Merge branch 'amd-xgbe-next'

Tom Lendacky says:

====================
amd-xgbe: AMD XGBE driver updates 2014-11-04

The following series of patches includes functional updates to the
driver as well as some trivial changes for function renaming and
spelling fixes.

- Move channel and ring structure allocation into the device open path
- Rename the pre_xmit function to dev_xmit
- Explicitly use the u32 data type for the device descriptors
- Use page allocation for the receive buffers
- Add support for split header/payload receive
- Add support for per DMA channel interrupts
- Add support for receive side scaling (RSS)
- Add support for ethtool receive side scaling commands
- Fix the spelling of descriptors
- After a PCS reset, sync the PCS and PHY modes
- Add dependency on HAS_IOMEM to both the amd-xgbe and amd-xgbe-phy
  drivers

This patch series is based on net-next.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

11 files changed, 955 insertions, 277 deletions

diff --git a/Documentation/devicetree/bindings/net/amd-xgbe.txt b/Documentation/devicetree/bindings/net/amd-xgbe.txt
index 41354f730beb..26efd526d16c 100644
--- a/Documentation/devicetree/bindings/net/amd-xgbe.txt
+++ b/Documentation/devicetree/bindings/net/amd-xgbe.txt
@@ -7,7 +7,10 @@ Required properties:
    - PCS registers
 - interrupt-parent: Should be the phandle for the interrupt controller
   that services interrupts for this device
-- interrupts: Should contain the amd-xgbe interrupt
+- interrupts: Should contain the amd-xgbe interrupt(s). The first interrupt
+  listed is required and is the general device interrupt. If the optional
+  amd,per-channel-interrupt property is specified, then one additional
+  interrupt for each DMA channel supported by the device should be specified
 - clocks:
    - DMA clock for the amd-xgbe device (used for calculating the
      correct Rx interrupt watchdog timer value on a DMA channel
@@ -23,6 +26,9 @@ Optional properties:
 - mac-address: mac address to be assigned to the device. Can be overridden
   by UEFI.
 - dma-coherent: Present if dma operations are coherent
+- amd,per-channel-interrupt: Indicates that Rx and Tx complete will generate
+  a unique interrupt for each DMA channel - this requires an additional
+  interrupt be configured for each DMA channel
 
 Example:
 	xgbe@e0700000 {
@@ -30,7 +36,9 @@ Example:
 		reg = <0 0xe0700000 0 0x80000>,
 		      <0 0xe0780000 0 0x80000>;
 		interrupt-parent = <&gic>;
-		interrupts = <0 325 4>;
+		interrupts = <0 325 4>,
+			     <0 326 1>, <0 327 1>, <0 328 1>, <0 329 1>;
+		amd,per-channel-interrupt;
 		clocks = <&xgbe_dma_clk>, <&xgbe_ptp_clk>;
 		clock-names = "dma_clk", "ptp_clk";
 		phy-handle = <&phy>;
diff --git a/drivers/net/ethernet/amd/Kconfig b/drivers/net/ethernet/amd/Kconfig
index 8319c99331b0..7a5e4aa5415e 100644
--- a/drivers/net/ethernet/amd/Kconfig
+++ b/drivers/net/ethernet/amd/Kconfig
@@ -179,7 +179,7 @@ config SUNLANCE
 
 config AMD_XGBE
 	tristate "AMD 10GbE Ethernet driver"
-	depends on OF_NET
+	depends on OF_NET && HAS_IOMEM
 	select PHYLIB
 	select AMD_XGBE_PHY
 	select BITREVERSE
diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-common.h b/drivers/net/ethernet/amd/xgbe/xgbe-common.h
index caade30820d5..2fe8fc71fe01 100644
--- a/drivers/net/ethernet/amd/xgbe/xgbe-common.h
+++ b/drivers/net/ethernet/amd/xgbe/xgbe-common.h
@@ -207,6 +207,8 @@
 /* DMA channel register entry bit positions and sizes */
 #define DMA_CH_CR_PBLX8_INDEX		16
 #define DMA_CH_CR_PBLX8_WIDTH		1
+#define DMA_CH_CR_SPH_INDEX		24
+#define DMA_CH_CR_SPH_WIDTH		1
 #define DMA_CH_IER_AIE_INDEX		15
 #define DMA_CH_IER_AIE_WIDTH		1
 #define DMA_CH_IER_FBEE_INDEX		12
@@ -306,6 +308,9 @@
 #define MAC_MACA0LR			0x0304
 #define MAC_MACA1HR			0x0308
 #define MAC_MACA1LR			0x030c
+#define MAC_RSSCR			0x0c80
+#define MAC_RSSAR			0x0c88
+#define MAC_RSSDR			0x0c8c
 #define MAC_TSCR			0x0d00
 #define MAC_SSIR			0x0d04
 #define MAC_STSR			0x0d08
@@ -429,6 +434,8 @@
 #define MAC_RCR_CST_WIDTH		1
 #define MAC_RCR_DCRCC_INDEX		3
 #define MAC_RCR_DCRCC_WIDTH		1
+#define MAC_RCR_HDSMS_INDEX		12
+#define MAC_RCR_HDSMS_WIDTH		3
 #define MAC_RCR_IPC_INDEX		9
 #define MAC_RCR_IPC_WIDTH		1
 #define MAC_RCR_JE_INDEX		8
@@ -445,6 +452,24 @@
 #define MAC_RFCR_UP_WIDTH		1
 #define MAC_RQC0R_RXQ0EN_INDEX		0
 #define MAC_RQC0R_RXQ0EN_WIDTH		2
+#define MAC_RSSAR_ADDRT_INDEX		2
+#define MAC_RSSAR_ADDRT_WIDTH		1
+#define MAC_RSSAR_CT_INDEX		1
+#define MAC_RSSAR_CT_WIDTH		1
+#define MAC_RSSAR_OB_INDEX		0
+#define MAC_RSSAR_OB_WIDTH		1
+#define MAC_RSSAR_RSSIA_INDEX		8
+#define MAC_RSSAR_RSSIA_WIDTH		8
+#define MAC_RSSCR_IP2TE_INDEX		1
+#define MAC_RSSCR_IP2TE_WIDTH		1
+#define MAC_RSSCR_RSSE_INDEX		0
+#define MAC_RSSCR_RSSE_WIDTH		1
+#define MAC_RSSCR_TCP4TE_INDEX		2
+#define MAC_RSSCR_TCP4TE_WIDTH		1
+#define MAC_RSSCR_UDP4TE_INDEX		3
+#define MAC_RSSCR_UDP4TE_WIDTH		1
+#define MAC_RSSDR_DMCH_INDEX		0
+#define MAC_RSSDR_DMCH_WIDTH		4
 #define MAC_SSIR_SNSINC_INDEX		8
 #define MAC_SSIR_SNSINC_WIDTH		8
 #define MAC_SSIR_SSINC_INDEX		16
@@ -844,9 +869,13 @@
 #define RX_PACKET_ATTRIBUTES_CONTEXT_WIDTH	1
 #define RX_PACKET_ATTRIBUTES_RX_TSTAMP_INDEX	5
 #define RX_PACKET_ATTRIBUTES_RX_TSTAMP_WIDTH	1
+#define RX_PACKET_ATTRIBUTES_RSS_HASH_INDEX	6
+#define RX_PACKET_ATTRIBUTES_RSS_HASH_WIDTH	1
 
