1 files changed, 1176 insertions, 344 deletions

diff --git a/drivers/net/ixgbe/ixgbe_main.c b/drivers/net/ixgbe/ixgbe_main.c
index c2095ce531c..cb371a8c24a 100644
--- a/drivers/net/ixgbe/ixgbe_main.c
+++ b/drivers/net/ixgbe/ixgbe_main.c
@@ -48,7 +48,7 @@ char ixgbe_driver_name[] = "ixgbe";
 static const char ixgbe_driver_string[] =
 	"Intel(R) 10 Gigabit PCI Express Network Driver";
 
-#define DRV_VERSION "1.1.18"
+#define DRV_VERSION "1.3.18-k2"
 const char ixgbe_driver_version[] = DRV_VERSION;
 static const char ixgbe_copyright[] =
 	 "Copyright (c) 1999-2007 Intel Corporation.";
@@ -80,6 +80,16 @@ static struct pci_device_id ixgbe_pci_tbl[] = {
 };
 MODULE_DEVICE_TABLE(pci, ixgbe_pci_tbl);
 
+#ifdef CONFIG_DCA
+static int ixgbe_notify_dca(struct notifier_block *, unsigned long event,
+			    void *p);
+static struct notifier_block dca_notifier = {
+	.notifier_call = ixgbe_notify_dca,
+	.next          = NULL,
+	.priority      = 0
+};
+#endif
+
 MODULE_AUTHOR("Intel Corporation, <linux.nics@intel.com>");
 MODULE_DESCRIPTION("Intel(R) 10 Gigabit PCI Express Network Driver");
 MODULE_LICENSE("GPL");
@@ -256,26 +266,125 @@ static bool ixgbe_clean_tx_irq(struct ixgbe_adapter *adapter,
 		 * sees the new next_to_clean.
 		 */
 		smp_mb();
+#ifdef CONFIG_NETDEVICES_MULTIQUEUE
+		if (__netif_subqueue_stopped(netdev, tx_ring->queue_index) &&
+		    !test_bit(__IXGBE_DOWN, &adapter->state)) {
+			netif_wake_subqueue(netdev, tx_ring->queue_index);
+			adapter->restart_queue++;
+		}
+#else
 		if (netif_queue_stopped(netdev) &&
 		    !test_bit(__IXGBE_DOWN, &adapter->state)) {
 			netif_wake_queue(netdev);
 			adapter->restart_queue++;
 		}
+#endif
 	}
 
 	if (adapter->detect_tx_hung)
 		if (ixgbe_check_tx_hang(adapter, tx_ring, eop, eop_desc))
+#ifdef CONFIG_NETDEVICES_MULTIQUEUE
+			netif_stop_subqueue(netdev, tx_ring->queue_index);
+#else
 			netif_stop_queue(netdev);
+#endif
 
 	if (total_tx_packets >= tx_ring->work_limit)
 		IXGBE_WRITE_REG(&adapter->hw, IXGBE_EICS, tx_ring->eims_value);
 
+	tx_ring->total_bytes += total_tx_bytes;
+	tx_ring->total_packets += total_tx_packets;
 	adapter->net_stats.tx_bytes += total_tx_bytes;
 	adapter->net_stats.tx_packets += total_tx_packets;
 	cleaned = total_tx_packets ? true : false;
 	return cleaned;
 }
 
+#ifdef CONFIG_DCA
+static void ixgbe_update_rx_dca(struct ixgbe_adapter *adapter,
+				struct ixgbe_ring *rxr)
+{
+	u32 rxctrl;
+	int cpu = get_cpu();
+	int q = rxr - adapter->rx_ring;
+
+	if (rxr->cpu != cpu) {
+		rxctrl = IXGBE_READ_REG(&adapter->hw, IXGBE_DCA_RXCTRL(q));
+		rxctrl &= ~IXGBE_DCA_RXCTRL_CPUID_MASK;
+		rxctrl |= dca_get_tag(cpu);
+		rxctrl |= IXGBE_DCA_RXCTRL_DESC_DCA_EN;
+		rxctrl |= IXGBE_DCA_RXCTRL_HEAD_DCA_EN;
+		IXGBE_WRITE_REG(&adapter->hw, IXGBE_DCA_RXCTRL(q), rxctrl);
+		rxr->cpu = cpu;
+	}
+	put_cpu();
+}
+
+static void ixgbe_update_tx_dca(struct ixgbe_adapter *adapter,
+				struct ixgbe_ring *txr)
+{
+	u32 txctrl;
+	int cpu = get_cpu();
+	int q = txr - adapter->tx_ring;
+
+	if (txr->cpu != cpu) {
+		txctrl = IXGBE_READ_REG(&adapter->hw, IXGBE_DCA_TXCTRL(q));
+		txctrl &= ~IXGBE_DCA_TXCTRL_CPUID_MASK;
+		txctrl |= dca_get_tag(cpu);
+		txctrl |= IXGBE_DCA_TXCTRL_DESC_DCA_EN;
+		IXGBE_WRITE_REG(&adapter->hw, IXGBE_DCA_TXCTRL(q), txctrl);
+		txr->cpu = cpu;
+	}
+	put_cpu();
+}
+
+static void ixgbe_setup_dca(struct ixgbe_adapter *adapter)
+{
+	int i;
+
+	if (!(adapter->flags & IXGBE_FLAG_DCA_ENABLED))
+		return;
+
+	for (i = 0; i < adapter->num_tx_queues; i++) {
+		adapter->tx_ring[i].cpu = -1;
+		ixgbe_update_tx_dca(adapter, &adapter->tx_ring[i]);
+	}
+	for (i = 0; i < adapter->num_rx_queues; i++) {
+		adapter->rx_ring[i].cpu = -1;
+		ixgbe_update_rx_dca(adapter, &adapter->rx_ring[i]);
+	}
+}
+
+static int __ixgbe_notify_dca(struct device *dev, void *data)
+{
+	struct net_device *netdev = dev_get_drvdata(dev);
+	struct ixgbe_adapter *adapter = netdev_priv(netdev);
+	unsigned long event = *(unsigned long *)data;
+
+	switch (event) {
+	case DCA_PROVIDER_ADD:
+		adapter->flags |= IXGBE_FLAG_DCA_ENABLED;
+		/* Always use CB2 mode, difference is masked
+		 * in the CB driver. */
+		IXGBE_WRITE_REG(&adapter->hw, IXGBE_DCA_CTRL, 2);
+		if (dca_add_requester(dev) == 0) {
+			ixgbe_setup_dca(adapter);
+			break;
+		}
+		/* Fall Through since DCA is disabled. */
+	case DCA_PROVIDER_REMOVE:
+		if (adapter->flags & IXGBE_FLAG_DCA_ENABLED) {
+			dca_remove_requester(dev);
+			adapter->flags &= ~IXGBE_FLAG_DCA_ENABLED;
+			IXGBE_WRITE_REG(&adapter->hw, IXGBE_DCA_CTRL, 1);
+		}
+		break;
+	}
+
+	return 0;
+}
+
+#endif /* CONFIG_DCA */
 /**
  * ixgbe_receive_skb - Send a completed packet up the stack
  * @adapter: board private structure
@@ -556,10 +665,15 @@ next_desc:
 	adapter->net_stats.rx_bytes += total_rx_bytes;
 	adapter->net_stats.rx_packets += total_rx_packets;
 
