i40e: allow vector Rx and Tx usage

To support FVL PMD can select which RX and TX function should be used
according to the queue config.

Signed-off-by: Zhe Tao <zhe.tao@intel.com>
Acked-by: Cunming Liang <cunming.liang@intel.com>

diff --git a/doc/guides/rel_notes/release_2_2.rst b/doc/guides/rel_notes/release_2_2.rst
index 01f69de..0c0d2ff 100644 (file)
--- a/doc/guides/rel_notes/release_2_2.rst
+++ b/doc/guides/rel_notes/release_2_2.rst
@@ -18,6 +18,8 @@ New Features
   * Add 2 new flow director modes on x550.
   * One is MAC VLAN mode, the other is tunnel mode.
 
+* **Added i40e vector RX/TX.**
+
 * **Added i40e flow control support.**
 
 * **Added fm10k TSO support for both PF and VF.**

diff --git a/drivers/net/i40e/i40e_ethdev.c b/drivers/net/i40e/i40e_ethdev.c
index 497fee7..016838a 100644 (file)
--- a/drivers/net/i40e/i40e_ethdev.c
+++ b/drivers/net/i40e/i40e_ethdev.c
@@ -451,8 +451,8 @@ eth_i40e_dev_init(struct rte_eth_dev *dev)
         * has already done this work. Only check we don't need a different
         * RX function */
        if (rte_eal_process_type() != RTE_PROC_PRIMARY){
-               if (dev->data->scattered_rx)
-                       dev->rx_pkt_burst = i40e_recv_scattered_pkts;
+               i40e_set_rx_function(dev);
+               i40e_set_tx_function(dev);
                return 0;
        }
        pci_dev = dev->pci_dev;
@@ -724,10 +724,20 @@ eth_i40e_dev_uninit(struct rte_eth_dev *dev)
 static int
 i40e_dev_configure(struct rte_eth_dev *dev)
 {
+       struct i40e_adapter *ad =
+               I40E_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
        struct i40e_pf *pf = I40E_DEV_PRIVATE_TO_PF(dev->data->dev_private);
        enum rte_eth_rx_mq_mode mq_mode = dev->data->dev_conf.rxmode.mq_mode;
        int ret;
 
+       /* Initialize to TRUE. If any of Rx queues doesn't meet the
+        * bulk allocation or vector Rx preconditions we will reset it.
+        */
+       ad->rx_bulk_alloc_allowed = true;
+       ad->rx_vec_allowed = true;
+       ad->tx_simple_allowed = true;
+       ad->tx_vec_allowed = true;
+
        if (dev->data->dev_conf.fdir_conf.mode == RTE_FDIR_MODE_PERFECT) {
                ret = i40e_fdir_setup(pf);
                if (ret != I40E_SUCCESS) {
@@ -3901,6 +3911,9 @@ i40e_dev_tx_init(struct i40e_pf *pf)
                if (ret != I40E_SUCCESS)
                        break;
        }
+       if (ret == I40E_SUCCESS)
+               i40e_set_tx_function(container_of(pf, struct i40e_adapter, pf)
+                                    ->eth_dev);
 
        return ret;
 }
@@ -3927,6 +3940,9 @@ i40e_dev_rx_init(struct i40e_pf *pf)
                        break;
                }
        }
+       if (ret == I40E_SUCCESS)
+               i40e_set_rx_function(container_of(pf, struct i40e_adapter, pf)
+                                    ->eth_dev);
 
        return ret;
 }

diff --git a/drivers/net/i40e/i40e_ethdev.h b/drivers/net/i40e/i40e_ethdev.h
index 70804e1..832c036 100644 (file)
--- a/drivers/net/i40e/i40e_ethdev.h
+++ b/drivers/net/i40e/i40e_ethdev.h
@@ -479,6 +479,12 @@ struct i40e_adapter {
                struct i40e_pf pf;
                struct i40e_vf vf;
        };
+
+       /* for vector PMD */
+       bool rx_bulk_alloc_allowed;
+       bool rx_vec_allowed;
+       bool tx_simple_allowed;
+       bool tx_vec_allowed;
 };
 
 int i40e_dev_switch_queues(struct i40e_pf *pf, bool on);

diff --git a/drivers/net/i40e/i40e_ethdev_vf.c b/drivers/net/i40e/i40e_ethdev_vf.c
index 1324281..48b904e 100644 (file)
--- a/drivers/net/i40e/i40e_ethdev_vf.c
+++ b/drivers/net/i40e/i40e_ethdev_vf.c
@@ -1197,8 +1197,8 @@ i40evf_dev_init(struct rte_eth_dev *eth_dev)
         * has already done this work.
         */
        if (rte_eal_process_type() != RTE_PROC_PRIMARY){
-               if (eth_dev->data->scattered_rx)
-                       eth_dev->rx_pkt_burst = i40e_recv_scattered_pkts;
+               i40e_set_rx_function(eth_dev);
+               i40e_set_tx_function(eth_dev);
                return 0;
        }
 
