> -----Original Message-----
> From: Xing, Beilei <beilei.x...@intel.com>
> Sent: Friday, April 21, 2023 2:51 PM
> To: Wu, Jingjing <jingjing...@intel.com>
> Cc: dev@dpdk.org; Liu, Mingxia <mingxia....@intel.com>; Xing, Beilei
> <beilei.x...@intel.com>; Wang, Xiao W <xiao.w.w...@intel.com>
> Subject: [PATCH 06/10] net/cpfl: support hairpin queue configuration
>
> From: Beilei Xing <beilei.x...@intel.com>
>
> This patch supports Rx/Tx hairpin queue configuration.
>
> Signed-off-by: Xiao Wang <xiao.w.w...@intel.com>
> Signed-off-by: Mingxia Liu <mingxia....@intel.com>
> Signed-off-by: Beilei Xing <beilei.x...@intel.com>
> ---
> drivers/common/idpf/idpf_common_virtchnl.c | 70 +++++++++++
> drivers/common/idpf/idpf_common_virtchnl.h | 6 +
> drivers/common/idpf/version.map | 2 +
> drivers/net/cpfl/cpfl_ethdev.c | 136 ++++++++++++++++++++-
> drivers/net/cpfl/cpfl_rxtx.c | 80 ++++++++++++
> drivers/net/cpfl/cpfl_rxtx.h | 7 ++
> 6 files changed, 297 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/common/idpf/idpf_common_virtchnl.c
> b/drivers/common/idpf/idpf_common_virtchnl.c
> index 76a658bb26..50cd43a8dd 100644
> --- a/drivers/common/idpf/idpf_common_virtchnl.c
> +++ b/drivers/common/idpf/idpf_common_virtchnl.c
> @@ -1050,6 +1050,41 @@ idpf_vc_rxq_config(struct idpf_vport *vport, struct
> idpf_rx_queue *rxq)
> return err;
> }
>
> +int idpf_vc_rxq_config_by_info(struct idpf_vport *vport, struct
> virtchnl2_rxq_info *rxq_info,
> + uint16_t num_qs)
> +{
> + struct idpf_adapter *adapter = vport->adapter;
> + struct virtchnl2_config_rx_queues *vc_rxqs = NULL;
> + struct idpf_cmd_info args;
> + int size, err, i;
> +
> + size = sizeof(*vc_rxqs) + (num_qs - 1) *
> + sizeof(struct virtchnl2_rxq_info);
> + vc_rxqs = rte_zmalloc("cfg_rxqs", size, 0);
> + if (vc_rxqs == NULL) {
> + DRV_LOG(ERR, "Failed to allocate virtchnl2_config_rx_queues");
> + err = -ENOMEM;
> + return err;
> + }
> + vc_rxqs->vport_id = vport->vport_id;
> + vc_rxqs->num_qinfo = num_qs;
> + memcpy(vc_rxqs->qinfo, rxq_info, num_qs * sizeof(struct
> +virtchnl2_rxq_info));
> +
> + memset(&args, 0, sizeof(args));
> + args.ops = VIRTCHNL2_OP_CONFIG_RX_QUEUES;
> + args.in_args = (uint8_t *)vc_rxqs;
> + args.in_args_size = size;
> + args.out_buffer = adapter->mbx_resp;
> + args.out_size = IDPF_DFLT_MBX_BUF_SIZE;
> +
> + err = idpf_vc_cmd_execute(adapter, &args);
> + rte_free(vc_rxqs);
> + if (err != 0)
> + DRV_LOG(ERR, "Failed to execute command of
> +VIRTCHNL2_OP_CONFIG_RX_QUEUES");
> +
> + return err;
> +}
> +
> int
> idpf_vc_txq_config(struct idpf_vport *vport, struct idpf_tx_queue *txq) {
> @@ -
> 1121,6 +1156,41 @@ idpf_vc_txq_config(struct idpf_vport *vport, struct
> idpf_tx_queue *txq)
> return err;
> }
>
> +int
> +idpf_vc_txq_config_by_info(struct idpf_vport *vport, struct
> virtchnl2_txq_info