+	rx_ring->total_packets += total_rx_packets;
+	rx_ring->total_bytes += total_rx_bytes;
+	adapter->net_stats.rx_bytes += total_rx_bytes;
+	adapter->net_stats.rx_packets += total_rx_packets;
+
 	return cleaned;
 }
 
-#define IXGBE_MAX_INTR 10
+static int ixgbe_clean_rxonly(struct napi_struct *, int);
 /**
  * ixgbe_configure_msix - Configure MSI-X hardware
  * @adapter: board private structure
@@ -569,28 +683,195 @@ next_desc:
  **/
 static void ixgbe_configure_msix(struct ixgbe_adapter *adapter)
 {
-	int i, vector = 0;
+	struct ixgbe_q_vector *q_vector;
+	int i, j, q_vectors, v_idx, r_idx;
+	u32 mask;
 
-	for (i = 0; i < adapter->num_tx_queues; i++) {
-		ixgbe_set_ivar(adapter, IXGBE_IVAR_TX_QUEUE(i),
-			       IXGBE_MSIX_VECTOR(vector));
-		writel(EITR_INTS_PER_SEC_TO_REG(adapter->tx_eitr),
-		       adapter->hw.hw_addr + adapter->tx_ring[i].itr_register);
-		vector++;
-	}
+	q_vectors = adapter->num_msix_vectors - NON_Q_VECTORS;
 
-	for (i = 0; i < adapter->num_rx_queues; i++) {
-		ixgbe_set_ivar(adapter, IXGBE_IVAR_RX_QUEUE(i),
-			       IXGBE_MSIX_VECTOR(vector));
-		writel(EITR_INTS_PER_SEC_TO_REG(adapter->rx_eitr),
-		       adapter->hw.hw_addr + adapter->rx_ring[i].itr_register);
-		vector++;
+	/* Populate the IVAR table and set the ITR values to the
+	 * corresponding register.
+	 */
+	for (v_idx = 0; v_idx < q_vectors; v_idx++) {
+		q_vector = &adapter->q_vector[v_idx];
+		/* XXX for_each_bit(...) */
+		r_idx = find_first_bit(q_vector->rxr_idx,
+				      adapter->num_rx_queues);
+
+		for (i = 0; i < q_vector->rxr_count; i++) {
+			j = adapter->rx_ring[r_idx].reg_idx;
+			ixgbe_set_ivar(adapter, IXGBE_IVAR_RX_QUEUE(j), v_idx);
+			r_idx = find_next_bit(q_vector->rxr_idx,
+					      adapter->num_rx_queues,
+					      r_idx + 1);
+		}
+		r_idx = find_first_bit(q_vector->txr_idx,
+				       adapter->num_tx_queues);
+
+		for (i = 0; i < q_vector->txr_count; i++) {
+			j = adapter->tx_ring[r_idx].reg_idx;
+			ixgbe_set_ivar(adapter, IXGBE_IVAR_TX_QUEUE(j), v_idx);
+			r_idx = find_next_bit(q_vector->txr_idx,
+					      adapter->num_tx_queues,
+					      r_idx + 1);
+		}
+
+		/* if this is a tx only vector use half the irq (tx) rate */
+		if (q_vector->txr_count && !q_vector->rxr_count)
+			q_vector->eitr = adapter->tx_eitr;
+		else
+			/* rx only or mixed */
+			q_vector->eitr = adapter->rx_eitr;
+
+		IXGBE_WRITE_REG(&adapter->hw, IXGBE_EITR(v_idx),
+				EITR_INTS_PER_SEC_TO_REG(q_vector->eitr));
 	}
 
