[dpdk-dev] [PATCH v7 14/14] net/avf: enable Rx interrupt support

Wenzhuo Lu wenzhuo.lu at intel.com
Wed Jan 10 14:02:06 CET 2018


From: Jingjing Wu <jingjing.wu at intel.com>

Update the doc for the AVF features either.

Signed-off-by: Jingjing Wu <jingjing.wu at intel.com>
---
 doc/guides/nics/features/avf.ini       |   1 +
 doc/guides/nics/features/avf_vec.ini   |   1 +
 doc/guides/nics/intel_vf.rst           |  20 +++-
 doc/guides/rel_notes/release_18_02.rst |  16 +++
 drivers/net/avf/avf_ethdev.c           | 204 +++++++++++++++++++++++++++------
 5 files changed, 204 insertions(+), 38 deletions(-)

diff --git a/doc/guides/nics/features/avf.ini b/doc/guides/nics/features/avf.ini
index da4d81b..ccb9edd 100644
--- a/doc/guides/nics/features/avf.ini
+++ b/doc/guides/nics/features/avf.ini
@@ -7,6 +7,7 @@
 Speed capabilities   = Y
 Link status          = Y
 Link status event    = Y
+Rx interrupt         = Y
 Queue start/stop     = Y
 MTU update           = Y
 Jumbo frame          = Y
diff --git a/doc/guides/nics/features/avf_vec.ini b/doc/guides/nics/features/avf_vec.ini
index 45dd5e5..8924994 100644
--- a/doc/guides/nics/features/avf_vec.ini
+++ b/doc/guides/nics/features/avf_vec.ini
@@ -7,6 +7,7 @@
 Speed capabilities   = Y
 Link status          = Y
 Link status event    = Y
+Rx interrupt         = Y
 Queue start/stop     = Y
 MTU update           = Y
 Jumbo frame          = Y
diff --git a/doc/guides/nics/intel_vf.rst b/doc/guides/nics/intel_vf.rst
index 1e83bf6..66f90b1 100644
--- a/doc/guides/nics/intel_vf.rst
+++ b/doc/guides/nics/intel_vf.rst
@@ -28,8 +28,8 @@
     (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
     OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
-I40E/IXGBE/IGB Virtual Function Driver
-======================================
+Intel Virtual Function Driver
+=============================
 
 Supported Intel® Ethernet Controllers (see the *DPDK Release Notes* for details)
 support the following modes of operation in a virtualized environment:
@@ -93,6 +93,22 @@ and the Physical Function operates on the global resources on behalf of the Virt
 For this out-of-band communication, an SR-IOV enabled NIC provides a memory buffer for each Virtual Function,
 which is called a "Mailbox".
 
+Intel® Ethernet Adaptive Virtual Function
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+Adaptive Virtual Function (AVF) is a SR-IOV Virtual Function with the same device id (8086:1889) on different Intel Ethernet Controller.
+AVF Driver is VF driver which supports for all future Intel devices without requiring a VM update. And since this happens to be an adaptive VF driver,
+every new drop of the VF driver would add more and more advanced features that can be turned on in the VM if the underlying HW device supports those
+advanced features based on a device agnostic way without ever compromising on the base functionality. AVF provides generic hardware interface and
+interface between AVF driver and a compliant PF driver is specified.
+
+Intel products starting Ethernet Controller 710 Series to support Adaptive Virtual Function.
+
+The way to generate Virtual Function is like normal, and the resource of VF assignment depends on the NIC Infrastructure.
+
+For more detail on SR-IOV, please refer to the following documents:
+
+*   `Intel® AVF HAS <https://www.intel.com/content/dam/www/public/us/en/documents/product-specifications/ethernet-adaptive-virtual-function-hardware-spec.pdf>`_
+
 The PCIE host-interface of Intel Ethernet Switch FM10000 Series VF infrastructure
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
diff --git a/doc/guides/rel_notes/release_18_02.rst b/doc/guides/rel_notes/release_18_02.rst
index 24b67bb..0672b0e 100644
--- a/doc/guides/rel_notes/release_18_02.rst
+++ b/doc/guides/rel_notes/release_18_02.rst
@@ -41,6 +41,22 @@ New Features
      Also, make sure to start the actual text at the margin.
      =========================================================
 
+   * **Add AVF (Adaptive Virtual Function) net PMD.**
+
+     A new net PMD has been added, which supports Intel® Ethernet Adaptive
+     Virtual Function (AVF) with features list below:
+
+     * Basic Rx/Tx burst
+     * SSE vectorized Rx/Tx burst
+     * Promiscuous mode
+     * MAC/VLAN offload
+     * Checksum offload
+     * TSO offload
+     * Jumbo frame and MTU setting
+     * RSS configuration
+     * stats
+     * Rx/Tx descriptor status
+     * Link status update/event
 