@@ -1292,6 +1292,17 @@ PMD_REGISTER_DRIVER(rte_i40evf_driver);
 static int
 i40evf_dev_configure(struct rte_eth_dev *dev)
 {
+       struct i40e_adapter *ad =
+               I40E_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
+
+       /* Initialize to TRUE. If any of Rx queues doesn't meet the bulk
+        * allocation or vector Rx preconditions we will reset it.
+        */
+       ad->rx_bulk_alloc_allowed = true;
+       ad->rx_vec_allowed = true;
+       ad->tx_simple_allowed = true;
+       ad->tx_vec_allowed = true;
+
        return i40evf_init_vlan(dev);
 }
 
@@ -1523,7 +1534,6 @@ i40evf_rxq_init(struct rte_eth_dev *dev, struct i40e_rx_queue *rxq)
        if (dev_data->dev_conf.rxmode.enable_scatter ||
            (rxq->max_pkt_len + 2 * I40E_VLAN_TAG_SIZE) > buf_size) {
                dev_data->scattered_rx = 1;
-               dev->rx_pkt_burst = i40e_recv_scattered_pkts;
        }
 
        return 0;
@@ -1534,6 +1544,7 @@ i40evf_rx_init(struct rte_eth_dev *dev)
 {
        struct i40e_vf *vf = I40EVF_DEV_PRIVATE_TO_VF(dev->data->dev_private);
        uint16_t i;
+       int ret = I40E_SUCCESS;
        struct i40e_rx_queue **rxq =
                (struct i40e_rx_queue **)dev->data->rx_queues;
 
@@ -1541,11 +1552,14 @@ i40evf_rx_init(struct rte_eth_dev *dev)
        for (i = 0; i < dev->data->nb_rx_queues; i++) {
                if (!rxq[i] || !rxq[i]->q_set)
                        continue;
-               if (i40evf_rxq_init(dev, rxq[i]) < 0)
-                       return -EFAULT;
+               ret = i40evf_rxq_init(dev, rxq[i]);
+               if (ret != I40E_SUCCESS)
+                       break;
        }
+       if (ret == I40E_SUCCESS)
+               i40e_set_rx_function(dev);
 
-       return 0;
+       return ret;
 }
 
 static void
@@ -1558,6 +1572,8 @@ i40evf_tx_init(struct rte_eth_dev *dev)
 
        for (i = 0; i < dev->data->nb_tx_queues; i++)
                txq[i]->qtx_tail = hw->hw_addr + I40E_QTX_TAIL1(i);
+
+       i40e_set_tx_function(dev);
 }
 
 static inline void

diff --git a/drivers/net/i40e/i40e_rxtx.c b/drivers/net/i40e/i40e_rxtx.c
index d37cbde..8731712 100644 (file)
--- a/drivers/net/i40e/i40e_rxtx.c
+++ b/drivers/net/i40e/i40e_rxtx.c
@@ -2105,6 +2105,8 @@ i40e_dev_rx_queue_setup(struct rte_eth_dev *dev,
        struct i40e_vsi *vsi;
        struct i40e_hw *hw = I40E_DEV_PRIVATE_TO_HW(dev->data->dev_private);
        struct i40e_pf *pf = I40E_DEV_PRIVATE_TO_PF(dev->data->dev_private);
+       struct i40e_adapter *ad =
+               I40E_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
        struct i40e_rx_queue *rxq;
        const struct rte_memzone *rz;
        uint32_t ring_size;
@@ -2213,13 +2215,12 @@ i40e_dev_rx_queue_setup(struct rte_eth_dev *dev,
 
        use_def_burst_func = check_rx_burst_bulk_alloc_preconditions(rxq);
 