-	vector = adapter->num_tx_queues + adapter->num_rx_queues;
-	ixgbe_set_ivar(adapter, IXGBE_IVAR_OTHER_CAUSES_INDEX,
-		       IXGBE_MSIX_VECTOR(vector));
-	IXGBE_WRITE_REG(&adapter->hw, IXGBE_EITR(vector), 1950);
+	ixgbe_set_ivar(adapter, IXGBE_IVAR_OTHER_CAUSES_INDEX, v_idx);
+	IXGBE_WRITE_REG(&adapter->hw, IXGBE_EITR(v_idx), 1950);
+
+	/* set up to autoclear timer, lsc, and the vectors */
+	mask = IXGBE_EIMS_ENABLE_MASK;
+	mask &= ~IXGBE_EIMS_OTHER;
+	IXGBE_WRITE_REG(&adapter->hw, IXGBE_EIAC, mask);
+}
+
+enum latency_range {
+	lowest_latency = 0,
+	low_latency = 1,
+	bulk_latency = 2,
+	latency_invalid = 255
+};
+
+/**
+ * ixgbe_update_itr - update the dynamic ITR value based on statistics
+ * @adapter: pointer to adapter
+ * @eitr: eitr setting (ints per sec) to give last timeslice
+ * @itr_setting: current throttle rate in ints/second
+ * @packets: the number of packets during this measurement interval
+ * @bytes: the number of bytes during this measurement interval
+ *
+ *      Stores a new ITR value based on packets and byte
+ *      counts during the last interrupt.  The advantage of per interrupt
+ *      computation is faster updates and more accurate ITR for the current
+ *      traffic pattern.  Constants in this function were computed
+ *      based on theoretical maximum wire speed and thresholds were set based
+ *      on testing data as well as attempting to minimize response time
+ *      while increasing bulk throughput.
+ *      this functionality is controlled by the InterruptThrottleRate module
+ *      parameter (see ixgbe_param.c)
+ **/
+static u8 ixgbe_update_itr(struct ixgbe_adapter *adapter,
+			   u32 eitr, u8 itr_setting,
+			   int packets, int bytes)
+{
+	unsigned int retval = itr_setting;
+	u32 timepassed_us;
+	u64 bytes_perint;
+
+	if (packets == 0)
+		goto update_itr_done;
+
+
+	/* simple throttlerate management
+	 *    0-20MB/s lowest (100000 ints/s)
+	 *   20-100MB/s low   (20000 ints/s)
+	 *  100-1249MB/s bulk (8000 ints/s)
+	 */
+	/* what was last interrupt timeslice? */
+	timepassed_us = 1000000/eitr;
+	bytes_perint = bytes / timepassed_us; /* bytes/usec */
+
+	switch (itr_setting) {
+	case lowest_latency:
+		if (bytes_perint > adapter->eitr_low)
+			retval = low_latency;
+		break;
+	case low_latency:
+		if (bytes_perint > adapter->eitr_high)
+			retval = bulk_latency;
+		else if (bytes_perint <= adapter->eitr_low)
+			retval = lowest_latency;
+		break;
+	case bulk_latency:
+		if (bytes_perint <= adapter->eitr_high)
+			retval = low_latency;
+		break;
+	}
+
+update_itr_done:
+	return retval;
+}
+
+static void ixgbe_set_itr_msix(struct ixgbe_q_vector *q_vector)
+{
+	struct ixgbe_adapter *adapter = q_vector->adapter;
+	struct ixgbe_hw *hw = &adapter->hw;
+	u32 new_itr;
+	u8 current_itr, ret_itr;
+	int i, r_idx, v_idx = ((void *)q_vector - (void *)(adapter->q_vector)) /
+			      sizeof(struct ixgbe_q_vector);
+	struct ixgbe_ring *rx_ring, *tx_ring;
+
+	r_idx = find_first_bit(q_vector->txr_idx, adapter->num_tx_queues);
+	for (i = 0; i < q_vector->txr_count; i++) {
+		tx_ring = &(adapter->tx_ring[r_idx]);
+		ret_itr = ixgbe_update_itr(adapter, q_vector->eitr,
+					   q_vector->tx_eitr,
+					   tx_ring->total_packets,
+					   tx_ring->total_bytes);
+		/* if the result for this queue would decrease interrupt
+		 * rate for this vector then use that result */
+		q_vector->tx_eitr = ((q_vector->tx_eitr > ret_itr) ?
+				    q_vector->tx_eitr - 1 : ret_itr);
+		r_idx = find_next_bit(q_vector->txr_idx, adapter->num_tx_queues,
+				      r_idx + 1);
+	}
+
+	r_idx = find_first_bit(q_vector->rxr_idx, adapter->num_rx_queues);
+	for (i = 0; i < q_vector->rxr_count; i++) {
+		rx_ring = &(adapter->rx_ring[r_idx]);
+		ret_itr = ixgbe_update_itr(adapter, q_vector->eitr,
+					   q_vector->rx_eitr,
+					   rx_ring->total_packets,
+					   rx_ring->total_bytes);
+		/* if the result for this queue would decrease interrupt
+		 * rate for this vector then use that result */
+		q_vector->rx_eitr = ((q_vector->rx_eitr > ret_itr) ?
+				    q_vector->rx_eitr - 1 : ret_itr);
+		r_idx = find_next_bit(q_vector->rxr_idx, adapter->num_rx_queues,
+				      r_idx + 1);
+	}
+
+	current_itr = max(q_vector->rx_eitr, q_vector->tx_eitr);
+
+	switch (current_itr) {
+	/* counts and packets in update_itr are dependent on these numbers */
+	case lowest_latency:
+		new_itr = 100000;
+		break;
+	case low_latency:
+		new_itr = 20000; /* aka hwitr = ~200 */
+		break;
+	case bulk_latency:
+	default:
+		new_itr = 8000;
+		break;
+	}
+
+	if (new_itr != q_vector->eitr) {
+		u32 itr_reg;
+		/* do an exponential smoothing */
+		new_itr = ((q_vector->eitr * 90)/100) + ((new_itr * 10)/100);
+		q_vector->eitr = new_itr;
+		itr_reg = EITR_INTS_PER_SEC_TO_REG(new_itr);
+		/* must write high and low 16 bits to reset counter */
+		DPRINTK(TX_ERR, DEBUG, "writing eitr(%d): %08X\n", v_idx,
+			itr_reg);
+		IXGBE_WRITE_REG(hw, IXGBE_EITR(v_idx), itr_reg | (itr_reg)<<16);
+	}
+
+	return;
 }
 
 static irqreturn_t ixgbe_msix_lsc(int irq, void *data)
@@ -614,153 +895,302 @@ static irqreturn_t ixgbe_msix_lsc(int irq, void *data)
 
 static irqreturn_t ixgbe_msix_clean_tx(int irq, void *data)
 {
-	struct ixgbe_ring *txr = data;
-	struct ixgbe_adapter *adapter = txr->adapter;
+	struct ixgbe_q_vector *q_vector = data;
+	struct ixgbe_adapter  *adapter = q_vector->adapter;
+	struct ixgbe_ring     *txr;
+	int i, r_idx;
 