 API Changes
 -----------
diff --git a/drivers/net/avf/avf_ethdev.c b/drivers/net/avf/avf_ethdev.c
index d9f7cea..13f6329 100644
--- a/drivers/net/avf/avf_ethdev.c
+++ b/drivers/net/avf/avf_ethdev.c
@@ -67,9 +67,14 @@ static int avf_dev_rss_hash_conf_get(struct rte_eth_dev *dev,
 static int avf_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu);
 static void avf_dev_set_default_mac_addr(struct rte_eth_dev *dev,
 					 struct ether_addr *mac_addr);
+static int avf_dev_rx_queue_intr_enable(struct rte_eth_dev *dev,
+					uint16_t queue_id);
+static int avf_dev_rx_queue_intr_disable(struct rte_eth_dev *dev,
+					 uint16_t queue_id);
 
 int avf_logtype_init;
 int avf_logtype_driver;
+
 static const struct rte_pci_id pci_id_avf_map[] = {
 	{ RTE_PCI_DEVICE(AVF_INTEL_VENDOR_ID, AVF_DEV_ID_ADAPTIVE_VF) },
 	{ .vendor_id = 0, /* sentinel */ },
@@ -111,6 +116,8 @@ static void avf_dev_set_default_mac_addr(struct rte_eth_dev *dev,
 	.rx_descriptor_status       = avf_dev_rx_desc_status,
 	.tx_descriptor_status       = avf_dev_tx_desc_status,
 	.mtu_set                    = avf_dev_mtu_set,
+	.rx_queue_intr_enable       = avf_dev_rx_queue_intr_enable,
+	.rx_queue_intr_disable      = avf_dev_rx_queue_intr_disable,
 };
 
 static int
@@ -275,6 +282,99 @@ static void avf_dev_set_default_mac_addr(struct rte_eth_dev *dev,
 	return ret;
 }
 
+static int avf_config_rx_queues_irqs(struct rte_eth_dev *dev,
+				     struct rte_intr_handle *intr_handle)
+{
+	struct avf_adapter *adapter =
+		AVF_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
+	struct avf_info *vf = AVF_DEV_PRIVATE_TO_VF(adapter);
+	struct avf_hw *hw = AVF_DEV_PRIVATE_TO_HW(adapter);
+	uint16_t interval, i;
+	int vec;
+
+	if (dev->data->dev_conf.intr_conf.rxq != 0) {
+		if (rte_intr_efd_enable(intr_handle, dev->data->nb_rx_queues))
+			return -1;
+	}
+
+	if (rte_intr_dp_is_en(intr_handle) && !intr_handle->intr_vec) {
+		intr_handle->intr_vec =
+			rte_zmalloc("intr_vec",
+				    dev->data->nb_rx_queues * sizeof(int), 0);
+		if (!intr_handle->intr_vec) {
+			PMD_DRV_LOG(ERR, "Failed to allocate %d rx intr_vec",
+				    dev->data->nb_rx_queues);
+			return -1;
+		}
+	}
+
+	if (!dev->data->dev_conf.intr_conf.rxq) {
+		/* Rx interrupt disabled, Map interrupt only for writeback */
+		vf->nb_msix = 1;
+		if (vf->vf_res->vf_cap_flags &
+		    VIRTCHNL_VF_OFFLOAD_WB_ON_ITR) {
+			/* If WB_ON_ITR supports, enable it */
+			vf->msix_base = AVF_RX_VEC_START;
+			AVF_WRITE_REG(hw, AVFINT_DYN_CTLN1(vf->msix_base - 1),
+				      AVFINT_DYN_CTLN1_ITR_INDX_MASK |
+				      AVFINT_DYN_CTLN1_WB_ON_ITR_MASK);
+		} else {
+			/* If no WB_ON_ITR offload flags, need to set
+			 * interrupt for descriptor write back.
+			 */
+			vf->msix_base = AVF_MISC_VEC_ID;
+
+			/* set ITR to max */
+			interval = avf_calc_itr_interval(
+					AVF_QUEUE_ITR_INTERVAL_MAX);
+			AVF_WRITE_REG(hw, AVFINT_DYN_CTL01,
+				      AVFINT_DYN_CTL01_INTENA_MASK |
+				      (AVF_ITR_INDEX_DEFAULT <<
+				       AVFINT_DYN_CTL01_ITR_INDX_SHIFT) |
+				      (interval <<
+				       AVFINT_DYN_CTL01_INTERVAL_SHIFT));
+		}
+		AVF_WRITE_FLUSH(hw);
+		/* map all queues to the same interrupt */
+		for (i = 0; i < dev->data->nb_rx_queues; i++)
+			vf->rxq_map[0] |= 1 << i;
+	} else {
+		if (!rte_intr_allow_others(intr_handle)) {
+			vf->nb_msix = 1;
+			vf->msix_base = AVF_MISC_VEC_ID;
+			for (i = 0; i < dev->data->nb_rx_queues; i++) {
+				vf->rxq_map[0] |= 1 << i;
+				intr_handle->intr_vec[i] = AVF_MISC_VEC_ID;
+			}
+			PMD_DRV_LOG(DEBUG,
+				    "vector 0 are mapping to all Rx queues");
+		} else {
+			/* If Rx interrupt is reuquired, and we can use
+			 * multi interrupts, then the vec is from 1
+			 */
+			vf->nb_msix = RTE_MIN(vf->vf_res->max_vectors,
+					      intr_handle->nb_efd);
+			vf->msix_base = AVF_RX_VEC_START;
+			vec = AVF_RX_VEC_START;
+			for (i = 0; i < dev->data->nb_rx_queues; i++) {
+				vf->rxq_map[vec] |= 1 << i;
+				intr_handle->intr_vec[i] = vec++;
+				if (vec >= vf->nb_msix)
+					vec = AVF_RX_VEC_START;
+			}
+			PMD_DRV_LOG(DEBUG,
+				    "%u vectors are mapping to %u Rx queues",
+				    vf->nb_msix, dev->data->nb_rx_queues);
+		}
+	}
+
+	if (avf_config_irq_map(adapter)) {
+		PMD_DRV_LOG(ERR, "config interrupt mapping failed");
+		return -1;
+	}
+	return 0;
+}
+
 static int
 avf_start_queues(struct rte_eth_dev *dev)
 {
@@ -314,8 +414,6 @@ static void avf_dev_set_default_mac_addr(struct rte_eth_dev *dev,
 	struct avf_hw *hw = AVF_DEV_PRIVATE_TO_HW(dev->data->dev_private);
 	struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
 	struct rte_intr_handle *intr_handle = dev->intr_handle;
-	uint16_t interval;
-	int i;
 
