Merge branch 'stmmac-mq-part3'

Joao Pinto says:

====================
net: stmmac: adding multiple buffers and routing

As agreed with David Miller, this patch-set is the third and last to enable
multiple queues in stmmac.

This third one focuses on:

a) Enable multiple buffering to the driver and queue independent data
b) Configuration of RX and TX queues' priority
c) Configuration of RX queues' routing
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Documentation/devicetree/bindings/net/stmmac.txt

@@ -83,6 +83,13 @@ Optional properties:
 			- snps,dcb-algorithm: Queue to be enabled as DCB
 			- snps,avb-algorithm: Queue to be enabled as AVB
 		- snps,map-to-dma-channel: Channel to map
+		- Specifiy specific packet routing:
+			- snps,route-avcp: AV Untagged Control packets
+			- snps,route-ptp: PTP Packets
+			- snps,route-dcbcp: DCB Control Packets
+			- snps,route-up: Untagged Packets
+			- snps,route-multi-broad: Multicast & Broadcast Packets
+		- snps,priority: RX queue priority (Range: 0x0 to 0xF)
 - Multiple TX Queues parameters: below the list of all the parameters to
 				 configure the multiple TX queues:
 	- snps,tx-queues-to-use: number of TX queues to be used in the driver
@@ -101,6 +108,7 @@ Optional properties:
 			- snps,idle_slope: unlock on WoL
 			- snps,high_credit: max write outstanding req. limit
 			- snps,low_credit: max read outstanding req. limit
+		- snps,priority: TX queue priority (Range: 0x0 to 0xF)
 Examples:
 
 	stmmac_axi_setup: stmmac-axi-config {
@@ -115,6 +123,7 @@ Examples:
 		queue0 {
 			snps,dcb-algorithm;
 			snps,map-to-dma-channel = <0x0>;
+			snps,priority = <0x0>;
 		};
 	};
 
@@ -124,6 +133,7 @@ Examples:
 		queue0 {
 			snps,weight = <0x10>;
 			snps,dcb-algorithm;
+			snps,priority = <0x0>;
 		};
 
 		queue1 {
@@ -132,6 +142,7 @@ Examples:
 			snps,idle_slope = <0x1000>;
 			snps,high_credit = <0x3E800>;
 			snps,low_credit = <0xFFC18000>;
+			snps,priority = <0x1>;
 		};
 	};
 

drivers/net/ethernet/stmicro/stmmac/chain_mode.c

@@ -26,12 +26,15 @@
 
 static int stmmac_jumbo_frm(void *p, struct sk_buff *skb, int csum)
 {
-	struct stmmac_priv *priv = (struct stmmac_priv *)p;
-	unsigned int entry = priv->cur_tx;
-	struct dma_desc *desc = priv->dma_tx + entry;
+	struct stmmac_tx_queue *tx_q = (struct stmmac_tx_queue *)p;
 	unsigned int nopaged_len = skb_headlen(skb);
+	struct stmmac_priv *priv = tx_q->priv_data;
+	unsigned int entry = tx_q->cur_tx;
 	unsigned int bmax, des2;
 	unsigned int i = 1, len;
+	struct dma_desc *desc;
+
+	desc = tx_q->dma_tx + entry;
 
 	if (priv->plat->enh_desc)
 		bmax = BUF_SIZE_8KiB;
@@ -45,16 +48,16 @@ static int stmmac_jumbo_frm(void *p, struct sk_buff *skb, int csum)
 	desc->des2 = cpu_to_le32(des2);
 	if (dma_mapping_error(priv->device, des2))
 		return -1;
-	priv->tx_skbuff_dma[entry].buf = des2;
-	priv->tx_skbuff_dma[entry].len = bmax;
+	tx_q->tx_skbuff_dma[entry].buf = des2;
+	tx_q->tx_skbuff_dma[entry].len = bmax;
 	/* do not close the descriptor and do not set own bit */
 	priv->hw->desc->prepare_tx_desc(desc, 1, bmax, csum, STMMAC_CHAIN_MODE,
 					0, false);
 