-	ixgbe_clean_tx_irq(adapter, txr);
+	if (!q_vector->txr_count)
+		return IRQ_HANDLED;
+
+	r_idx = find_first_bit(q_vector->txr_idx, adapter->num_tx_queues);
+	for (i = 0; i < q_vector->txr_count; i++) {
+		txr = &(adapter->tx_ring[r_idx]);
+#ifdef CONFIG_DCA
+		if (adapter->flags & IXGBE_FLAG_DCA_ENABLED)
+			ixgbe_update_tx_dca(adapter, txr);
+#endif
+		txr->total_bytes = 0;
+		txr->total_packets = 0;
+		ixgbe_clean_tx_irq(adapter, txr);
+		r_idx = find_next_bit(q_vector->txr_idx, adapter->num_tx_queues,
+				      r_idx + 1);
+	}
 
 	return IRQ_HANDLED;
 }
 
+/**
+ * ixgbe_msix_clean_rx - single unshared vector rx clean (all queues)
+ * @irq: unused
+ * @data: pointer to our q_vector struct for this interrupt vector
+ **/
 static irqreturn_t ixgbe_msix_clean_rx(int irq, void *data)
 {
-	struct ixgbe_ring *rxr = data;
-	struct ixgbe_adapter *adapter = rxr->adapter;
+	struct ixgbe_q_vector *q_vector = data;
+	struct ixgbe_adapter  *adapter = q_vector->adapter;
+	struct ixgbe_ring  *rxr;
+	int r_idx;
+
+	r_idx = find_first_bit(q_vector->rxr_idx, adapter->num_rx_queues);
+	if (!q_vector->rxr_count)
+		return IRQ_HANDLED;
+
+	rxr = &(adapter->rx_ring[r_idx]);
+	/* disable interrupts on this vector only */
+	IXGBE_WRITE_REG(&adapter->hw, IXGBE_EIMC, rxr->v_idx);
+	rxr->total_bytes = 0;
+	rxr->total_packets = 0;
+	netif_rx_schedule(adapter->netdev, &q_vector->napi);
+
+	return IRQ_HANDLED;
+}
+
+static irqreturn_t ixgbe_msix_clean_many(int irq, void *data)
+{
+	ixgbe_msix_clean_rx(irq, data);
+	ixgbe_msix_clean_tx(irq, data);
 
-	IXGBE_WRITE_REG(&adapter->hw, IXGBE_EIMC, rxr->eims_value);
-	netif_rx_schedule(adapter->netdev, &adapter->napi);
 	return IRQ_HANDLED;
 }
 
+/**
+ * ixgbe_clean_rxonly - msix (aka one shot) rx clean routine
+ * @napi: napi struct with our devices info in it
+ * @budget: amount of work driver is allowed to do this pass, in packets
+ *
+ **/
 static int ixgbe_clean_rxonly(struct napi_struct *napi, int budget)
 {
-	struct ixgbe_adapter *adapter = container_of(napi,
-					struct ixgbe_adapter, napi);
-	struct net_device *netdev = adapter->netdev;
+	struct ixgbe_q_vector *q_vector =
+			       container_of(napi, struct ixgbe_q_vector, napi);
+	struct ixgbe_adapter *adapter = q_vector->adapter;
+	struct ixgbe_ring *rxr;
 	int work_done = 0;
-	struct ixgbe_ring *rxr = adapter->rx_ring;
+	long r_idx;
 
-	/* Keep link state information with original netdev */
-	if (!netif_carrier_ok(netdev))
-		goto quit_polling;
+	r_idx = find_first_bit(q_vector->rxr_idx, adapter->num_rx_queues);
+	rxr = &(adapter->rx_ring[r_idx]);
+#ifdef CONFIG_DCA
+	if (adapter->flags & IXGBE_FLAG_DCA_ENABLED)
+		ixgbe_update_rx_dca(adapter, rxr);
+#endif
 
 	ixgbe_clean_rx_irq(adapter, rxr, &work_done, budget);
 
-	/* If no Tx and not enough Rx work done, exit the polling mode */
-	if ((work_done < budget) || !netif_running(netdev)) {
-quit_polling:
-		netif_rx_complete(netdev, napi);
+	/* If all Rx work done, exit the polling mode */
+	if (work_done < budget) {
+		netif_rx_complete(adapter->netdev, napi);
+		if (adapter->rx_eitr < IXGBE_MIN_ITR_USECS)
+			ixgbe_set_itr_msix(q_vector);
 		if (!test_bit(__IXGBE_DOWN, &adapter->state))
-			IXGBE_WRITE_REG(&adapter->hw, IXGBE_EIMS,
-					rxr->eims_value);
+			IXGBE_WRITE_REG(&adapter->hw, IXGBE_EIMS, rxr->v_idx);
 	}
 
 	return work_done;
 }
 
+static inline void map_vector_to_rxq(struct ixgbe_adapter *a, int v_idx,
+				     int r_idx)
+{
+	a->q_vector[v_idx].adapter = a;
+	set_bit(r_idx, a->q_vector[v_idx].rxr_idx);
+	a->q_vector[v_idx].rxr_count++;
+	a->rx_ring[r_idx].v_idx = 1 << v_idx;
+}
+
+static inline void map_vector_to_txq(struct ixgbe_adapter *a, int v_idx,
+				     int r_idx)
+{
+	a->q_vector[v_idx].adapter = a;
+	set_bit(r_idx, a->q_vector[v_idx].txr_idx);
+	a->q_vector[v_idx].txr_count++;
+	a->tx_ring[r_idx].v_idx = 1 << v_idx;
+}
+
 /**
- * ixgbe_setup_msix - Initialize MSI-X interrupts
+ * ixgbe_map_rings_to_vectors - Maps descriptor rings to vectors
+ * @adapter: board private structure to initialize
+ * @vectors: allotted vector count for descriptor rings
  *
- * ixgbe_setup_msix allocates MSI-X vectors and requests
- * interrutps from the kernel.
+ * This function maps descriptor rings to the queue-specific vectors
+ * we were allotted through the MSI-X enabling code.  Ideally, we'd have
+ * one vector per ring/queue, but on a constrained vector budget, we
+ * group the rings as "efficiently" as possible.  You would add new
+ * mapping configurations in here.
  **/
-static int ixgbe_setup_msix(struct ixgbe_adapter *adapter)
-{
-	struct net_device *netdev = adapter->netdev;
-	int i, int_vector = 0, err = 0;
-	int max_msix_count;
+static int ixgbe_map_rings_to_vectors(struct ixgbe_adapter *adapter,
+				      int vectors)
+{
+	int v_start = 0;
+	int rxr_idx = 0, txr_idx = 0;
+	int rxr_remaining = adapter->num_rx_queues;
+	int txr_remaining = adapter->num_tx_queues;
+	int i, j;
+	int rqpv, tqpv;
+	int err = 0;
+
+	/* No mapping required if MSI-X is disabled. */
+	if (!(adapter->flags & IXGBE_FLAG_MSIX_ENABLED))
+		goto out;
 
