Merge branch 'next' into for-linus

Prepare input updates for 5.9 merge window.

5 files changed, 446 insertions, 207 deletions

diff --git a/drivers/net/ethernet/xilinx/Kconfig b/drivers/net/ethernet/xilinx/Kconfig
index 6304ebd8b5c6..0810af8193cb 100644
--- a/drivers/net/ethernet/xilinx/Kconfig
+++ b/drivers/net/ethernet/xilinx/Kconfig
@@ -32,7 +32,6 @@ config XILINX_AXI_EMAC
 
 config XILINX_LL_TEMAC
 	tristate "Xilinx LL TEMAC (LocalLink Tri-mode Ethernet MAC) driver"
-	depends on PPC || MICROBLAZE || X86 || COMPILE_TEST
 	select PHYLIB
 	---help---
 	  This driver supports the Xilinx 10/100/1000 LocalLink TEMAC
diff --git a/drivers/net/ethernet/xilinx/ll_temac.h b/drivers/net/ethernet/xilinx/ll_temac.h
index 53fb8141f1a6..4a73127e10a6 100644
--- a/drivers/net/ethernet/xilinx/ll_temac.h
+++ b/drivers/net/ethernet/xilinx/ll_temac.h
@@ -369,18 +369,20 @@ struct temac_local {
 	/* Buffer descriptors */
 	struct cdmac_bd *tx_bd_v;
 	dma_addr_t tx_bd_p;
+	u32 tx_bd_num;
 	struct cdmac_bd *rx_bd_v;
 	dma_addr_t rx_bd_p;
+	u32 rx_bd_num;
 	int tx_bd_ci;
-	int tx_bd_next;
 	int tx_bd_tail;
 	int rx_bd_ci;
 	int rx_bd_tail;
 
 	/* DMA channel control setup */
-	u32 tx_chnl_ctrl;
-	u32 rx_chnl_ctrl;
+	u8 coalesce_count_tx;
+	u8 coalesce_delay_tx;
 	u8 coalesce_count_rx;
+	u8 coalesce_delay_rx;
 
 	struct delayed_work restart_work;
 };
diff --git a/drivers/net/ethernet/xilinx/ll_temac_main.c b/drivers/net/ethernet/xilinx/ll_temac_main.c
index 9461acec6f70..3e313e71ae36 100644
--- a/drivers/net/ethernet/xilinx/ll_temac_main.c
+++ b/drivers/net/ethernet/xilinx/ll_temac_main.c
@@ -58,8 +58,11 @@
 
 #include "ll_temac.h"
 
-#define TX_BD_NUM   64
-#define RX_BD_NUM   128
+/* Descriptors defines for Tx and Rx DMA */
+#define TX_BD_NUM_DEFAULT		64
+#define RX_BD_NUM_DEFAULT		1024
+#define TX_BD_NUM_MAX			4096
+#define RX_BD_NUM_MAX			4096
 
 /* ---------------------------------------------------------------------
  * Low level register access functions
@@ -301,7 +304,7 @@ static void temac_dma_bd_release(struct net_device *ndev)
 	/* Reset Local Link (DMA) */
 	lp->dma_out(lp, DMA_CONTROL_REG, DMA_CONTROL_RST);
 
-	for (i = 0; i < RX_BD_NUM; i++) {
+	for (i = 0; i < lp->rx_bd_num; i++) {
 		if (!lp->rx_skb[i])
 			break;
 		else {
@@ -312,12 +315,12 @@ static void temac_dma_bd_release(struct net_device *ndev)
 	}
 	if (lp->rx_bd_v)
 		dma_free_coherent(ndev->dev.parent,
-				sizeof(*lp->rx_bd_v) * RX_BD_NUM,
-				lp->rx_bd_v, lp->rx_bd_p);
+				  sizeof(*lp->rx_bd_v) * lp->rx_bd_num,
+				  lp->rx_bd_v, lp->rx_bd_p);
 	if (lp->tx_bd_v)
 		dma_free_coherent(ndev->dev.parent,
-				sizeof(*lp->tx_bd_v) * TX_BD_NUM,
-				lp->tx_bd_v, lp->tx_bd_p);
+				  sizeof(*lp->tx_bd_v) * lp->tx_bd_num,
+				  lp->tx_bd_v, lp->tx_bd_p);
 }
 
 /**
@@ -330,33 +333,33 @@ static int temac_dma_bd_init(struct net_device *ndev)
 	dma_addr_t skb_dma_addr;
 	int i;
 
-	lp->rx_skb = devm_kcalloc(&ndev->dev, RX_BD_NUM, sizeof(*lp->rx_skb),
-				  GFP_KERNEL);
+	lp->rx_skb = devm_kcalloc(&ndev->dev, lp->rx_bd_num,
+				  sizeof(*lp->rx_skb), GFP_KERNEL);
 	if (!lp->rx_skb)
 		goto out;
 
 	/* allocate the tx and rx ring buffer descriptors. */
 	/* returns a virtual address and a physical address. */
 	lp->tx_bd_v = dma_alloc_coherent(ndev->dev.parent,
-					 sizeof(*lp->tx_bd_v) * TX_BD_NUM,
+					 sizeof(*lp->tx_bd_v) * lp->tx_bd_num,
 					 &lp->tx_bd_p, GFP_KERNEL);
 	if (!lp->tx_bd_v)
 		goto out;
 
 	lp->rx_bd_v = dma_alloc_coherent(ndev->dev.parent,
-					 sizeof(*lp->rx_bd_v) * RX_BD_NUM,
+					 sizeof(*lp->rx_bd_v) * lp->rx_bd_num,
 					 &lp->rx_bd_p, GFP_KERNEL);
 	if (!lp->rx_bd_v)
 		goto out;
 
-	for (i = 0; i < TX_BD_NUM; i++) {
+	for (i = 0; i < lp->tx_bd_num; i++) {
 		lp->tx_bd_v[i].next = cpu_to_be32(lp->tx_bd_p
-				+ sizeof(*lp->tx_bd_v) * ((i + 1) % TX_BD_NUM));
+			+ sizeof(*lp->tx_bd_v) * ((i + 1) % lp->tx_bd_num));
 	}
 
-	for (i = 0; i < RX_BD_NUM; i++) {
+	for (i = 0; i < lp->rx_bd_num; i++) {
 		lp->rx_bd_v[i].next = cpu_to_be32(lp->rx_bd_p
-				+ sizeof(*lp->rx_bd_v) * ((i + 1) % RX_BD_NUM));
+			+ sizeof(*lp->rx_bd_v) * ((i + 1) % lp->rx_bd_num));
 
 		skb = netdev_alloc_skb_ip_align(ndev,
 						XTE_MAX_JUMBO_FRAME_SIZE);
@@ -376,21 +379,22 @@ static int temac_dma_bd_init(struct net_device *ndev)
 	}
 
 	/* Configure DMA channel (irq setup) */
-	lp->dma_out(lp, TX_CHNL_CTRL, lp->tx_chnl_ctrl |
+	lp->dma_out(lp, TX_CHNL_CTRL,
+		    lp->coalesce_delay_tx << 24 | lp->coalesce_count_tx << 16 |
 		    0x00000400 | // Use 1 Bit Wide Counters. Currently Not Used!
 		    CHNL_CTRL_IRQ_EN | CHNL_CTRL_IRQ_ERR_EN |
 		    CHNL_CTRL_IRQ_DLY_EN | CHNL_CTRL_IRQ_COAL_EN);
-	lp->dma_out(lp, RX_CHNL_CTRL, lp->rx_chnl_ctrl |
+	lp->dma_out(lp, RX_CHNL_CTRL,
+		    lp->coalesce_delay_rx << 24 | lp->coalesce_count_rx << 16 |
 		    CHNL_CTRL_IRQ_IOE |
 		    CHNL_CTRL_IRQ_EN | CHNL_CTRL_IRQ_ERR_EN |
 		    CHNL_CTRL_IRQ_DLY_EN | CHNL_CTRL_IRQ_COAL_EN);
 
