[v3,04/15] net/idpf: add queue setup and release in split queue model

Message ID 20221018111245.890651-5-junfeng.guo@intel.com (mailing list archive)
State Superseded, archived
Delegated to: Andrew Rybchenko
Headers
Series add support for idpf PMD in DPDK |

Checks

Context Check Description
ci/checkpatch success coding style OK

Commit Message

Junfeng Guo Oct. 18, 2022, 11:12 a.m. UTC
  Add support for queue operations in split queue model:
 - rx_queue_setup
 - rx_queue_release
 - tx_queue_setup
 - tx_queue_release

In the split queue model, "RX buffer queues" are used to pass
descriptor buffers from SW to HW while Rx queues are used only to
pass the descriptor completions, that is, descriptors that point
to completed buffers, from HW to SW. This is contrary to the single
queue model in which Rx queues are used for both purposes.

Signed-off-by: Beilei Xing <beilei.xing@intel.com>
Signed-off-by: Xiaoyun Li <xiaoyun.li@intel.com>
Signed-off-by: Junfeng Guo <junfeng.guo@intel.com>
---
 drivers/net/idpf/idpf_ethdev.c |  34 ++-
 drivers/net/idpf/idpf_ethdev.h |  11 +
 drivers/net/idpf/idpf_rxtx.c   | 485 ++++++++++++++++++++++++++++++++-
 drivers/net/idpf/idpf_vchnl.c  | 127 ++++++++-
 4 files changed, 644 insertions(+), 13 deletions(-)
  

Patch

diff --git a/drivers/net/idpf/idpf_ethdev.c b/drivers/net/idpf/idpf_ethdev.c
index c0a6b3e90d..21520fa648 100644
--- a/drivers/net/idpf/idpf_ethdev.c
+++ b/drivers/net/idpf/idpf_ethdev.c
@@ -69,13 +69,25 @@  idpf_init_vport_req_info(struct rte_eth_dev *dev)
 		(struct virtchnl2_create_vport *)adapter->vport_req_info[idx];
 
 	vport_info->vport_type = rte_cpu_to_le_16(VIRTCHNL2_VPORT_TYPE_DEFAULT);
-	if (adapter->txq_model) {
+	if (!adapter->txq_model) {
+		vport_info->txq_model =
+			rte_cpu_to_le_16(VIRTCHNL2_QUEUE_MODEL_SPLIT);
+		vport_info->num_tx_q = IDPF_DEFAULT_TXQ_NUM;
+		vport_info->num_tx_complq =
+			IDPF_DEFAULT_TXQ_NUM * IDPF_TX_COMPLQ_PER_GRP;
+	} else {
 		vport_info->txq_model =
 			rte_cpu_to_le_16(VIRTCHNL2_QUEUE_MODEL_SINGLE);
 		vport_info->num_tx_q = IDPF_DEFAULT_TXQ_NUM;
 		vport_info->num_tx_complq = 0;
 	}
-	if (adapter->rxq_model) {
+	if (!adapter->rxq_model) {
+		vport_info->rxq_model =
+			rte_cpu_to_le_16(VIRTCHNL2_QUEUE_MODEL_SPLIT);
+		vport_info->num_rx_q = IDPF_DEFAULT_RXQ_NUM;
+		vport_info->num_rx_bufq =
+			IDPF_DEFAULT_RXQ_NUM * IDPF_RX_BUFQ_PER_GRP;
+	} else {
 		vport_info->rxq_model =
 			rte_cpu_to_le_16(VIRTCHNL2_QUEUE_MODEL_SINGLE);
 		vport_info->num_rx_q = IDPF_DEFAULT_RXQ_NUM;
@@ -113,7 +125,9 @@  idpf_init_vport(struct rte_eth_dev *dev)
 	vport->txq_model = vport_info->txq_model;
 	vport->rxq_model = vport_info->rxq_model;
 	vport->num_tx_q = vport_info->num_tx_q;
+	vport->num_tx_complq = vport_info->num_tx_complq;
 	vport->num_rx_q = vport_info->num_rx_q;
+	vport->num_rx_bufq = vport_info->num_rx_bufq;
 	vport->max_mtu = vport_info->max_mtu;
 	rte_memcpy(vport->default_mac_addr,
 		   vport_info->default_mac_addr, ETH_ALEN);
@@ -136,6 +150,22 @@  idpf_init_vport(struct rte_eth_dev *dev)
 				vport_info->chunks.chunks[i].qtail_reg_start;
 			vport->chunks_info.rx_qtail_spacing =
 				vport_info->chunks.chunks[i].qtail_reg_spacing;
+		} else if (vport_info->chunks.chunks[i].type ==
+			 VIRTCHNL2_QUEUE_TYPE_TX_COMPLETION) {
+			vport->chunks_info.tx_compl_start_qid =
+				vport_info->chunks.chunks[i].start_queue_id;
+			vport->chunks_info.tx_compl_qtail_start =
+				vport_info->chunks.chunks[i].qtail_reg_start;
+			vport->chunks_info.tx_compl_qtail_spacing =
+				vport_info->chunks.chunks[i].qtail_reg_spacing;
+		} else if (vport_info->chunks.chunks[i].type ==
+			 VIRTCHNL2_QUEUE_TYPE_RX_BUFFER) {
+			vport->chunks_info.rx_buf_start_qid =
+				vport_info->chunks.chunks[i].start_queue_id;
+			vport->chunks_info.rx_buf_qtail_start =
+				vport_info->chunks.chunks[i].qtail_reg_start;
+			vport->chunks_info.rx_buf_qtail_spacing =
+				vport_info->chunks.chunks[i].qtail_reg_spacing;
 		}
 	}
 