-	/* +1 for the LSC interrupt */
-	max_msix_count = adapter->num_rx_queues + adapter->num_tx_queues + 1;
-	adapter->msix_entries = kcalloc(max_msix_count,
-					sizeof(struct msix_entry), GFP_KERNEL);
-	if (!adapter->msix_entries)
-		return -ENOMEM;
+	/*
+	 * The ideal configuration...
+	 * We have enough vectors to map one per queue.
+	 */
+	if (vectors == adapter->num_rx_queues + adapter->num_tx_queues) {
+		for (; rxr_idx < rxr_remaining; v_start++, rxr_idx++)
+			map_vector_to_rxq(adapter, v_start, rxr_idx);
 
-	for (i = 0; i < max_msix_count; i++)
-		adapter->msix_entries[i].entry = i;
+		for (; txr_idx < txr_remaining; v_start++, txr_idx++)
+			map_vector_to_txq(adapter, v_start, txr_idx);
 
-	err = pci_enable_msix(adapter->pdev, adapter->msix_entries,
-			      max_msix_count);
-	if (err)
 		goto out;
+	}
 
-	for (i = 0; i < adapter->num_tx_queues; i++) {
-		sprintf(adapter->tx_ring[i].name, "%s-tx%d", netdev->name, i);
-		err = request_irq(adapter->msix_entries[int_vector].vector,
-				  &ixgbe_msix_clean_tx,
-				  0,
-				  adapter->tx_ring[i].name,
-				  &(adapter->tx_ring[i]));
-		if (err) {
-			DPRINTK(PROBE, ERR,
-				"request_irq failed for MSIX interrupt "
-				"Error: %d\n", err);
-			goto release_irqs;
+	/*
+	 * If we don't have enough vectors for a 1-to-1
+	 * mapping, we'll have to group them so there are
+	 * multiple queues per vector.
+	 */
+	/* Re-adjusting *qpv takes care of the remainder. */
+	for (i = v_start; i < vectors; i++) {
+		rqpv = DIV_ROUND_UP(rxr_remaining, vectors - i);
+		for (j = 0; j < rqpv; j++) {
+			map_vector_to_rxq(adapter, i, rxr_idx);
+			rxr_idx++;
+			rxr_remaining--;
+		}
+	}
+	for (i = v_start; i < vectors; i++) {
+		tqpv = DIV_ROUND_UP(txr_remaining, vectors - i);
+		for (j = 0; j < tqpv; j++) {
+			map_vector_to_txq(adapter, i, txr_idx);
+			txr_idx++;
+			txr_remaining--;
 		}
-		adapter->tx_ring[i].eims_value =
-		    (1 << IXGBE_MSIX_VECTOR(int_vector));
-		adapter->tx_ring[i].itr_register = IXGBE_EITR(int_vector);
-		int_vector++;
 	}
 
-	for (i = 0; i < adapter->num_rx_queues; i++) {
-		if (strlen(netdev->name) < (IFNAMSIZ - 5))
-			sprintf(adapter->rx_ring[i].name,
-				"%s-rx%d", netdev->name, i);
-		else
-			memcpy(adapter->rx_ring[i].name,
-			       netdev->name, IFNAMSIZ);
-		err = request_irq(adapter->msix_entries[int_vector].vector,
-				  &ixgbe_msix_clean_rx, 0,
-				  adapter->rx_ring[i].name,
-				  &(adapter->rx_ring[i]));
+out:
+	return err;
+}
+
+/**
+ * ixgbe_request_msix_irqs - Initialize MSI-X interrupts
+ * @adapter: board private structure
+ *
+ * ixgbe_request_msix_irqs allocates MSI-X vectors and requests
+ * interrupts from the kernel.
+ **/
+static int ixgbe_request_msix_irqs(struct ixgbe_adapter *adapter)
+{
+	struct net_device *netdev = adapter->netdev;
+	irqreturn_t (*handler)(int, void *);
+	int i, vector, q_vectors, err;
+
+	/* Decrement for Other and TCP Timer vectors */
+	q_vectors = adapter->num_msix_vectors - NON_Q_VECTORS;
+
+	/* Map the Tx/Rx rings to the vectors we were allotted. */
+	err = ixgbe_map_rings_to_vectors(adapter, q_vectors);
+	if (err)
+		goto out;
+
+#define SET_HANDLER(_v) ((!(_v)->rxr_count) ? &ixgbe_msix_clean_tx : \
+			 (!(_v)->txr_count) ? &ixgbe_msix_clean_rx : \
+			 &ixgbe_msix_clean_many)
+	for (vector = 0; vector < q_vectors; vector++) {
+		handler = SET_HANDLER(&adapter->q_vector[vector]);
+		sprintf(adapter->name[vector], "%s:v%d-%s",
+			netdev->name, vector,
+			(handler == &ixgbe_msix_clean_rx) ? "Rx" :
+			 ((handler == &ixgbe_msix_clean_tx) ? "Tx" : "TxRx"));
+		err = request_irq(adapter->msix_entries[vector].vector,
+				  handler, 0, adapter->name[vector],
+				  &(adapter->q_vector[vector]));
 		if (err) {
 			DPRINTK(PROBE, ERR,
 				"request_irq failed for MSIX interrupt "
 				"Error: %d\n", err);
-			goto release_irqs;
+			goto free_queue_irqs;
 		}
-
-		adapter->rx_ring[i].eims_value =
-		    (1 << IXGBE_MSIX_VECTOR(int_vector));
-		adapter->rx_ring[i].itr_register = IXGBE_EITR(int_vector);
-		int_vector++;
 	}
 
-	sprintf(adapter->lsc_name, "%s-lsc", netdev->name);
-	err = request_irq(adapter->msix_entries[int_vector].vector,
-			  &ixgbe_msix_lsc, 0, adapter->lsc_name, netdev);
+	sprintf(adapter->name[vector], "%s:lsc", netdev->name);
+	err = request_irq(adapter->msix_entries[vector].vector,
+			  &ixgbe_msix_lsc, 0, adapter->name[vector], netdev);
 	if (err) {
 		DPRINTK(PROBE, ERR,
 			"request_irq for msix_lsc failed: %d\n", err);
-		goto release_irqs;
+		goto free_queue_irqs;
 	}
 
