+ tmpl->type = MLX5_RXQ_OBJ_TYPE_DEVX_HAIRPIN;
+ tmpl->rxq_ctrl = rxq_ctrl;
+ attr.hairpin = 1;
+ max_wq_data = priv->config.hca_attr.log_max_hairpin_wq_data_sz;
+ /* Jumbo frames > 9KB should be supported, and more packets. */
+ if (priv->config.log_hp_size != (uint32_t)MLX5_ARG_UNSET) {
+ if (priv->config.log_hp_size > max_wq_data) {
+ DRV_LOG(ERR, "total data size %u power of 2 is "
+ "too large for hairpin",
+ priv->config.log_hp_size);
+ rte_errno = ERANGE;
+ return NULL;
+ }
+ attr.wq_attr.log_hairpin_data_sz = priv->config.log_hp_size;
+ } else {
+ attr.wq_attr.log_hairpin_data_sz =
+ (max_wq_data < MLX5_HAIRPIN_JUMBO_LOG_SIZE) ?
+ max_wq_data : MLX5_HAIRPIN_JUMBO_LOG_SIZE;
+ }
+ /* Set the packets number to the maximum value for performance. */
+ attr.wq_attr.log_hairpin_num_packets =
+ attr.wq_attr.log_hairpin_data_sz -
+ MLX5_HAIRPIN_QUEUE_STRIDE;
+ tmpl->rq = mlx5_devx_cmd_create_rq(priv->sh->ctx, &attr,
+ rxq_ctrl->socket);
+ if (!tmpl->rq) {
+ DRV_LOG(ERR,
+ "port %u Rx hairpin queue %u can't create rq object",
+ dev->data->port_id, idx);
+ rte_errno = errno;
+ goto error;
+ }
+ DRV_LOG(DEBUG, "port %u rxq %u updated with %p", dev->data->port_id,
+ idx, (void *)&tmpl);
+ rte_atomic32_inc(&tmpl->refcnt);
+ LIST_INSERT_HEAD(&priv->rxqsobj, tmpl, next);
+ priv->verbs_alloc_ctx.type = MLX5_VERBS_ALLOC_TYPE_NONE;
+ return tmpl;
+error:
+ ret = rte_errno; /* Save rte_errno before cleanup. */
+ if (tmpl->rq)
+ mlx5_devx_cmd_destroy(tmpl->rq);
+ rte_errno = ret; /* Restore rte_errno. */
+ return NULL;
+}
+
+/**
+ * Create the Rx queue Verbs/DevX object.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ * @param idx
+ * Queue index in DPDK Rx queue array
+ * @param type
+ * Type of Rx queue object to create.
+ *
+ * @return
+ * The Verbs/DevX object initialised, NULL otherwise and rte_errno is set.
+ */
+struct mlx5_rxq_obj *
+mlx5_rxq_obj_new(struct rte_eth_dev *dev, uint16_t idx,
+ enum mlx5_rxq_obj_type type)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_rxq_data *rxq_data = (*priv->rxqs)[idx];
+ struct mlx5_rxq_ctrl *rxq_ctrl =
+ container_of(rxq_data, struct mlx5_rxq_ctrl, rxq);
+ struct ibv_wq_attr mod;
+ unsigned int cqe_n;
+ unsigned int wqe_n = 1 << rxq_data->elts_n;
+ struct mlx5_rxq_obj *tmpl = NULL;
+ struct mlx5dv_cq cq_info;
+ struct mlx5dv_rwq rwq;
+ int ret = 0;
+ struct mlx5dv_obj obj;
+
+ MLX5_ASSERT(rxq_data);
+ MLX5_ASSERT(!rxq_ctrl->obj);
+ if (type == MLX5_RXQ_OBJ_TYPE_DEVX_HAIRPIN)
+ return mlx5_rxq_obj_hairpin_new(dev, idx);
+ priv->verbs_alloc_ctx.type = MLX5_VERBS_ALLOC_TYPE_RX_QUEUE;
+ priv->verbs_alloc_ctx.obj = rxq_ctrl;
+ tmpl = rte_calloc_socket(__func__, 1, sizeof(*tmpl), 0,
+ rxq_ctrl->socket);
+ if (!tmpl) {
+ DRV_LOG(ERR,
+ "port %u Rx queue %u cannot allocate verbs resources",
+ dev->data->port_id, rxq_data->idx);
+ rte_errno = ENOMEM;
+ goto error;
+ }
+ tmpl->type = type;
+ tmpl->rxq_ctrl = rxq_ctrl;
+ if (rxq_ctrl->irq) {
+ tmpl->channel = mlx5_glue->create_comp_channel(priv->sh->ctx);
+ if (!tmpl->channel) {
+ DRV_LOG(ERR, "port %u: comp channel creation failure",
+ dev->data->port_id);
+ rte_errno = ENOMEM;
+ goto error;
+ }
+ }
+ if (mlx5_rxq_mprq_enabled(rxq_data))
+ cqe_n = wqe_n * (1 << rxq_data->strd_num_n) - 1;
+ else