 	/* Init descriptor indexes */
 	lp->tx_bd_ci = 0;
-	lp->tx_bd_next = 0;
 	lp->tx_bd_tail = 0;
 	lp->rx_bd_ci = 0;
-	lp->rx_bd_tail = RX_BD_NUM - 1;
+	lp->rx_bd_tail = lp->rx_bd_num - 1;
 
 	/* Enable RX DMA transfers */
 	wmb();
@@ -785,7 +789,7 @@ static void temac_start_xmit_done(struct net_device *ndev)
 		ndev->stats.tx_bytes += be32_to_cpu(cur_p->len);
 
 		lp->tx_bd_ci++;
-		if (lp->tx_bd_ci >= TX_BD_NUM)
+		if (lp->tx_bd_ci >= lp->tx_bd_num)
 			lp->tx_bd_ci = 0;
 
 		cur_p = &lp->tx_bd_v[lp->tx_bd_ci];
@@ -811,7 +815,7 @@ static inline int temac_check_tx_bd_space(struct temac_local *lp, int num_frag)
 			return NETDEV_TX_BUSY;
 
 		tail++;
-		if (tail >= TX_BD_NUM)
+		if (tail >= lp->tx_bd_num)
 			tail = 0;
 
 		cur_p = &lp->tx_bd_v[tail];
@@ -826,14 +830,13 @@ temac_start_xmit(struct sk_buff *skb, struct net_device *ndev)
 {
 	struct temac_local *lp = netdev_priv(ndev);
 	struct cdmac_bd *cur_p;
-	dma_addr_t start_p, tail_p, skb_dma_addr;
+	dma_addr_t tail_p, skb_dma_addr;
 	int ii;
 	unsigned long num_frag;
 	skb_frag_t *frag;
 
 	num_frag = skb_shinfo(skb)->nr_frags;
 	frag = &skb_shinfo(skb)->frags[0];
-	start_p = lp->tx_bd_p + sizeof(*lp->tx_bd_v) * lp->tx_bd_tail;
 	cur_p = &lp->tx_bd_v[lp->tx_bd_tail];
 
 	if (temac_check_tx_bd_space(lp, num_frag + 1)) {
@@ -876,7 +879,7 @@ temac_start_xmit(struct sk_buff *skb, struct net_device *ndev)
 	ptr_to_txbd((void *)skb, cur_p);
 
 	for (ii = 0; ii < num_frag; ii++) {
-		if (++lp->tx_bd_tail >= TX_BD_NUM)
+		if (++lp->tx_bd_tail >= lp->tx_bd_num)
 			lp->tx_bd_tail = 0;
 
 		cur_p = &lp->tx_bd_v[lp->tx_bd_tail];
@@ -886,7 +889,7 @@ temac_start_xmit(struct sk_buff *skb, struct net_device *ndev)
 					      DMA_TO_DEVICE);
 		if (dma_mapping_error(ndev->dev.parent, skb_dma_addr)) {
 			if (--lp->tx_bd_tail < 0)
-				lp->tx_bd_tail = TX_BD_NUM - 1;
+				lp->tx_bd_tail = lp->tx_bd_num - 1;
 			cur_p = &lp->tx_bd_v[lp->tx_bd_tail];
 			while (--ii >= 0) {
 				--frag;
@@ -895,7 +898,7 @@ temac_start_xmit(struct sk_buff *skb, struct net_device *ndev)
 						 skb_frag_size(frag),
 						 DMA_TO_DEVICE);
 				if (--lp->tx_bd_tail < 0)
-					lp->tx_bd_tail = TX_BD_NUM - 1;
+					lp->tx_bd_tail = lp->tx_bd_num - 1;
 				cur_p = &lp->tx_bd_v[lp->tx_bd_tail];
 			}
 			dma_unmap_single(ndev->dev.parent,
@@ -914,7 +917,7 @@ temac_start_xmit(struct sk_buff *skb, struct net_device *ndev)
 
 	tail_p = lp->tx_bd_p + sizeof(*lp->tx_bd_v) * lp->tx_bd_tail;
 	lp->tx_bd_tail++;
-	if (lp->tx_bd_tail >= TX_BD_NUM)
+	if (lp->tx_bd_tail >= lp->tx_bd_num)
 		lp->tx_bd_tail = 0;
 
 	skb_tx_timestamp(skb);
@@ -934,7 +937,7 @@ static int ll_temac_recv_buffers_available(struct temac_local *lp)
 		return 0;
 	available = 1 + lp->rx_bd_tail - lp->rx_bd_ci;
 	if (available <= 0)
-		available += RX_BD_NUM;
+		available += lp->rx_bd_num;
 	return available;
 }
 
@@ -1003,7 +1006,7 @@ static void ll_temac_recv(struct net_device *ndev)
 		ndev->stats.rx_bytes += length;
 
 		rx_bd = lp->rx_bd_ci;
-		if (++lp->rx_bd_ci >= RX_BD_NUM)
+		if (++lp->rx_bd_ci >= lp->rx_bd_num)
 			lp->rx_bd_ci = 0;
 	} while (rx_bd != lp->rx_bd_tail);
 
@@ -1034,7 +1037,7 @@ static void ll_temac_recv(struct net_device *ndev)
 		dma_addr_t skb_dma_addr;
 
 		rx_bd = lp->rx_bd_tail + 1;
-		if (rx_bd >= RX_BD_NUM)
+		if (rx_bd >= lp->rx_bd_num)
 			rx_bd = 0;
 		bd = &lp->rx_bd_v[rx_bd];
 
@@ -1250,13 +1253,96 @@ static const struct attribute_group temac_attr_group = {
 	.attrs = temac_device_attrs,
 };
 
-/* ethtool support */
+/* ---------------------------------------------------------------------
+ * ethtool support
+ */
+
+static void ll_temac_ethtools_get_ringparam(struct net_device *ndev,
+					    struct ethtool_ringparam *ering)
+{
+	struct temac_local *lp = netdev_priv(ndev);
+
+	ering->rx_max_pending = RX_BD_NUM_MAX;
+	ering->rx_mini_max_pending = 0;
+	ering->rx_jumbo_max_pending = 0;
+	ering->tx_max_pending = TX_BD_NUM_MAX;
+	ering->rx_pending = lp->rx_bd_num;
+	ering->rx_mini_pending = 0;
+	ering->rx_jumbo_pending = 0;
+	ering->tx_pending = lp->tx_bd_num;
+}
+
+static int ll_temac_ethtools_set_ringparam(struct net_device *ndev,
+					   struct ethtool_ringparam *ering)
+{
+	struct temac_local *lp = netdev_priv(ndev);
+
+	if (ering->rx_pending > RX_BD_NUM_MAX ||
+	    ering->rx_mini_pending ||
+	    ering->rx_jumbo_pending ||
+	    ering->rx_pending > TX_BD_NUM_MAX)
+		return -EINVAL;
+
+	if (netif_running(ndev))
+		return -EBUSY;
+
+	lp->rx_bd_num = ering->rx_pending;
+	lp->tx_bd_num = ering->tx_pending;
+	return 0;
+}
+
+static int ll_temac_ethtools_get_coalesce(struct net_device *ndev,
+					  struct ethtool_coalesce *ec)
+{
+	struct temac_local *lp = netdev_priv(ndev);
+
+	ec->rx_max_coalesced_frames = lp->coalesce_count_rx;
+	ec->tx_max_coalesced_frames = lp->coalesce_count_tx;
+	ec->rx_coalesce_usecs = (lp->coalesce_delay_rx * 512) / 100;
+	ec->tx_coalesce_usecs = (lp->coalesce_delay_tx * 512) / 100;
+	return 0;
+}
+
+static int ll_temac_ethtools_set_coalesce(struct net_device *ndev,
+					  struct ethtool_coalesce *ec)
+{
+	struct temac_local *lp = netdev_priv(ndev);
+
+	if (netif_running(ndev)) {
+		netdev_err(ndev,
+			   "Please stop netif before applying configuration\n");
+		return -EFAULT;
+	}
+
+	if (ec->rx_max_coalesced_frames)
+		lp->coalesce_count_rx = ec->rx_max_coalesced_frames;
+	if (ec->tx_max_coalesced_frames)
+		lp->coalesce_count_tx = ec->tx_max_coalesced_frames;
+	/* With typical LocalLink clock speed of 200 MHz and
+	 * C_PRESCALAR=1023, each delay count corresponds to 5.12 us.
+	 */
+	if (ec->rx_coalesce_usecs)
+		lp->coalesce_delay_rx =
+			min(255U, (ec->rx_coalesce_usecs * 100) / 512);
+	if (ec->tx_coalesce_usecs)
+		lp->coalesce_delay_tx =
+			min(255U, (ec->tx_coalesce_usecs * 100) / 512);
+
+	return 0;
+}
+
 static const struct ethtool_ops temac_ethtool_ops = {
+	.supported_coalesce_params = ETHTOOL_COALESCE_USECS |
+				     ETHTOOL_COALESCE_MAX_FRAMES,
 	.nway_reset = phy_ethtool_nway_reset,
 	.get_link = ethtool_op_get_link,
 	.get_ts_info = ethtool_op_get_ts_info,
 	.get_link_ksettings = phy_ethtool_get_link_ksettings,
 	.set_link_ksettings = phy_ethtool_set_link_ksettings,
+	.get_ringparam	= ll_temac_ethtools_get_ringparam,
+	.set_ringparam	= ll_temac_ethtools_set_ringparam,
+	.get_coalesce	= ll_temac_ethtools_get_coalesce,
+	.set_coalesce	= ll_temac_ethtools_set_coalesce,
 };
 