 	PMD_INIT_FUNC_TRACE();
 
@@ -325,8 +423,6 @@ static void avf_dev_set_default_mac_addr(struct rte_eth_dev *dev,
 	vf->num_queue_pairs = RTE_MAX(dev->data->nb_rx_queues,
 				      dev->data->nb_tx_queues);
 
-	/* TODO: Rx interrupt */
-
 	if (avf_init_queues(dev) != 0) {
 		PMD_DRV_LOG(ERR, "failed to do Queue init");
 		return -1;
@@ -344,36 +440,15 @@ static void avf_dev_set_default_mac_addr(struct rte_eth_dev *dev,
 		goto err_queue;
 	}
 
-	/* Map interrupt for writeback */
-	vf->nb_msix = 1;
-	if (vf->vf_res->vf_cap_flags & VIRTCHNL_VF_OFFLOAD_WB_ON_ITR) {
-		/* If WB_ON_ITR supports, enable it */
-		vf->msix_base = AVF_RX_VEC_START;
-		AVF_WRITE_REG(hw, AVFINT_DYN_CTLN1(vf->msix_base - 1),
-			      AVFINT_DYN_CTLN1_ITR_INDX_MASK |
-			      AVFINT_DYN_CTLN1_WB_ON_ITR_MASK);
-	} else {
-		/* If no WB_ON_ITR offload flags, need to set interrupt for
-		 * descriptor write back.
-		 */
-		vf->msix_base = AVF_MISC_VEC_ID;
-
-		/* set ITR to max */
-		interval = avf_calc_itr_interval(AVF_QUEUE_ITR_INTERVAL_MAX);
-		AVF_WRITE_REG(hw, AVFINT_DYN_CTL01,
-			      AVFINT_DYN_CTL01_INTENA_MASK |
-			      (AVF_ITR_INDEX_DEFAULT <<
-			       AVFINT_DYN_CTL01_ITR_INDX_SHIFT) |
-			      (interval << AVFINT_DYN_CTL01_INTERVAL_SHIFT));
-	}
-	AVF_WRITE_FLUSH(hw);
-	/* map all queues to the same interrupt */
-	for (i = 0; i < dev->data->nb_rx_queues; i++)
-		vf->rxq_map[0] |= 1 << i;
-	if (avf_config_irq_map(adapter)) {
-		PMD_DRV_LOG(ERR, "config interrupt mapping failed");
+	if (avf_config_rx_queues_irqs(dev, intr_handle) != 0) {
+		PMD_DRV_LOG(ERR, "configure irq failed");
 		goto err_queue;
 	}
+	/* re-enable intr again, because efd assign may change */
+	if (dev->data->dev_conf.intr_conf.rxq != 0) {
+		rte_intr_disable(intr_handle);
+		rte_intr_enable(intr_handle);
+	}
 
