RDMA/nes: Fix SFP+ link down detection issue with switch port disable

In case of SFP+ PHY, link status check at interrupt processing can
give false results.  For proper link status change detection a delayed
recheck is needed to give nes registers time to settle.  Add a
periodic link status recheck scheduled at interrupt to detect
potential delayed registers state changes.

Addresses: http://bugs.openfabrics.org/bugzilla/show_bug.cgi?id=2117
Signed-off-by: Maciej Sosnowski <maciej.sosnowski@intel.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

4 files changed, 100 insertions, 0 deletions

diff --git a/drivers/infiniband/hw/nes/nes.c b/drivers/infiniband/hw/nes/nes.c
index e17f52c..3b4ec32 100644
--- a/drivers/infiniband/hw/nes/nes.c
+++ b/drivers/infiniband/hw/nes/nes.c
@@ -674,6 +674,8 @@ static int __devinit nes_probe(struct pci_dev *pcidev, const struct pci_device_i
 	}
 	nes_notifiers_registered++;
 
+	INIT_DELAYED_WORK(&nesdev->work, nes_recheck_link_status);
+
 	/* Initialize network devices */
 	if ((netdev = nes_netdev_init(nesdev, mmio_regs)) == NULL)
 		goto bail7;
@@ -756,6 +758,7 @@ static void __devexit nes_remove(struct pci_dev *pcidev)
 	struct nes_device *nesdev = pci_get_drvdata(pcidev);
 	struct net_device *netdev;
 	int netdev_index = 0;
+	unsigned long flags;
 
 		if (nesdev->netdev_count) {
 			netdev = nesdev->netdev[netdev_index];
@@ -782,6 +785,14 @@ static void __devexit nes_remove(struct pci_dev *pcidev)
 	free_irq(pcidev->irq, nesdev);
 	tasklet_kill(&nesdev->dpc_tasklet);
 
+	spin_lock_irqsave(&nesdev->nesadapter->phy_lock, flags);
+	if (nesdev->link_recheck) {
+		spin_unlock_irqrestore(&nesdev->nesadapter->phy_lock, flags);
+		cancel_delayed_work_sync(&nesdev->work);
+	} else {
+		spin_unlock_irqrestore(&nesdev->nesadapter->phy_lock, flags);
+	}
+
 	/* Deallocate the Adapter Structure */
 	nes_destroy_adapter(nesdev->nesadapter);
 
diff --git a/drivers/infiniband/hw/nes/nes.h b/drivers/infiniband/hw/nes/nes.h
index b3d145e..6fe7987 100644
--- a/drivers/infiniband/hw/nes/nes.h
+++ b/drivers/infiniband/hw/nes/nes.h
@@ -268,6 +268,9 @@ struct nes_device {
 	u8                     napi_isr_ran;
 	u8                     disable_rx_flow_control;
 	u8                     disable_tx_flow_control;
+
+	struct delayed_work    work;
+	u8                     link_recheck;
 };
 
 
@@ -507,6 +510,7 @@ void nes_nic_ce_handler(struct nes_device *, struct nes_hw_nic_cq *);
 void nes_iwarp_ce_handler(struct nes_device *, struct nes_hw_cq *);
 int nes_destroy_cqp(struct nes_device *);
 int nes_nic_cm_xmit(struct sk_buff *, struct net_device *);
+void nes_recheck_link_status(struct work_struct *work);
 
 /* nes_nic.c */
 struct net_device *nes_netdev_init(struct nes_device *, void __iomem *);
diff --git a/drivers/infiniband/hw/nes/nes_hw.c b/drivers/infiniband/hw/nes/nes_hw.c
index 2b89b06..8b606fd 100644
--- a/drivers/infiniband/hw/nes/nes_hw.c
+++ b/drivers/infiniband/hw/nes/nes_hw.c
@@ -2650,6 +2650,13 @@ static void nes_process_mac_intr(struct nes_device *nesdev, u32 mac_number)
 				}
 			}
 		}
+		if (nesadapter->phy_type[mac_index] == NES_PHY_TYPE_SFP_D) {
+			if (nesdev->link_recheck)
+				cancel_delayed_work(&nesdev->work);
+			nesdev->link_recheck = 1;
+			schedule_delayed_work(&nesdev->work,
+					      NES_LINK_RECHECK_DELAY);
+		}
 	}
 
 	spin_unlock_irqrestore(&nesadapter->phy_lock, flags);
@@ -2657,6 +2664,80 @@ static void nes_process_mac_intr(struct nes_device *nesdev, u32 mac_number)
 	nesadapter->mac_sw_state[mac_number] = NES_MAC_SW_IDLE;
 }
 