-       if (!use_def_burst_func && !dev->data->scattered_rx) {
+       if (!use_def_burst_func) {
 #ifdef RTE_LIBRTE_I40E_RX_ALLOW_BULK_ALLOC
                PMD_INIT_LOG(DEBUG, "Rx Burst Bulk Alloc Preconditions are "
                             "satisfied. Rx Burst Bulk Alloc function will be "
                             "used on port=%d, queue=%d.",
                             rxq->port_id, rxq->queue_id);
-               dev->rx_pkt_burst = i40e_recv_pkts_bulk_alloc;
 #endif /* RTE_LIBRTE_I40E_RX_ALLOW_BULK_ALLOC */
        } else {
                PMD_INIT_LOG(DEBUG, "Rx Burst Bulk Alloc Preconditions are "
@@ -2227,6 +2228,7 @@ i40e_dev_rx_queue_setup(struct rte_eth_dev *dev,
                             "or RTE_LIBRTE_I40E_RX_ALLOW_BULK_ALLOC is "
                             "not enabled on port=%d, queue=%d.",
                             rxq->port_id, rxq->queue_id);
+               ad->rx_bulk_alloc_allowed = false;
        }
 
        return 0;
@@ -2488,14 +2490,7 @@ i40e_dev_tx_queue_setup(struct rte_eth_dev *dev,
        dev->data->tx_queues[queue_idx] = txq;
 
        /* Use a simple TX queue without offloads or multi segs if possible */
-       if (((txq->txq_flags & I40E_SIMPLE_FLAGS) == I40E_SIMPLE_FLAGS) &&
-                               (txq->tx_rs_thresh >= I40E_TX_MAX_BURST)) {
-               PMD_INIT_LOG(INFO, "Using simple tx path");
-               dev->tx_pkt_burst = i40e_xmit_pkts_simple;
-       } else {
-               PMD_INIT_LOG(INFO, "Using full-featured tx path");
-               dev->tx_pkt_burst = i40e_xmit_pkts;
-       }
+       i40e_set_tx_function_flag(dev, txq);
 
        return 0;
 }
@@ -2564,6 +2559,12 @@ i40e_rx_queue_release_mbufs(struct i40e_rx_queue *rxq)
 {
        uint16_t i;
 
+       /* SSE Vector driver has a different way of releasing mbufs. */
+       if (rxq->rx_using_sse) {
+               i40e_rx_queue_release_mbufs_vec(rxq);
+               return;
+       }
+
        if (!rxq || !rxq->sw_ring) {
                PMD_DRV_LOG(DEBUG, "Pointer to rxq or sw_ring is NULL");
                return;
@@ -2837,7 +2838,6 @@ i40e_rx_queue_init(struct i40e_rx_queue *rxq)
        int err = I40E_SUCCESS;
        struct i40e_hw *hw = I40E_VSI_TO_HW(rxq->vsi);
        struct rte_eth_dev_data *dev_data = I40E_VSI_TO_DEV_DATA(rxq->vsi);
-       struct rte_eth_dev *dev = I40E_VSI_TO_ETH_DEV(rxq->vsi);
        uint16_t pf_q = rxq->reg_idx;
        uint16_t buf_size;
        struct i40e_hmc_obj_rxq rx_ctx;
@@ -2893,7 +2893,6 @@ i40e_rx_queue_init(struct i40e_rx_queue *rxq)
        /* Check if scattered RX needs to be used. */
        if ((rxq->max_pkt_len + 2 * I40E_VLAN_TAG_SIZE) > buf_size) {
                dev_data->scattered_rx = 1;
-               dev->rx_pkt_burst = i40e_recv_scattered_pkts;
        }
 
        /* Init the RX tail regieter. */
@@ -3064,7 +3063,159 @@ i40e_fdir_setup_rx_resources(struct i40e_pf *pf)
        return I40E_SUCCESS;
 }
 