+ cqe_n = wqe_n - 1;
+ tmpl->cq = mlx5_ibv_cq_new(dev, priv, rxq_data, cqe_n, tmpl);
+ if (!tmpl->cq) {
+ DRV_LOG(ERR, "port %u Rx queue %u CQ creation failure",
+ dev->data->port_id, idx);
+ rte_errno = ENOMEM;
+ goto error;
+ }
+ obj.cq.in = tmpl->cq;
+ obj.cq.out = &cq_info;
+ ret = mlx5_glue->dv_init_obj(&obj, MLX5DV_OBJ_CQ);
+ if (ret) {
+ rte_errno = ret;
+ goto error;
+ }
+ if (cq_info.cqe_size != RTE_CACHE_LINE_SIZE) {
+ DRV_LOG(ERR,
+ "port %u wrong MLX5_CQE_SIZE environment variable"
+ " value: it should be set to %u",
+ dev->data->port_id, RTE_CACHE_LINE_SIZE);
+ rte_errno = EINVAL;
+ goto error;
+ }
+ DRV_LOG(DEBUG, "port %u device_attr.max_qp_wr is %d",
+ dev->data->port_id, priv->sh->device_attr.orig_attr.max_qp_wr);
+ DRV_LOG(DEBUG, "port %u device_attr.max_sge is %d",
+ dev->data->port_id, priv->sh->device_attr.orig_attr.max_sge);
+ /* Allocate door-bell for types created with DevX. */
+ if (tmpl->type != MLX5_RXQ_OBJ_TYPE_IBV) {
+ struct mlx5_devx_dbr_page *dbr_page;
+ int64_t dbr_offset;
+
+ dbr_offset = mlx5_get_dbr(dev, &dbr_page);
+ if (dbr_offset < 0)
+ goto error;
+ rxq_ctrl->dbr_offset = dbr_offset;
+ rxq_ctrl->dbr_umem_id = dbr_page->umem->umem_id;
+ rxq_ctrl->dbr_umem_id_valid = 1;
+ rxq_data->rq_db = (uint32_t *)((uintptr_t)dbr_page->dbrs +
+ (uintptr_t)rxq_ctrl->dbr_offset);
+ }
+ if (tmpl->type == MLX5_RXQ_OBJ_TYPE_IBV) {
+ tmpl->wq = mlx5_ibv_wq_new(dev, priv, rxq_data, idx, wqe_n,
+ tmpl);
+ if (!tmpl->wq) {
+ DRV_LOG(ERR, "port %u Rx queue %u WQ creation failure",
+ dev->data->port_id, idx);
+ rte_errno = ENOMEM;
+ goto error;
+ }
+ /* Change queue state to ready. */
+ mod = (struct ibv_wq_attr){
+ .attr_mask = IBV_WQ_ATTR_STATE,
+ .wq_state = IBV_WQS_RDY,
+ };
+ ret = mlx5_glue->modify_wq(tmpl->wq, &mod);
+ if (ret) {
+ DRV_LOG(ERR,
+ "port %u Rx queue %u WQ state to IBV_WQS_RDY"
+ " failed", dev->data->port_id, idx);
+ rte_errno = ret;
+ goto error;
+ }
+ obj.rwq.in = tmpl->wq;
+ obj.rwq.out = &rwq;
+ ret = mlx5_glue->dv_init_obj(&obj, MLX5DV_OBJ_RWQ);
+ if (ret) {
+ rte_errno = ret;
+ goto error;
+ }
+ rxq_data->wqes = rwq.buf;
+ rxq_data->rq_db = rwq.dbrec;
+ } else if (tmpl->type == MLX5_RXQ_OBJ_TYPE_DEVX_RQ) {
+ struct mlx5_devx_modify_rq_attr rq_attr;
+
+ memset(&rq_attr, 0, sizeof(rq_attr));
+ tmpl->rq = mlx5_devx_rq_new(dev, idx, cq_info.cqn);
+ if (!tmpl->rq) {
+ DRV_LOG(ERR, "port %u Rx queue %u RQ creation failure",
+ dev->data->port_id, idx);
+ rte_errno = ENOMEM;
+ goto error;
+ }
+ /* Change queue state to ready. */
+ rq_attr.rq_state = MLX5_RQC_STATE_RST;
+ rq_attr.state = MLX5_RQC_STATE_RDY;
+ ret = mlx5_devx_cmd_modify_rq(tmpl->rq, &rq_attr);
+ if (ret)
+ goto error;
+ }
+ /* Fill the rings. */
+ rxq_data->cqe_n = log2above(cq_info.cqe_cnt);
+ rxq_data->cq_db = cq_info.dbrec;
+ rxq_data->cqes = (volatile struct mlx5_cqe (*)[])(uintptr_t)cq_info.buf;
+ rxq_data->cq_uar = cq_info.cq_uar;
+ rxq_data->cqn = cq_info.cqn;
+ rxq_data->cq_arm_sn = 0;
+ mlx5_rxq_initialize(rxq_data);
+ rxq_data->cq_ci = 0;
+ DRV_LOG(DEBUG, "port %u rxq %u updated with %p", dev->data->port_id,
+ idx, (void *)&tmpl);
+ rte_atomic32_inc(&tmpl->refcnt);
+ LIST_INSERT_HEAD(&priv->rxqsobj, tmpl, next);
+ priv->verbs_alloc_ctx.type = MLX5_VERBS_ALLOC_TYPE_NONE;
