVICE); + if (ib_dma_mapping_error(ctx->ib_dev, tun_qp->ring[i].map)) { + kfree(tun_qp->ring[i].addr); + goto err; + } } for (i = 0; i < MLX4_NUM_TUNNEL_BUFS; i++) { @@ -1342,6 +1505,11 @@ tun_qp->tx_ring[i].buf.addr, tx_buf_size, DMA_TO_DEVICE); + if (ib_dma_mapping_error(ctx->ib_dev, + tun_qp->tx_ring[i].buf.map)) { + kfree(tun_qp->tx_ring[i].buf.addr); + goto tx_err; + } tun_qp->tx_ring[i].ah = NULL; } spin_lock_init(&tun_qp->tx_lock); @@ -1523,8 +1691,14 @@ memset(&attr, 0, sizeof attr); attr.qp_state = IB_QPS_INIT; - attr.pkey_index = - to_mdev(ctx->ib_dev)->pkeys.virt2phys_pkey[ctx->slave][ctx->port - 1][0]; + ret = 0; + if (create_tun) + ret = find_slave_port_pkey_ix(to_mdev(ctx->ib_dev), ctx->slave, + ctx->port, IB_DEFAULT_PKEY_FULL, + &attr.pkey_index); + if (ret || !create_tun) + attr.pkey_index = + to_mdev(ctx->ib_dev)->pkeys.virt2phys_pkey[ctx->slave][ctx->port - 1][0]; attr.qkey = IB_QP1_QKEY; attr.port_num = ctx->port; ret = ib_modify_qp(tun_qp->qp, &attr, qp_attr_mask_INIT); @@ -1658,14 +1832,15 @@ int create_tun, struct mlx4_ib_demux_pv_ctx *ctx) { int ret, cq_size; + struct ib_cq_init_attr cq_attr = {}; if (ctx->state != DEMUX_PV_STATE_DOWN) return -EEXIST; ctx->state = DEMUX_PV_STATE_STARTING; - /* have QP0 only on port owner, and only if link layer is IB */ - if (ctx->slave == mlx4_master_func_num(to_mdev(ctx->ib_dev)->dev) && - rdma_port_get_link_layer(ibdev, ctx->port) == IB_LINK_LAYER_INFINIBAND) + /* have QP0 only if link layer is IB */ + if (rdma_port_get_link_layer(ibdev, ctx->port) == + IB_LINK_LAYER_INFINIBAND) ctx->has_smi = 1; if (ctx->has_smi) { @@ -1686,8 +1861,9 @@ if (ctx->has_smi) cq_size *= 2; + cq_attr.cqe = cq_size; ctx->cq = ib_create_cq(ctx->ib_dev, mlx4_ib_tunnel_comp_handler, - NULL, ctx, cq_size, 0); + NULL, ctx, &cq_attr); if (IS_ERR(ctx->cq)) { ret = PTR_ERR(ctx->cq); pr_err("Couldn't create tunnel CQ (%d)\n", ret); @@ -1701,19 +1877,12 @@ goto err_cq; } - ctx->mr = ib_get_dma_mr(ctx->pd, IB_ACCESS_LOCAL_WRITE); - if (IS_ERR(ctx->mr)) { - ret = PTR_ERR(ctx->mr); - pr_err("Couldn't get tunnel DMA MR (%d)\n", ret); - goto err_pd; - } - if (ctx->has_smi) { ret = create_pv_sqp(ctx, IB_QPT_SMI, create_tun); if (ret) { pr_err("Couldn't create %s QP0 (%d)\n", create_tun ? "tunnel for" : "", ret); - goto err_mr; + goto err_pd; } } @@ -1750,10 +1919,6 @@ ib_destroy_qp(ctx->qp[0].qp); ctx->qp[0].qp = NULL; -err_mr: - ib_dereg_mr(ctx->mr); - ctx->mr = NULL; - err_pd: ib_dealloc_pd(ctx->pd); ctx->pd = NULL; @@ -1790,8 +1955,6 @@ ib_destroy_qp(ctx->qp[1].qp); ctx->qp[1].qp = NULL; mlx4_ib_free_pv_qp_bufs(ctx, IB_QPT_GSI, 1); - ib_dereg_mr(ctx->mr); - ctx->mr = NULL; ib_dealloc_pd(ctx->pd); ctx->pd = NULL; ib_destroy_cq(ctx->cq); @@ -1856,7 +2019,16 @@ ctx->port = port; ctx->ib_dev = &dev->ib_dev; - for (i = 0; i < dev->dev->caps.sqp_demux; i++) { + for (i = 0; + i < min(dev->dev->caps.sqp_demux, + (u16)(dev->dev->persist->num_vfs + 1)); + i++) { + struct mlx4_active_ports actv_ports = + mlx4_get_active_ports(dev->dev, i); + + if (!test_bit(port - 1, actv_ports.ports)) + continue; + ret = alloc_pv_object(dev, i, port, &ctx->tun[i]); if (ret) { ret = -ENOMEM; @@ -1915,8 +2087,6 @@ ib_destroy_qp(sqp_ctx->qp[1].qp); sqp_ctx->qp[1].qp = NULL; mlx4_ib_free_pv_qp_bufs(sqp_ctx, IB_QPT_GSI, 0); - ib_dereg_mr(sqp_ctx->mr); - sqp_ctx->mr = NULL; ib_dealloc_pd(sqp_ctx->pd); sqp_ctx->pd = NULL; ib_destroy_cq(sqp_ctx->cq); @@ -2005,6 +2175,8 @@ if (err) goto demux_err; dev->sriov.demux[i].guid_cache[0] = gid.global.interface_id; + atomic64_set(&dev->sriov.demux[i].subnet_prefix, + be64_to_cpu(gid.global.subnet_prefix)); err = alloc_pv_object(dev, mlx4_master_func_num(dev->dev), i + 1, &dev->sriov.sqps[i]); if (err)