+void nes_recheck_link_status(struct work_struct *work)
+{
+	unsigned long flags;
+	struct nes_device *nesdev = container_of(work, struct nes_device, work.work);
+	struct nes_adapter *nesadapter = nesdev->nesadapter;
+	struct nes_vnic *nesvnic;
+	u32 mac_index = nesdev->mac_index;
+	u16 phy_data;
+	u16 temp_phy_data;
+
+	spin_lock_irqsave(&nesadapter->phy_lock, flags);
+
+	/* check link status */
+	nes_read_10G_phy_reg(nesdev, nesadapter->phy_index[mac_index], 1, 0x9003);
+	temp_phy_data = (u16)nes_read_indexed(nesdev, NES_IDX_MAC_MDIO_CONTROL);
+
+	nes_read_10G_phy_reg(nesdev, nesadapter->phy_index[mac_index], 3, 0x0021);
+	nes_read_indexed(nesdev, NES_IDX_MAC_MDIO_CONTROL);
+	nes_read_10G_phy_reg(nesdev, nesadapter->phy_index[mac_index], 3, 0x0021);
+	phy_data = (u16)nes_read_indexed(nesdev, NES_IDX_MAC_MDIO_CONTROL);
+
+	phy_data = (!temp_phy_data && (phy_data == 0x8000)) ? 0x4 : 0x0;
+
+	nes_debug(NES_DBG_PHY, "%s: Phy data = 0x%04X, link was %s.\n",
+		__func__, phy_data,
+		nesadapter->mac_link_down[mac_index] ? "DOWN" : "UP");
+
+	if (phy_data & 0x0004) {
+		nesadapter->mac_link_down[mac_index] = 0;
+		list_for_each_entry(nesvnic, &nesadapter->nesvnic_list[mac_index], list) {
+			if (nesvnic->linkup == 0) {
+				printk(PFX "The Link is now up for port %s, netdev %p.\n",
+						nesvnic->netdev->name, nesvnic->netdev);
+				if (netif_queue_stopped(nesvnic->netdev))
+					netif_start_queue(nesvnic->netdev);
+				nesvnic->linkup = 1;
+				netif_carrier_on(nesvnic->netdev);
+
+				spin_lock(&nesvnic->port_ibevent_lock);
+				if (nesdev->iw_status == 0) {
+					nesdev->iw_status = 1;
+					nes_port_ibevent(nesvnic);
+				}
+				spin_unlock(&nesvnic->port_ibevent_lock);
+			}
+		}
+
+	} else {
+		nesadapter->mac_link_down[mac_index] = 1;
+		list_for_each_entry(nesvnic, &nesadapter->nesvnic_list[mac_index], list) {
+			if (nesvnic->linkup == 1) {
+				printk(PFX "The Link is now down for port %s, netdev %p.\n",
+						nesvnic->netdev->name, nesvnic->netdev);
+				if (!(netif_queue_stopped(nesvnic->netdev)))
+					netif_stop_queue(nesvnic->netdev);
+				nesvnic->linkup = 0;
+				netif_carrier_off(nesvnic->netdev);
+
+				spin_lock(&nesvnic->port_ibevent_lock);
+				if (nesdev->iw_status == 1) {
+					nesdev->iw_status = 0;
+					nes_port_ibevent(nesvnic);
+				}
+				spin_unlock(&nesvnic->port_ibevent_lock);
+			}
+		}
+	}
+	if (nesdev->link_recheck++ < NES_LINK_RECHECK_MAX)
+		schedule_delayed_work(&nesdev->work, NES_LINK_RECHECK_DELAY);
+	else
+		nesdev->link_recheck = 0;
+
+	spin_unlock_irqrestore(&nesadapter->phy_lock, flags);
+}
 
 
 static void nes_nic_napi_ce_handler(struct nes_device *nesdev, struct nes_hw_nic_cq *cq)
diff --git a/drivers/infiniband/hw/nes/nes_hw.h b/drivers/infiniband/hw/nes/nes_hw.h
index 8a9ea9a..d2abe07 100644
--- a/drivers/infiniband/hw/nes/nes_hw.h
+++ b/drivers/infiniband/hw/nes/nes_hw.h
@@ -1354,6 +1354,10 @@ struct nes_terminate_hdr {
 #define BAD_FRAME_OFFSET	64
 #define CQE_MAJOR_DRV		0x8000
 
+/* Used for link status recheck after interrupt processing */
+#define NES_LINK_RECHECK_DELAY	msecs_to_jiffies(50)
+#define NES_LINK_RECHECK_MAX	60
+
 #define nes_vlan_rx vlan_hwaccel_receive_skb
 #define nes_netif_rx netif_receive_skb