+ return tmpl;
+error:
+ if (tmpl) {
+ ret = rte_errno; /* Save rte_errno before cleanup. */
+ if (tmpl->type == MLX5_RXQ_OBJ_TYPE_IBV && tmpl->wq)
+ claim_zero(mlx5_glue->destroy_wq(tmpl->wq));
+ else if (tmpl->type == MLX5_RXQ_OBJ_TYPE_DEVX_RQ && tmpl->rq)
+ claim_zero(mlx5_devx_cmd_destroy(tmpl->rq));
+ if (tmpl->cq)
+ claim_zero(mlx5_glue->destroy_cq(tmpl->cq));
+ if (tmpl->channel)
+ claim_zero(mlx5_glue->destroy_comp_channel
+ (tmpl->channel));
+ rte_free(tmpl);
+ rte_errno = ret; /* Restore rte_errno. */
+ }
+ if (type == MLX5_RXQ_OBJ_TYPE_DEVX_RQ)
+ rxq_release_rq_resources(rxq_ctrl);
+ priv->verbs_alloc_ctx.type = MLX5_VERBS_ALLOC_TYPE_NONE;
+ return NULL;
+}
+
+/**
+ * Verify the Rx queue objects list is empty
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ *
+ * @return
+ * The number of objects not released.
+ */
+int
+mlx5_rxq_obj_verify(struct rte_eth_dev *dev)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ int ret = 0;
+ struct mlx5_rxq_obj *rxq_obj;
+
+ LIST_FOREACH(rxq_obj, &priv->rxqsobj, next) {
+ DRV_LOG(DEBUG, "port %u Rx queue %u still referenced",
+ dev->data->port_id, rxq_obj->rxq_ctrl->rxq.idx);
+ ++ret;
+ }
+ return ret;
+}
+
+/**
+ * Callback function to initialize mbufs for Multi-Packet RQ.
+ */
+static inline void
+mlx5_mprq_buf_init(struct rte_mempool *mp, void *opaque_arg,
+ void *_m, unsigned int i __rte_unused)
+{
+ struct mlx5_mprq_buf *buf = _m;
+ struct rte_mbuf_ext_shared_info *shinfo;
+ unsigned int strd_n = (unsigned int)(uintptr_t)opaque_arg;
+ unsigned int j;
+
+ memset(_m, 0, sizeof(*buf));
+ buf->mp = mp;
+ rte_atomic16_set(&buf->refcnt, 1);
+ for (j = 0; j != strd_n; ++j) {
+ shinfo = &buf->shinfos[j];
+ shinfo->free_cb = mlx5_mprq_buf_free_cb;
+ shinfo->fcb_opaque = buf;
+ }
+}
+
+/**
+ * Free mempool of Multi-Packet RQ.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ *
+ * @return
+ * 0 on success, negative errno value on failure.
+ */
+int
+mlx5_mprq_free_mp(struct rte_eth_dev *dev)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct rte_mempool *mp = priv->mprq_mp;
+ unsigned int i;
+
+ if (mp == NULL)
+ return 0;
+ DRV_LOG(DEBUG, "port %u freeing mempool (%s) for Multi-Packet RQ",
+ dev->data->port_id, mp->name);
+ /*
+ * If a buffer in the pool has been externally attached to a mbuf and it
+ * is still in use by application, destroying the Rx queue can spoil
+ * the packet. It is unlikely to happen but if application dynamically
+ * creates and destroys with holding Rx packets, this can happen.
+ *
+ * TODO: It is unavoidable for now because the mempool for Multi-Packet
+ * RQ isn't provided by application but managed by PMD.
+ */
+ if (!rte_mempool_full(mp)) {
+ DRV_LOG(ERR,
+ "port %u mempool for Multi-Packet RQ is still in use",
+ dev->data->port_id);
+ rte_errno = EBUSY;
+ return -rte_errno;
+ }
+ rte_mempool_free(mp);
+ /* Unset mempool for each Rx queue. */
+ for (i = 0; i != priv->rxqs_n; ++i) {
+ struct mlx5_rxq_data *rxq = (*priv->rxqs)[i];
+
+ if (rxq == NULL)
+ continue;
+ rxq->mprq_mp = NULL;
+ }
+ priv->mprq_mp = NULL;
+ return 0;
+}
+
+/**
+ * Allocate a mempool for Multi-Packet RQ. All configured Rx queues share the
+ * mempool. If already allocated, reuse it if there're enough elements.