diff --git a/drivers/net/idpf/idpf_ethdev.h b/drivers/net/idpf/idpf_ethdev.h
index d90bcfa63a..9f8fcba4ca 100644
--- a/drivers/net/idpf/idpf_ethdev.h
+++ b/drivers/net/idpf/idpf_ethdev.h
@@ -73,11 +73,18 @@  enum idpf_vc_result {
 struct idpf_chunks_info {
 	uint32_t tx_start_qid;
 	uint32_t rx_start_qid;
+	/* Valid only if split queue model */
+	uint32_t tx_compl_start_qid;
+	uint32_t rx_buf_start_qid;
 
 	uint64_t tx_qtail_start;
 	uint32_t tx_qtail_spacing;
 	uint64_t rx_qtail_start;
 	uint32_t rx_qtail_spacing;
+	uint64_t tx_compl_qtail_start;
+	uint32_t tx_compl_qtail_spacing;
+	uint64_t rx_buf_qtail_start;
+	uint32_t rx_buf_qtail_spacing;
 };
 
 struct idpf_vport {
@@ -86,7 +93,11 @@  struct idpf_vport {
 	uint32_t txq_model;
 	uint32_t rxq_model;
 	uint16_t num_tx_q;
+	/* valid only if txq_model is split Q */
+	uint16_t num_tx_complq;
 	uint16_t num_rx_q;
+	/* valid only if rxq_model is split Q */
+	uint16_t num_rx_bufq;
 
 	uint16_t max_mtu;
 	uint8_t default_mac_addr[VIRTCHNL_ETH_LENGTH_OF_ADDRESS];
diff --git a/drivers/net/idpf/idpf_rxtx.c b/drivers/net/idpf/idpf_rxtx.c
index 9fef467990..07ee8f1e14 100644
--- a/drivers/net/idpf/idpf_rxtx.c
+++ b/drivers/net/idpf/idpf_rxtx.c
@@ -97,7 +97,10 @@  release_txq_mbufs(struct idpf_tx_queue *txq)
 		return;
 	}
 
-	if (!txq->sw_nb_desc) {
+	if (txq->sw_nb_desc) {
+		/* For split queue model, descriptor ring */
+		nb_desc = txq->sw_nb_desc;
+	} else {
 		/* For single queue model */
 		nb_desc = txq->nb_tx_desc;
 	}
@@ -125,6 +128,21 @@  idpf_rx_queue_release(void *rxq)
 	if (!q)
 		return;
 
+	/* Split queue */
+	if (q->bufq1 && q->bufq2) {
+		q->bufq1->ops->release_mbufs(q->bufq1);
+		rte_free(q->bufq1->sw_ring);
+		rte_memzone_free(q->bufq1->mz);
+		rte_free(q->bufq1);
+		q->bufq2->ops->release_mbufs(q->bufq2);
+		rte_free(q->bufq2->sw_ring);
+		rte_memzone_free(q->bufq2->mz);
+		rte_free(q->bufq2);
+		rte_memzone_free(q->mz);
+		rte_free(q);
+		return;
+	}
+
 	/* Single queue */
 	q->ops->release_mbufs(q);
 	rte_free(q->sw_ring);
@@ -147,6 +165,65 @@  idpf_tx_queue_release(void *txq)
 	rte_free(q);
 }
 
+static inline void
+reset_split_rx_descq(struct idpf_rx_queue *rxq)
+{
+	uint16_t len;
+	uint32_t i;
+
+	if (!rxq)
+		return;
+
+	len = rxq->nb_rx_desc + IDPF_RX_MAX_BURST;
+
+	for (i = 0; i < len * sizeof(struct virtchnl2_rx_flex_desc_adv_nic_3);
+	     i++)
+		((volatile char *)rxq->rx_ring)[i] = 0;
+
+	rxq->rx_tail = 0;
+	rxq->expected_gen_id = 1;
+}
+
+static inline void
+reset_split_rx_bufq(struct idpf_rx_queue *rxq)
+{
+	uint16_t len;
+	uint32_t i;
+
+	if (!rxq)
+		return;
+
+	len = rxq->nb_rx_desc + IDPF_RX_MAX_BURST;
+
+	for (i = 0; i < len * sizeof(struct virtchnl2_splitq_rx_buf_desc);
+	     i++)
+		((volatile char *)rxq->rx_ring)[i] = 0;
+
+	memset(&rxq->fake_mbuf, 0x0, sizeof(rxq->fake_mbuf));
+
+	for (i = 0; i < IDPF_RX_MAX_BURST; i++)
+		rxq->sw_ring[rxq->nb_rx_desc + i] = &rxq->fake_mbuf;
+
+	/* The next descriptor id which can be received. */
+	rxq->rx_next_avail = 0;
+
+	/* The next descriptor id which can be refilled. */
+	rxq->rx_tail = 0;
+	/* The number of descriptors which can be refilled. */
+	rxq->nb_rx_hold = rxq->nb_rx_desc - 1;
+
+	rxq->bufq1 = NULL;
+	rxq->bufq2 = NULL;
+}
+
+static inline void
+reset_split_rx_queue(struct idpf_rx_queue *rxq)
+{
+	reset_split_rx_descq(rxq);
+	reset_split_rx_bufq(rxq->bufq1);
+	reset_split_rx_bufq(rxq->bufq2);
+}
+
 static inline void
 reset_single_rx_queue(struct idpf_rx_queue *rxq)
 {
@@ -179,6 +256,58 @@  reset_single_rx_queue(struct idpf_rx_queue *rxq)
 	rxq->rxrearm_nb = 0;
 }
 