 #define RX_NORMAL_DESC0_OVT_INDEX		0
 #define RX_NORMAL_DESC0_OVT_WIDTH		16
+#define RX_NORMAL_DESC2_HL_INDEX		0
+#define RX_NORMAL_DESC2_HL_WIDTH		10
 #define RX_NORMAL_DESC3_CDA_INDEX		27
 #define RX_NORMAL_DESC3_CDA_WIDTH		1
 #define RX_NORMAL_DESC3_CTXT_INDEX		30
@@ -855,14 +884,27 @@
 #define RX_NORMAL_DESC3_ES_WIDTH		1
 #define RX_NORMAL_DESC3_ETLT_INDEX		16
 #define RX_NORMAL_DESC3_ETLT_WIDTH		4
+#define RX_NORMAL_DESC3_FD_INDEX		29
+#define RX_NORMAL_DESC3_FD_WIDTH		1
 #define RX_NORMAL_DESC3_INTE_INDEX		30
 #define RX_NORMAL_DESC3_INTE_WIDTH		1
+#define RX_NORMAL_DESC3_L34T_INDEX		20
+#define RX_NORMAL_DESC3_L34T_WIDTH		4
 #define RX_NORMAL_DESC3_LD_INDEX		28
 #define RX_NORMAL_DESC3_LD_WIDTH		1
 #define RX_NORMAL_DESC3_OWN_INDEX		31
 #define RX_NORMAL_DESC3_OWN_WIDTH		1
 #define RX_NORMAL_DESC3_PL_INDEX		0
 #define RX_NORMAL_DESC3_PL_WIDTH		14
+#define RX_NORMAL_DESC3_RSV_INDEX		26
+#define RX_NORMAL_DESC3_RSV_WIDTH		1
+
+#define RX_DESC3_L34T_IPV4_TCP			1
+#define RX_DESC3_L34T_IPV4_UDP			2
+#define RX_DESC3_L34T_IPV4_ICMP			3
+#define RX_DESC3_L34T_IPV6_TCP			9
+#define RX_DESC3_L34T_IPV6_UDP			10
+#define RX_DESC3_L34T_IPV6_ICMP			11
 
 #define RX_CONTEXT_DESC3_TSA_INDEX		4
 #define RX_CONTEXT_DESC3_TSA_WIDTH		1
diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-desc.c b/drivers/net/ethernet/amd/xgbe/xgbe-desc.c
index 6fc5da01437d..e6b9f54b9697 100644
--- a/drivers/net/ethernet/amd/xgbe/xgbe-desc.c
+++ b/drivers/net/ethernet/amd/xgbe/xgbe-desc.c
@@ -117,7 +117,7 @@
 #include "xgbe.h"
 #include "xgbe-common.h"
 
-static void xgbe_unmap_skb(struct xgbe_prv_data *, struct xgbe_ring_data *);
+static void xgbe_unmap_rdata(struct xgbe_prv_data *, struct xgbe_ring_data *);
 
 static void xgbe_free_ring(struct xgbe_prv_data *pdata,
 			   struct xgbe_ring *ring)
@@ -131,13 +131,35 @@ static void xgbe_free_ring(struct xgbe_prv_data *pdata,
 	if (ring->rdata) {
 		for (i = 0; i < ring->rdesc_count; i++) {
 			rdata = XGBE_GET_DESC_DATA(ring, i);
-			xgbe_unmap_skb(pdata, rdata);
+			xgbe_unmap_rdata(pdata, rdata);
 		}
 
 		kfree(ring->rdata);
 		ring->rdata = NULL;
 	}
 
+	if (ring->rx_hdr_pa.pages) {
+		dma_unmap_page(pdata->dev, ring->rx_hdr_pa.pages_dma,
+			       ring->rx_hdr_pa.pages_len, DMA_FROM_DEVICE);
+		put_page(ring->rx_hdr_pa.pages);
+
+		ring->rx_hdr_pa.pages = NULL;
+		ring->rx_hdr_pa.pages_len = 0;
+		ring->rx_hdr_pa.pages_offset = 0;
+		ring->rx_hdr_pa.pages_dma = 0;
+	}
+
+	if (ring->rx_buf_pa.pages) {
+		dma_unmap_page(pdata->dev, ring->rx_buf_pa.pages_dma,
+			       ring->rx_buf_pa.pages_len, DMA_FROM_DEVICE);
+		put_page(ring->rx_buf_pa.pages);
+
+		ring->rx_buf_pa.pages = NULL;
+		ring->rx_buf_pa.pages_len = 0;
+		ring->rx_buf_pa.pages_offset = 0;
+		ring->rx_buf_pa.pages_dma = 0;
+	}
+
 	if (ring->rdesc) {
 		dma_free_coherent(pdata->dev,
 				  (sizeof(struct xgbe_ring_desc) *
@@ -233,6 +255,96 @@ err_ring:
 	return ret;
 }
 