+ * Otherwise, resize it.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ *
+ * @return
+ * 0 on success, negative errno value on failure.
+ */
+int
+mlx5_mprq_alloc_mp(struct rte_eth_dev *dev)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct rte_mempool *mp = priv->mprq_mp;
+ char name[RTE_MEMPOOL_NAMESIZE];
+ unsigned int desc = 0;
+ unsigned int buf_len;
+ unsigned int obj_num;
+ unsigned int obj_size;
+ unsigned int strd_num_n = 0;
+ unsigned int strd_sz_n = 0;
+ unsigned int i;
+ unsigned int n_ibv = 0;
+
+ if (!mlx5_mprq_enabled(dev))
+ return 0;
+ /* Count the total number of descriptors configured. */
+ for (i = 0; i != priv->rxqs_n; ++i) {
+ struct mlx5_rxq_data *rxq = (*priv->rxqs)[i];
+ struct mlx5_rxq_ctrl *rxq_ctrl = container_of
+ (rxq, struct mlx5_rxq_ctrl, rxq);
+
+ if (rxq == NULL || rxq_ctrl->type != MLX5_RXQ_TYPE_STANDARD)
+ continue;
+ n_ibv++;
+ desc += 1 << rxq->elts_n;
+ /* Get the max number of strides. */
+ if (strd_num_n < rxq->strd_num_n)
+ strd_num_n = rxq->strd_num_n;
+ /* Get the max size of a stride. */
+ if (strd_sz_n < rxq->strd_sz_n)
+ strd_sz_n = rxq->strd_sz_n;
+ }
+ MLX5_ASSERT(strd_num_n && strd_sz_n);
+ buf_len = (1 << strd_num_n) * (1 << strd_sz_n);
+ obj_size = sizeof(struct mlx5_mprq_buf) + buf_len + (1 << strd_num_n) *
+ sizeof(struct rte_mbuf_ext_shared_info) + RTE_PKTMBUF_HEADROOM;
+ /*
+ * Received packets can be either memcpy'd or externally referenced. In
+ * case that the packet is attached to an mbuf as an external buffer, as
+ * it isn't possible to predict how the buffers will be queued by
+ * application, there's no option to exactly pre-allocate needed buffers
+ * in advance but to speculatively prepares enough buffers.
+ *
+ * In the data path, if this Mempool is depleted, PMD will try to memcpy
+ * received packets to buffers provided by application (rxq->mp) until
+ * this Mempool gets available again.
+ */
+ desc *= 4;
+ obj_num = desc + MLX5_MPRQ_MP_CACHE_SZ * n_ibv;
+ /*
+ * rte_mempool_create_empty() has sanity check to refuse large cache
+ * size compared to the number of elements.
+ * CACHE_FLUSHTHRESH_MULTIPLIER is defined in a C file, so using a
+ * constant number 2 instead.
+ */
+ obj_num = RTE_MAX(obj_num, MLX5_MPRQ_MP_CACHE_SZ * 2);
+ /* Check a mempool is already allocated and if it can be resued. */
+ if (mp != NULL && mp->elt_size >= obj_size && mp->size >= obj_num) {
+ DRV_LOG(DEBUG, "port %u mempool %s is being reused",
+ dev->data->port_id, mp->name);
+ /* Reuse. */
+ goto exit;
+ } else if (mp != NULL) {
+ DRV_LOG(DEBUG, "port %u mempool %s should be resized, freeing it",
+ dev->data->port_id, mp->name);
+ /*
+ * If failed to free, which means it may be still in use, no way
+ * but to keep using the existing one. On buffer underrun,
+ * packets will be memcpy'd instead of external buffer
+ * attachment.