+static inline void
+reset_split_tx_descq(struct idpf_tx_queue *txq)
+{
+	struct idpf_tx_entry *txe;
+	uint32_t i, size;
+	uint16_t prev;
+
+	if (!txq) {
+		PMD_DRV_LOG(DEBUG, "Pointer to txq is NULL");
+		return;
+	}
+
+	size = sizeof(struct idpf_flex_tx_sched_desc) * txq->nb_tx_desc;
+	for (i = 0; i < size; i++)
+		((volatile char *)txq->desc_ring)[i] = 0;
+
+	txe = txq->sw_ring;
+	prev = (uint16_t)(txq->sw_nb_desc - 1);
+	for (i = 0; i < txq->sw_nb_desc; i++) {
+		txe[i].mbuf = NULL;
+		txe[i].last_id = i;
+		txe[prev].next_id = i;
+		prev = i;
+	}
+
+	txq->tx_tail = 0;
+	txq->nb_used = 0;
+
+	/* Use this as next to clean for split desc queue */
+	txq->last_desc_cleaned = 0;
+	txq->sw_tail = 0;
+	txq->nb_free = txq->nb_tx_desc - 1;
+}
+
+static inline void
+reset_split_tx_complq(struct idpf_tx_queue *cq)
+{
+	uint32_t i, size;
+
+	if (!cq) {
+		PMD_DRV_LOG(DEBUG, "Pointer to complq is NULL");
+		return;
+	}
+
+	size = sizeof(struct idpf_splitq_tx_compl_desc) * cq->nb_tx_desc;
+	for (i = 0; i < size; i++)
+		((volatile char *)cq->compl_ring)[i] = 0;
+
+	cq->tx_tail = 0;
+	cq->expected_gen_id = 1;
+}
+
 static inline void
 reset_single_tx_queue(struct idpf_tx_queue *txq)
 {
@@ -216,6 +345,224 @@  reset_single_tx_queue(struct idpf_tx_queue *txq)
 	txq->next_rs = txq->rs_thresh - 1;
 }
 