+static int xgbe_alloc_pages(struct xgbe_prv_data *pdata,
+			    struct xgbe_page_alloc *pa, gfp_t gfp, int order)
+{
+	struct page *pages = NULL;
+	dma_addr_t pages_dma;
+	int ret;
+
+	/* Try to obtain pages, decreasing order if necessary */
+	gfp |= __GFP_COLD | __GFP_COMP;
+	while (order >= 0) {
+		pages = alloc_pages(gfp, order);
+		if (pages)
+			break;
+
+		order--;
+	}
+	if (!pages)
+		return -ENOMEM;
+
+	/* Map the pages */
+	pages_dma = dma_map_page(pdata->dev, pages, 0,
+				 PAGE_SIZE << order, DMA_FROM_DEVICE);
+	ret = dma_mapping_error(pdata->dev, pages_dma);
+	if (ret) {
+		put_page(pages);
+		return ret;
+	}
+
+	pa->pages = pages;
+	pa->pages_len = PAGE_SIZE << order;
+	pa->pages_offset = 0;
+	pa->pages_dma = pages_dma;
+
+	return 0;
+}
+
+static void xgbe_set_buffer_data(struct xgbe_buffer_data *bd,
+				 struct xgbe_page_alloc *pa,
+				 unsigned int len)
+{
+	get_page(pa->pages);
+	bd->pa = *pa;
+
+	bd->dma = pa->pages_dma + pa->pages_offset;
+	bd->dma_len = len;
+
+	pa->pages_offset += len;
+	if ((pa->pages_offset + len) > pa->pages_len) {
+		/* This data descriptor is responsible for unmapping page(s) */
+		bd->pa_unmap = *pa;
+
+		/* Get a new allocation next time */
+		pa->pages = NULL;
+		pa->pages_len = 0;
+		pa->pages_offset = 0;
+		pa->pages_dma = 0;
+	}
+}
+
+static int xgbe_map_rx_buffer(struct xgbe_prv_data *pdata,
+			      struct xgbe_ring *ring,
+			      struct xgbe_ring_data *rdata)
+{
+	int order, ret;
+
+	if (!ring->rx_hdr_pa.pages) {
+		ret = xgbe_alloc_pages(pdata, &ring->rx_hdr_pa, GFP_ATOMIC, 0);
+		if (ret)
+			return ret;
+	}
+
+	if (!ring->rx_buf_pa.pages) {
+		order = max_t(int, PAGE_ALLOC_COSTLY_ORDER - 1, 0);
+		ret = xgbe_alloc_pages(pdata, &ring->rx_buf_pa, GFP_ATOMIC,
+				       order);
+		if (ret)
+			return ret;
+	}
+
+	/* Set up the header page info */
+	xgbe_set_buffer_data(&rdata->rx_hdr, &ring->rx_hdr_pa,
+			     XGBE_SKB_ALLOC_SIZE);
+
+	/* Set up the buffer page info */
+	xgbe_set_buffer_data(&rdata->rx_buf, &ring->rx_buf_pa,
+			     pdata->rx_buf_size);
+
+	return 0;
+}
+
 static void xgbe_wrapper_tx_descriptor_init(struct xgbe_prv_data *pdata)
 {
 	struct xgbe_hw_if *hw_if = &pdata->hw_if;
@@ -281,8 +393,7 @@ static void xgbe_wrapper_rx_descriptor_init(struct xgbe_prv_data *pdata)
 	struct xgbe_ring *ring;
 	struct xgbe_ring_desc *rdesc;
 	struct xgbe_ring_data *rdata;
-	dma_addr_t rdesc_dma, skb_dma;
-	struct sk_buff *skb = NULL;
+	dma_addr_t rdesc_dma;
 	unsigned int i, j;
 
 	DBGPR("-->xgbe_wrapper_rx_descriptor_init\n");
@@ -302,22 +413,8 @@ static void xgbe_wrapper_rx_descriptor_init(struct xgbe_prv_data *pdata)
 			rdata->rdesc = rdesc;
 			rdata->rdesc_dma = rdesc_dma;
 
-			/* Allocate skb & assign to each rdesc */
-			skb = dev_alloc_skb(pdata->rx_buf_size);
-			if (skb == NULL)
-				break;
-			skb_dma = dma_map_single(pdata->dev, skb->data,
-						 pdata->rx_buf_size,
-						 DMA_FROM_DEVICE);
-			if (dma_mapping_error(pdata->dev, skb_dma)) {
-				netdev_alert(pdata->netdev,
-					     "failed to do the dma map\n");
-				dev_kfree_skb_any(skb);
+			if (xgbe_map_rx_buffer(pdata, ring, rdata))
 				break;
-			}
-			rdata->skb = skb;
-			rdata->skb_dma = skb_dma;
-			rdata->skb_dma_len = pdata->rx_buf_size;
 
 			rdesc++;
 			rdesc_dma += sizeof(struct xgbe_ring_desc);
@@ -334,8 +431,8 @@ static void xgbe_wrapper_rx_descriptor_init(struct xgbe_prv_data *pdata)
 	DBGPR("<--xgbe_wrapper_rx_descriptor_init\n");
 }
 