+void __attribute__((cold))
+i40e_set_rx_function(struct rte_eth_dev *dev)
+{
+       struct i40e_adapter *ad =
+               I40E_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
+       uint16_t rx_using_sse, i;
+       /* In order to allow Vector Rx there are a few configuration
+        * conditions to be met and Rx Bulk Allocation should be allowed.
+        */
+       if (rte_eal_process_type() == RTE_PROC_PRIMARY) {
+               if (i40e_rx_vec_dev_conf_condition_check(dev) ||
+                   !ad->rx_bulk_alloc_allowed) {
+                       PMD_INIT_LOG(DEBUG, "Port[%d] doesn't meet"
+                                    " Vector Rx preconditions",
+                                    dev->data->port_id);
+
+                       ad->rx_vec_allowed = false;
+               }
+               if (ad->rx_vec_allowed) {
+                       for (i = 0; i < dev->data->nb_rx_queues; i++) {
+                               struct i40e_rx_queue *rxq =
+                                       dev->data->rx_queues[i];
+
+                               if (i40e_rxq_vec_setup(rxq)) {
+                                       ad->rx_vec_allowed = false;
+                                       break;
+                               }
+                       }
+               }
+       }
+
+       if (dev->data->scattered_rx) {
+               /* Set the non-LRO scattered callback: there are Vector and
+                * single allocation versions.
+                */
+               if (ad->rx_vec_allowed) {
+                       PMD_INIT_LOG(DEBUG, "Using Vector Scattered Rx "
+                                           "callback (port=%d).",
+                                    dev->data->port_id);
+
+                       dev->rx_pkt_burst = i40e_recv_scattered_pkts_vec;
+               } else {
+                       PMD_INIT_LOG(DEBUG, "Using a Scattered with bulk "
+                                          "allocation callback (port=%d).",
+                                    dev->data->port_id);
+                       dev->rx_pkt_burst = i40e_recv_scattered_pkts;
+               }
+       /* If parameters allow we are going to choose between the following
+        * callbacks:
+        *    - Vector
+        *    - Bulk Allocation
+        *    - Single buffer allocation (the simplest one)
+        */
+       } else if (ad->rx_vec_allowed) {
+               PMD_INIT_LOG(DEBUG, "Vector rx enabled, please make sure RX "
+                                   "burst size no less than %d (port=%d).",
+                            RTE_I40E_DESCS_PER_LOOP,
+                            dev->data->port_id);
+
+               dev->rx_pkt_burst = i40e_recv_pkts_vec;
+       } else if (ad->rx_bulk_alloc_allowed) {
+               PMD_INIT_LOG(DEBUG, "Rx Burst Bulk Alloc Preconditions are "
+                                   "satisfied. Rx Burst Bulk Alloc function "
+                                   "will be used on port=%d.",
+                            dev->data->port_id);
+
+               dev->rx_pkt_burst = i40e_recv_pkts_bulk_alloc;
+       } else {
+               PMD_INIT_LOG(DEBUG, "Rx Burst Bulk Alloc Preconditions are not "
+                                   "satisfied, or Scattered Rx is requested "
+                                   "(port=%d).",
+                            dev->data->port_id);
+
+               dev->rx_pkt_burst = i40e_recv_pkts;
+       }
+
+       /* Propagate information about RX function choice through all queues. */
+       if (rte_eal_process_type() == RTE_PROC_PRIMARY) {
+               rx_using_sse =
+                       (dev->rx_pkt_burst == i40e_recv_scattered_pkts_vec ||
+                        dev->rx_pkt_burst == i40e_recv_pkts_vec);
+
+               for (i = 0; i < dev->data->nb_rx_queues; i++) {
+                       struct i40e_rx_queue *rxq = dev->data->rx_queues[i];
+
+                       rxq->rx_using_sse = rx_using_sse;
+               }
+       }
+}
+
+void __attribute__((cold))
+i40e_set_tx_function_flag(struct rte_eth_dev *dev, struct i40e_tx_queue *txq)
+{
+       struct i40e_adapter *ad =
+               I40E_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
+
+       /* Use a simple Tx queue (no offloads, no multi segs) if possible */
+       if (((txq->txq_flags & I40E_SIMPLE_FLAGS) == I40E_SIMPLE_FLAGS)
+                       && (txq->tx_rs_thresh >= RTE_PMD_I40E_TX_MAX_BURST)) {
+               if (txq->tx_rs_thresh <= RTE_I40E_TX_MAX_FREE_BUF_SZ) {
+                       PMD_INIT_LOG(DEBUG, "Vector tx"
+                                    " can be enabled on this txq.");
+
+               } else {
+                       ad->tx_vec_allowed = false;
+               }
+       } else {
+               ad->tx_simple_allowed = false;
+       }
+}
+
+void __attribute__((cold))
+i40e_set_tx_function(struct rte_eth_dev *dev)
+{
+       struct i40e_adapter *ad =
+               I40E_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
+       int i;
+
+       if (rte_eal_process_type() == RTE_PROC_PRIMARY) {
+               if (ad->tx_vec_allowed) {
+                       for (i = 0; i < dev->data->nb_tx_queues; i++) {
+                               struct i40e_tx_queue *txq =
+                                       dev->data->tx_queues[i];
+
+                               if (i40e_txq_vec_setup(txq)) {
+                                       ad->tx_vec_allowed = false;
+                                       break;
+                               }
+                       }
+               }
+       }
+
+       if (ad->tx_simple_allowed) {
+               if (ad->tx_vec_allowed) {
+                       PMD_INIT_LOG(DEBUG, "Vector tx finally be used.");
+                       dev->tx_pkt_burst = i40e_xmit_pkts_vec;
+               } else {
+                       PMD_INIT_LOG(DEBUG, "Simple tx finally be used.");
+                       dev->tx_pkt_burst = i40e_xmit_pkts_simple;
+               }
+       } else {
+               PMD_INIT_LOG(DEBUG, "Xmit tx finally be used.");
+               dev->tx_pkt_burst = i40e_xmit_pkts;
+       }
+}
+
 /* Stubs needed for linkage when CONFIG_RTE_I40E_INC_VECTOR is set to 'n' */
+int __attribute__((weak))
+i40e_rx_vec_dev_conf_condition_check(struct rte_eth_dev __rte_unused *dev)
+{
+       return -1;
+}
+
 uint16_t __attribute__((weak))
 i40e_recv_pkts_vec(
        void __rte_unused *rx_queue,
@@ -3089,6 +3240,12 @@ i40e_rxq_vec_setup(struct i40e_rx_queue __rte_unused *rxq)
        return -1;
 }
 