+static int
+idpf_rx_split_bufq_setup(struct rte_eth_dev *dev, struct idpf_rx_queue *bufq,
+			 uint16_t queue_idx, uint16_t rx_free_thresh,
+			 uint16_t nb_desc, unsigned int socket_id,
+			 const struct rte_eth_rxconf *rx_conf,
+			 struct rte_mempool *mp)
+{
+	struct idpf_vport *vport = dev->data->dev_private;
+	struct idpf_adapter *adapter = vport->adapter;
+	struct idpf_hw *hw = &adapter->hw;
+	const struct rte_memzone *mz;
+	uint32_t ring_size;
+	uint16_t len;
+
+	bufq->mp = mp;
+	bufq->nb_rx_desc = nb_desc;
+	bufq->rx_free_thresh = rx_free_thresh;
+	bufq->queue_id = vport->chunks_info.rx_buf_start_qid + queue_idx;
+	bufq->port_id = dev->data->port_id;
+	bufq->rx_deferred_start = rx_conf->rx_deferred_start;
+	bufq->rx_hdr_len = 0;
+	bufq->adapter = adapter;
+
+	len = rte_pktmbuf_data_room_size(bufq->mp) - RTE_PKTMBUF_HEADROOM;
+	bufq->rx_buf_len = len;
+
+	/* Allocate the software ring. */
+	len = nb_desc + IDPF_RX_MAX_BURST;
+	bufq->sw_ring =
+		rte_zmalloc_socket("idpf rx bufq sw ring",
+				   sizeof(struct rte_mbuf *) * len,
+				   RTE_CACHE_LINE_SIZE,
+				   socket_id);
+	if (!bufq->sw_ring) {
+		PMD_INIT_LOG(ERR, "Failed to allocate memory for SW ring");
+		return -ENOMEM;
+	}
+
+	/* Allocate a liitle more to support bulk allocate. */
+	len = nb_desc + IDPF_RX_MAX_BURST;
+	ring_size = RTE_ALIGN(len *
+			      sizeof(struct virtchnl2_splitq_rx_buf_desc),
+			      IDPF_DMA_MEM_ALIGN);
+	mz = rte_eth_dma_zone_reserve(dev, "rx_buf_ring", queue_idx,
+				      ring_size, IDPF_RING_BASE_ALIGN,
+				      socket_id);
+	if (!mz) {
+		PMD_INIT_LOG(ERR, "Failed to reserve DMA memory for RX buffer queue.");
+		rte_free(bufq->sw_ring);
+		return -ENOMEM;
+	}
+
+	/* Zero all the descriptors in the ring. */
+	memset(mz->addr, 0, ring_size);
+	bufq->rx_ring_phys_addr = mz->iova;
+	bufq->rx_ring = mz->addr;
+
+	bufq->mz = mz;
+	reset_split_rx_bufq(bufq);
+	bufq->q_set = true;
+	bufq->qrx_tail = hw->hw_addr + (vport->chunks_info.rx_buf_qtail_start +
+			 queue_idx * vport->chunks_info.rx_buf_qtail_spacing);
+	bufq->ops = &def_rxq_ops;
+
+	/* TODO: allow bulk or vec */
+
+	return 0;
+}
+
+static int
+idpf_rx_split_queue_setup(struct rte_eth_dev *dev, uint16_t queue_idx,
+			  uint16_t nb_desc, unsigned int socket_id,
+			  const struct rte_eth_rxconf *rx_conf,
+			  struct rte_mempool *mp)
+{
+	struct idpf_vport *vport = dev->data->dev_private;
+	struct idpf_adapter *adapter = vport->adapter;
+	struct idpf_rx_queue *rxq;
+	struct idpf_rx_queue *bufq1, *bufq2;
+	const struct rte_memzone *mz;
+	uint16_t rx_free_thresh;
+	uint32_t ring_size;
+	uint16_t qid;
+	uint16_t len;
+	uint64_t offloads;
+	int ret;
+
+	PMD_INIT_FUNC_TRACE();
+
+	offloads = rx_conf->offloads | dev->data->dev_conf.rxmode.offloads;
+
+	if (nb_desc % IDPF_ALIGN_RING_DESC != 0 ||
+	    nb_desc > IDPF_MAX_RING_DESC ||
+	    nb_desc < IDPF_MIN_RING_DESC) {
+		PMD_INIT_LOG(ERR, "Number (%u) of receive descriptors is invalid", nb_desc);
+		return -EINVAL;
+	}
+
+	/* Check free threshold */
+	rx_free_thresh = (rx_conf->rx_free_thresh == 0) ?
+		IDPF_DEFAULT_RX_FREE_THRESH :
+		rx_conf->rx_free_thresh;
+	if (check_rx_thresh(nb_desc, rx_free_thresh))
+		return -EINVAL;
+
+	/* Free memory if needed */
+	if (dev->data->rx_queues[queue_idx]) {
+		idpf_rx_queue_release(dev->data->rx_queues[queue_idx]);
+		dev->data->rx_queues[queue_idx] = NULL;
+	}
+
+	/* Setup Rx description queue */
+	rxq = rte_zmalloc_socket("idpf rxq",
+				 sizeof(struct idpf_rx_queue),
+				 RTE_CACHE_LINE_SIZE,
+				 socket_id);
+	if (!rxq) {
+		PMD_INIT_LOG(ERR, "Failed to allocate memory for rx queue data structure");