-static void xgbe_unmap_skb(struct xgbe_prv_data *pdata,
-			   struct xgbe_ring_data *rdata)
+static void xgbe_unmap_rdata(struct xgbe_prv_data *pdata,
+			     struct xgbe_ring_data *rdata)
 {
 	if (rdata->skb_dma) {
 		if (rdata->mapped_as_page) {
@@ -354,6 +451,29 @@ static void xgbe_unmap_skb(struct xgbe_prv_data *pdata,
 		rdata->skb = NULL;
 	}
 
+	if (rdata->rx_hdr.pa.pages)
+		put_page(rdata->rx_hdr.pa.pages);
+
+	if (rdata->rx_hdr.pa_unmap.pages) {
+		dma_unmap_page(pdata->dev, rdata->rx_hdr.pa_unmap.pages_dma,
+			       rdata->rx_hdr.pa_unmap.pages_len,
+			       DMA_FROM_DEVICE);
+		put_page(rdata->rx_hdr.pa_unmap.pages);
+	}
+
+	if (rdata->rx_buf.pa.pages)
+		put_page(rdata->rx_buf.pa.pages);
+
+	if (rdata->rx_buf.pa_unmap.pages) {
+		dma_unmap_page(pdata->dev, rdata->rx_buf.pa_unmap.pages_dma,
+			       rdata->rx_buf.pa_unmap.pages_len,
+			       DMA_FROM_DEVICE);
+		put_page(rdata->rx_buf.pa_unmap.pages);
+	}
+
+	memset(&rdata->rx_hdr, 0, sizeof(rdata->rx_hdr));
+	memset(&rdata->rx_buf, 0, sizeof(rdata->rx_buf));
+
 	rdata->tso_header = 0;
 	rdata->len = 0;
 	rdata->interrupt = 0;
@@ -494,7 +614,7 @@ static int xgbe_map_tx_skb(struct xgbe_channel *channel, struct sk_buff *skb)
 err_out:
 	while (start_index < cur_index) {
 		rdata = XGBE_GET_DESC_DATA(ring, start_index++);
-		xgbe_unmap_skb(pdata, rdata);
+		xgbe_unmap_rdata(pdata, rdata);
 	}
 
 	DBGPR("<--xgbe_map_tx_skb: count=0\n");
@@ -502,40 +622,25 @@ err_out:
 	return 0;
 }
 
-static void xgbe_realloc_skb(struct xgbe_channel *channel)
+static void xgbe_realloc_rx_buffer(struct xgbe_channel *channel)
 {
 	struct xgbe_prv_data *pdata = channel->pdata;
 	struct xgbe_hw_if *hw_if = &pdata->hw_if;
 	struct xgbe_ring *ring = channel->rx_ring;
 	struct xgbe_ring_data *rdata;
-	struct sk_buff *skb = NULL;
-	dma_addr_t skb_dma;
 	int i;
 
-	DBGPR("-->xgbe_realloc_skb: rx_ring->rx.realloc_index = %u\n",
+	DBGPR("-->xgbe_realloc_rx_buffer: rx_ring->rx.realloc_index = %u\n",
 	      ring->rx.realloc_index);
 
 	for (i = 0; i < ring->dirty; i++) {
 		rdata = XGBE_GET_DESC_DATA(ring, ring->rx.realloc_index);
 
 		/* Reset rdata values */
-		xgbe_unmap_skb(pdata, rdata);
+		xgbe_unmap_rdata(pdata, rdata);
 
-		/* Allocate skb & assign to each rdesc */
-		skb = dev_alloc_skb(pdata->rx_buf_size);
-		if (skb == NULL)
+		if (xgbe_map_rx_buffer(pdata, ring, rdata))
 			break;
-		skb_dma = dma_map_single(pdata->dev, skb->data,
-					 pdata->rx_buf_size, DMA_FROM_DEVICE);
-		if (dma_mapping_error(pdata->dev, skb_dma)) {
-			netdev_alert(pdata->netdev,
-				     "failed to do the dma map\n");
-			dev_kfree_skb_any(skb);
-			break;
-		}
-		rdata->skb = skb;
-		rdata->skb_dma = skb_dma;
-		rdata->skb_dma_len = pdata->rx_buf_size;
 
 		hw_if->rx_desc_reset(rdata);
 
@@ -543,7 +648,7 @@ static void xgbe_realloc_skb(struct xgbe_channel *channel)
 	}
 	ring->dirty = 0;
 
-	DBGPR("<--xgbe_realloc_skb\n");
+	DBGPR("<--xgbe_realloc_rx_buffer\n");
 }
 
 void xgbe_init_function_ptrs_desc(struct xgbe_desc_if *desc_if)
@@ -553,8 +658,8 @@ void xgbe_init_function_ptrs_desc(struct xgbe_desc_if *desc_if)
 	desc_if->alloc_ring_resources = xgbe_alloc_ring_resources;
 	desc_if->free_ring_resources = xgbe_free_ring_resources;
 	desc_if->map_tx_skb = xgbe_map_tx_skb;
-	desc_if->realloc_skb = xgbe_realloc_skb;
-	desc_if->unmap_skb = xgbe_unmap_skb;
+	desc_if->realloc_rx_buffer = xgbe_realloc_rx_buffer;
+	desc_if->unmap_rdata = xgbe_unmap_rdata;
 	desc_if->wrapper_tx_desc_init = xgbe_wrapper_tx_descriptor_init;
 	desc_if->wrapper_rx_desc_init = xgbe_wrapper_rx_descriptor_init;
 
diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-dev.c b/drivers/net/ethernet/amd/xgbe/xgbe-dev.c
index 9da3a03e8c07..7daa2cd9af5f 100644
--- a/drivers/net/ethernet/amd/xgbe/xgbe-dev.c
+++ b/drivers/net/ethernet/amd/xgbe/xgbe-dev.c
@@ -335,6 +335,161 @@ static void xgbe_config_tso_mode(struct xgbe_prv_data *pdata)
 	}
 }
 
