nfp: add link status interrupt
authorAlejandro Lucero <alejandro.lucero@netronome.com>
Mon, 30 Nov 2015 10:25:41 +0000 (10:25 +0000)
committerThomas Monjalon <thomas.monjalon@6wind.com>
Tue, 8 Dec 2015 02:00:42 +0000 (03:00 +0100)
Signed-off-by: Alejandro Lucero <alejandro.lucero@netronome.com>
Signed-off-by: Rolf Neugebauer <rolf.neugebauer@netronome.com>
drivers/net/nfp/nfp_net.c

index ff9a8d6..bc2089f 100644 (file)
@@ -73,6 +73,9 @@
 /* Prototypes */
 static void nfp_net_close(struct rte_eth_dev *dev);
 static int nfp_net_configure(struct rte_eth_dev *dev);
+static void nfp_net_dev_interrupt_handler(struct rte_intr_handle *handle,
+                                         void *param);
+static void nfp_net_dev_interrupt_delayed_handler(void *param);
 static int nfp_net_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu);
 static void nfp_net_infos_get(struct rte_eth_dev *dev,
                              struct rte_eth_dev_info *dev_info);
@@ -731,6 +734,7 @@ nfp_net_close(struct rte_eth_dev *dev)
 
        nfp_net_stop(dev);
 
+       rte_intr_disable(&dev->pci_dev->intr_handle);
        nn_cfg_writeb(hw, NFP_NET_CFG_LSC, 0xff);
 
        /*
@@ -1115,6 +1119,114 @@ nfp_net_rx_queue_count(struct rte_eth_dev *dev, uint16_t queue_idx)
        return count;
 }
 
+static void
+nfp_net_dev_link_status_print(struct rte_eth_dev *dev)
+{
+       struct rte_eth_link link;
+
+       memset(&link, 0, sizeof(link));
+       nfp_net_dev_atomic_read_link_status(dev, &link);
+       if (link.link_status)
+               RTE_LOG(INFO, PMD, "Port %d: Link Up - speed %u Mbps - %s\n",
+                       (int)(dev->data->port_id), (unsigned)link.link_speed,
+                       link.link_duplex == ETH_LINK_FULL_DUPLEX
+                       ? "full-duplex" : "half-duplex");
+       else
+               RTE_LOG(INFO, PMD, " Port %d: Link Down\n",
+                       (int)(dev->data->port_id));
+
+       RTE_LOG(INFO, PMD, "PCI Address: %04d:%02d:%02d:%d\n",
+               dev->pci_dev->addr.domain, dev->pci_dev->addr.bus,
+               dev->pci_dev->addr.devid, dev->pci_dev->addr.function);
+}
+
+/* Interrupt configuration and handling */
+
+/*
+ * nfp_net_irq_unmask - Unmask an interrupt
+ *
+ * If MSI-X auto-masking is enabled clear the mask bit, otherwise
+ * clear the ICR for the entry.
+ */
+static void
+nfp_net_irq_unmask(struct rte_eth_dev *dev)
+{
+       struct nfp_net_hw *hw;
+
+       hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+       if (hw->ctrl & NFP_NET_CFG_CTRL_MSIXAUTO) {
+               /* If MSI-X auto-masking is used, clear the entry */
+               rte_wmb();
+               rte_intr_enable(&dev->pci_dev->intr_handle);
+       } else {
+               /* Make sure all updates are written before un-masking */
+               rte_wmb();
+               nn_cfg_writeb(hw, NFP_NET_CFG_ICR(NFP_NET_IRQ_LSC_IDX),
+                             NFP_NET_CFG_ICR_UNMASKED);
+       }
+}
+
+static void
+nfp_net_dev_interrupt_handler(__rte_unused struct rte_intr_handle *handle,
+                             void *param)
+{
+       int64_t timeout;
+       struct rte_eth_link link;
+       struct rte_eth_dev *dev = (struct rte_eth_dev *)param;
+
+       PMD_DRV_LOG(DEBUG, "We got a LSC interrupt!!!\n");
+
+       /* get the link status */
+       memset(&link, 0, sizeof(link));
+       nfp_net_dev_atomic_read_link_status(dev, &link);
+
+       nfp_net_link_update(dev, 0);
+
+       /* likely to up */
+       if (!link.link_status) {
+               /* handle it 1 sec later, wait it being stable */
+               timeout = NFP_NET_LINK_UP_CHECK_TIMEOUT;
+               /* likely to down */
+       } else {
+               /* handle it 4 sec later, wait it being stable */
+               timeout = NFP_NET_LINK_DOWN_CHECK_TIMEOUT;
+       }
+
+       if (rte_eal_alarm_set(timeout * 1000,
+                             nfp_net_dev_interrupt_delayed_handler,
+                             (void *)dev) < 0) {
+               RTE_LOG(ERR, PMD, "Error setting alarm");
+               /* Unmasking */
+               nfp_net_irq_unmask(dev);
+       }
+}
+
+/*
+ * Interrupt handler which shall be registered for alarm callback for delayed
+ * handling specific interrupt to wait for the stable nic state. As the NIC
+ * interrupt state is not stable for nfp after link is just down, it needs
+ * to wait 4 seconds to get the stable status.
+ *
+ * @param handle   Pointer to interrupt handle.
+ * @param param    The address of parameter (struct rte_eth_dev *)
+ *
+ * @return  void
+ */
+static void
+nfp_net_dev_interrupt_delayed_handler(void *param)
+{
+       struct rte_eth_dev *dev = (struct rte_eth_dev *)param;
+
+       nfp_net_link_update(dev, 0);
+       _rte_eth_dev_callback_process(dev, RTE_ETH_EVENT_INTR_LSC);
+
+       nfp_net_dev_link_status_print(dev);
+
+       /* Unmasking */
+       nfp_net_irq_unmask(dev);
+}
+
 static int
 nfp_net_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu)
 {
@@ -2315,6 +2427,17 @@ nfp_net_init(struct rte_eth_dev *eth_dev)
                     hw->mac_addr[0], hw->mac_addr[1], hw->mac_addr[2],
                     hw->mac_addr[3], hw->mac_addr[4], hw->mac_addr[5]);
 
+       /* Registering LSC interrupt handler */
+       rte_intr_callback_register(&pci_dev->intr_handle,
+                                  nfp_net_dev_interrupt_handler,
+                                  (void *)eth_dev);
+
+       /* enable uio intr after callback register */
+       rte_intr_enable(&pci_dev->intr_handle);
+
+       /* Telling the firmware about the LSC interrupt entry */
+       nn_cfg_writeb(hw, NFP_NET_CFG_LSC, NFP_NET_IRQ_LSC_IDX);
+
        /* Recording current stats counters values */
        nfp_net_stats_reset(eth_dev);