+		return -ENOMEM;
+	}
+
+	rxq->mp = mp;
+	rxq->nb_rx_desc = nb_desc;
+	rxq->rx_free_thresh = rx_free_thresh;
+	rxq->queue_id = vport->chunks_info.rx_start_qid + queue_idx;
+	rxq->port_id = dev->data->port_id;
+	rxq->rx_deferred_start = rx_conf->rx_deferred_start;
+	rxq->rx_hdr_len = 0;
+	rxq->adapter = adapter;
+	rxq->offloads = offloads;
+
+	len = rte_pktmbuf_data_room_size(rxq->mp) - RTE_PKTMBUF_HEADROOM;
+	rxq->rx_buf_len = len;
+
+	len = rxq->nb_rx_desc + IDPF_RX_MAX_BURST;
+	ring_size = RTE_ALIGN(len *
+			      sizeof(struct virtchnl2_rx_flex_desc_adv_nic_3),
+			      IDPF_DMA_MEM_ALIGN);
+	mz = rte_eth_dma_zone_reserve(dev, "rx_cpmpl_ring", queue_idx,
+				      ring_size, IDPF_RING_BASE_ALIGN,
+				      socket_id);
+
+	if (!mz) {
+		PMD_INIT_LOG(ERR, "Failed to reserve DMA memory for RX");
+		ret = -ENOMEM;
+		goto free_rxq;
+	}
+
+	/* Zero all the descriptors in the ring. */
+	memset(mz->addr, 0, ring_size);
+	rxq->rx_ring_phys_addr = mz->iova;
+	rxq->rx_ring = mz->addr;
+
+	rxq->mz = mz;
+	reset_split_rx_descq(rxq);
+
+	/* TODO: allow bulk or vec */
+
+	/* setup Rx buffer queue */
+	bufq1 = rte_zmalloc_socket("idpf bufq1",
+				   sizeof(struct idpf_rx_queue),
+				   RTE_CACHE_LINE_SIZE,
+				   socket_id);
+	if (!bufq1) {
+		PMD_INIT_LOG(ERR, "Failed to allocate memory for rx buffer queue 1.");
+		ret = -ENOMEM;
+		goto free_mz;
+	}
+	qid = 2 * queue_idx;
+	ret = idpf_rx_split_bufq_setup(dev, bufq1, qid, rx_free_thresh,
+				       nb_desc, socket_id, rx_conf, mp);
+	if (ret) {
+		PMD_INIT_LOG(ERR, "Failed to setup buffer queue 1");
+		ret = -EINVAL;
+		goto free_bufq1;
+	}
+	rxq->bufq1 = bufq1;
+
+	bufq2 = rte_zmalloc_socket("idpf bufq2",
+				   sizeof(struct idpf_rx_queue),
+				   RTE_CACHE_LINE_SIZE,
+				   socket_id);
+	if (!bufq2) {
+		PMD_INIT_LOG(ERR, "Failed to allocate memory for rx buffer queue 2.");
+		rte_free(bufq1->sw_ring);
+		rte_memzone_free(bufq1->mz);
+		ret = -ENOMEM;
+		goto free_bufq1;
+	}
+	qid = 2 * queue_idx + 1;
+	ret = idpf_rx_split_bufq_setup(dev, bufq2, qid, rx_free_thresh,
+				       nb_desc, socket_id, rx_conf, mp);
+	if (ret) {
+		PMD_INIT_LOG(ERR, "Failed to setup buffer queue 2");
+		rte_free(bufq1->sw_ring);
+		rte_memzone_free(bufq1->mz);
+		ret = -EINVAL;
+		goto free_bufq2;
+	}
+	rxq->bufq2 = bufq2;
+
+	rxq->q_set = true;
+	dev->data->rx_queues[queue_idx] = rxq;
+
+	return 0;
+
+free_bufq2:
+	rte_free(bufq2);
+free_bufq1:
+	rte_free(bufq1);
+free_mz:
+	rte_memzone_free(mz);
+free_rxq:
+	rte_free(rxq);
+
+	return ret;
+}
+
 static int
 idpf_rx_single_queue_setup(struct rte_eth_dev *dev, uint16_t queue_idx,
 			   uint16_t nb_desc, unsigned int socket_id,
@@ -335,7 +682,138 @@  idpf_rx_queue_setup(struct rte_eth_dev *dev, uint16_t queue_idx,
 		return idpf_rx_single_queue_setup(dev, queue_idx, nb_desc,
 						  socket_id, rx_conf, mp);
 	else
-		return -1;
+		return idpf_rx_split_queue_setup(dev, queue_idx, nb_desc,
+						 socket_id, rx_conf, mp);
+}
+
+static int
+idpf_tx_split_queue_setup(struct rte_eth_dev *dev, uint16_t queue_idx,
+			  uint16_t nb_desc, unsigned int socket_id,
+			  const struct rte_eth_txconf *tx_conf)
+{
+	struct idpf_vport *vport = dev->data->dev_private;
+	struct idpf_adapter *adapter = vport->adapter;
+	struct idpf_hw *hw = &adapter->hw;
+	struct idpf_tx_queue *txq, *cq;
+	const struct rte_memzone *mz;
+	uint32_t ring_size;
+	uint16_t tx_rs_thresh, tx_free_thresh;
+	uint64_t offloads;
+
+	PMD_INIT_FUNC_TRACE();
+
+	offloads = tx_conf->offloads | dev->data->dev_conf.txmode.offloads;
+
+	if (nb_desc % IDPF_ALIGN_RING_DESC != 0 ||