> *txq_info,
> + uint16_t num_qs)
> +{
> + struct idpf_adapter *adapter = vport->adapter;
> + struct virtchnl2_config_tx_queues *vc_txqs = NULL;
> + struct idpf_cmd_info args;
> + int size, err;
> +
> + size = sizeof(*vc_txqs) + (num_qs - 1) * sizeof(struct
> virtchnl2_txq_info);
> + vc_txqs = rte_zmalloc("cfg_txqs", size, 0);
> + if (vc_txqs == NULL) {
> + DRV_LOG(ERR, "Failed to allocate virtchnl2_config_tx_queues");
> + err = -ENOMEM;
> + return err;
> + }
> + vc_txqs->vport_id = vport->vport_id;
> + vc_txqs->num_qinfo = num_qs;
> + memcpy(vc_txqs->qinfo, txq_info, num_qs * sizeof(struct
> +virtchnl2_txq_info));
> +
> + memset(&args, 0, sizeof(args));
> + args.ops = VIRTCHNL2_OP_CONFIG_TX_QUEUES;
> + args.in_args = (uint8_t *)vc_txqs;
> + args.in_args_size = size;
> + args.out_buffer = adapter->mbx_resp;
> + args.out_size = IDPF_DFLT_MBX_BUF_SIZE;
> +
> + err = idpf_vc_cmd_execute(adapter, &args);
> + rte_free(vc_txqs);
> + if (err != 0)
> + DRV_LOG(ERR, "Failed to execute command of
> +VIRTCHNL2_OP_CONFIG_TX_QUEUES");
> +
> + return err;
> +}
> +
> int
> idpf_vc_ctlq_recv(struct idpf_ctlq_info *cq, u16 *num_q_msg,
> struct idpf_ctlq_msg *q_msg)
> diff --git a/drivers/common/idpf/idpf_common_virtchnl.h
> b/drivers/common/idpf/idpf_common_virtchnl.h
> index bf1d014c8d..277235ba7d 100644
> --- a/drivers/common/idpf/idpf_common_virtchnl.h
> +++ b/drivers/common/idpf/idpf_common_virtchnl.h
> @@ -65,6 +65,12 @@ __rte_internal
> int idpf_vc_ctlq_post_rx_buffs(struct idpf_hw *hw, struct idpf_ctlq_info *cq,
> u16 *buff_count, struct idpf_dma_mem **buffs);
> __rte_internal
> +int idpf_vc_rxq_config_by_info(struct idpf_vport *vport, struct
> virtchnl2_rxq_info *rxq_info,
> + uint16_t num_qs);
> +__rte_internal
> +int idpf_vc_txq_config_by_info(struct idpf_vport *vport, struct
> virtchnl2_txq_info *txq_info,
> + uint16_t num_qs);
> +__rte_internal
> int idpf_vc_queue_grps_del(struct idpf_vport *vport,
> uint16_t num_q_grps,
> struct virtchnl2_queue_group_id *qg_ids); diff --git
> a/drivers/common/idpf/version.map b/drivers/common/idpf/version.map index
> aa67f7ee27..a339a4bf8e 100644
> --- a/drivers/common/idpf/version.map
> +++ b/drivers/common/idpf/version.map
> @@ -59,8 +59,10 @@ INTERNAL {
> idpf_vc_rss_lut_get;
> idpf_vc_rss_lut_set;
> idpf_vc_rxq_config;
> + idpf_vc_rxq_config_by_info;
> idpf_vc_stats_query;
> idpf_vc_txq_config;
> + idpf_vc_txq_config_by_info;
> idpf_vc_vectors_alloc;
> idpf_vc_vectors_dealloc;
> idpf_vc_vport_create;
> diff --git a/drivers/net/cpfl/cpfl_ethdev.c b/drivers/net/cpfl/cpfl_ethdev.c
> index
> d3300f17cc..13edf2e706 100644
> --- a/drivers/net/cpfl/cpfl_ethdev.c
> +++ b/drivers/net/cpfl/cpfl_ethdev.c
> @@ -737,32 +737,160 @@ cpfl_config_rx_queues_irqs(struct rte_eth_dev *dev)
> return idpf_vport_irq_map_config(vport, nb_rx_queues); }
>