-	/* FIXME: implement netif_napi_remove() instead */
-	adapter->napi.poll = ixgbe_clean_rxonly;
-	adapter->flags |= IXGBE_FLAG_MSIX_ENABLED;
 	return 0;
 
-release_irqs:
-	int_vector--;
-	for (; int_vector >= adapter->num_tx_queues; int_vector--)
-		free_irq(adapter->msix_entries[int_vector].vector,
-			 &(adapter->rx_ring[int_vector -
-					    adapter->num_tx_queues]));
-
-	for (; int_vector >= 0; int_vector--)
-		free_irq(adapter->msix_entries[int_vector].vector,
-			 &(adapter->tx_ring[int_vector]));
-out:
+free_queue_irqs:
+	for (i = vector - 1; i >= 0; i--)
+		free_irq(adapter->msix_entries[--vector].vector,
+			 &(adapter->q_vector[i]));
+	adapter->flags &= ~IXGBE_FLAG_MSIX_ENABLED;
+	pci_disable_msix(adapter->pdev);
 	kfree(adapter->msix_entries);
 	adapter->msix_entries = NULL;
-	adapter->flags &= ~IXGBE_FLAG_MSIX_ENABLED;
+out:
 	return err;
 }
 
+static void ixgbe_set_itr(struct ixgbe_adapter *adapter)
+{
+	struct ixgbe_hw *hw = &adapter->hw;
+	struct ixgbe_q_vector *q_vector = adapter->q_vector;
+	u8 current_itr;
+	u32 new_itr = q_vector->eitr;
+	struct ixgbe_ring *rx_ring = &adapter->rx_ring[0];
+	struct ixgbe_ring *tx_ring = &adapter->tx_ring[0];
+
+	q_vector->tx_eitr = ixgbe_update_itr(adapter, new_itr,
+					     q_vector->tx_eitr,
+					     tx_ring->total_packets,
+					     tx_ring->total_bytes);
+	q_vector->rx_eitr = ixgbe_update_itr(adapter, new_itr,
+					     q_vector->rx_eitr,
+					     rx_ring->total_packets,
+					     rx_ring->total_bytes);
+
+	current_itr = max(q_vector->rx_eitr, q_vector->tx_eitr);
+
+	switch (current_itr) {
+	/* counts and packets in update_itr are dependent on these numbers */
+	case lowest_latency:
+		new_itr = 100000;
+		break;
+	case low_latency:
+		new_itr = 20000; /* aka hwitr = ~200 */
+		break;
+	case bulk_latency:
+		new_itr = 8000;
+		break;
+	default:
+		break;
+	}
+
+	if (new_itr != q_vector->eitr) {
+		u32 itr_reg;
+		/* do an exponential smoothing */
+		new_itr = ((q_vector->eitr * 90)/100) + ((new_itr * 10)/100);
+		q_vector->eitr = new_itr;
+		itr_reg = EITR_INTS_PER_SEC_TO_REG(new_itr);
+		/* must write high and low 16 bits to reset counter */
+		IXGBE_WRITE_REG(hw, IXGBE_EITR(0), itr_reg | (itr_reg)<<16);
+	}
+
+	return;
+}
+
+static inline void ixgbe_irq_enable(struct ixgbe_adapter *adapter);
+
 /**
- * ixgbe_intr - Interrupt Handler
+ * ixgbe_intr - legacy mode Interrupt Handler
  * @irq: interrupt number
  * @data: pointer to a network interface device structure
  * @pt_regs: CPU registers structure
@@ -772,8 +1202,10 @@ static irqreturn_t ixgbe_intr(int irq, void *data)
 	struct ixgbe_hw *hw = &adapter->hw;
 	u32 eicr;
 
-	eicr = IXGBE_READ_REG(hw, IXGBE_EICR);
 
+	/* for NAPI, using EIAM to auto-mask tx/rx interrupt bits on read
+	 * therefore no explict interrupt disable is necessary */
+	eicr = IXGBE_READ_REG(hw, IXGBE_EICR);
 	if (!eicr)
 		return IRQ_NONE;	/* Not our interrupt */
 
@@ -782,16 +1214,33 @@ static irqreturn_t ixgbe_intr(int irq, void *data)
 		if (!test_bit(__IXGBE_DOWN, &adapter->state))
 			mod_timer(&adapter->watchdog_timer, jiffies);
 	}
-	if (netif_rx_schedule_prep(netdev, &adapter->napi)) {
-		/* Disable interrupts and register for poll. The flush of the
-		 * posted write is intentionally left out. */
-		IXGBE_WRITE_REG(&adapter->hw, IXGBE_EIMC, ~0);
-		__netif_rx_schedule(netdev, &adapter->napi);
+
+
+	if (netif_rx_schedule_prep(netdev, &adapter->q_vector[0].napi)) {
+		adapter->tx_ring[0].total_packets = 0;
+		adapter->tx_ring[0].total_bytes = 0;
+		adapter->rx_ring[0].total_packets = 0;
+		adapter->rx_ring[0].total_bytes = 0;
+		/* would disable interrupts here but EIAM disabled it */
+		__netif_rx_schedule(netdev, &adapter->q_vector[0].napi);
 	}
 
 	return IRQ_HANDLED;
 }
 
+static inline void ixgbe_reset_q_vectors(struct ixgbe_adapter *adapter)
+{
+	int i, q_vectors = adapter->num_msix_vectors - NON_Q_VECTORS;
+
+	for (i = 0; i < q_vectors; i++) {
+		struct ixgbe_q_vector *q_vector = &adapter->q_vector[i];
+		bitmap_zero(q_vector->rxr_idx, MAX_RX_QUEUES);
+		bitmap_zero(q_vector->txr_idx, MAX_TX_QUEUES);
+		q_vector->rxr_count = 0;
+		q_vector->txr_count = 0;
+	}
+}
+
 /**
  * ixgbe_request_irq - initialize interrupts
  * @adapter: board private structure
@@ -799,40 +1248,24 @@ static irqreturn_t ixgbe_intr(int irq, void *data)
  * Attempts to configure interrupts using the best available
  * capabilities of the hardware and kernel.
  **/
-static int ixgbe_request_irq(struct ixgbe_adapter *adapter, u32 *num_rx_queues)
+static int ixgbe_request_irq(struct ixgbe_adapter *adapter)
 {
 	struct net_device *netdev = adapter->netdev;
-	int flags, err;
-	irq_handler_t handler = ixgbe_intr;
-
-	flags = IRQF_SHARED;
-
-	err = ixgbe_setup_msix(adapter);
-	if (!err)
-		goto request_done;
-
-	/*
-	 * if we can't do MSI-X, fall through and try MSI
-	 * No need to reallocate memory since we're decreasing the number of
-	 * queues. We just won't use the other ones, also it is freed correctly
-	 * on ixgbe_remove.
-	 */
-	*num_rx_queues = 1;
+	int err;
 