 static int temac_probe(struct platform_device *pdev)
@@ -1300,6 +1386,8 @@ static int temac_probe(struct platform_device *pdev)
 	lp->ndev = ndev;
 	lp->dev = &pdev->dev;
 	lp->options = XTE_OPTION_DEFAULTS;
+	lp->rx_bd_num = RX_BD_NUM_DEFAULT;
+	lp->tx_bd_num = TX_BD_NUM_DEFAULT;
 	spin_lock_init(&lp->rx_lock);
 	INIT_DELAYED_WORK(&lp->restart_work, ll_temac_restart_work_func);
 
@@ -1364,6 +1452,14 @@ static int temac_probe(struct platform_device *pdev)
 		/* Can checksum TCP/UDP over IPv4. */
 		ndev->features |= NETIF_F_IP_CSUM;
 
+	/* Defaults for IRQ delay/coalescing setup.  These are
+	 * configuration values, so does not belong in device-tree.
+	 */
+	lp->coalesce_delay_tx = 0x10;
+	lp->coalesce_count_tx = 0x22;
+	lp->coalesce_delay_rx = 0xff;
+	lp->coalesce_count_rx = 0x07;
+
 	/* Setup LocalLink DMA */
 	if (temac_np) {
 		/* Find the DMA node, map the DMA registers, and
@@ -1402,14 +1498,6 @@ static int temac_probe(struct platform_device *pdev)
 		lp->rx_irq = irq_of_parse_and_map(dma_np, 0);
 		lp->tx_irq = irq_of_parse_and_map(dma_np, 1);
 
-		/* Use defaults for IRQ delay/coalescing setup.  These
-		 * are configuration values, so does not belong in
-		 * device-tree.
-		 */
-		lp->tx_chnl_ctrl = 0x10220000;
-		lp->rx_chnl_ctrl = 0xff070000;
-		lp->coalesce_count_rx = 0x07;
-
 		/* Finished with the DMA node; drop the reference */
 		of_node_put(dma_np);
 	} else if (pdata) {
@@ -1435,18 +1523,13 @@ static int temac_probe(struct platform_device *pdev)
 		lp->tx_irq = platform_get_irq(pdev, 1);
 
 		/* IRQ delay/coalescing setup */
-		if (pdata->tx_irq_timeout || pdata->tx_irq_count)
-			lp->tx_chnl_ctrl = (pdata->tx_irq_timeout << 24) |
-				(pdata->tx_irq_count << 16);
-		else
-			lp->tx_chnl_ctrl = 0x10220000;
+		if (pdata->tx_irq_timeout || pdata->tx_irq_count) {
+			lp->coalesce_delay_tx = pdata->tx_irq_timeout;
+			lp->coalesce_count_tx = pdata->tx_irq_count;
+		}
 		if (pdata->rx_irq_timeout || pdata->rx_irq_count) {
-			lp->rx_chnl_ctrl = (pdata->rx_irq_timeout << 24) |
-				(pdata->rx_irq_count << 16);
+			lp->coalesce_delay_rx = pdata->rx_irq_timeout;
 			lp->coalesce_count_rx = pdata->rx_irq_count;
-		} else {
-			lp->rx_chnl_ctrl = 0xff070000;
-			lp->coalesce_count_rx = 0x07;
 		}
 	}
 
diff --git a/drivers/net/ethernet/xilinx/xilinx_axienet.h b/drivers/net/ethernet/xilinx/xilinx_axienet.h
index 2dacfc85b3ba..fbaf3c987d9c 100644
--- a/drivers/net/ethernet/xilinx/xilinx_axienet.h
+++ b/drivers/net/ethernet/xilinx/xilinx_axienet.h
@@ -161,17 +161,11 @@
 #define XAE_FCC_OFFSET		0x0000040C /* Flow Control Configuration */
 #define XAE_EMMC_OFFSET		0x00000410 /* EMAC mode configuration */
 #define XAE_PHYC_OFFSET		0x00000414 /* RGMII/SGMII configuration */
+#define XAE_ID_OFFSET		0x000004F8 /* Identification register */
 #define XAE_MDIO_MC_OFFSET	0x00000500 /* MII Management Config */
 #define XAE_MDIO_MCR_OFFSET	0x00000504 /* MII Management Control */
 #define XAE_MDIO_MWD_OFFSET	0x00000508 /* MII Management Write Data */
 #define XAE_MDIO_MRD_OFFSET	0x0000050C /* MII Management Read Data */
-#define XAE_MDIO_MIS_OFFSET	0x00000600 /* MII Management Interrupt Status */
-/* MII Mgmt Interrupt Pending register offset */
-#define XAE_MDIO_MIP_OFFSET	0x00000620
-/* MII Management Interrupt Enable register offset */
-#define XAE_MDIO_MIE_OFFSET	0x00000640
-/* MII Management Interrupt Clear register offset. */
-#define XAE_MDIO_MIC_OFFSET	0x00000660
 #define XAE_UAW0_OFFSET		0x00000700 /* Unicast address word 0 */
 #define XAE_UAW1_OFFSET		0x00000704 /* Unicast address word 1 */
 #define XAE_FMI_OFFSET		0x00000708 /* Filter Mask Index */
@@ -335,6 +329,7 @@
 #define XAE_FEATURE_PARTIAL_TX_CSUM	(1 << 1)
 #define XAE_FEATURE_FULL_RX_CSUM	(1 << 2)
 #define XAE_FEATURE_FULL_TX_CSUM	(1 << 3)
+#define XAE_FEATURE_DMA_64BIT		(1 << 4)
 
 #define XAE_NO_CSUM_OFFLOAD		0
 
@@ -347,9 +342,9 @@
 /**
  * struct axidma_bd - Axi Dma buffer descriptor layout
  * @next:         MM2S/S2MM Next Descriptor Pointer
- * @reserved1:    Reserved and not used
+ * @next_msb:     MM2S/S2MM Next Descriptor Pointer (high 32 bits)
  * @phys:         MM2S/S2MM Buffer Address
- * @reserved2:    Reserved and not used
+ * @phys_msb:     MM2S/S2MM Buffer Address (high 32 bits)
  * @reserved3:    Reserved and not used
  * @reserved4:    Reserved and not used
  * @cntrl:        MM2S/S2MM Control value
@@ -362,9 +357,9 @@
  */
 struct axidma_bd {
 	u32 next;	/* Physical address of next buffer descriptor */
-	u32 reserved1;
+	u32 next_msb;	/* high 32 bits for IP >= v7.1, reserved on older IP */
 	u32 phys;
-	u32 reserved2;
+	u32 phys_msb;	/* for IP >= v7.1, reserved for older IP */
 	u32 reserved3;
 	u32 reserved4;
 	u32 cntrl;
@@ -435,7 +430,7 @@ struct axienet_local {
 	void __iomem *regs;
 	void __iomem *dma_regs;
 