 	/* Set all mac addrs */
 	avf_add_del_all_mac_addr(adapter, TRUE);
@@ -383,7 +458,6 @@ static void avf_dev_set_default_mac_addr(struct rte_eth_dev *dev,
 		goto err_mac;
 	}
 
-	/* TODO: enable interrupt for RX interrupt */
 	return 0;
 
 err_mac:
@@ -399,6 +473,8 @@ static void avf_dev_set_default_mac_addr(struct rte_eth_dev *dev,
 	struct avf_adapter *adapter =
 		AVF_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
 	struct avf_hw *hw = AVF_DEV_PRIVATE_TO_HW(dev);
+	struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
+	struct rte_intr_handle *intr_handle = dev->intr_handle;
 	int ret, i;
 
 	PMD_INIT_FUNC_TRACE();
@@ -408,9 +484,13 @@ static void avf_dev_set_default_mac_addr(struct rte_eth_dev *dev,
 
 	avf_stop_queues(dev);
 
-	/*TODO: Disable the interrupt for Rx*/
-
-	/* TODO: Rx interrupt vector mapping free */
+	/* Disable the interrupt for Rx */
+	rte_intr_efd_disable(intr_handle);
+	/* Rx interrupt vector mapping free */
+	if (intr_handle->intr_vec) {
+		rte_free(intr_handle->intr_vec);
+		intr_handle->intr_vec = NULL;
+	}
 
 	/* remove all mac addrs */
 	avf_add_del_all_mac_addr(adapter, FALSE);
@@ -913,6 +993,58 @@ static void avf_dev_set_default_mac_addr(struct rte_eth_dev *dev,
 }
 
 static int
+avf_dev_rx_queue_intr_enable(struct rte_eth_dev *dev, uint16_t queue_id)
+{
+	struct avf_adapter *adapter =
+		AVF_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
+	struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
+	struct avf_hw *hw = AVF_DEV_PRIVATE_TO_HW(adapter);
+	uint16_t msix_intr;
+
+	msix_intr = pci_dev->intr_handle.intr_vec[queue_id];
+	if (msix_intr == AVF_MISC_VEC_ID) {
+		PMD_DRV_LOG(INFO, "MISC is also enabled for control");
+		AVF_WRITE_REG(hw, AVFINT_DYN_CTL01,
+			      AVFINT_DYN_CTL01_INTENA_MASK |
+			      AVFINT_DYN_CTL01_ITR_INDX_MASK);
+	} else {
+		AVF_WRITE_REG(hw,
+			      AVFINT_DYN_CTLN1(msix_intr - AVF_RX_VEC_START),
+			      AVFINT_DYN_CTLN1_INTENA_MASK |
+			      AVFINT_DYN_CTLN1_ITR_INDX_MASK);
+	}
+
+	AVF_WRITE_FLUSH(hw);
+
+	rte_intr_enable(&pci_dev->intr_handle);
+
+	return 0;
+}
+
+static int
+avf_dev_rx_queue_intr_disable(struct rte_eth_dev *dev, uint16_t queue_id)
+{
+	struct avf_adapter *adapter =
+		AVF_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
+	struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
+	struct avf_hw *hw = AVF_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+	uint16_t msix_intr;
+
+	msix_intr = pci_dev->intr_handle.intr_vec[queue_id];
+	if (msix_intr == AVF_MISC_VEC_ID) {
+		PMD_DRV_LOG(ERR, "MISC is used for control, cannot disable it");
+		return -EIO;
+	}
+
+	AVF_WRITE_REG(hw,
+		      AVFINT_DYN_CTLN1(msix_intr - AVF_RX_VEC_START),
+		      0);
+
+	AVF_WRITE_FLUSH(hw);
+	return 0;
+}
+
+static int
 avf_check_vf_reset_done(struct avf_hw *hw)
 {
 	int i, reset;
-- 
1.9.3



More information about the dev mailing list