+static void xgbe_config_sph_mode(struct xgbe_prv_data *pdata)
+{
+	struct xgbe_channel *channel;
+	unsigned int i;
+
+	channel = pdata->channel;
+	for (i = 0; i < pdata->channel_count; i++, channel++) {
+		if (!channel->rx_ring)
+			break;
+
+		XGMAC_DMA_IOWRITE_BITS(channel, DMA_CH_CR, SPH, 1);
+	}
+
+	XGMAC_IOWRITE_BITS(pdata, MAC_RCR, HDSMS, XGBE_SPH_HDSMS_SIZE);
+}
+
+static int xgbe_write_rss_reg(struct xgbe_prv_data *pdata, unsigned int type,
+			      unsigned int index, unsigned int val)
+{
+	unsigned int wait;
+	int ret = 0;
+
+	mutex_lock(&pdata->rss_mutex);
+
+	if (XGMAC_IOREAD_BITS(pdata, MAC_RSSAR, OB)) {
+		ret = -EBUSY;
+		goto unlock;
+	}
+
+	XGMAC_IOWRITE(pdata, MAC_RSSDR, val);
+
+	XGMAC_IOWRITE_BITS(pdata, MAC_RSSAR, RSSIA, index);
+	XGMAC_IOWRITE_BITS(pdata, MAC_RSSAR, ADDRT, type);
+	XGMAC_IOWRITE_BITS(pdata, MAC_RSSAR, CT, 0);
+	XGMAC_IOWRITE_BITS(pdata, MAC_RSSAR, OB, 1);
+
+	wait = 1000;
+	while (wait--) {
+		if (!XGMAC_IOREAD_BITS(pdata, MAC_RSSAR, OB))
+			goto unlock;
+
+		usleep_range(1000, 1500);
+	}
+
+	ret = -EBUSY;
+
+unlock:
+	mutex_unlock(&pdata->rss_mutex);
+
+	return ret;
+}
+
+static int xgbe_write_rss_hash_key(struct xgbe_prv_data *pdata)
+{
+	unsigned int key_regs = sizeof(pdata->rss_key) / sizeof(u32);
+	unsigned int *key = (unsigned int *)&pdata->rss_key;
+	int ret;
+
+	while (key_regs--) {
+		ret = xgbe_write_rss_reg(pdata, XGBE_RSS_HASH_KEY_TYPE,
+					 key_regs, *key++);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int xgbe_write_rss_lookup_table(struct xgbe_prv_data *pdata)
+{
+	unsigned int i;
+	int ret;
+
+	for (i = 0; i < ARRAY_SIZE(pdata->rss_table); i++) {
+		ret = xgbe_write_rss_reg(pdata,
+					 XGBE_RSS_LOOKUP_TABLE_TYPE, i,
+					 pdata->rss_table[i]);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int xgbe_set_rss_hash_key(struct xgbe_prv_data *pdata, const u8 *key)
+{
+	memcpy(pdata->rss_key, key, sizeof(pdata->rss_key));
+
+	return xgbe_write_rss_hash_key(pdata);
+}
+
+static int xgbe_set_rss_lookup_table(struct xgbe_prv_data *pdata,
+				     const u32 *table)
+{
+	unsigned int i;
+
+	for (i = 0; i < ARRAY_SIZE(pdata->rss_table); i++)
+		XGMAC_SET_BITS(pdata->rss_table[i], MAC_RSSDR, DMCH, table[i]);
+
+	return xgbe_write_rss_lookup_table(pdata);
+}
+
+static int xgbe_enable_rss(struct xgbe_prv_data *pdata)
+{
+	int ret;
+
+	if (!pdata->hw_feat.rss)
+		return -EOPNOTSUPP;
+
+	/* Program the hash key */
+	ret = xgbe_write_rss_hash_key(pdata);
+	if (ret)
+		return ret;
+
+	/* Program the lookup table */
+	ret = xgbe_write_rss_lookup_table(pdata);
+	if (ret)
+		return ret;
+
+	/* Set the RSS options */
+	XGMAC_IOWRITE(pdata, MAC_RSSCR, pdata->rss_options);
+
+	/* Enable RSS */
+	XGMAC_IOWRITE_BITS(pdata, MAC_RSSCR, RSSE, 1);
+
+	return 0;
+}
+
+static int xgbe_disable_rss(struct xgbe_prv_data *pdata)
+{
+	if (!pdata->hw_feat.rss)
+		return -EOPNOTSUPP;
+
+	XGMAC_IOWRITE_BITS(pdata, MAC_RSSCR, RSSE, 0);
+
+	return 0;
+}
+
+static void xgbe_config_rss(struct xgbe_prv_data *pdata)
+{
+	int ret;
+
+	if (!pdata->hw_feat.rss)
+		return;
+
+	if (pdata->netdev->features & NETIF_F_RXHASH)
+		ret = xgbe_enable_rss(pdata);
+	else
+		ret = xgbe_disable_rss(pdata);
+
+	if (ret)
+		netdev_err(pdata->netdev,
+			   "error configuring RSS, RSS disabled\n");
+}
+
 static int xgbe_disable_tx_flow_control(struct xgbe_prv_data *pdata)
 {
 	unsigned int max_q_count, q_count;
@@ -465,17 +620,21 @@ static void xgbe_enable_dma_interrupts(struct xgbe_prv_data *pdata)
 
 		if (channel->tx_ring) {
 			/* Enable the following Tx interrupts
-			 *   TIE  - Transmit Interrupt Enable (unless polling)
+			 *   TIE  - Transmit Interrupt Enable (unless using
+			 *          per channel interrupts)
 			 */
-			XGMAC_SET_BITS(dma_ch_ier, DMA_CH_IER, TIE, 1);
+			if (!pdata->per_channel_irq)
+				XGMAC_SET_BITS(dma_ch_ier, DMA_CH_IER, TIE, 1);
 		}
 		if (channel->rx_ring) {
 			/* Enable following Rx interrupts
 			 *   RBUE - Receive Buffer Unavailable Enable
-			 *   RIE  - Receive Interrupt Enable
+			 *   RIE  - Receive Interrupt Enable (unless using
+			 *          per channel interrupts)
 			 */
 			XGMAC_SET_BITS(dma_ch_ier, DMA_CH_IER, RBUE, 1);
-			XGMAC_SET_BITS(dma_ch_ier, DMA_CH_IER, RIE, 1);
+			if (!pdata->per_channel_irq)
+				XGMAC_SET_BITS(dma_ch_ier, DMA_CH_IER, RIE, 1);
 		}
 