-	struct tasklet_struct dma_err_tasklet;
+	struct work_struct dma_err_task;
 
 	int tx_irq;
 	int rx_irq;
diff --git a/drivers/net/ethernet/xilinx/xilinx_axienet_main.c b/drivers/net/ethernet/xilinx/xilinx_axienet_main.c
index 20746b801959..fa5dc2993520 100644
--- a/drivers/net/ethernet/xilinx/xilinx_axienet_main.c
+++ b/drivers/net/ethernet/xilinx/xilinx_axienet_main.c
@@ -147,6 +147,34 @@ static inline void axienet_dma_out32(struct axienet_local *lp,
 	iowrite32(value, lp->dma_regs + reg);
 }
 
+static void axienet_dma_out_addr(struct axienet_local *lp, off_t reg,
+				 dma_addr_t addr)
+{
+	axienet_dma_out32(lp, reg, lower_32_bits(addr));
+
+	if (lp->features & XAE_FEATURE_DMA_64BIT)
+		axienet_dma_out32(lp, reg + 4, upper_32_bits(addr));
+}
+
+static void desc_set_phys_addr(struct axienet_local *lp, dma_addr_t addr,
+			       struct axidma_bd *desc)
+{
+	desc->phys = lower_32_bits(addr);
+	if (lp->features & XAE_FEATURE_DMA_64BIT)
+		desc->phys_msb = upper_32_bits(addr);
+}
+
+static dma_addr_t desc_get_phys_addr(struct axienet_local *lp,
+				     struct axidma_bd *desc)
+{
+	dma_addr_t ret = desc->phys;
+
+	if (lp->features & XAE_FEATURE_DMA_64BIT)
+		ret |= ((dma_addr_t)desc->phys_msb << 16) << 16;
+
+	return ret;
+}
+
 /**
  * axienet_dma_bd_release - Release buffer descriptor rings
  * @ndev:	Pointer to the net_device structure
@@ -160,24 +188,41 @@ static void axienet_dma_bd_release(struct net_device *ndev)
 	int i;
 	struct axienet_local *lp = netdev_priv(ndev);
 
+	/* If we end up here, tx_bd_v must have been DMA allocated. */
+	dma_free_coherent(ndev->dev.parent,
+			  sizeof(*lp->tx_bd_v) * lp->tx_bd_num,
+			  lp->tx_bd_v,
+			  lp->tx_bd_p);
+
+	if (!lp->rx_bd_v)
+		return;
+
 	for (i = 0; i < lp->rx_bd_num; i++) {
-		dma_unmap_single(ndev->dev.parent, lp->rx_bd_v[i].phys,
-				 lp->max_frm_size, DMA_FROM_DEVICE);
+		dma_addr_t phys;
+
+		/* A NULL skb means this descriptor has not been initialised
+		 * at all.
+		 */
+		if (!lp->rx_bd_v[i].skb)
+			break;
+
 		dev_kfree_skb(lp->rx_bd_v[i].skb);
-	}
 
-	if (lp->rx_bd_v) {
-		dma_free_coherent(ndev->dev.parent,
-				  sizeof(*lp->rx_bd_v) * lp->rx_bd_num,
-				  lp->rx_bd_v,
-				  lp->rx_bd_p);
-	}
-	if (lp->tx_bd_v) {
-		dma_free_coherent(ndev->dev.parent,
-				  sizeof(*lp->tx_bd_v) * lp->tx_bd_num,
-				  lp->tx_bd_v,
-				  lp->tx_bd_p);
+		/* For each descriptor, we programmed cntrl with the (non-zero)
+		 * descriptor size, after it had been successfully allocated.
+		 * So a non-zero value in there means we need to unmap it.
+		 */
+		if (lp->rx_bd_v[i].cntrl) {
+			phys = desc_get_phys_addr(lp, &lp->rx_bd_v[i]);
+			dma_unmap_single(ndev->dev.parent, phys,
+					 lp->max_frm_size, DMA_FROM_DEVICE);
+		}
 	}
+
+	dma_free_coherent(ndev->dev.parent,
+			  sizeof(*lp->rx_bd_v) * lp->rx_bd_num,
+			  lp->rx_bd_v,
+			  lp->rx_bd_p);
 }
 
 /**
@@ -207,7 +252,7 @@ static int axienet_dma_bd_init(struct net_device *ndev)
 					 sizeof(*lp->tx_bd_v) * lp->tx_bd_num,
 					 &lp->tx_bd_p, GFP_KERNEL);
 	if (!lp->tx_bd_v)
-		goto out;
+		return -ENOMEM;
 
 	lp->rx_bd_v = dma_alloc_coherent(ndev->dev.parent,
 					 sizeof(*lp->rx_bd_v) * lp->rx_bd_num,
@@ -216,25 +261,37 @@ static int axienet_dma_bd_init(struct net_device *ndev)
 		goto out;
 
 	for (i = 0; i < lp->tx_bd_num; i++) {
-		lp->tx_bd_v[i].next = lp->tx_bd_p +
-				      sizeof(*lp->tx_bd_v) *
-				      ((i + 1) % lp->tx_bd_num);
+		dma_addr_t addr = lp->tx_bd_p +
+				  sizeof(*lp->tx_bd_v) *
+				  ((i + 1) % lp->tx_bd_num);
+
+		lp->tx_bd_v[i].next = lower_32_bits(addr);
+		if (lp->features & XAE_FEATURE_DMA_64BIT)
+			lp->tx_bd_v[i].next_msb = upper_32_bits(addr);
 	}
 
 	for (i = 0; i < lp->rx_bd_num; i++) {
-		lp->rx_bd_v[i].next = lp->rx_bd_p +
-				      sizeof(*lp->rx_bd_v) *
-				      ((i + 1) % lp->rx_bd_num);
+		dma_addr_t addr;
+
+		addr = lp->rx_bd_p + sizeof(*lp->rx_bd_v) *
+			((i + 1) % lp->rx_bd_num);
+		lp->rx_bd_v[i].next = lower_32_bits(addr);
+		if (lp->features & XAE_FEATURE_DMA_64BIT)
+			lp->rx_bd_v[i].next_msb = upper_32_bits(addr);
 
 		skb = netdev_alloc_skb_ip_align(ndev, lp->max_frm_size);
 		if (!skb)
 			goto out;
 
 		lp->rx_bd_v[i].skb = skb;
-		lp->rx_bd_v[i].phys = dma_map_single(ndev->dev.parent,
-						     skb->data,
-						     lp->max_frm_size,
-						     DMA_FROM_DEVICE);
+		addr = dma_map_single(ndev->dev.parent, skb->data,
+				      lp->max_frm_size, DMA_FROM_DEVICE);
+		if (dma_mapping_error(ndev->dev.parent, addr)) {
+			netdev_err(ndev, "DMA mapping error\n");
+			goto out;
+		}
+		desc_set_phys_addr(lp, addr, &lp->rx_bd_v[i]);
+
 		lp->rx_bd_v[i].cntrl = lp->max_frm_size;
 	}
 
@@ -267,18 +324,18 @@ static int axienet_dma_bd_init(struct net_device *ndev)
 	/* Populate the tail pointer and bring the Rx Axi DMA engine out of
 	 * halted state. This will make the Rx side ready for reception.
 	 */
-	axienet_dma_out32(lp, XAXIDMA_RX_CDESC_OFFSET, lp->rx_bd_p);
+	axienet_dma_out_addr(lp, XAXIDMA_RX_CDESC_OFFSET, lp->rx_bd_p);
 	cr = axienet_dma_in32(lp, XAXIDMA_RX_CR_OFFSET);
 	axienet_dma_out32(lp, XAXIDMA_RX_CR_OFFSET,
 			  cr | XAXIDMA_CR_RUNSTOP_MASK);
-	axienet_dma_out32(lp, XAXIDMA_RX_TDESC_OFFSET, lp->rx_bd_p +
-			  (sizeof(*lp->rx_bd_v) * (lp->rx_bd_num - 1)));
+	axienet_dma_out_addr(lp, XAXIDMA_RX_TDESC_OFFSET, lp->rx_bd_p +
+			     (sizeof(*lp->rx_bd_v) * (lp->rx_bd_num - 1)));
 