+	    nb_desc > IDPF_MAX_RING_DESC ||
+	    nb_desc < IDPF_MIN_RING_DESC) {
+		PMD_INIT_LOG(ERR, "Number (%u) of transmit descriptors is invalid",
+			     nb_desc);
+		return -EINVAL;
+	}
+
+	tx_rs_thresh = (uint16_t)((tx_conf->tx_rs_thresh) ?
+		tx_conf->tx_rs_thresh : IDPF_DEFAULT_TX_RS_THRESH);
+	tx_free_thresh = (uint16_t)((tx_conf->tx_free_thresh) ?
+		tx_conf->tx_free_thresh : IDPF_DEFAULT_TX_FREE_THRESH);
+	if (check_tx_thresh(nb_desc, tx_rs_thresh, tx_free_thresh))
+		return -EINVAL;
+
+	/* Free memory if needed. */
+	if (dev->data->tx_queues[queue_idx]) {
+		idpf_tx_queue_release(dev->data->tx_queues[queue_idx]);
+		dev->data->tx_queues[queue_idx] = NULL;
+	}
+
+	/* Allocate the TX queue data structure. */
+	txq = rte_zmalloc_socket("idpf split txq",
+				 sizeof(struct idpf_tx_queue),
+				 RTE_CACHE_LINE_SIZE,
+				 socket_id);
+	if (!txq) {
+		PMD_INIT_LOG(ERR, "Failed to allocate memory for tx queue structure");
+		return -ENOMEM;
+	}
+
+	txq->nb_tx_desc = nb_desc;
+	txq->rs_thresh = tx_rs_thresh;
+	txq->free_thresh = tx_free_thresh;
+	txq->queue_id = vport->chunks_info.tx_start_qid + queue_idx;
+	txq->port_id = dev->data->port_id;
+	txq->offloads = offloads;
+	txq->tx_deferred_start = tx_conf->tx_deferred_start;
+
+	/* Allocate software ring */
+	txq->sw_nb_desc = 2 * nb_desc;
+	txq->sw_ring =
+		rte_zmalloc_socket("idpf split tx sw ring",
+				   sizeof(struct idpf_tx_entry) *
+				   txq->sw_nb_desc,
+				   RTE_CACHE_LINE_SIZE,
+				   socket_id);
+	if (!txq->sw_ring) {
+		PMD_INIT_LOG(ERR, "Failed to allocate memory for SW TX ring");
+		rte_free(txq);
+		return -ENOMEM;
+	}
+
+	/* Allocate TX hardware ring descriptors. */
+	ring_size = sizeof(struct idpf_flex_tx_sched_desc) * txq->nb_tx_desc;
+	ring_size = RTE_ALIGN(ring_size, IDPF_DMA_MEM_ALIGN);
+	mz = rte_eth_dma_zone_reserve(dev, "split_tx_ring", queue_idx,
+				      ring_size, IDPF_RING_BASE_ALIGN,
+				      socket_id);
+	if (!mz) {
+		PMD_INIT_LOG(ERR, "Failed to reserve DMA memory for TX");
+		rte_free(txq->sw_ring);
+		rte_free(txq);
+		return -ENOMEM;
+	}
+	txq->tx_ring_phys_addr = mz->iova;
+	txq->desc_ring = (struct idpf_flex_tx_sched_desc *)mz->addr;
+
+	txq->mz = mz;
+	reset_split_tx_descq(txq);
+	txq->qtx_tail = hw->hw_addr + (vport->chunks_info.tx_qtail_start +
+			queue_idx * vport->chunks_info.tx_qtail_spacing);
+	txq->ops = &def_txq_ops;
+
+	/* Allocate the TX completion queue data structure. */
+	txq->complq = rte_zmalloc_socket("idpf splitq cq",
+					 sizeof(struct idpf_tx_queue),
+					 RTE_CACHE_LINE_SIZE,
+					 socket_id);
+	cq = txq->complq;
+	if (!cq) {
+		PMD_INIT_LOG(ERR, "Failed to allocate memory for tx queue structure");
+		return -ENOMEM;
+	}
+	cq->nb_tx_desc = 2 * nb_desc;
+	cq->queue_id = vport->chunks_info.tx_compl_start_qid + queue_idx;
+	cq->port_id = dev->data->port_id;
+	cq->txqs = dev->data->tx_queues;
+	cq->tx_start_qid = vport->chunks_info.tx_start_qid;
+
+	ring_size = sizeof(struct idpf_splitq_tx_compl_desc) * cq->nb_tx_desc;
+	ring_size = RTE_ALIGN(ring_size, IDPF_DMA_MEM_ALIGN);
+	mz = rte_eth_dma_zone_reserve(dev, "tx_split_compl_ring", queue_idx,
+				      ring_size, IDPF_RING_BASE_ALIGN,
+				      socket_id);
+	if (!mz) {
+		PMD_INIT_LOG(ERR, "Failed to reserve DMA memory for TX completion queue");
+		rte_free(txq->sw_ring);
+		rte_free(txq);
+		return -ENOMEM;
+	}
+	cq->tx_ring_phys_addr = mz->iova;
+	cq->compl_ring = (struct idpf_splitq_tx_compl_desc *)mz->addr;
+	cq->mz = mz;
+	reset_split_tx_complq(cq);
+
+	txq->q_set = true;
+	dev->data->tx_queues[queue_idx] = txq;
+
+	return 0;
 }
 