+ */
+ if (mlx5_mprq_free_mp(dev)) {
+ if (mp->elt_size >= obj_size)
+ goto exit;
+ else
+ return -rte_errno;
+ }
+ }
+ snprintf(name, sizeof(name), "port-%u-mprq", dev->data->port_id);
+ mp = rte_mempool_create(name, obj_num, obj_size, MLX5_MPRQ_MP_CACHE_SZ,
+ 0, NULL, NULL, mlx5_mprq_buf_init,
+ (void *)(uintptr_t)(1 << strd_num_n),
+ dev->device->numa_node, 0);
+ if (mp == NULL) {
+ DRV_LOG(ERR,
+ "port %u failed to allocate a mempool for"
+ " Multi-Packet RQ, count=%u, size=%u",
+ dev->data->port_id, obj_num, obj_size);
+ rte_errno = ENOMEM;
+ return -rte_errno;
+ }
+ priv->mprq_mp = mp;
+exit:
+ /* Set mempool for each Rx queue. */
+ for (i = 0; i != priv->rxqs_n; ++i) {
+ struct mlx5_rxq_data *rxq = (*priv->rxqs)[i];
+ struct mlx5_rxq_ctrl *rxq_ctrl = container_of
+ (rxq, struct mlx5_rxq_ctrl, rxq);
+
+ if (rxq == NULL || rxq_ctrl->type != MLX5_RXQ_TYPE_STANDARD)
+ continue;
+ rxq->mprq_mp = mp;
+ }
+ DRV_LOG(INFO, "port %u Multi-Packet RQ is configured",
+ dev->data->port_id);
+ return 0;
+}
+
+#define MLX5_MAX_TCP_HDR_OFFSET ((unsigned int)(sizeof(struct rte_ether_hdr) + \
+ sizeof(struct rte_vlan_hdr) * 2 + \
+ sizeof(struct rte_ipv6_hdr)))
+#define MAX_TCP_OPTION_SIZE 40u
+#define MLX5_MAX_LRO_HEADER_FIX ((unsigned int)(MLX5_MAX_TCP_HDR_OFFSET + \
+ sizeof(struct rte_tcp_hdr) + \
+ MAX_TCP_OPTION_SIZE))
+
+/**
+ * Adjust the maximum LRO massage size.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ * @param idx
+ * RX queue index.
+ * @param max_lro_size
+ * The maximum size for LRO packet.
+ */
+static void
+mlx5_max_lro_msg_size_adjust(struct rte_eth_dev *dev, uint16_t idx,
+ uint32_t max_lro_size)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+
+ if (priv->config.hca_attr.lro_max_msg_sz_mode ==
+ MLX5_LRO_MAX_MSG_SIZE_START_FROM_L4 && max_lro_size >
+ MLX5_MAX_TCP_HDR_OFFSET)
+ max_lro_size -= MLX5_MAX_TCP_HDR_OFFSET;
+ max_lro_size = RTE_MIN(max_lro_size, MLX5_MAX_LRO_SIZE);
+ MLX5_ASSERT(max_lro_size >= MLX5_LRO_SEG_CHUNK_SIZE);
+ max_lro_size /= MLX5_LRO_SEG_CHUNK_SIZE;
+ if (priv->max_lro_msg_size)
+ priv->max_lro_msg_size =
+ RTE_MIN((uint32_t)priv->max_lro_msg_size, max_lro_size);
+ else
+ priv->max_lro_msg_size = max_lro_size;
+ DRV_LOG(DEBUG,
+ "port %u Rx Queue %u max LRO message size adjusted to %u bytes",
+ dev->data->port_id, idx,
+ priv->max_lro_msg_size * MLX5_LRO_SEG_CHUNK_SIZE);
+}
+
+/**
+ * Create a DPDK Rx queue.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ * @param idx
+ * RX queue index.
+ * @param desc
+ * Number of descriptors to configure in queue.
+ * @param socket
+ * NUMA socket on which memory must be allocated.
+ *
+ * @return
+ * A DPDK queue object on success, NULL otherwise and rte_errno is set.
+ */
+struct mlx5_rxq_ctrl *
+mlx5_rxq_new(struct rte_eth_dev *dev, uint16_t idx, uint16_t desc,
+ unsigned int socket, const struct rte_eth_rxconf *conf,
+ struct rte_mempool *mp)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_rxq_ctrl *tmpl;
+ unsigned int mb_len = rte_pktmbuf_data_room_size(mp);
+ unsigned int mprq_stride_size;
+ struct mlx5_dev_config *config = &priv->config;
+ unsigned int strd_headroom_en;
+ /*
+ * Always allocate extra slots, even if eventually
+ * the vector Rx will not be used.
+ */
+ uint16_t desc_n =
+ desc + config->rx_vec_en * MLX5_VPMD_DESCS_PER_LOOP;
+ uint64_t offloads = conf->offloads |
+ dev->data->dev_conf.rxmode.offloads;
+ unsigned int lro_on_queue = !!(offloads & DEV_RX_OFFLOAD_TCP_LRO);
+ const int mprq_en = mlx5_check_mprq_support(dev) > 0;
+ unsigned int max_rx_pkt_len = lro_on_queue ?