 	while (len != 0) {
-		priv->tx_skbuff[entry] = NULL;
+		tx_q->tx_skbuff[entry] = NULL;
 		entry = STMMAC_GET_ENTRY(entry, DMA_TX_SIZE);
-		desc = priv->dma_tx + entry;
+		desc = tx_q->dma_tx + entry;
 
 		if (len > bmax) {
 			des2 = dma_map_single(priv->device,
@@ -63,8 +66,8 @@ static int stmmac_jumbo_frm(void *p, struct sk_buff *skb, int csum)
 			desc->des2 = cpu_to_le32(des2);
 			if (dma_mapping_error(priv->device, des2))
 				return -1;
-			priv->tx_skbuff_dma[entry].buf = des2;
-			priv->tx_skbuff_dma[entry].len = bmax;
+			tx_q->tx_skbuff_dma[entry].buf = des2;
+			tx_q->tx_skbuff_dma[entry].len = bmax;
 			priv->hw->desc->prepare_tx_desc(desc, 0, bmax, csum,
 							STMMAC_CHAIN_MODE, 1,
 							false);
@@ -77,8 +80,8 @@ static int stmmac_jumbo_frm(void *p, struct sk_buff *skb, int csum)
 			desc->des2 = cpu_to_le32(des2);
 			if (dma_mapping_error(priv->device, des2))
 				return -1;
-			priv->tx_skbuff_dma[entry].buf = des2;
-			priv->tx_skbuff_dma[entry].len = len;
+			tx_q->tx_skbuff_dma[entry].buf = des2;
+			tx_q->tx_skbuff_dma[entry].len = len;
 			/* last descriptor can be set now */
 			priv->hw->desc->prepare_tx_desc(desc, 0, len, csum,
 							STMMAC_CHAIN_MODE, 1,
@@ -87,7 +90,7 @@ static int stmmac_jumbo_frm(void *p, struct sk_buff *skb, int csum)
 		}
 	}
 
-	priv->cur_tx = entry;
+	tx_q->cur_tx = entry;
 
 	return entry;
 }
@@ -136,32 +139,34 @@ static void stmmac_init_dma_chain(void *des, dma_addr_t phy_addr,
 
 static void stmmac_refill_desc3(void *priv_ptr, struct dma_desc *p)
 {
-	struct stmmac_priv *priv = (struct stmmac_priv *)priv_ptr;
+	struct stmmac_rx_queue *rx_q = (struct stmmac_rx_queue *)priv_ptr;
+	struct stmmac_priv *priv = rx_q->priv_data;
 
 	if (priv->hwts_rx_en && !priv->extend_desc)
 		/* NOTE: Device will overwrite des3 with timestamp value if
 		 * 1588-2002 time stamping is enabled, hence reinitialize it
 		 * to keep explicit chaining in the descriptor.
 		 */
-		p->des3 = cpu_to_le32((unsigned int)(priv->dma_rx_phy +
-				      (((priv->dirty_rx) + 1) %
+		p->des3 = cpu_to_le32((unsigned int)(rx_q->dma_rx_phy +
+				      (((rx_q->dirty_rx) + 1) %
 				       DMA_RX_SIZE) *
 				      sizeof(struct dma_desc)));
 }
 
 static void stmmac_clean_desc3(void *priv_ptr, struct dma_desc *p)
 {
-	struct stmmac_priv *priv = (struct stmmac_priv *)priv_ptr;
-	unsigned int entry = priv->dirty_tx;
+	struct stmmac_tx_queue *tx_q = (struct stmmac_tx_queue *)priv_ptr;
+	struct stmmac_priv *priv = tx_q->priv_data;
+	unsigned int entry = tx_q->dirty_tx;
 
-	if (priv->tx_skbuff_dma[entry].last_segment && !priv->extend_desc &&
+	if (tx_q->tx_skbuff_dma[entry].last_segment && !priv->extend_desc &&
 	    priv->hwts_tx_en)
 		/* NOTE: Device will overwrite des3 with timestamp value if
 		 * 1588-2002 time stamping is enabled, hence reinitialize it
 		 * to keep explicit chaining in the descriptor.
 		 */
-		p->des3 = cpu_to_le32((unsigned int)((priv->dma_tx_phy +
-				      ((priv->dirty_tx + 1) % DMA_TX_SIZE))
+		p->des3 = cpu_to_le32((unsigned int)((tx_q->dma_tx_phy +
+				      ((tx_q->dirty_tx + 1) % DMA_TX_SIZE))
 				      * sizeof(struct dma_desc)));
 }
 

drivers/net/ethernet/stmicro/stmmac/common.h

@@ -246,6 +246,15 @@ struct stmmac_extra_stats {
 #define STMMAC_TX_MAX_FRAMES	256
 #define STMMAC_TX_FRAMES	64
 