 	/* Write to the RS (Run-stop) bit in the Tx channel control register.
 	 * Tx channel is now ready to run. But only after we write to the
 	 * tail pointer register that the Tx channel will start transmitting.
 	 */
-	axienet_dma_out32(lp, XAXIDMA_TX_CDESC_OFFSET, lp->tx_bd_p);
+	axienet_dma_out_addr(lp, XAXIDMA_TX_CDESC_OFFSET, lp->tx_bd_p);
 	cr = axienet_dma_in32(lp, XAXIDMA_TX_CR_OFFSET);
 	axienet_dma_out32(lp, XAXIDMA_TX_CR_OFFSET,
 			  cr | XAXIDMA_CR_RUNSTOP_MASK);
@@ -437,9 +494,10 @@ static void axienet_setoptions(struct net_device *ndev, u32 options)
 	lp->options |= options;
 }
 
-static void __axienet_device_reset(struct axienet_local *lp)
+static int __axienet_device_reset(struct axienet_local *lp)
 {
 	u32 timeout;
+
 	/* Reset Axi DMA. This would reset Axi Ethernet core as well. The reset
 	 * process of Axi DMA takes a while to complete as all pending
 	 * commands/transfers will be flushed or completed during this
@@ -455,9 +513,11 @@ static void __axienet_device_reset(struct axienet_local *lp)
 		if (--timeout == 0) {
 			netdev_err(lp->ndev, "%s: DMA reset timeout!\n",
 				   __func__);
-			break;
+			return -ETIMEDOUT;
 		}
 	}
+
+	return 0;
 }
 
 /**
@@ -470,13 +530,17 @@ static void __axienet_device_reset(struct axienet_local *lp)
  * areconnected to Axi Ethernet reset lines, this in turn resets the Axi
  * Ethernet core. No separate hardware reset is done for the Axi Ethernet
  * core.
+ * Returns 0 on success or a negative error number otherwise.
  */
-static void axienet_device_reset(struct net_device *ndev)
+static int axienet_device_reset(struct net_device *ndev)
 {
 	u32 axienet_status;
 	struct axienet_local *lp = netdev_priv(ndev);
+	int ret;
 
-	__axienet_device_reset(lp);
+	ret = __axienet_device_reset(lp);
+	if (ret)
+		return ret;
 
 	lp->max_frm_size = XAE_MAX_VLAN_FRAME_SIZE;
 	lp->options |= XAE_OPTION_VLAN;
@@ -491,9 +555,11 @@ static void axienet_device_reset(struct net_device *ndev)
 			lp->options |= XAE_OPTION_JUMBO;
 	}
 
-	if (axienet_dma_bd_init(ndev)) {
+	ret = axienet_dma_bd_init(ndev);
+	if (ret) {
 		netdev_err(ndev, "%s: descriptor allocation failed\n",
 			   __func__);
+		return ret;
 	}
 
 	axienet_status = axienet_ior(lp, XAE_RCW1_OFFSET);
@@ -518,36 +584,54 @@ static void axienet_device_reset(struct net_device *ndev)
 	axienet_setoptions(ndev, lp->options);
 
 	netif_trans_update(ndev);
+
+	return 0;
 }
 
 /**
- * axienet_start_xmit_done - Invoked once a transmit is completed by the
- * Axi DMA Tx channel.
+ * axienet_free_tx_chain - Clean up a series of linked TX descriptors.
  * @ndev:	Pointer to the net_device structure
+ * @first_bd:	Index of first descriptor to clean up
+ * @nr_bds:	Number of descriptors to clean up, can be -1 if unknown.
+ * @sizep:	Pointer to a u32 filled with the total sum of all bytes
+ * 		in all cleaned-up descriptors. Ignored if NULL.
  *
- * This function is invoked from the Axi DMA Tx isr to notify the completion
- * of transmit operation. It clears fields in the corresponding Tx BDs and
- * unmaps the corresponding buffer so that CPU can regain ownership of the
- * buffer. It finally invokes "netif_wake_queue" to restart transmission if
- * required.
+ * Would either be called after a successful transmit operation, or after
+ * there was an error when setting up the chain.
+ * Returns the number of descriptors handled.
  */
-static void axienet_start_xmit_done(struct net_device *ndev)
+static int axienet_free_tx_chain(struct net_device *ndev, u32 first_bd,
+				 int nr_bds, u32 *sizep)
 {
-	u32 size = 0;
-	u32 packets = 0;
 	struct axienet_local *lp = netdev_priv(ndev);
 	struct axidma_bd *cur_p;
-	unsigned int status = 0;
-
-	cur_p = &lp->tx_bd_v[lp->tx_bd_ci];
-	status = cur_p->status;
-	while (status & XAXIDMA_BD_STS_COMPLETE_MASK) {
-		dma_unmap_single(ndev->dev.parent, cur_p->phys,
-				(cur_p->cntrl & XAXIDMA_BD_CTRL_LENGTH_MASK),
-				DMA_TO_DEVICE);
-		if (cur_p->skb)
+	int max_bds = nr_bds;
+	unsigned int status;
+	dma_addr_t phys;
+	int i;
+
+	if (max_bds == -1)
+		max_bds = lp->tx_bd_num;
+
+	for (i = 0; i < max_bds; i++) {
+		cur_p = &lp->tx_bd_v[(first_bd + i) % lp->tx_bd_num];
+		status = cur_p->status;
+
+		/* If no number is given, clean up *all* descriptors that have
+		 * been completed by the MAC.
+		 */
+		if (nr_bds == -1 && !(status & XAXIDMA_BD_STS_COMPLETE_MASK))
+			break;
+
+		phys = desc_get_phys_addr(lp, cur_p);
+		dma_unmap_single(ndev->dev.parent, phys,
+				 (cur_p->cntrl & XAXIDMA_BD_CTRL_LENGTH_MASK),
+				 DMA_TO_DEVICE);
+
+		if (cur_p->skb && (status & XAXIDMA_BD_STS_COMPLETE_MASK))
 			dev_consume_skb_irq(cur_p->skb);
-		/*cur_p->phys = 0;*/
+
+		cur_p->cntrl = 0;
 		cur_p->app0 = 0;
 		cur_p->app1 = 0;
 		cur_p->app2 = 0;
@@ -555,15 +639,36 @@ static void axienet_start_xmit_done(struct net_device *ndev)
 		cur_p->status = 0;
 		cur_p->skb = NULL;
 
-		size += status & XAXIDMA_BD_STS_ACTUAL_LEN_MASK;
-		packets++;
-
-		if (++lp->tx_bd_ci >= lp->tx_bd_num)
-			lp->tx_bd_ci = 0;
-		cur_p = &lp->tx_bd_v[lp->tx_bd_ci];
-		status = cur_p->status;
+		if (sizep)
+			*sizep += status & XAXIDMA_BD_STS_ACTUAL_LEN_MASK;
 	}
 
+	return i;
+}
+
+/**
+ * axienet_start_xmit_done - Invoked once a transmit is completed by the
+ * Axi DMA Tx channel.
+ * @ndev:	Pointer to the net_device structure
+ *
+ * This function is invoked from the Axi DMA Tx isr to notify the completion
+ * of transmit operation. It clears fields in the corresponding Tx BDs and
+ * unmaps the corresponding buffer so that CPU can regain ownership of the
+ * buffer. It finally invokes "netif_wake_queue" to restart transmission if
+ * required.
+ */
+static void axienet_start_xmit_done(struct net_device *ndev)
+{
+	struct axienet_local *lp = netdev_priv(ndev);
+	u32 packets = 0;
+	u32 size = 0;
+
+	packets = axienet_free_tx_chain(ndev, lp->tx_bd_ci, -1, &size);
+
+	lp->tx_bd_ci += packets;
+	if (lp->tx_bd_ci >= lp->tx_bd_num)
+		lp->tx_bd_ci -= lp->tx_bd_num;
+
 	ndev->stats.tx_packets += packets;
 	ndev->stats.tx_bytes += size;
 