+ dev->data->dev_conf.rxmode.max_lro_pkt_size :
+ dev->data->dev_conf.rxmode.max_rx_pkt_len;
+ unsigned int non_scatter_min_mbuf_size = max_rx_pkt_len +
+ RTE_PKTMBUF_HEADROOM;
+ unsigned int max_lro_size = 0;
+ unsigned int first_mb_free_size = mb_len - RTE_PKTMBUF_HEADROOM;
+
+ if (non_scatter_min_mbuf_size > mb_len && !(offloads &
+ DEV_RX_OFFLOAD_SCATTER)) {
+ DRV_LOG(ERR, "port %u Rx queue %u: Scatter offload is not"
+ " configured and no enough mbuf space(%u) to contain "
+ "the maximum RX packet length(%u) with head-room(%u)",
+ dev->data->port_id, idx, mb_len, max_rx_pkt_len,
+ RTE_PKTMBUF_HEADROOM);
+ rte_errno = ENOSPC;
+ return NULL;
+ }
+ tmpl = rte_calloc_socket("RXQ", 1,
+ sizeof(*tmpl) +
+ desc_n * sizeof(struct rte_mbuf *),
+ 0, socket);
+ if (!tmpl) {
+ rte_errno = ENOMEM;
+ return NULL;
+ }
+ tmpl->type = MLX5_RXQ_TYPE_STANDARD;
+ if (mlx5_mr_btree_init(&tmpl->rxq.mr_ctrl.cache_bh,
+ MLX5_MR_BTREE_CACHE_N, socket)) {
+ /* rte_errno is already set. */
+ goto error;
+ }
+ tmpl->socket = socket;
+ if (dev->data->dev_conf.intr_conf.rxq)
+ tmpl->irq = 1;
+ /*
+ * LRO packet may consume all the stride memory, hence we cannot
+ * guaranty head-room near the packet memory in the stride.
+ * In this case scatter is, for sure, enabled and an empty mbuf may be
+ * added in the start for the head-room.
+ */
+ if (lro_on_queue && RTE_PKTMBUF_HEADROOM > 0 &&
+ non_scatter_min_mbuf_size > mb_len) {
+ strd_headroom_en = 0;
+ mprq_stride_size = RTE_MIN(max_rx_pkt_len,
+ 1u << config->mprq.max_stride_size_n);
+ } else {
+ strd_headroom_en = 1;
+ mprq_stride_size = non_scatter_min_mbuf_size;
+ }
+ /*
+ * This Rx queue can be configured as a Multi-Packet RQ if all of the
+ * following conditions are met:
+ * - MPRQ is enabled.
+ * - The number of descs is more than the number of strides.
+ * - max_rx_pkt_len plus overhead is less than the max size of a
+ * stride.
+ * Otherwise, enable Rx scatter if necessary.
+ */
+ if (mprq_en &&
+ desc > (1U << config->mprq.stride_num_n) &&
+ mprq_stride_size <= (1U << config->mprq.max_stride_size_n)) {
+ /* TODO: Rx scatter isn't supported yet. */
+ tmpl->rxq.sges_n = 0;
+ /* Trim the number of descs needed. */
+ desc >>= config->mprq.stride_num_n;
+ tmpl->rxq.strd_num_n = config->mprq.stride_num_n;
+ tmpl->rxq.strd_sz_n = RTE_MAX(log2above(mprq_stride_size),
+ config->mprq.min_stride_size_n);
+ tmpl->rxq.strd_shift_en = MLX5_MPRQ_TWO_BYTE_SHIFT;
+ tmpl->rxq.strd_headroom_en = strd_headroom_en;
+ tmpl->rxq.mprq_max_memcpy_len = RTE_MIN(first_mb_free_size,
+ config->mprq.max_memcpy_len);
+ max_lro_size = RTE_MIN(max_rx_pkt_len,
+ (1u << tmpl->rxq.strd_num_n) *
+ (1u << tmpl->rxq.strd_sz_n));
+ DRV_LOG(DEBUG,
+ "port %u Rx queue %u: Multi-Packet RQ is enabled"
+ " strd_num_n = %u, strd_sz_n = %u",
+ dev->data->port_id, idx,
+ tmpl->rxq.strd_num_n, tmpl->rxq.strd_sz_n);
+ } else if (max_rx_pkt_len <= first_mb_free_size) {
+ tmpl->rxq.sges_n = 0;
+ max_lro_size = max_rx_pkt_len;
+ } else if (offloads & DEV_RX_OFFLOAD_SCATTER) {
+ unsigned int size = non_scatter_min_mbuf_size;
+ unsigned int sges_n;
+
+ if (lro_on_queue && first_mb_free_size <
+ MLX5_MAX_LRO_HEADER_FIX) {
+ DRV_LOG(ERR, "Not enough space in the first segment(%u)"
+ " to include the max header size(%u) for LRO",
+ first_mb_free_size, MLX5_MAX_LRO_HEADER_FIX);
+ rte_errno = ENOTSUP;
+ goto error;
+ }
+ /*
+ * Determine the number of SGEs needed for a full packet
+ * and round it to the next power of two.