 		XGMAC_DMA_IOWRITE(channel, DMA_CH_IER, dma_ch_ier);
@@ -880,13 +1039,15 @@ static void xgbe_tx_desc_reset(struct xgbe_ring_data *rdata)
 	rdesc->desc1 = 0;
 	rdesc->desc2 = 0;
 	rdesc->desc3 = 0;
+
+	/* Make sure ownership is written to the descriptor */
+	wmb();
 }
 
 static void xgbe_tx_desc_init(struct xgbe_channel *channel)
 {
 	struct xgbe_ring *ring = channel->tx_ring;
 	struct xgbe_ring_data *rdata;
-	struct xgbe_ring_desc *rdesc;
 	int i;
 	int start_index = ring->cur;
 
@@ -895,26 +1056,11 @@ static void xgbe_tx_desc_init(struct xgbe_channel *channel)
 	/* Initialze all descriptors */
 	for (i = 0; i < ring->rdesc_count; i++) {
 		rdata = XGBE_GET_DESC_DATA(ring, i);
-		rdesc = rdata->rdesc;
 
-		/* Initialize Tx descriptor
-		 *   Set buffer 1 (lo) address to zero
-		 *   Set buffer 1 (hi) address to zero
-		 *   Reset all other control bits (IC, TTSE, B2L & B1L)
-		 *   Reset all other control bits (OWN, CTXT, FD, LD, CPC, CIC,
-		 *     etc)
-		 */
-		rdesc->desc0 = 0;
-		rdesc->desc1 = 0;
-		rdesc->desc2 = 0;
-		rdesc->desc3 = 0;
+		/* Initialize Tx descriptor */
+		xgbe_tx_desc_reset(rdata);
 	}
 
-	/* Make sure everything is written to the descriptor(s) before
-	 * telling the device about them
-	 */
-	wmb();
-
 	/* Update the total number of Tx descriptors */
 	XGMAC_DMA_IOWRITE(channel, DMA_CH_TDRLR, ring->rdesc_count - 1);
 
@@ -933,19 +1079,19 @@ static void xgbe_rx_desc_reset(struct xgbe_ring_data *rdata)
 	struct xgbe_ring_desc *rdesc = rdata->rdesc;
 
 	/* Reset the Rx descriptor
-	 *   Set buffer 1 (lo) address to dma address (lo)
-	 *   Set buffer 1 (hi) address to dma address (hi)
-	 *   Set buffer 2 (lo) address to zero
-	 *   Set buffer 2 (hi) address to zero and set control bits
-	 *     OWN and INTE
+	 *   Set buffer 1 (lo) address to header dma address (lo)
+	 *   Set buffer 1 (hi) address to header dma address (hi)
+	 *   Set buffer 2 (lo) address to buffer dma address (lo)
+	 *   Set buffer 2 (hi) address to buffer dma address (hi) and
+	 *     set control bits OWN and INTE
 	 */
-	rdesc->desc0 = cpu_to_le32(lower_32_bits(rdata->skb_dma));
-	rdesc->desc1 = cpu_to_le32(upper_32_bits(rdata->skb_dma));
-	rdesc->desc2 = 0;
+	rdesc->desc0 = cpu_to_le32(lower_32_bits(rdata->rx_hdr.dma));
+	rdesc->desc1 = cpu_to_le32(upper_32_bits(rdata->rx_hdr.dma));
+	rdesc->desc2 = cpu_to_le32(lower_32_bits(rdata->rx_buf.dma));
+	rdesc->desc3 = cpu_to_le32(upper_32_bits(rdata->rx_buf.dma));
 
-	rdesc->desc3 = 0;
-	if (rdata->interrupt)
-		XGMAC_SET_BITS_LE(rdesc->desc3, RX_NORMAL_DESC3, INTE, 1);
+	XGMAC_SET_BITS_LE(rdesc->desc3, RX_NORMAL_DESC3, INTE,
+			  rdata->interrupt ? 1 : 0);
 
 	/* Since the Rx DMA engine is likely running, make sure everything
 	 * is written to the descriptor(s) before setting the OWN bit
@@ -964,7 +1110,6 @@ static void xgbe_rx_desc_init(struct xgbe_channel *channel)
 	struct xgbe_prv_data *pdata = channel->pdata;
 	struct xgbe_ring *ring = channel->rx_ring;
 	struct xgbe_ring_data *rdata;
-	struct xgbe_ring_desc *rdesc;
 	unsigned int start_index = ring->cur;
 	unsigned int rx_coalesce, rx_frames;
 	unsigned int i;
@@ -977,34 +1122,16 @@ static void xgbe_rx_desc_init(struct xgbe_channel *channel)
 	/* Initialize all descriptors */
 	for (i = 0; i < ring->rdesc_count; i++) {
 		rdata = XGBE_GET_DESC_DATA(ring, i);
-		rdesc = rdata->rdesc;
 