@@ -617,9 +722,10 @@ axienet_start_xmit(struct sk_buff *skb, struct net_device *ndev)
 	u32 csum_start_off;
 	u32 csum_index_off;
 	skb_frag_t *frag;
-	dma_addr_t tail_p;
+	dma_addr_t tail_p, phys;
 	struct axienet_local *lp = netdev_priv(ndev);
 	struct axidma_bd *cur_p;
+	u32 orig_tail_ptr = lp->tx_bd_tail;
 
 	num_frag = skb_shinfo(skb)->nr_frags;
 	cur_p = &lp->tx_bd_v[lp->tx_bd_tail];
@@ -655,19 +761,37 @@ axienet_start_xmit(struct sk_buff *skb, struct net_device *ndev)
 		cur_p->app0 |= 2; /* Tx Full Checksum Offload Enabled */
 	}
 
+	phys = dma_map_single(ndev->dev.parent, skb->data,
+			      skb_headlen(skb), DMA_TO_DEVICE);
+	if (unlikely(dma_mapping_error(ndev->dev.parent, phys))) {
+		if (net_ratelimit())
+			netdev_err(ndev, "TX DMA mapping error\n");
+		ndev->stats.tx_dropped++;
+		return NETDEV_TX_OK;
+	}
+	desc_set_phys_addr(lp, phys, cur_p);
 	cur_p->cntrl = skb_headlen(skb) | XAXIDMA_BD_CTRL_TXSOF_MASK;
-	cur_p->phys = dma_map_single(ndev->dev.parent, skb->data,
-				     skb_headlen(skb), DMA_TO_DEVICE);
 
 	for (ii = 0; ii < num_frag; ii++) {
 		if (++lp->tx_bd_tail >= lp->tx_bd_num)
 			lp->tx_bd_tail = 0;
 		cur_p = &lp->tx_bd_v[lp->tx_bd_tail];
 		frag = &skb_shinfo(skb)->frags[ii];
-		cur_p->phys = dma_map_single(ndev->dev.parent,
-					     skb_frag_address(frag),
-					     skb_frag_size(frag),
-					     DMA_TO_DEVICE);
+		phys = dma_map_single(ndev->dev.parent,
+				      skb_frag_address(frag),
+				      skb_frag_size(frag),
+				      DMA_TO_DEVICE);
+		if (unlikely(dma_mapping_error(ndev->dev.parent, phys))) {
+			if (net_ratelimit())
+				netdev_err(ndev, "TX DMA mapping error\n");
+			ndev->stats.tx_dropped++;
+			axienet_free_tx_chain(ndev, orig_tail_ptr, ii + 1,
+					      NULL);
+			lp->tx_bd_tail = orig_tail_ptr;
+
+			return NETDEV_TX_OK;
+		}
+		desc_set_phys_addr(lp, phys, cur_p);
 		cur_p->cntrl = skb_frag_size(frag);
 	}
 
@@ -676,7 +800,7 @@ axienet_start_xmit(struct sk_buff *skb, struct net_device *ndev)
 
 	tail_p = lp->tx_bd_p + sizeof(*lp->tx_bd_v) * lp->tx_bd_tail;
 	/* Start the transfer */
-	axienet_dma_out32(lp, XAXIDMA_TX_TDESC_OFFSET, tail_p);
+	axienet_dma_out_addr(lp, XAXIDMA_TX_TDESC_OFFSET, tail_p);
 	if (++lp->tx_bd_tail >= lp->tx_bd_num)
 		lp->tx_bd_tail = 0;
 
@@ -706,10 +830,12 @@ static void axienet_recv(struct net_device *ndev)
 	cur_p = &lp->rx_bd_v[lp->rx_bd_ci];
 
 	while ((cur_p->status & XAXIDMA_BD_STS_COMPLETE_MASK)) {
+		dma_addr_t phys;
+
 		tail_p = lp->rx_bd_p + sizeof(*lp->rx_bd_v) * lp->rx_bd_ci;
 
-		dma_unmap_single(ndev->dev.parent, cur_p->phys,
-				 lp->max_frm_size,
+		phys = desc_get_phys_addr(lp, cur_p);
+		dma_unmap_single(ndev->dev.parent, phys, lp->max_frm_size,
 				 DMA_FROM_DEVICE);
 
 		skb = cur_p->skb;
@@ -745,9 +871,17 @@ static void axienet_recv(struct net_device *ndev)
 		if (!new_skb)
 			return;
 
-		cur_p->phys = dma_map_single(ndev->dev.parent, new_skb->data,
-					     lp->max_frm_size,
-					     DMA_FROM_DEVICE);
+		phys = dma_map_single(ndev->dev.parent, new_skb->data,
+				      lp->max_frm_size,
+				      DMA_FROM_DEVICE);
+		if (unlikely(dma_mapping_error(ndev->dev.parent, phys))) {
+			if (net_ratelimit())
+				netdev_err(ndev, "RX DMA mapping error\n");
+			dev_kfree_skb(new_skb);
+			return;
+		}
+		desc_set_phys_addr(lp, phys, cur_p);
+
 		cur_p->cntrl = lp->max_frm_size;
 		cur_p->status = 0;
 		cur_p->skb = new_skb;
@@ -761,7 +895,7 @@ static void axienet_recv(struct net_device *ndev)
 	ndev->stats.rx_bytes += size;
 
 	if (tail_p)
-		axienet_dma_out32(lp, XAXIDMA_RX_TDESC_OFFSET, tail_p);
+		axienet_dma_out_addr(lp, XAXIDMA_RX_TDESC_OFFSET, tail_p);
 }
 
 /**
@@ -791,7 +925,8 @@ static irqreturn_t axienet_tx_irq(int irq, void *_ndev)
 		return IRQ_NONE;
 	if (status & XAXIDMA_IRQ_ERROR_MASK) {
 		dev_err(&ndev->dev, "DMA Tx error 0x%x\n", status);
-		dev_err(&ndev->dev, "Current BD is at: 0x%x\n",
+		dev_err(&ndev->dev, "Current BD is at: 0x%x%08x\n",
+			(lp->tx_bd_v[lp->tx_bd_ci]).phys_msb,
 			(lp->tx_bd_v[lp->tx_bd_ci]).phys);
 
 		cr = axienet_dma_in32(lp, XAXIDMA_TX_CR_OFFSET);
@@ -806,7 +941,7 @@ static irqreturn_t axienet_tx_irq(int irq, void *_ndev)
 		/* Write to the Rx channel control register */
 		axienet_dma_out32(lp, XAXIDMA_RX_CR_OFFSET, cr);
 
-		tasklet_schedule(&lp->dma_err_tasklet);
+		schedule_work(&lp->dma_err_task);
 		axienet_dma_out32(lp, XAXIDMA_TX_SR_OFFSET, status);
 	}
 out:
@@ -840,7 +975,8 @@ static irqreturn_t axienet_rx_irq(int irq, void *_ndev)
 		return IRQ_NONE;
 	if (status & XAXIDMA_IRQ_ERROR_MASK) {
 		dev_err(&ndev->dev, "DMA Rx error 0x%x\n", status);
-		dev_err(&ndev->dev, "Current BD is at: 0x%x\n",
+		dev_err(&ndev->dev, "Current BD is at: 0x%x%08x\n",
+			(lp->rx_bd_v[lp->rx_bd_ci]).phys_msb,
 			(lp->rx_bd_v[lp->rx_bd_ci]).phys);
 
 		cr = axienet_dma_in32(lp, XAXIDMA_TX_CR_OFFSET);
@@ -855,7 +991,7 @@ static irqreturn_t axienet_rx_irq(int irq, void *_ndev)
 		/* write to the Rx channel control register */
 		axienet_dma_out32(lp, XAXIDMA_RX_CR_OFFSET, cr);
 
-		tasklet_schedule(&lp->dma_err_tasklet);
+		schedule_work(&lp->dma_err_task);
 		axienet_dma_out32(lp, XAXIDMA_RX_SR_OFFSET, status);
 	}
 out:
@@ -891,7 +1027,7 @@ static irqreturn_t axienet_eth_irq(int irq, void *_ndev)
 	return IRQ_HANDLED;
 }
 