> +/* Update hairpin_info for dev's tx hairpin queue */ static int
> +cpfl_txq_hairpin_info_update(struct rte_eth_dev *dev, uint16_t rx_port)
> +{
> + struct cpfl_vport *cpfl_tx_vport = dev->data->dev_private;
> + struct rte_eth_dev *peer_dev = &rte_eth_devices[rx_port];
> + struct cpfl_vport *cpfl_rx_vport = peer_dev->data->dev_private;
> + struct cpfl_txq_hairpin_info *hairpin_info;
> + struct cpfl_tx_queue *cpfl_txq;
> + int i;
> +
> + for (i = cpfl_tx_vport->nb_data_txq; i < dev->data->nb_tx_queues; i++) {
> + cpfl_txq = dev->data->tx_queues[i];
> + hairpin_info = &cpfl_txq->hairpin_info;
> + if (hairpin_info->peer_rxp != rx_port) {
> + PMD_DRV_LOG(ERR, "port %d is not the peer port",
> rx_port);
> + return -EINVAL;
> + }
> + hairpin_info->peer_rxq_id =
> + cpfl_hw_qid_get(cpfl_rx_vport-
> >p2p_q_chunks_info.rx_start_qid,
> + hairpin_info->peer_rxq_id -
> cpfl_rx_vport->nb_data_rxq);
> + }
> +
> + return 0;
> +}
> +
> +/* Bind Rx hairpin queue's memory zone to peer Tx hairpin queue's
> +memory zone */ static void cpfl_rxq_hairpin_mz_bind(struct rte_eth_dev
> +*dev) {
> + struct cpfl_vport *cpfl_rx_vport = dev->data->dev_private;
> + struct idpf_vport *vport = &cpfl_rx_vport->base;
> + struct idpf_adapter *adapter = vport->adapter;
> + struct idpf_hw *hw = &adapter->hw;
> + struct cpfl_rx_queue *cpfl_rxq;
> + struct cpfl_tx_queue *cpfl_txq;
> + struct rte_eth_dev *peer_dev;
> + const struct rte_memzone *mz;
> + uint16_t peer_tx_port;
> + uint16_t peer_tx_qid;
> + int i;
> +
> + for (i = cpfl_rx_vport->nb_data_rxq; i < dev->data->nb_rx_queues; i++) {
> + cpfl_rxq = dev->data->rx_queues[i];
> + peer_tx_port = cpfl_rxq->hairpin_info.peer_txp;
> + peer_tx_qid = cpfl_rxq->hairpin_info.peer_txq_id;
> + peer_dev = &rte_eth_devices[peer_tx_port];
> + cpfl_txq = peer_dev->data->tx_queues[peer_tx_qid];
> +
> + /* bind rx queue */
> + mz = cpfl_txq->base.mz;
> + cpfl_rxq->base.rx_ring_phys_addr = mz->iova;
> + cpfl_rxq->base.rx_ring = mz->addr;
> + cpfl_rxq->base.mz = mz;
> +
> + /* bind rx buffer queue */
> + mz = cpfl_txq->base.complq->mz;
> + cpfl_rxq->base.bufq1->rx_ring_phys_addr = mz->iova;
> + cpfl_rxq->base.bufq1->rx_ring = mz->addr;
> + cpfl_rxq->base.bufq1->mz = mz;
> + cpfl_rxq->base.bufq1->qrx_tail = hw->hw_addr +
> + cpfl_hw_qtail_get(cpfl_rx_vport-
> >p2p_q_chunks_info.rx_buf_qtail_start,
> + 0, cpfl_rx_vport-
> >p2p_q_chunks_info.rx_buf_qtail_spacing);
> + }
> +}
> +
> static int
> cpfl_start_queues(struct rte_eth_dev *dev) {
> + struct cpfl_vport *cpfl_vport = dev->data->dev_private;
> + struct idpf_vport *vport = &cpfl_vport->base;
> struct cpfl_rx_queue *cpfl_rxq;
> struct cpfl_tx_queue *cpfl_txq;
> + int tx_cmplq_flag = 0;
> + int rx_bufq_flag = 0;
> + int flag = 0;
> int err = 0;
> int i;
>
> + /* For normal data queues, configure, init and enale Txq.