-		/* Initialize Rx descriptor
-		 *   Set buffer 1 (lo) address to dma address (lo)
-		 *   Set buffer 1 (hi) address to dma address (hi)
-		 *   Set buffer 2 (lo) address to zero
-		 *   Set buffer 2 (hi) address to zero and set control
-		 *     bits OWN and INTE appropriateley
-		 */
-		rdesc->desc0 = cpu_to_le32(lower_32_bits(rdata->skb_dma));
-		rdesc->desc1 = cpu_to_le32(upper_32_bits(rdata->skb_dma));
-		rdesc->desc2 = 0;
-		rdesc->desc3 = 0;
-		XGMAC_SET_BITS_LE(rdesc->desc3, RX_NORMAL_DESC3, OWN, 1);
-		XGMAC_SET_BITS_LE(rdesc->desc3, RX_NORMAL_DESC3, INTE, 1);
-		rdata->interrupt = 1;
-		if (rx_coalesce && (!rx_frames || ((i + 1) % rx_frames))) {
-			/* Clear interrupt on completion bit */
-			XGMAC_SET_BITS_LE(rdesc->desc3, RX_NORMAL_DESC3, INTE,
-					  0);
+		/* Set interrupt on completion bit as appropriate */
+		if (rx_coalesce && (!rx_frames || ((i + 1) % rx_frames)))
 			rdata->interrupt = 0;
-		}
-	}
+		else
+			rdata->interrupt = 1;
 
-	/* Make sure everything is written to the descriptors before
-	 * telling the device about them
-	 */
-	wmb();
+		/* Initialize Rx descriptor */
+		xgbe_rx_desc_reset(rdata);
+	}
 
 	/* Update the total number of Rx descriptors */
 	XGMAC_DMA_IOWRITE(channel, DMA_CH_RDRLR, ring->rdesc_count - 1);
@@ -1198,7 +1325,7 @@ static void xgbe_config_dcb_pfc(struct xgbe_prv_data *pdata)
 	xgbe_config_flow_control(pdata);
 }
 
-static void xgbe_pre_xmit(struct xgbe_channel *channel)
+static void xgbe_dev_xmit(struct xgbe_channel *channel)
 {
 	struct xgbe_prv_data *pdata = channel->pdata;
 	struct xgbe_ring *ring = channel->tx_ring;
@@ -1211,7 +1338,7 @@ static void xgbe_pre_xmit(struct xgbe_channel *channel)
 	int start_index = ring->cur;
 	int i;
 
-	DBGPR("-->xgbe_pre_xmit\n");
+	DBGPR("-->xgbe_dev_xmit\n");
 
 	csum = XGMAC_GET_BITS(packet->attributes, TX_PACKET_ATTRIBUTES,
 			      CSUM_ENABLE);
@@ -1410,7 +1537,7 @@ static void xgbe_pre_xmit(struct xgbe_channel *channel)
 	      channel->name, start_index & (ring->rdesc_count - 1),
 	      (ring->cur - 1) & (ring->rdesc_count - 1));
 
-	DBGPR("<--xgbe_pre_xmit\n");
+	DBGPR("<--xgbe_dev_xmit\n");
 }
 
 static int xgbe_dev_read(struct xgbe_channel *channel)
@@ -1420,7 +1547,7 @@ static int xgbe_dev_read(struct xgbe_channel *channel)
 	struct xgbe_ring_desc *rdesc;
 	struct xgbe_packet_data *packet = &ring->packet_data;
 	struct net_device *netdev = channel->pdata->netdev;
-	unsigned int err, etlt;
+	unsigned int err, etlt, l34t;
 
 	DBGPR("-->xgbe_dev_read: cur = %d\n", ring->cur);
 
@@ -1454,6 +1581,31 @@ static int xgbe_dev_read(struct xgbe_channel *channel)
 		XGMAC_SET_BITS(packet->attributes, RX_PACKET_ATTRIBUTES,
 			       CONTEXT_NEXT, 1);
 
+	/* Get the header length */
+	if (XGMAC_GET_BITS_LE(rdesc->desc3, RX_NORMAL_DESC3, FD))
+		rdata->hdr_len = XGMAC_GET_BITS_LE(rdesc->desc2,
+						   RX_NORMAL_DESC2, HL);
+
+	/* Get the RSS hash */
+	if (XGMAC_GET_BITS_LE(rdesc->desc3, RX_NORMAL_DESC3, RSV)) {
+		XGMAC_SET_BITS(packet->attributes, RX_PACKET_ATTRIBUTES,
+			       RSS_HASH, 1);
+
+		packet->rss_hash = le32_to_cpu(rdesc->desc1);
+
+		l34t = XGMAC_GET_BITS_LE(rdesc->desc3, RX_NORMAL_DESC3, L34T);
+		switch (l34t) {
+		case RX_DESC3_L34T_IPV4_TCP:
+		case RX_DESC3_L34T_IPV4_UDP:
+		case RX_DESC3_L34T_IPV6_TCP:
+		case RX_DESC3_L34T_IPV6_UDP:
+			packet->rss_hash_type = PKT_HASH_TYPE_L4;
+
+		default:
+			packet->rss_hash_type = PKT_HASH_TYPE_L3;
+		}
+	}
+
 	/* Get the packet length */
 	rdata->len = XGMAC_GET_BITS_LE(rdesc->desc3, RX_NORMAL_DESC3, PL);
 
@@ -2485,6 +2637,8 @@ static int xgbe_init(struct xgbe_prv_data *pdata)
 	xgbe_config_tx_coalesce(pdata);
 	xgbe_config_rx_buffer_size(pdata);
 	xgbe_config_tso_mode(pdata);
+	xgbe_config_sph_mode(pdata);
+	xgbe_config_rss(pdata);
 	desc_if->wrapper_tx_desc_init(pdata);
 	desc_if->wrapper_rx_desc_init(pdata);
 	xgbe_enable_dma_interrupts(pdata);
@@ -2561,7 +2715,7 @@ void xgbe_init_function_ptrs_dev(struct xgbe_hw_if *hw_if)
 	hw_if->powerup_rx = xgbe_powerup_rx;
 	hw_if->powerdown_rx = xgbe_powerdown_rx;
 