-static void axienet_dma_err_handler(unsigned long data);
+static void axienet_dma_err_handler(struct work_struct *work);
 
 /**
  * axienet_open - Driver open routine.
@@ -921,8 +1057,9 @@ static int axienet_open(struct net_device *ndev)
 	 */
 	mutex_lock(&lp->mii_bus->mdio_lock);
 	axienet_mdio_disable(lp);
-	axienet_device_reset(ndev);
-	ret = axienet_mdio_enable(lp);
+	ret = axienet_device_reset(ndev);
+	if (ret == 0)
+		ret = axienet_mdio_enable(lp);
 	mutex_unlock(&lp->mii_bus->mdio_lock);
 	if (ret < 0)
 		return ret;
@@ -935,9 +1072,8 @@ static int axienet_open(struct net_device *ndev)
 
 	phylink_start(lp->phylink);
 
-	/* Enable tasklets for Axi DMA error handling */
-	tasklet_init(&lp->dma_err_tasklet, axienet_dma_err_handler,
-		     (unsigned long) lp);
+	/* Enable worker thread for Axi DMA error handling */
+	INIT_WORK(&lp->dma_err_task, axienet_dma_err_handler);
 
 	/* Enable interrupts for Axi DMA Tx */
 	ret = request_irq(lp->tx_irq, axienet_tx_irq, IRQF_SHARED,
@@ -966,7 +1102,7 @@ err_rx_irq:
 err_tx_irq:
 	phylink_stop(lp->phylink);
 	phylink_disconnect_phy(lp->phylink);
-	tasklet_kill(&lp->dma_err_tasklet);
+	cancel_work_sync(&lp->dma_err_task);
 	dev_err(lp->dev, "request_irq() failed\n");
 	return ret;
 }
@@ -1025,7 +1161,7 @@ static int axienet_stop(struct net_device *ndev)
 	axienet_mdio_enable(lp);
 	mutex_unlock(&lp->mii_bus->mdio_lock);
 
-	tasklet_kill(&lp->dma_err_tasklet);
+	cancel_work_sync(&lp->dma_err_task);
 
 	if (lp->eth_irq > 0)
 		free_irq(lp->eth_irq, ndev);
@@ -1083,6 +1219,16 @@ static void axienet_poll_controller(struct net_device *ndev)
 }
 #endif
 
+static int axienet_ioctl(struct net_device *dev, struct ifreq *rq, int cmd)
+{
+	struct axienet_local *lp = netdev_priv(dev);
+
+	if (!netif_running(dev))
+		return -EINVAL;
+
+	return phylink_mii_ioctl(lp->phylink, rq, cmd);
+}
+
 static const struct net_device_ops axienet_netdev_ops = {
 	.ndo_open = axienet_open,
 	.ndo_stop = axienet_stop,
@@ -1090,6 +1236,7 @@ static const struct net_device_ops axienet_netdev_ops = {
 	.ndo_change_mtu	= axienet_change_mtu,
 	.ndo_set_mac_address = netdev_set_mac_address,
 	.ndo_validate_addr = eth_validate_addr,
+	.ndo_do_ioctl = axienet_ioctl,
 	.ndo_set_rx_mode = axienet_set_multicast_list,
 #ifdef CONFIG_NET_POLL_CONTROLLER
 	.ndo_poll_controller = axienet_poll_controller,
@@ -1170,10 +1317,6 @@ static void axienet_ethtools_get_regs(struct net_device *ndev,
 	data[20] = axienet_ior(lp, XAE_MDIO_MCR_OFFSET);
 	data[21] = axienet_ior(lp, XAE_MDIO_MWD_OFFSET);
 	data[22] = axienet_ior(lp, XAE_MDIO_MRD_OFFSET);
-	data[23] = axienet_ior(lp, XAE_MDIO_MIS_OFFSET);
-	data[24] = axienet_ior(lp, XAE_MDIO_MIP_OFFSET);
-	data[25] = axienet_ior(lp, XAE_MDIO_MIE_OFFSET);
-	data[26] = axienet_ior(lp, XAE_MDIO_MIC_OFFSET);
 	data[27] = axienet_ior(lp, XAE_UAW0_OFFSET);
 	data[28] = axienet_ior(lp, XAE_UAW1_OFFSET);
 	data[29] = axienet_ior(lp, XAE_FMI_OFFSET);
@@ -1309,27 +1452,6 @@ static int axienet_ethtools_set_coalesce(struct net_device *ndev,
 		return -EFAULT;
 	}
 
-	if ((ecoalesce->rx_coalesce_usecs) ||
-	    (ecoalesce->rx_coalesce_usecs_irq) ||
-	    (ecoalesce->rx_max_coalesced_frames_irq) ||
-	    (ecoalesce->tx_coalesce_usecs) ||
-	    (ecoalesce->tx_coalesce_usecs_irq) ||
-	    (ecoalesce->tx_max_coalesced_frames_irq) ||
-	    (ecoalesce->stats_block_coalesce_usecs) ||
-	    (ecoalesce->use_adaptive_rx_coalesce) ||
-	    (ecoalesce->use_adaptive_tx_coalesce) ||
-	    (ecoalesce->pkt_rate_low) ||
-	    (ecoalesce->rx_coalesce_usecs_low) ||
-	    (ecoalesce->rx_max_coalesced_frames_low) ||
-	    (ecoalesce->tx_coalesce_usecs_low) ||
-	    (ecoalesce->tx_max_coalesced_frames_low) ||
-	    (ecoalesce->pkt_rate_high) ||
-	    (ecoalesce->rx_coalesce_usecs_high) ||
-	    (ecoalesce->rx_max_coalesced_frames_high) ||
-	    (ecoalesce->tx_coalesce_usecs_high) ||
-	    (ecoalesce->tx_max_coalesced_frames_high) ||
-	    (ecoalesce->rate_sample_interval))
-		return -EOPNOTSUPP;
 	if (ecoalesce->rx_max_coalesced_frames)
 		lp->coalesce_count_rx = ecoalesce->rx_max_coalesced_frames;
 	if (ecoalesce->tx_max_coalesced_frames)
@@ -1357,6 +1479,7 @@ axienet_ethtools_set_link_ksettings(struct net_device *ndev,
 }
 
 static const struct ethtool_ops axienet_ethtool_ops = {
+	.supported_coalesce_params = ETHTOOL_COALESCE_MAX_FRAMES,
 	.get_drvinfo    = axienet_ethtools_get_drvinfo,
 	.get_regs_len   = axienet_ethtools_get_regs_len,
 	.get_regs       = axienet_ethtools_get_regs,
@@ -1441,6 +1564,22 @@ static void axienet_mac_an_restart(struct phylink_config *config)
 static void axienet_mac_config(struct phylink_config *config, unsigned int mode,
 			       const struct phylink_link_state *state)
 {
+	/* nothing meaningful to do */
+}
+
+static void axienet_mac_link_down(struct phylink_config *config,
+				  unsigned int mode,
+				  phy_interface_t interface)
+{
+	/* nothing meaningful to do */
+}
+
+static void axienet_mac_link_up(struct phylink_config *config,
+				struct phy_device *phy,
+				unsigned int mode, phy_interface_t interface,
+				int speed, int duplex,
+				bool tx_pause, bool rx_pause)
+{
 	struct net_device *ndev = to_net_dev(config->dev);
 	struct axienet_local *lp = netdev_priv(ndev);
 	u32 emmc_reg, fcc_reg;
@@ -1448,7 +1587,7 @@ static void axienet_mac_config(struct phylink_config *config, unsigned int mode,
 	emmc_reg = axienet_ior(lp, XAE_EMMC_OFFSET);
 	emmc_reg &= ~XAE_EMMC_LINKSPEED_MASK;
 