 static int
@@ -447,7 +925,8 @@  idpf_tx_queue_setup(struct rte_eth_dev *dev, uint16_t queue_idx,
 		return idpf_tx_single_queue_setup(dev, queue_idx, nb_desc,
 						  socket_id, tx_conf);
 	else
-		return -1;
+		return idpf_tx_split_queue_setup(dev, queue_idx, nb_desc,
+						 socket_id, tx_conf);
 }
 
 void
diff --git a/drivers/net/idpf/idpf_vchnl.c b/drivers/net/idpf/idpf_vchnl.c
index 02a55e0658..87a024b6db 100644
--- a/drivers/net/idpf/idpf_vchnl.c
+++ b/drivers/net/idpf/idpf_vchnl.c
@@ -481,10 +481,10 @@  idpf_vc_config_rxqs(struct idpf_vport *vport)
 	struct virtchnl2_rxq_info *rxq_info;
 	struct idpf_cmd_info args;
 	uint16_t total_qs, num_qs;
-	int size, err, i;
+	int size, err, i, j;
 	int k = 0;
 
-	total_qs = vport->num_rx_q;
+	total_qs = vport->num_rx_q + vport->num_rx_bufq;
 	while (total_qs) {
 		if (total_qs > adapter->max_rxq_per_msg) {
 			num_qs = adapter->max_rxq_per_msg;
@@ -520,7 +520,46 @@  idpf_vc_config_rxqs(struct idpf_vport *vport)
 				rxq_info->ring_len = rxq[k]->nb_rx_desc;
 			}
 		} else {
-			return -1;
+			for (i = 0; i < num_qs / 3; i++, k++) {
+				/* Rx queue */
+				rxq_info = &vc_rxqs->qinfo[i * 3];
+				rxq_info->dma_ring_addr =
+					rxq[k]->rx_ring_phys_addr;
+				rxq_info->type = VIRTCHNL2_QUEUE_TYPE_RX;
+				rxq_info->queue_id = rxq[k]->queue_id;
+				rxq_info->model = VIRTCHNL2_QUEUE_MODEL_SPLIT;
+				rxq_info->data_buffer_size = rxq[k]->rx_buf_len;
+				rxq_info->max_pkt_size = vport->max_pkt_len;
+
+				rxq_info->desc_ids = VIRTCHNL2_RXDID_2_FLEX_SPLITQ_M;
+				rxq_info->qflags |= VIRTCHNL2_RX_DESC_SIZE_32BYTE;
+
+				rxq_info->ring_len = rxq[k]->nb_rx_desc;
+				rxq_info->rx_bufq1_id = rxq[k]->bufq1->queue_id;
+				rxq_info->rx_bufq2_id = rxq[k]->bufq2->queue_id;
+				rxq_info->rx_buffer_low_watermark = 64;
+
+				/* Buffer queue */
+				for (j = 1; j <= IDPF_RX_BUFQ_PER_GRP; j++) {
+					struct idpf_rx_queue *bufq = j == 1 ?
+						rxq[k]->bufq1 : rxq[k]->bufq2;
+					rxq_info = &vc_rxqs->qinfo[i * 3 + j];
+					rxq_info->dma_ring_addr =
+						bufq->rx_ring_phys_addr;
+					rxq_info->type =
+						VIRTCHNL2_QUEUE_TYPE_RX_BUFFER;
+					rxq_info->queue_id = bufq->queue_id;
+					rxq_info->model = VIRTCHNL2_QUEUE_MODEL_SPLIT;
+					rxq_info->data_buffer_size = bufq->rx_buf_len;
+					rxq_info->desc_ids =
+						VIRTCHNL2_RXDID_2_FLEX_SPLITQ_M;
+					rxq_info->ring_len = bufq->nb_rx_desc;
+
+					rxq_info->buffer_notif_stride =
+						IDPF_RX_BUF_STRIDE;
+					rxq_info->rx_buffer_low_watermark = 64;
+				}
+			}
 		}
 		memset(&args, 0, sizeof(args));
 		args.ops = VIRTCHNL2_OP_CONFIG_RX_QUEUES;
@@ -550,7 +589,7 @@  idpf_vc_config_rxq(struct idpf_vport *vport, uint16_t rxq_id)
 	struct virtchnl2_rxq_info *rxq_info;
 	struct idpf_cmd_info args;
 	uint16_t num_qs;
-	int size, err;
+	int size, err, i;
 
 	if (vport->rxq_model == VIRTCHNL2_QUEUE_MODEL_SINGLE)
 		num_qs = IDPF_RXQ_PER_GRP;
@@ -581,7 +620,39 @@  idpf_vc_config_rxq(struct idpf_vport *vport, uint16_t rxq_id)
 
 		rxq_info->ring_len = rxq[rxq_id]->nb_rx_desc;
 	}  else {
-		return -1;
+		/* Rx queue */
+		rxq_info = &vc_rxqs->qinfo[0];
+		rxq_info->dma_ring_addr = rxq[rxq_id]->rx_ring_phys_addr;
+		rxq_info->type = VIRTCHNL2_QUEUE_TYPE_RX;
+		rxq_info->queue_id = rxq[rxq_id]->queue_id;
+		rxq_info->model = VIRTCHNL2_QUEUE_MODEL_SPLIT;
+		rxq_info->data_buffer_size = rxq[rxq_id]->rx_buf_len;
+		rxq_info->max_pkt_size = vport->max_pkt_len;
+
+		rxq_info->desc_ids = VIRTCHNL2_RXDID_2_FLEX_SPLITQ_M;
+		rxq_info->qflags |= VIRTCHNL2_RX_DESC_SIZE_32BYTE;
+
+		rxq_info->ring_len = rxq[rxq_id]->nb_rx_desc;
+		rxq_info->rx_bufq1_id = rxq[rxq_id]->bufq1->queue_id;
+		rxq_info->rx_bufq2_id = rxq[rxq_id]->bufq2->queue_id;
+		rxq_info->rx_buffer_low_watermark = 64;
+
+		/* Buffer queue */
+		for (i = 1; i <= IDPF_RX_BUFQ_PER_GRP; i++) {
+			struct idpf_rx_queue *bufq =
+				i == 1 ? rxq[rxq_id]->bufq1 : rxq[rxq_id]->bufq2;
+			rxq_info = &vc_rxqs->qinfo[i];
+			rxq_info->dma_ring_addr = bufq->rx_ring_phys_addr;
+			rxq_info->type = VIRTCHNL2_QUEUE_TYPE_RX_BUFFER;
+			rxq_info->queue_id = bufq->queue_id;
+			rxq_info->model = VIRTCHNL2_QUEUE_MODEL_SPLIT;
+			rxq_info->data_buffer_size = bufq->rx_buf_len;
+			rxq_info->desc_ids = VIRTCHNL2_RXDID_2_FLEX_SPLITQ_M;
+			rxq_info->ring_len = bufq->nb_rx_desc;
+
+			rxq_info->buffer_notif_stride = IDPF_RX_BUF_STRIDE;
+			rxq_info->rx_buffer_low_watermark = 64;
+		}
 	}
 