-	/* do MSI */
-	err = pci_enable_msi(adapter->pdev);
-	if (!err) {
-		adapter->flags |= IXGBE_FLAG_MSI_ENABLED;
-		flags &= ~IRQF_SHARED;
-		handler = &ixgbe_intr;
+	if (adapter->flags & IXGBE_FLAG_MSIX_ENABLED) {
+		err = ixgbe_request_msix_irqs(adapter);
+	} else if (adapter->flags & IXGBE_FLAG_MSI_ENABLED) {
+		err = request_irq(adapter->pdev->irq, &ixgbe_intr, 0,
+				  netdev->name, netdev);
+	} else {
+		err = request_irq(adapter->pdev->irq, &ixgbe_intr, IRQF_SHARED,
+				  netdev->name, netdev);
 	}
 
-	err = request_irq(adapter->pdev->irq, handler, flags,
-			  netdev->name, netdev);
 	if (err)
 		DPRINTK(PROBE, ERR, "request_irq failed, Error %d\n", err);
 
-request_done:
 	return err;
 }
 
@@ -841,28 +1274,22 @@ static void ixgbe_free_irq(struct ixgbe_adapter *adapter)
 	struct net_device *netdev = adapter->netdev;
 
 	if (adapter->flags & IXGBE_FLAG_MSIX_ENABLED) {
-		int i;
+		int i, q_vectors;
 
-		for (i = 0; i < adapter->num_tx_queues; i++)
-			free_irq(adapter->msix_entries[i].vector,
-				 &(adapter->tx_ring[i]));
-		for (i = 0; i < adapter->num_rx_queues; i++)
-			free_irq(adapter->msix_entries[i +
-						adapter->num_tx_queues].vector,
-				&(adapter->rx_ring[i]));
-		i = adapter->num_rx_queues + adapter->num_tx_queues;
+		q_vectors = adapter->num_msix_vectors;
+
+		i = q_vectors - 1;
 		free_irq(adapter->msix_entries[i].vector, netdev);
-		pci_disable_msix(adapter->pdev);
-		kfree(adapter->msix_entries);
-		adapter->msix_entries = NULL;
-		adapter->flags &= ~IXGBE_FLAG_MSIX_ENABLED;
-		return;
-	}
 
-	free_irq(adapter->pdev->irq, netdev);
-	if (adapter->flags & IXGBE_FLAG_MSI_ENABLED) {
-		pci_disable_msi(adapter->pdev);
-		adapter->flags &= ~IXGBE_FLAG_MSI_ENABLED;
+		i--;
+		for (; i >= 0; i--) {
+			free_irq(adapter->msix_entries[i].vector,
+				 &(adapter->q_vector[i]));
+		}
+
+		ixgbe_reset_q_vectors(adapter);
+	} else {
+		free_irq(adapter->pdev->irq, netdev);
 	}
 }
 
@@ -874,7 +1301,13 @@ static inline void ixgbe_irq_disable(struct ixgbe_adapter *adapter)
 {
 	IXGBE_WRITE_REG(&adapter->hw, IXGBE_EIMC, ~0);
 	IXGBE_WRITE_FLUSH(&adapter->hw);
-	synchronize_irq(adapter->pdev->irq);
+	if (adapter->flags & IXGBE_FLAG_MSIX_ENABLED) {
+		int i;
+		for (i = 0; i < adapter->num_msix_vectors; i++)
+			synchronize_irq(adapter->msix_entries[i].vector);
+	} else {
+		synchronize_irq(adapter->pdev->irq);
+	}
 }
 
 /**
@@ -883,12 +1316,9 @@ static inline void ixgbe_irq_disable(struct ixgbe_adapter *adapter)
  **/
 static inline void ixgbe_irq_enable(struct ixgbe_adapter *adapter)
 {
-	if (adapter->flags & IXGBE_FLAG_MSIX_ENABLED)
-		IXGBE_WRITE_REG(&adapter->hw, IXGBE_EIAC,
-				(IXGBE_EIMS_ENABLE_MASK &
-				 ~(IXGBE_EIMS_OTHER | IXGBE_EIMS_LSC)));
-	IXGBE_WRITE_REG(&adapter->hw, IXGBE_EIMS,
-			IXGBE_EIMS_ENABLE_MASK);
+	u32 mask;
+	mask = IXGBE_EIMS_ENABLE_MASK;
+	IXGBE_WRITE_REG(&adapter->hw, IXGBE_EIMS, mask);
 	IXGBE_WRITE_FLUSH(&adapter->hw);
 }
 
@@ -898,20 +1328,18 @@ static inline void ixgbe_irq_enable(struct ixgbe_adapter *adapter)
  **/
 static void ixgbe_configure_msi_and_legacy(struct ixgbe_adapter *adapter)
 {
-	int i;
 	struct ixgbe_hw *hw = &adapter->hw;
 