+ */
+ sges_n = log2above((size / mb_len) + !!(size % mb_len));
+ if (sges_n > MLX5_MAX_LOG_RQ_SEGS) {
+ DRV_LOG(ERR,
+ "port %u too many SGEs (%u) needed to handle"
+ " requested maximum packet size %u, the maximum"
+ " supported are %u", dev->data->port_id,
+ 1 << sges_n, max_rx_pkt_len,
+ 1u << MLX5_MAX_LOG_RQ_SEGS);
+ rte_errno = ENOTSUP;
+ goto error;
+ }
+ tmpl->rxq.sges_n = sges_n;
+ max_lro_size = max_rx_pkt_len;
+ }
+ if (mprq_en && !mlx5_rxq_mprq_enabled(&tmpl->rxq))
+ DRV_LOG(WARNING,
+ "port %u MPRQ is requested but cannot be enabled"
+ " (requested: desc = %u, stride_sz = %u,"
+ " supported: min_stride_num = %u, max_stride_sz = %u).",
+ dev->data->port_id, desc, mprq_stride_size,
+ (1 << config->mprq.stride_num_n),
+ (1 << config->mprq.max_stride_size_n));
+ DRV_LOG(DEBUG, "port %u maximum number of segments per packet: %u",
+ dev->data->port_id, 1 << tmpl->rxq.sges_n);
+ if (desc % (1 << tmpl->rxq.sges_n)) {
+ DRV_LOG(ERR,
+ "port %u number of Rx queue descriptors (%u) is not a"
+ " multiple of SGEs per packet (%u)",
+ dev->data->port_id,
+ desc,
+ 1 << tmpl->rxq.sges_n);
+ rte_errno = EINVAL;
+ goto error;
+ }
+ mlx5_max_lro_msg_size_adjust(dev, idx, max_lro_size);
+ /* Toggle RX checksum offload if hardware supports it. */
+ tmpl->rxq.csum = !!(offloads & DEV_RX_OFFLOAD_CHECKSUM);
+ tmpl->rxq.hw_timestamp = !!(offloads & DEV_RX_OFFLOAD_TIMESTAMP);
+ /* Configure VLAN stripping. */
+ tmpl->rxq.vlan_strip = !!(offloads & DEV_RX_OFFLOAD_VLAN_STRIP);
+ /* By default, FCS (CRC) is stripped by hardware. */
+ tmpl->rxq.crc_present = 0;
+ tmpl->rxq.lro = lro_on_queue;
+ if (offloads & DEV_RX_OFFLOAD_KEEP_CRC) {
+ if (config->hw_fcs_strip) {
+ /*
+ * RQs used for LRO-enabled TIRs should not be
+ * configured to scatter the FCS.
+ */
+ if (lro_on_queue)
+ DRV_LOG(WARNING,
+ "port %u CRC stripping has been "
+ "disabled but will still be performed "
+ "by hardware, because LRO is enabled",
+ dev->data->port_id);
+ else
+ tmpl->rxq.crc_present = 1;
+ } else {
+ DRV_LOG(WARNING,
+ "port %u CRC stripping has been disabled but will"
+ " still be performed by hardware, make sure MLNX_OFED"
+ " and firmware are up to date",
+ dev->data->port_id);
+ }
+ }
+ DRV_LOG(DEBUG,
+ "port %u CRC stripping is %s, %u bytes will be subtracted from"
+ " incoming frames to hide it",
+ dev->data->port_id,
+ tmpl->rxq.crc_present ? "disabled" : "enabled",
+ tmpl->rxq.crc_present << 2);
+ /* Save port ID. */
+ tmpl->rxq.rss_hash = !!priv->rss_conf.rss_hf &&
+ (!!(dev->data->dev_conf.rxmode.mq_mode & ETH_MQ_RX_RSS));
+ tmpl->rxq.port_id = dev->data->port_id;
+ tmpl->priv = priv;
+ tmpl->rxq.mp = mp;
+ tmpl->rxq.elts_n = log2above(desc);
+ tmpl->rxq.rq_repl_thresh =
+ MLX5_VPMD_RXQ_RPLNSH_THRESH(1 << tmpl->rxq.elts_n);
+ tmpl->rxq.elts =
+ (struct rte_mbuf *(*)[1 << tmpl->rxq.elts_n])(tmpl + 1);
+#ifndef RTE_ARCH_64
+ tmpl->rxq.uar_lock_cq = &priv->uar_lock_cq;
+#endif
+ tmpl->rxq.idx = idx;
+ rte_atomic32_inc(&tmpl->refcnt);
+ LIST_INSERT_HEAD(&priv->rxqsctrl, tmpl, next);
+ return tmpl;
+error:
+ rte_free(tmpl);
+ return NULL;
+}
+
+/**
+ * Create a DPDK Rx hairpin queue.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ * @param idx
+ * RX queue index.
+ * @param desc
+ * Number of descriptors to configure in queue.
+ * @param hairpin_conf
+ * The hairpin binding configuration.
+ *
+ * @return
+ * A DPDK queue object on success, NULL otherwise and rte_errno is set.