 	memset(&args, 0, sizeof(args));
@@ -612,7 +683,7 @@  idpf_vc_config_txqs(struct idpf_vport *vport)
 	int size, err, i;
 	int k = 0;
 
-	total_qs = vport->num_tx_q;
+	total_qs = vport->num_tx_q + vport->num_tx_complq;
 	while (total_qs) {
 		if (total_qs > adapter->max_txq_per_msg) {
 			num_qs = adapter->max_txq_per_msg;
@@ -642,7 +713,29 @@  idpf_vc_config_txqs(struct idpf_vport *vport)
 				txq_info->ring_len = txq[k]->nb_tx_desc;
 			}
 		} else {
-			return -1;
+			for (i = 0; i < num_qs / 2; i++, k++) {
+				/* txq info */
+				txq_info = &vc_txqs->qinfo[2 * i];
+				txq_info->dma_ring_addr = txq[k]->tx_ring_phys_addr;
+				txq_info->type = VIRTCHNL2_QUEUE_TYPE_TX;
+				txq_info->queue_id = txq[k]->queue_id;
+				txq_info->model = VIRTCHNL2_QUEUE_MODEL_SPLIT;
+				txq_info->sched_mode = VIRTCHNL2_TXQ_SCHED_MODE_FLOW;
+				txq_info->ring_len = txq[k]->nb_tx_desc;
+				txq_info->tx_compl_queue_id =
+					txq[k]->complq->queue_id;
+				txq_info->relative_queue_id = txq_info->queue_id;
+
+				/* tx completion queue info */
+				txq_info = &vc_txqs->qinfo[2 * i + 1];
+				txq_info->dma_ring_addr =
+					txq[k]->complq->tx_ring_phys_addr;
+				txq_info->type = VIRTCHNL2_QUEUE_TYPE_TX_COMPLETION;
+				txq_info->queue_id = txq[k]->complq->queue_id;
+				txq_info->model = VIRTCHNL2_QUEUE_MODEL_SPLIT;
+				txq_info->sched_mode = VIRTCHNL2_TXQ_SCHED_MODE_FLOW;
+				txq_info->ring_len = txq[k]->complq->nb_tx_desc;
+			}
 		}
 
 		memset(&args, 0, sizeof(args));
@@ -700,7 +793,25 @@  idpf_vc_config_txq(struct idpf_vport *vport, uint16_t txq_id)
 		txq_info->sched_mode = VIRTCHNL2_TXQ_SCHED_MODE_QUEUE;
 		txq_info->ring_len = txq[txq_id]->nb_tx_desc;
 	} else {
-		return -1;
+		/* txq info */
+		txq_info = &vc_txqs->qinfo[0];
+		txq_info->dma_ring_addr = txq[txq_id]->tx_ring_phys_addr;
+		txq_info->type = VIRTCHNL2_QUEUE_TYPE_TX;
+		txq_info->queue_id = txq[txq_id]->queue_id;
+		txq_info->model = VIRTCHNL2_QUEUE_MODEL_SPLIT;
+		txq_info->sched_mode = VIRTCHNL2_TXQ_SCHED_MODE_FLOW;
+		txq_info->ring_len = txq[txq_id]->nb_tx_desc;
+		txq_info->tx_compl_queue_id = txq[txq_id]->complq->queue_id;
+		txq_info->relative_queue_id = txq_info->queue_id;
+
+		/* tx completion queue info */
+		txq_info = &vc_txqs->qinfo[1];
+		txq_info->dma_ring_addr = txq[txq_id]->complq->tx_ring_phys_addr;
+		txq_info->type = VIRTCHNL2_QUEUE_TYPE_TX_COMPLETION;
+		txq_info->queue_id = txq[txq_id]->complq->queue_id;
+		txq_info->model = VIRTCHNL2_QUEUE_MODEL_SPLIT;
+		txq_info->sched_mode = VIRTCHNL2_TXQ_SCHED_MODE_FLOW;
+		txq_info->ring_len = txq[txq_id]->complq->nb_tx_desc;
 	}
 
 	memset(&args, 0, sizeof(args));