> + * For non-cross vport hairpin queues, configure Txq.
> + */
> for (i = 0; i < dev->data->nb_tx_queues; i++) {
> cpfl_txq = dev->data->tx_queues[i];
> if (cpfl_txq == NULL || cpfl_txq->base.tx_deferred_start)
> continue;
> - err = cpfl_tx_queue_start(dev, i);
> + if (!cpfl_txq->hairpin_info.hairpin_q) {
> + err = cpfl_tx_queue_start(dev, i);
> + if (err != 0) {
> + PMD_DRV_LOG(ERR, "Fail to start Tx
> queue %u", i);
> + return err;
> + }
> + } else if (!cpfl_txq->hairpin_info.manual_bind) {
> + if (flag == 0) {
> + err = cpfl_txq_hairpin_info_update(dev,
> + cpfl_txq-
> >hairpin_info.peer_rxp);
> + if (err != 0) {
> + PMD_DRV_LOG(ERR, "Fail to update Tx
> hairpin queue info");
> + return err;
> + }
> + flag = 1;
[Liu, Mingxia] The variable flag is not been used, can it be removed?
> + }
> + err = cpfl_hairpin_txq_config(vport, cpfl_txq);
> + if (err != 0) {
> + PMD_DRV_LOG(ERR, "Fail to configure hairpin
> Tx queue %u", i);
> + return err;
> + }
> + tx_cmplq_flag = 1;
> + }
> + }
> +
> + /* For non-cross vport hairpin queues, configure Tx completion queue
> first.*/
> + if (tx_cmplq_flag == 1 && cpfl_vport->p2p_tx_complq != NULL) {
> + err = cpfl_hairpin_tx_complq_config(cpfl_vport);
> if (err != 0) {
> - PMD_DRV_LOG(ERR, "Fail to start Tx queue %u", i);
> + PMD_DRV_LOG(ERR, "Fail to config Tx completion
> queue");
> return err;
> }
> }
>
[Liu, Mingxia] Better to move this code next to
+ err = cpfl_hairpin_txq_config(vport, cpfl_txq);
+ if (err != 0) {
+ PMD_DRV_LOG(ERR, "Fail to configure hairpin Tx
queue %u", i);
+ return err;
+ }
When cpfl_rxq->hairpin_info.hairpin_q is true, then cpfl_vport->p2p_tx_complq
is not null, right ?
And remove tx_cmplq_flag?
> + /* For normal data queues, configure, init and enale Rxq.
> + * For non-cross vport hairpin queues, configure Rxq, and then init Rxq.
> + */
> + cpfl_rxq_hairpin_mz_bind(dev);
> for (i = 0; i < dev->data->nb_rx_queues; i++) {
> cpfl_rxq = dev->data->rx_queues[i];
> if (cpfl_rxq == NULL || cpfl_rxq->base.rx_deferred_start)
> continue;
> - err = cpfl_rx_queue_start(dev, i);
> + if (!cpfl_rxq->hairpin_info.hairpin_q) {
> + err = cpfl_rx_queue_start(dev, i);
> + if (err != 0) {
> + PMD_DRV_LOG(ERR, "Fail to start Rx
> queue %u", i);
> + return err;
> + }
> + } else if (!cpfl_rxq->hairpin_info.manual_bind) {
> + err = cpfl_hairpin_rxq_config(vport, cpfl_rxq);
> + if (err != 0) {
> + PMD_DRV_LOG(ERR, "Fail to configure hairpin
> Rx queue %u", i);
> + return err;
> + }
> + err = cpfl_rx_queue_init(dev, i);
> + if (err != 0) {
> + PMD_DRV_LOG(ERR, "Fail to init hairpin Rx
> queue %u", i);
> + return err;
> + }
> + rx_bufq_flag = 1;
> + }
> + }
> +
> + /* For non-cross vport hairpin queues, configure Rx buffer queue.*/
> + if (rx_bufq_flag == 1 && cpfl_vport->p2p_rx_bufq != NULL) {
> + err = cpfl_hairpin_rx_bufq_config(cpfl_vport);
> if (err != 0) {
> - PMD_DRV_LOG(ERR, "Fail to start Rx queue %u", i);
> + PMD_DRV_LOG(ERR, "Fail to config Rx buffer queue");
> return err;
> }
> }
[Liu, Mingxia] Similar to above.