+ */
+struct mlx5_rxq_ctrl *
+mlx5_rxq_hairpin_new(struct rte_eth_dev *dev, uint16_t idx, uint16_t desc,
+ const struct rte_eth_hairpin_conf *hairpin_conf)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_rxq_ctrl *tmpl;
+
+ tmpl = rte_calloc_socket("RXQ", 1, sizeof(*tmpl), 0, SOCKET_ID_ANY);
+ if (!tmpl) {
+ rte_errno = ENOMEM;
+ return NULL;
+ }
+ tmpl->type = MLX5_RXQ_TYPE_HAIRPIN;
+ tmpl->socket = SOCKET_ID_ANY;
+ tmpl->rxq.rss_hash = 0;
+ tmpl->rxq.port_id = dev->data->port_id;
+ tmpl->priv = priv;
+ tmpl->rxq.mp = NULL;
+ tmpl->rxq.elts_n = log2above(desc);
+ tmpl->rxq.elts = NULL;
+ tmpl->rxq.mr_ctrl.cache_bh = (struct mlx5_mr_btree) { 0 };
+ tmpl->hairpin_conf = *hairpin_conf;
+ tmpl->rxq.idx = idx;
+ rte_atomic32_inc(&tmpl->refcnt);
+ LIST_INSERT_HEAD(&priv->rxqsctrl, tmpl, next);
+ return tmpl;
+}
+
+/**
+ * Get a Rx queue.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ * @param idx
+ * RX queue index.
+ *
+ * @return
+ * A pointer to the queue if it exists, NULL otherwise.
+ */
+struct mlx5_rxq_ctrl *
+mlx5_rxq_get(struct rte_eth_dev *dev, uint16_t idx)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_rxq_ctrl *rxq_ctrl = NULL;
+
+ if ((*priv->rxqs)[idx]) {
+ rxq_ctrl = container_of((*priv->rxqs)[idx],
+ struct mlx5_rxq_ctrl,
+ rxq);
+ mlx5_rxq_obj_get(dev, idx);
+ rte_atomic32_inc(&rxq_ctrl->refcnt);
+ }
+ return rxq_ctrl;
+}
+
+/**
+ * Release a Rx queue.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ * @param idx
+ * RX queue index.
+ *
+ * @return
+ * 1 while a reference on it exists, 0 when freed.
+ */
+int
+mlx5_rxq_release(struct rte_eth_dev *dev, uint16_t idx)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_rxq_ctrl *rxq_ctrl;
+
+ if (!(*priv->rxqs)[idx])
+ return 0;
+ rxq_ctrl = container_of((*priv->rxqs)[idx], struct mlx5_rxq_ctrl, rxq);
+ MLX5_ASSERT(rxq_ctrl->priv);
+ if (rxq_ctrl->obj && !mlx5_rxq_obj_release(rxq_ctrl->obj))
+ rxq_ctrl->obj = NULL;
+ if (rte_atomic32_dec_and_test(&rxq_ctrl->refcnt)) {
+ if (rxq_ctrl->dbr_umem_id_valid)
+ claim_zero(mlx5_release_dbr(dev, rxq_ctrl->dbr_umem_id,
+ rxq_ctrl->dbr_offset));
+ if (rxq_ctrl->type == MLX5_RXQ_TYPE_STANDARD)
+ mlx5_mr_btree_free(&rxq_ctrl->rxq.mr_ctrl.cache_bh);
+ LIST_REMOVE(rxq_ctrl, next);
+ rte_free(rxq_ctrl);
+ (*priv->rxqs)[idx] = NULL;
+ return 0;
+ }
+ return 1;
+}
+
+/**
+ * Verify the Rx Queue list is empty
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ *
+ * @return
+ * The number of object not released.
+ */
+int
+mlx5_rxq_verify(struct rte_eth_dev *dev)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_rxq_ctrl *rxq_ctrl;
+ int ret = 0;
+
+ LIST_FOREACH(rxq_ctrl, &priv->rxqsctrl, next) {
+ DRV_LOG(DEBUG, "port %u Rx Queue %u still referenced",
+ dev->data->port_id, rxq_ctrl->rxq.idx);
+ ++ret;
+ }
+ return ret;
+}
+
+/**
+ * Get a Rx queue type.
+ *
+ * @param dev
+ * Pointer to Ethernet device.
+ * @param idx
+ * Rx queue index.
+ *
+ * @return
+ * The Rx queue type.
+ */
+enum mlx5_rxq_type
+mlx5_rxq_get_type(struct rte_eth_dev *dev, uint16_t idx)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_rxq_ctrl *rxq_ctrl = NULL;
+
+ if (idx < priv->rxqs_n && (*priv->rxqs)[idx]) {
+ rxq_ctrl = container_of((*priv->rxqs)[idx],
+ struct mlx5_rxq_ctrl,
+ rxq);
+ return rxq_ctrl->type;