-	if (adapter->rx_eitr)
-		IXGBE_WRITE_REG(hw, IXGBE_EITR(0),
-				EITR_INTS_PER_SEC_TO_REG(adapter->rx_eitr));
-
-	/* for re-triggering the interrupt in non-NAPI mode */
-	adapter->rx_ring[0].eims_value = (1 << IXGBE_MSIX_VECTOR(0));
-	adapter->tx_ring[0].eims_value = (1 << IXGBE_MSIX_VECTOR(0));
+	IXGBE_WRITE_REG(hw, IXGBE_EITR(0),
+			EITR_INTS_PER_SEC_TO_REG(adapter->rx_eitr));
 
 	ixgbe_set_ivar(adapter, IXGBE_IVAR_RX_QUEUE(0), 0);
-	for (i = 0; i < adapter->num_tx_queues; i++)
-		ixgbe_set_ivar(adapter, IXGBE_IVAR_TX_QUEUE(i), i);
+	ixgbe_set_ivar(adapter, IXGBE_IVAR_TX_QUEUE(0), 0);
+
+	map_vector_to_rxq(adapter, 0, 0);
+	map_vector_to_txq(adapter, 0, 0);
+
+	DPRINTK(HW, INFO, "Legacy interrupt IVAR setup done\n");
 }
 
 /**
@@ -924,23 +1352,29 @@ static void ixgbe_configure_tx(struct ixgbe_adapter *adapter)
 {
 	u64 tdba;
 	struct ixgbe_hw *hw = &adapter->hw;
-	u32 i, tdlen;
+	u32 i, j, tdlen, txctrl;
 
 	/* Setup the HW Tx Head and Tail descriptor pointers */
 	for (i = 0; i < adapter->num_tx_queues; i++) {
+		j = adapter->tx_ring[i].reg_idx;
 		tdba = adapter->tx_ring[i].dma;
 		tdlen = adapter->tx_ring[i].count *
-		    sizeof(union ixgbe_adv_tx_desc);
-		IXGBE_WRITE_REG(hw, IXGBE_TDBAL(i), (tdba & DMA_32BIT_MASK));
-		IXGBE_WRITE_REG(hw, IXGBE_TDBAH(i), (tdba >> 32));
-		IXGBE_WRITE_REG(hw, IXGBE_TDLEN(i), tdlen);
-		IXGBE_WRITE_REG(hw, IXGBE_TDH(i), 0);
-		IXGBE_WRITE_REG(hw, IXGBE_TDT(i), 0);
-		adapter->tx_ring[i].head = IXGBE_TDH(i);
-		adapter->tx_ring[i].tail = IXGBE_TDT(i);
+			sizeof(union ixgbe_adv_tx_desc);
+		IXGBE_WRITE_REG(hw, IXGBE_TDBAL(j),
+				(tdba & DMA_32BIT_MASK));
+		IXGBE_WRITE_REG(hw, IXGBE_TDBAH(j), (tdba >> 32));
+		IXGBE_WRITE_REG(hw, IXGBE_TDLEN(j), tdlen);
+		IXGBE_WRITE_REG(hw, IXGBE_TDH(j), 0);
+		IXGBE_WRITE_REG(hw, IXGBE_TDT(j), 0);
+		adapter->tx_ring[i].head = IXGBE_TDH(j);
+		adapter->tx_ring[i].tail = IXGBE_TDT(j);
+		/* Disable Tx Head Writeback RO bit, since this hoses
+		 * bookkeeping if things aren't delivered in order.
+		 */
+		txctrl = IXGBE_READ_REG(hw, IXGBE_DCA_TXCTRL(i));
+		txctrl &= ~IXGBE_DCA_TXCTRL_TX_WB_RO_EN;
+		IXGBE_WRITE_REG(hw, IXGBE_DCA_TXCTRL(i), txctrl);
 	}
-
-	IXGBE_WRITE_REG(hw, IXGBE_TIPG, IXGBE_TIPG_FIBER_DEFAULT);
 }
 
 #define PAGE_USE_COUNT(S) (((S) >> PAGE_SHIFT) + \
@@ -959,13 +1393,12 @@ static void ixgbe_configure_rx(struct ixgbe_adapter *adapter)
 	struct ixgbe_hw *hw = &adapter->hw;
 	struct net_device *netdev = adapter->netdev;
 	int max_frame = netdev->mtu + ETH_HLEN + ETH_FCS_LEN;
+	int i, j;
 	u32 rdlen, rxctrl, rxcsum;
 	u32 random[10];
-	u32 reta, mrqc;
-	int i;
 	u32 fctrl, hlreg0;
-	u32 srrctl;
 	u32 pages;
+	u32 reta = 0, mrqc, srrctl;
 
 	/* Decide whether to use packet split mode or not */
 	if (netdev->mtu > ETH_DATA_LEN)
@@ -985,6 +1418,7 @@ static void ixgbe_configure_rx(struct ixgbe_adapter *adapter)
 
 	fctrl = IXGBE_READ_REG(&adapter->hw, IXGBE_FCTRL);
 	fctrl |= IXGBE_FCTRL_BAM;
+	fctrl |= IXGBE_FCTRL_DPF; /* discard pause frames when FC enabled */
 	IXGBE_WRITE_REG(&adapter->hw, IXGBE_FCTRL, fctrl);
 
 	hlreg0 = IXGBE_READ_REG(hw, IXGBE_HLREG0);
@@ -1036,37 +1470,23 @@ static void ixgbe_configure_rx(struct ixgbe_adapter *adapter)
 		adapter->rx_ring[i].tail = IXGBE_RDT(i);
 	}
 
-	if (adapter->num_rx_queues > 1) {
-		/* Random 40bytes used as random key in RSS hash function */
-		get_random_bytes(&random[0], 40);
-
-		switch (adapter->num_rx_queues) {
-		case 8:
-		case 4:
-			/* Bits [3:0] in each byte refers the Rx queue no */
-			reta = 0x00010203;
-			break;
-		case 2:
-			reta = 0x00010001;
-			break;
-		default:
-			reta = 0x00000000;
-			break;
-		}
-
+	if (adapter->flags & IXGBE_FLAG_RSS_ENABLED) {
 		/* Fill out redirection table */
-		for (i = 0; i < 32; i++) {
-			IXGBE_WRITE_REG_ARRAY(hw, IXGBE_RETA(0), i, reta);
-			if (adapter->num_rx_queues > 4) {
-				i++;
-				IXGBE_WRITE_REG_ARRAY(hw, IXGBE_RETA(0), i,
-						      0x04050607);
-			}
+		for (i = 0, j = 0; i < 128; i++, j++) {
+			if (j == adapter->ring_feature[RING_F_RSS].indices)
+				j = 0;
+			/