> diff --git a/drivers/net/cpfl/cpfl_rxtx.c b/drivers/net/cpfl/cpfl_rxtx.c index
> 64ed331a6d..040beb5bac 100644
> --- a/drivers/net/cpfl/cpfl_rxtx.c
> +++ b/drivers/net/cpfl/cpfl_rxtx.c
> @@ -930,6 +930,86 @@ cpfl_tx_hairpin_queue_setup(struct rte_eth_dev *dev,
> uint16_t queue_idx,
> return 0;
> }
>
> +int
> +cpfl_hairpin_rx_bufq_config(struct cpfl_vport *cpfl_vport) {
> + struct idpf_rx_queue *rx_bufq = cpfl_vport->p2p_rx_bufq;
> + struct virtchnl2_rxq_info rxq_info[1] = {0};
> +
> + rxq_info[0].type = VIRTCHNL2_QUEUE_TYPE_RX_BUFFER;
> + rxq_info[0].queue_id = rx_bufq->queue_id;
> + rxq_info[0].ring_len = rx_bufq->nb_rx_desc;
> + rxq_info[0].dma_ring_addr = rx_bufq->rx_ring_phys_addr;
> + rxq_info[0].desc_ids = VIRTCHNL2_RXDID_2_FLEX_SPLITQ_M;
> + rxq_info[0].rx_buffer_low_watermark =
> CPFL_RXBUF_LOW_WATERMARK;
> + rxq_info[0].model = VIRTCHNL2_QUEUE_MODEL_SPLIT;
> + rxq_info[0].data_buffer_size = rx_bufq->rx_buf_len;
> + rxq_info[0].buffer_notif_stride = CPFL_RX_BUF_STRIDE;
> +
> + return idpf_vc_rxq_config_by_info(&cpfl_vport->base, rxq_info, 1); }
> +
> +int
> +cpfl_hairpin_rxq_config(struct idpf_vport *vport, struct cpfl_rx_queue
> +*cpfl_rxq) {
> + struct virtchnl2_rxq_info rxq_info[1] = {0};
> + struct idpf_rx_queue *rxq = &cpfl_rxq->base;
> +
> + rxq_info[0].type = VIRTCHNL2_QUEUE_TYPE_RX;
> + rxq_info[0].queue_id = rxq->queue_id;
> + rxq_info[0].ring_len = rxq->nb_rx_desc;
> + rxq_info[0].dma_ring_addr = rxq->rx_ring_phys_addr;
> + rxq_info[0].rx_bufq1_id = rxq->bufq1->queue_id;
> + rxq_info[0].max_pkt_size = vport->max_pkt_len;
> + rxq_info[0].desc_ids = VIRTCHNL2_RXDID_2_FLEX_SPLITQ_M;
> + rxq_info[0].qflags |= VIRTCHNL2_RX_DESC_SIZE_16BYTE;
> +
> + rxq_info[0].data_buffer_size = rxq->rx_buf_len;
> + rxq_info[0].model = VIRTCHNL2_QUEUE_MODEL_SPLIT;
> + rxq_info[0].rx_buffer_low_watermark =
> CPFL_RXBUF_LOW_WATERMARK;
> +
> + PMD_DRV_LOG(NOTICE, "hairpin: vport %u, Rxq id 0x%x",
> + vport->vport_id, rxq_info[0].queue_id);
> +
> + return idpf_vc_rxq_config_by_info(vport, rxq_info, 1); }
> +
> +int
> +cpfl_hairpin_tx_complq_config(struct cpfl_vport *cpfl_vport) {
> + struct idpf_tx_queue *tx_complq = cpfl_vport->p2p_tx_complq;
> + struct virtchnl2_txq_info txq_info[1] = {0};
> +
> + txq_info[0].dma_ring_addr = tx_complq->tx_ring_phys_addr;
> + txq_info[0].type = VIRTCHNL2_QUEUE_TYPE_TX_COMPLETION;
> + txq_info[0].queue_id = tx_complq->queue_id;