+/* Packets types */
+enum packets_types {
+	PACKET_AVCPQ = 0x1, /* AV Untagged Control packets */
+	PACKET_PTPQ = 0x2, /* PTP Packets */
+	PACKET_DCBCPQ = 0x3, /* DCB Control Packets */
+	PACKET_UPQ = 0x4, /* Untagged Packets */
+	PACKET_MCBCQ = 0x5, /* Multicast & Broadcast Packets */
+};
+
 /* Rx IPC status */
 enum rx_frame_status {
 	good_frame = 0x0,
@@ -469,6 +478,13 @@ struct stmmac_ops {
 	int (*rx_ipc)(struct mac_device_info *hw);
 	/* Enable RX Queues */
 	void (*rx_queue_enable)(struct mac_device_info *hw, u8 mode, u32 queue);
+	/* RX Queues Priority */
+	void (*rx_queue_prio)(struct mac_device_info *hw, u32 prio, u32 queue);
+	/* TX Queues Priority */
+	void (*tx_queue_prio)(struct mac_device_info *hw, u32 prio, u32 queue);
+	/* RX Queues Routing */
+	void (*rx_queue_routing)(struct mac_device_info *hw, u8 packet,
+				 u32 queue);
 	/* Program RX Algorithms */
 	void (*prog_mtl_rx_algorithms)(struct mac_device_info *hw, u32 rx_alg);
 	/* Program TX Algorithms */
@@ -577,6 +593,11 @@ struct mac_device_info {
 	unsigned int ps;
 };
 
+struct stmmac_rx_routing {
+	u32 reg_mask;
+	u32 reg_shift;
+};
+
 struct mac_device_info *dwmac1000_setup(void __iomem *ioaddr, int mcbins,
 					int perfect_uc_entries,
 					int *synopsys_id);

drivers/net/ethernet/stmicro/stmmac/dwmac4.h

@@ -22,7 +22,12 @@
 #define GMAC_HASH_TAB_32_63		0x00000014
 #define GMAC_RX_FLOW_CTRL		0x00000090
 #define GMAC_QX_TX_FLOW_CTRL(x)		(0x70 + x * 4)
+#define GMAC_TXQ_PRTY_MAP0		0x98
+#define GMAC_TXQ_PRTY_MAP1		0x9C
 #define GMAC_RXQ_CTRL0			0x000000a0
+#define GMAC_RXQ_CTRL1			0x000000a4
+#define GMAC_RXQ_CTRL2			0x000000a8
+#define GMAC_RXQ_CTRL3			0x000000ac
 #define GMAC_INT_STATUS			0x000000b0
 #define GMAC_INT_EN			0x000000b4
 #define GMAC_1US_TIC_COUNTER		0x000000dc
@@ -39,6 +44,22 @@
 #define GMAC_ADDR_HIGH(reg)		(0x300 + reg * 8)
 #define GMAC_ADDR_LOW(reg)		(0x304 + reg * 8)
 
+/* RX Queues Routing */
+#define GMAC_RXQCTRL_AVCPQ_MASK		GENMASK(2, 0)
+#define GMAC_RXQCTRL_AVCPQ_SHIFT	0
+#define GMAC_RXQCTRL_PTPQ_MASK		GENMASK(6, 4)
+#define GMAC_RXQCTRL_PTPQ_SHIFT		4
+#define GMAC_RXQCTRL_DCBCPQ_MASK	GENMASK(10, 8)
+#define GMAC_RXQCTRL_DCBCPQ_SHIFT	8
+#define GMAC_RXQCTRL_UPQ_MASK		GENMASK(14, 12)
+#define GMAC_RXQCTRL_UPQ_SHIFT		12
+#define GMAC_RXQCTRL_MCBCQ_MASK		GENMASK(18, 16)
+#define GMAC_RXQCTRL_MCBCQ_SHIFT	16
+#define GMAC_RXQCTRL_MCBCQEN		BIT(20)
+#define GMAC_RXQCTRL_MCBCQEN_SHIFT	20
+#define GMAC_RXQCTRL_TACPQE		BIT(21)
+#define GMAC_RXQCTRL_TACPQE_SHIFT	21
+
 /* MAC Packet Filtering */
 #define GMAC_PACKET_FILTER_PR		BIT(0)
 #define GMAC_PACKET_FILTER_HMC		BIT(2)
@@ -54,6 +75,14 @@
 /* MAC Flow Control RX */
 #define GMAC_RX_FLOW_CTRL_RFE		BIT(0)
 