+int __attribute__((weak))
+i40e_txq_vec_setup(struct i40e_tx_queue __rte_unused *txq)
+{
+       return -1;
+}
+
 void __attribute__((weak))
 i40e_rx_queue_release_mbufs_vec(struct i40e_rx_queue __rte_unused*rxq)
 {
@@ -3102,3 +3259,4 @@ i40e_xmit_pkts_vec(void __rte_unused *tx_queue,
 {
        return 0;
 }
+

diff --git a/drivers/net/i40e/i40e_rxtx.h b/drivers/net/i40e/i40e_rxtx.h
index 98d6e90..39cb95a 100644 (file)
--- a/drivers/net/i40e/i40e_rxtx.h
+++ b/drivers/net/i40e/i40e_rxtx.h
@@ -132,6 +132,7 @@ struct i40e_rx_queue {
        uint8_t hs_mode; /* Header Split mode */
        bool q_set; /**< indicate if rx queue has been configured */
        bool rx_deferred_start; /**< don't start this queue in dev start */
+       uint16_t rx_using_sse; /**<flag indicate the usage of vPMD for rx */
 };
 
 struct i40e_tx_entry {
@@ -234,10 +235,15 @@ uint16_t i40e_recv_pkts_vec(void *rx_queue, struct rte_mbuf **rx_pkts,
 uint16_t i40e_recv_scattered_pkts_vec(void *rx_queue,
                                      struct rte_mbuf **rx_pkts,
                                      uint16_t nb_pkts);
+int i40e_rx_vec_dev_conf_condition_check(struct rte_eth_dev *dev);
 int i40e_rxq_vec_setup(struct i40e_rx_queue *rxq);
 int i40e_txq_vec_setup(struct i40e_tx_queue *txq);
 void i40e_rx_queue_release_mbufs_vec(struct i40e_rx_queue *rxq);
 uint16_t i40e_xmit_pkts_vec(void *tx_queue, struct rte_mbuf **tx_pkts,
                            uint16_t nb_pkts);
+void i40e_set_rx_function(struct rte_eth_dev *dev);
+void i40e_set_tx_function_flag(struct rte_eth_dev *dev,
+                              struct i40e_tx_queue *txq);
+void i40e_set_tx_function(struct rte_eth_dev *dev);
 
 #endif /* _I40E_RXTX_H_ */

diff --git a/drivers/net/i40e/i40e_rxtx_vec.c b/drivers/net/i40e/i40e_rxtx_vec.c
index f746209..047aff5 100644 (file)
--- a/drivers/net/i40e/i40e_rxtx_vec.c
+++ b/drivers/net/i40e/i40e_rxtx_vec.c
@@ -743,3 +743,35 @@ i40e_txq_vec_setup(struct i40e_tx_queue __rte_unused *txq)
 {
        return 0;
 }
+
+int __attribute__((cold))
+i40e_rx_vec_dev_conf_condition_check(struct rte_eth_dev *dev)
+{
+#ifndef RTE_LIBRTE_IEEE1588
+       struct rte_eth_rxmode *rxmode = &dev->data->dev_conf.rxmode;
+       struct rte_fdir_conf *fconf = &dev->data->dev_conf.fdir_conf;
+
+#ifndef RTE_LIBRTE_I40E_RX_OLFLAGS_ENABLE
+       /* whithout rx ol_flags, no VP flag report */
+       if (rxmode->hw_vlan_strip != 0 ||
+           rxmode->hw_vlan_extend != 0)
+               return -1;
+#endif
+
+       /* no fdir support */
+       if (fconf->mode != RTE_FDIR_MODE_NONE)
+               return -1;
+
+        /* - no csum error report support
+        * - no header split support
+        */
+       if (rxmode->hw_ip_checksum == 1 ||
+           rxmode->header_split == 1)
+               return -1;
+
+       return 0;
+#else
+       RTE_SET_USED(dev);
+       return -1;
+#endif
+}