-	switch (state->speed) {
+	switch (speed) {
 	case SPEED_1000:
 		emmc_reg |= XAE_EMMC_LINKSPD_1000;
 		break;
@@ -1467,32 +1606,17 @@ static void axienet_mac_config(struct phylink_config *config, unsigned int mode,
 	axienet_iow(lp, XAE_EMMC_OFFSET, emmc_reg);
 
 	fcc_reg = axienet_ior(lp, XAE_FCC_OFFSET);
-	if (state->pause & MLO_PAUSE_TX)
+	if (tx_pause)
 		fcc_reg |= XAE_FCC_FCTX_MASK;
 	else
 		fcc_reg &= ~XAE_FCC_FCTX_MASK;
-	if (state->pause & MLO_PAUSE_RX)
+	if (rx_pause)
 		fcc_reg |= XAE_FCC_FCRX_MASK;
 	else
 		fcc_reg &= ~XAE_FCC_FCRX_MASK;
 	axienet_iow(lp, XAE_FCC_OFFSET, fcc_reg);
 }
 
-static void axienet_mac_link_down(struct phylink_config *config,
-				  unsigned int mode,
-				  phy_interface_t interface)
-{
-	/* nothing meaningful to do */
-}
-
-static void axienet_mac_link_up(struct phylink_config *config,
-				unsigned int mode,
-				phy_interface_t interface,
-				struct phy_device *phy)
-{
-	/* nothing meaningful to do */
-}
-
 static const struct phylink_mac_ops axienet_phylink_ops = {
 	.validate = axienet_validate,
 	.mac_pcs_get_state = axienet_mac_pcs_get_state,
@@ -1503,17 +1627,18 @@ static const struct phylink_mac_ops axienet_phylink_ops = {
 };
 
 /**
- * axienet_dma_err_handler - Tasklet handler for Axi DMA Error
- * @data:	Data passed
+ * axienet_dma_err_handler - Work queue task for Axi DMA Error
+ * @work:	pointer to work_struct
  *
  * Resets the Axi DMA and Axi Ethernet devices, and reconfigures the
  * Tx/Rx BDs.
  */
-static void axienet_dma_err_handler(unsigned long data)
+static void axienet_dma_err_handler(struct work_struct *work)
 {
 	u32 axienet_status;
 	u32 cr, i;
-	struct axienet_local *lp = (struct axienet_local *) data;
+	struct axienet_local *lp = container_of(work, struct axienet_local,
+						dma_err_task);
 	struct net_device *ndev = lp->ndev;
 	struct axidma_bd *cur_p;
 
@@ -1533,14 +1658,18 @@ static void axienet_dma_err_handler(unsigned long data)
 
 	for (i = 0; i < lp->tx_bd_num; i++) {
 		cur_p = &lp->tx_bd_v[i];
-		if (cur_p->phys)
-			dma_unmap_single(ndev->dev.parent, cur_p->phys,
+		if (cur_p->cntrl) {
+			dma_addr_t addr = desc_get_phys_addr(lp, cur_p);
+
+			dma_unmap_single(ndev->dev.parent, addr,
 					 (cur_p->cntrl &
 					  XAXIDMA_BD_CTRL_LENGTH_MASK),
 					 DMA_TO_DEVICE);
+		}
 		if (cur_p->skb)
 			dev_kfree_skb_irq(cur_p->skb);
 		cur_p->phys = 0;
+		cur_p->phys_msb = 0;
 		cur_p->cntrl = 0;
 		cur_p->status = 0;
 		cur_p->app0 = 0;
@@ -1594,18 +1723,18 @@ static void axienet_dma_err_handler(unsigned long data)
 	/* Populate the tail pointer and bring the Rx Axi DMA engine out of
 	 * halted state. This will make the Rx side ready for reception.
 	 */
-	axienet_dma_out32(lp, XAXIDMA_RX_CDESC_OFFSET, lp->rx_bd_p);
+	axienet_dma_out_addr(lp, XAXIDMA_RX_CDESC_OFFSET, lp->rx_bd_p);
 	cr = axienet_dma_in32(lp, XAXIDMA_RX_CR_OFFSET);
 	axienet_dma_out32(lp, XAXIDMA_RX_CR_OFFSET,
 			  cr | XAXIDMA_CR_RUNSTOP_MASK);
-	axienet_dma_out32(lp, XAXIDMA_RX_TDESC_OFFSET, lp->rx_bd_p +
-			  (sizeof(*lp->rx_bd_v) * (lp->rx_bd_num - 1)));
+	axienet_dma_out_addr(lp, XAXIDMA_RX_TDESC_OFFSET, lp->rx_bd_p +
+			     (sizeof(*lp->rx_bd_v) * (lp->rx_bd_num - 1)));
 
 	/* Write to the RS (Run-stop) bit in the Tx channel control register.
 	 * Tx channel is now ready to run. But only after we write to the
 	 * tail pointer register that the Tx channel will start transmitting
 	 */
-	axienet_dma_out32(lp, XAXIDMA_TX_CDESC_OFFSET, lp->tx_bd_p);
+	axienet_dma_out_addr(lp, XAXIDMA_TX_CDESC_OFFSET, lp->tx_bd_p);
 	cr = axienet_dma_in32(lp, XAXIDMA_TX_CR_OFFSET);
 	axienet_dma_out32(lp, XAXIDMA_TX_CR_OFFSET,
 			  cr | XAXIDMA_CR_RUNSTOP_MASK);
@@ -1651,6 +1780,7 @@ static int axienet_probe(struct platform_device *pdev)
 	struct net_device *ndev;
 	const void *mac_addr;
 	struct resource *ethres;
+	int addr_width = 32;
 	u32 value;
 
 	ndev = alloc_etherdev(sizeof(*lp));
@@ -1781,7 +1911,7 @@ static int axienet_probe(struct platform_device *pdev)
 		lp->rx_irq = irq_of_parse_and_map(np, 1);
 		lp->tx_irq = irq_of_parse_and_map(np, 0);
 		of_node_put(np);
-		lp->eth_irq = platform_get_irq(pdev, 0);
+		lp->eth_irq = platform_get_irq_optional(pdev, 0);
 	} else {
 		/* Check for these resources directly on the Ethernet node. */
 		struct resource *res = platform_get_resource(pdev,
@@ -1789,7 +1919,7 @@ static int axienet_probe(struct platform_device *pdev)
 		lp->dma_regs = devm_ioremap_resource(&pdev->dev, res);
 		lp->rx_irq = platform_get_irq(pdev, 1);
 		lp->tx_irq = platform_get_irq(pdev, 0);
-		lp->eth_irq = platform_get_irq(pdev, 2);
+		lp->eth_irq = platform_get_irq_optional(pdev, 2);
 	}
 	if (IS_ERR(lp->dma_regs)) {
 		dev_err(&pdev->dev, "could not map DMA regs\n");
@@ -1802,6 +1932,36 @@ static int axienet_probe(struct platform_device *pdev)
 		goto free_netdev;
 	}
 
+	/* Autodetect the need for 64-bit DMA pointers.
+	 * When the IP is configured for a bus width bigger than 32 bits,
+	 * writing the MSB registers is mandatory, even if they are all 0.
+	 * We can detect this case by writing all 1's to one such register
+	 * and see if that sticks: when the IP is configured for 32 bits
+	 * only, those registers are RES0.
+	 * Those MSB registers were introduced in IP v7.1, which we check first.
+	 */
+	if ((axienet_ior(lp, XAE_ID_OFFSET) >> 24) >= 0x9) {
+		void __iomem *desc = lp->dma_regs + XAXIDMA_TX_CDESC_OFFSET + 4;
+
+		iowrite32(0x0, desc);
+		if (ioread32(desc) == 0) {	/* sanity check */
+			iowrite32(0xffffffff, desc);
+			if (ioread32(desc) > 0) {
+				lp->features |= XAE_FEATURE_DMA_64BIT;
+				addr_width = 64;
+				dev_info(&pdev->dev,
+					 "autodetected 64-bit DMA range\n");
+			}
+			iowrite32(0x0, desc);
+		}
+	}
+
+	ret = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(addr_width));
+	if (ret) {
+		dev_err(&pdev->dev, "No suitable DMA available\n");
+		goto free_netdev;
+	}
+
 	/* Check for Ethernet core IRQ (optional) */
 	if (lp->eth_irq <= 0)
 		dev_info(&pdev->dev, "Ethernet core IRQ not defined\n");