+/* RX Queues Priorities */
+#define GMAC_RXQCTRL_PSRQX_MASK(x)	GENMASK(7 + ((x) * 8), 0 + ((x) * 8))
+#define GMAC_RXQCTRL_PSRQX_SHIFT(x)	((x) * 8)
+
+/* TX Queues Priorities */
+#define GMAC_TXQCTRL_PSTQX_MASK(x)	GENMASK(7 + ((x) * 8), 0 + ((x) * 8))
+#define GMAC_TXQCTRL_PSTQX_SHIFT(x)	((x) * 8)
+
 /* MAC Flow Control TX */
 #define GMAC_TX_FLOW_CTRL_TFE		BIT(1)
 #define GMAC_TX_FLOW_CTRL_PT_SHIFT	16

drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c

@@ -74,6 +74,74 @@ static void dwmac4_rx_queue_enable(struct mac_device_info *hw,
 	writel(value, ioaddr + GMAC_RXQ_CTRL0);
 }
 
+static void dwmac4_rx_queue_priority(struct mac_device_info *hw,
+				     u32 prio, u32 queue)
+{
+	void __iomem *ioaddr = hw->pcsr;
+	u32 base_register;
+	u32 value;
+
+	base_register = (queue < 4) ? GMAC_RXQ_CTRL2 : GMAC_RXQ_CTRL3;
+
+	value = readl(ioaddr + base_register);
+
+	value &= ~GMAC_RXQCTRL_PSRQX_MASK(queue);
+	value |= (prio << GMAC_RXQCTRL_PSRQX_SHIFT(queue)) &
+						GMAC_RXQCTRL_PSRQX_MASK(queue);
+	writel(value, ioaddr + base_register);
+}
+
+static void dwmac4_tx_queue_priority(struct mac_device_info *hw,
+				     u32 prio, u32 queue)
+{
+	void __iomem *ioaddr = hw->pcsr;
+	u32 base_register;
+	u32 value;
+
+	base_register = (queue < 4) ? GMAC_TXQ_PRTY_MAP0 : GMAC_TXQ_PRTY_MAP1;
+
+	value = readl(ioaddr + base_register);
+
+	value &= ~GMAC_TXQCTRL_PSTQX_MASK(queue);
+	value |= (prio << GMAC_TXQCTRL_PSTQX_SHIFT(queue)) &
+						GMAC_TXQCTRL_PSTQX_MASK(queue);
+
+	writel(value, ioaddr + base_register);
+}
+
+static void dwmac4_tx_queue_routing(struct mac_device_info *hw,
+				    u8 packet, u32 queue)
+{
+	void __iomem *ioaddr = hw->pcsr;
+	u32 value;
+
+	const struct stmmac_rx_routing route_possibilities[] = {
+		{ GMAC_RXQCTRL_AVCPQ_MASK, GMAC_RXQCTRL_AVCPQ_SHIFT },
+		{ GMAC_RXQCTRL_PTPQ_MASK, GMAC_RXQCTRL_PTPQ_SHIFT },
+		{ GMAC_RXQCTRL_DCBCPQ_MASK, GMAC_RXQCTRL_DCBCPQ_SHIFT },
+		{ GMAC_RXQCTRL_UPQ_MASK, GMAC_RXQCTRL_UPQ_SHIFT },
+		{ GMAC_RXQCTRL_MCBCQ_MASK, GMAC_RXQCTRL_MCBCQ_SHIFT },
+	};
+
+	value = readl(ioaddr + GMAC_RXQ_CTRL1);
+
+	/* routing configuration */
+	value &= ~route_possibilities[packet - 1].reg_mask;
+	value |= (queue << route_possibilities[packet-1].reg_shift) &
+		 route_possibilities[packet - 1].reg_mask;
+
+	/* some packets require extra ops */
+	if (packet == PACKET_AVCPQ) {
+		value &= ~GMAC_RXQCTRL_TACPQE;
+		value |= 0x1 << GMAC_RXQCTRL_TACPQE_SHIFT;
+	} else if (packet == PACKET_MCBCQ) {
+		value &= ~GMAC_RXQCTRL_MCBCQEN;
+		value |= 0x1 << GMAC_RXQCTRL_MCBCQEN_SHIFT;
+	}
+
+	writel(value, ioaddr + GMAC_RXQ_CTRL1);
+}
+
 static void dwmac4_prog_mtl_rx_algorithms(struct mac_device_info *hw,
 					  u32 rx_alg)
 {
@@ -603,6 +671,9 @@ static const struct stmmac_ops dwmac4_ops = {
 	.core_init = dwmac4_core_init,
 	.rx_ipc = dwmac4_rx_ipc_enable,
 	.rx_queue_enable = dwmac4_rx_queue_enable,
+	.rx_queue_prio = dwmac4_rx_queue_priority,
+	.tx_queue_prio = dwmac4_tx_queue_priority,
+	.rx_queue_routing = dwmac4_tx_queue_routing,
 	.prog_mtl_rx_algorithms = dwmac4_prog_mtl_rx_algorithms,
 	.prog_mtl_tx_algorithms = dwmac4_prog_mtl_tx_algorithms,
 	.set_mtl_tx_queue_weight = dwmac4_set_mtl_tx_queue_weight,