> + txq_info[0].ring_len = tx_complq->nb_tx_desc;
> + txq_info[0].peer_rx_queue_id = cpfl_vport->p2p_rx_bufq->queue_id;
> + txq_info[0].model = VIRTCHNL2_QUEUE_MODEL_SPLIT;
> + txq_info[0].sched_mode = VIRTCHNL2_TXQ_SCHED_MODE_FLOW;
> +
> + return idpf_vc_txq_config_by_info(&cpfl_vport->base, txq_info, 1); }
> +
> +int
> +cpfl_hairpin_txq_config(struct idpf_vport *vport, struct cpfl_tx_queue
> +*cpfl_txq) {
> + struct idpf_tx_queue *txq = &cpfl_txq->base;
> + struct virtchnl2_txq_info txq_info[1] = {0};
> +
> + txq_info[0].dma_ring_addr = txq->tx_ring_phys_addr;
> + txq_info[0].type = VIRTCHNL2_QUEUE_TYPE_TX;
> + txq_info[0].queue_id = txq->queue_id;
> + txq_info[0].ring_len = txq->nb_tx_desc;
> + txq_info[0].tx_compl_queue_id = txq->complq->queue_id;
> + txq_info[0].relative_queue_id = txq->queue_id;
> + txq_info[0].peer_rx_queue_id = cpfl_txq->hairpin_info.peer_rxq_id;
> + txq_info[0].model = VIRTCHNL2_QUEUE_MODEL_SPLIT;
> + txq_info[0].sched_mode = VIRTCHNL2_TXQ_SCHED_MODE_FLOW;
> +
> + return idpf_vc_txq_config_by_info(vport, txq_info, 1); }
> +
> int
> cpfl_rx_queue_init(struct rte_eth_dev *dev, uint16_t rx_queue_id) { diff
> --git
> a/drivers/net/cpfl/cpfl_rxtx.h b/drivers/net/cpfl/cpfl_rxtx.h index
> d844c9f057..b01ce5edf9 100644
> --- a/drivers/net/cpfl/cpfl_rxtx.h
> +++ b/drivers/net/cpfl/cpfl_rxtx.h
> @@ -30,12 +30,15 @@
> #define CPFL_RING_BASE_ALIGN 128
>
> #define CPFL_DEFAULT_RX_FREE_THRESH 32
> +#define CPFL_RXBUF_LOW_WATERMARK 64
>
> #define CPFL_DEFAULT_TX_RS_THRESH 32
> #define CPFL_DEFAULT_TX_FREE_THRESH 32
>
> #define CPFL_SUPPORT_CHAIN_NUM 5
>
> +#define CPFL_RX_BUF_STRIDE 64
> +
> struct cpfl_rxq_hairpin_info {
> bool hairpin_q; /* if rx queue is a hairpin queue */
> bool manual_bind; /* for cross vport */
> @@ -85,4 +88,8 @@ int cpfl_rx_hairpin_queue_setup(struct rte_eth_dev *dev,
> uint16_t queue_idx, int cpfl_tx_hairpin_queue_setup(struct rte_eth_dev *dev,
> uint16_t queue_idx,
> uint16_t nb_desc,
> const struct rte_eth_hairpin_conf *conf);
> +int cpfl_hairpin_tx_complq_config(struct cpfl_vport *cpfl_vport); int
> +cpfl_hairpin_txq_config(struct idpf_vport *vport, struct cpfl_tx_queue
> +*cpfl_txq); int cpfl_hairpin_rx_bufq_config(struct cpfl_vport
> +*cpfl_vport); int cpfl_hairpin_rxq_config(struct idpf_vport *vport,
> +struct cpfl_rx_queue *cpfl_rxq);
> #endif /* _CPFL_RXTX_H_ */
> --
> 2.26.2