-	hw_if->pre_xmit = xgbe_pre_xmit;
+	hw_if->dev_xmit = xgbe_dev_xmit;
 	hw_if->dev_read = xgbe_dev_read;
 	hw_if->enable_int = xgbe_enable_int;
 	hw_if->disable_int = xgbe_disable_int;
@@ -2620,5 +2774,11 @@ void xgbe_init_function_ptrs_dev(struct xgbe_hw_if *hw_if)
 	hw_if->config_dcb_tc = xgbe_config_dcb_tc;
 	hw_if->config_dcb_pfc = xgbe_config_dcb_pfc;
 
+	/* For Receive Side Scaling */
+	hw_if->enable_rss = xgbe_enable_rss;
+	hw_if->disable_rss = xgbe_disable_rss;
+	hw_if->set_rss_hash_key = xgbe_set_rss_hash_key;
+	hw_if->set_rss_lookup_table = xgbe_set_rss_lookup_table;
+
 	DBGPR("<--xgbe_init_function_ptrs\n");
 }
diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-drv.c b/drivers/net/ethernet/amd/xgbe/xgbe-drv.c
index 2349ea970255..ced9f52eb45b 100644
--- a/drivers/net/ethernet/amd/xgbe/xgbe-drv.c
+++ b/drivers/net/ethernet/amd/xgbe/xgbe-drv.c
@@ -114,6 +114,7 @@
  *     THE POSSIBILITY OF SUCH DAMAGE.
  */
 
+#include <linux/platform_device.h>
 #include <linux/spinlock.h>
 #include <linux/tcp.h>
 #include <linux/if_vlan.h>
@@ -126,9 +127,99 @@
 #include "xgbe.h"
 #include "xgbe-common.h"
 
-static int xgbe_poll(struct napi_struct *, int);
+static int xgbe_one_poll(struct napi_struct *, int);
+static int xgbe_all_poll(struct napi_struct *, int);
 static void xgbe_set_rx_mode(struct net_device *);
 
+static int xgbe_alloc_channels(struct xgbe_prv_data *pdata)
+{
+	struct xgbe_channel *channel_mem, *channel;
+	struct xgbe_ring *tx_ring, *rx_ring;
+	unsigned int count, i;
+	int ret = -ENOMEM;
+
+	count = max_t(unsigned int, pdata->tx_ring_count, pdata->rx_ring_count);
+
+	channel_mem = kcalloc(count, sizeof(struct xgbe_channel), GFP_KERNEL);
+	if (!channel_mem)
+		goto err_channel;
+
+	tx_ring = kcalloc(pdata->tx_ring_count, sizeof(struct xgbe_ring),
+			  GFP_KERNEL);
+	if (!tx_ring)
+		goto err_tx_ring;
+
+	rx_ring = kcalloc(pdata->rx_ring_count, sizeof(struct xgbe_ring),
+			  GFP_KERNEL);
+	if (!rx_ring)
+		goto err_rx_ring;
+
+	for (i = 0, channel = channel_mem; i < count; i++, channel++) {
+		snprintf(channel->name, sizeof(channel->name), "channel-%d", i);
+		channel->pdata = pdata;
+		channel->queue_index = i;
+		channel->dma_regs = pdata->xgmac_regs + DMA_CH_BASE +
+				    (DMA_CH_INC * i);
+
+		if (pdata->per_channel_irq) {
+			/* Get the DMA interrupt (offset 1) */
+			ret = platform_get_irq(pdata->pdev, i + 1);
+			if (ret < 0) {
+				netdev_err(pdata->netdev,
+					   "platform_get_irq %u failed\n",
+					   i + 1);
+				goto err_irq;
+			}
+
+			channel->dma_irq = ret;
+		}
+
+		if (i < pdata->tx_ring_count) {
+			spin_lock_init(&tx_ring->lock);
+			channel->tx_ring = tx_ring++;
+		}
+
+		if (i < pdata->rx_ring_count) {
+			spin_lock_init(&rx_ring->lock);
+			channel->rx_ring = rx_ring++;
+		}
+
+		DBGPR("  %s: queue=%u, dma_regs=%p, dma_irq=%d, tx=%p, rx=%p\n",
+		      channel->name, channel->queue_index, channel->dma_regs,
+		      channel->dma_irq, channel->tx_ring, channel->rx_ring);
+	}
+
+	pdata->channel = channel_mem;
+	pdata->channel_count = count;
+
+	return 0;
+
+err_irq:
+	kfree(rx_ring);
+
+err_rx_ring:
+	kfree(tx_ring);
+
+err_tx_ring:
+	kfree(channel_mem);
+
+err_channel:
+	return ret;
+}
+
+static void xgbe_free_channels(struct xgbe_prv_data *pdata)
+{
+	if (!pdata->channel)
+		return;
+
+	kfree(pdata->channel->rx_ring);
+	kfree(pdata->channel->tx_ring);
+	kfree(pdata->channel);
+
+	pdata->channel = NULL;
+	pdata->channel_count = 0;
+}
+
 static inline unsigned int xgbe_tx_avail_desc(struct xgbe_ring *ring)
 {
 	return (ring->rdesc_count - (ring->cur - ring->dirty));
@@ -144,8 +235,8 @@ static int xgbe_calc_rx_buf_size(struct net_device *netdev, unsigned int mtu)
 	}
 
 	rx_buf_size = mtu + ETH_HLEN + ETH_FCS_LEN + VLAN_HLEN;
-	if (rx_buf_size < XGBE_RX_MIN_BUF_SIZE)
-		rx_buf_size = XGBE_RX_MIN_BUF_SIZE;
+	rx_buf_size = clamp_val(rx_buf_size, XGBE_RX_MIN_BUF_SIZE, PAGE_SIZE);
+
 	rx_buf_size = (rx_buf_size + XGBE_RX_BUF_ALIGN - 1) &
 		      ~(XGBE_RX_BUF_ALIGN - 1);
 
@@ -213,11 +304,7 @@ static irqr