drivers/net/ethernet/stmicro/stmmac/ring_mode.c

@@ -26,16 +26,17 @@
 
 static int stmmac_jumbo_frm(void *p, struct sk_buff *skb, int csum)
 {
-	struct stmmac_priv *priv = (struct stmmac_priv *)p;
-	unsigned int entry = priv->cur_tx;
-	struct dma_desc *desc;
+	struct stmmac_tx_queue *tx_q = (struct stmmac_tx_queue *)p;
 	unsigned int nopaged_len = skb_headlen(skb);
+	struct stmmac_priv *priv = tx_q->priv_data;
+	unsigned int entry = tx_q->cur_tx;
 	unsigned int bmax, len, des2;
+	struct dma_desc *desc;
 
 	if (priv->extend_desc)
-		desc = (struct dma_desc *)(priv->dma_etx + entry);
+		desc = (struct dma_desc *)(tx_q->dma_etx + entry);
 	else
-		desc = priv->dma_tx + entry;
+		desc = tx_q->dma_tx + entry;
 
 	if (priv->plat->enh_desc)
 		bmax = BUF_SIZE_8KiB;
@@ -52,29 +53,29 @@ static int stmmac_jumbo_frm(void *p, struct sk_buff *skb, int csum)
 		if (dma_mapping_error(priv->device, des2))
 			return -1;
 
-		priv->tx_skbuff_dma[entry].buf = des2;
-		priv->tx_skbuff_dma[entry].len = bmax;
-		priv->tx_skbuff_dma[entry].is_jumbo = true;
+		tx_q->tx_skbuff_dma[entry].buf = des2;
+		tx_q->tx_skbuff_dma[entry].len = bmax;
+		tx_q->tx_skbuff_dma[entry].is_jumbo = true;
 
 		desc->des3 = cpu_to_le32(des2 + BUF_SIZE_4KiB);
 		priv->hw->desc->prepare_tx_desc(desc, 1, bmax, csum,
 						STMMAC_RING_MODE, 0, false);
-		priv->tx_skbuff[entry] = NULL;
+		tx_q->tx_skbuff[entry] = NULL;
 		entry = STMMAC_GET_ENTRY(entry, DMA_TX_SIZE);
 
 		if (priv->extend_desc)
-			desc = (struct dma_desc *)(priv->dma_etx + entry);
+			desc = (struct dma_desc *)(tx_q->dma_etx + entry);
 		else
-			desc = priv->dma_tx + entry;
+			desc = tx_q->dma_tx + entry;
 
 		des2 = dma_map_single(priv->device, skb->data + bmax, len,
 				      DMA_TO_DEVICE);
 		desc->des2 = cpu_to_le32(des2);
 		if (dma_mapping_error(priv->device, des2))
 			return -1;
-		priv->tx_skbuff_dma[entry].buf = des2;
-		priv->tx_skbuff_dma[entry].len = len;
-		priv->tx_skbuff_dma[entry].is_jumbo = true;
+		tx_q->tx_skbuff_dma[entry].buf = des2;
+		tx_q->tx_skbuff_dma[entry].len = len;
+		tx_q->tx_skbuff_dma[entry].is_jumbo = true;
 
 		desc->des3 = cpu_to_le32(des2 + BUF_SIZE_4KiB);
 		priv->hw->desc->prepare_tx_desc(desc, 0, len, csum,
@@ -85,15 +86,15 @@ static int stmmac_jumbo_frm(void *p, struct sk_buff *skb, int csum)
 		desc->des2 = cpu_to_le32(des2);
 		if (dma_mapping_error(priv->device, des2))
 			return -1;
-		priv->tx_skbuff_dma[entry].buf = des2;
-		priv->tx_skbuff_dma[entry].len = nopaged_len;
-		priv->tx_skbuff_dma[entry].is_jumbo = true;
+		tx_q->tx_skbuff_dma[entry].buf = des2;
+		tx_q->tx_skbuff_dma[entry].len = nopaged_len;
+		tx_q->tx_skbuff_dma[entry].is_jumbo = true;
 		desc->des3 = cpu_to_le32(des2 + BUF_SIZE_4KiB);
 		priv->hw->desc->prepare_tx_desc(desc, 1, nopaged_len, csum,
 						STMMAC_RING_MODE, 0, true);
 	}
 
-	priv->cur_tx = entry;
+	tx_q->cur_tx = entry;
 
 	return entry;
 }
@@ -125,12 +126,13 @@ static void stmmac_init_desc3(struct dma_desc *p)
 
 static void stmmac_clean_desc3(void *priv_ptr, struct dma_desc *p)
 {
-	struct stmmac_priv *priv = (struct stmmac_priv *)priv_ptr;
-	unsigned int entry = priv->dirty_tx;
+	struct stmmac_tx_queue *tx_q = (struct stmmac_tx_queue *)priv_ptr;
+	struct stmmac_priv *priv = tx_q->priv_data;
+	unsigned int entry = tx_q->dirty_tx;
 
 	/* des3 is only used for jumbo frames tx or time stamping */
-	if (unlikely(priv->tx_skbuff_dma[entry].is_jumbo ||
-		     (priv->tx_skbuff_dma[entry].last_segment &&
+	if (unlikely(tx_q->tx_skbuff_dma[entry].is_jumbo ||
+		     (tx_q->tx_skbuff_dma[entry].last_segment &&
 		      !priv->extend_desc && priv->hwts_tx_en)))
 		p->des3 = 0;
 }

drivers/net/ethernet/stmicro/stmmac/stmmac.h

@@ -46,6 +46,35 @@ struct stmmac_tx_info {
 	bool is_jumbo;
 };
 
+/* Frequently used values are kept adjacent for cache effect */
+struct stmmac_tx_queue {
+	u32 queue_index;
+	struct stmmac_priv *priv_data;
+	struct dma_extended_desc *dma_etx ____cacheline_aligned_in_smp;
+	struct dma_desc *dma_tx;
+	struct sk_buff **tx_skbuff;
+	struct stmmac_tx_info *tx_skbuff_dma;
+	unsigned int cur_tx;
+	unsigned int dirty_tx;
+	dma_addr_t dma_tx_phy;
+	u32 tx_tail_addr;
+};
+
+struct stmmac_rx_queue {
+	u32 queue_index;
+	struct stmmac_priv *priv_data;
+	struct dma_extended_desc *dma_erx;
+	struct dma_desc *dma_rx ____cacheline_aligned_in_smp;
+	struct sk_buff **rx_skbuff;
+	dma_addr_t *rx_skbuff_dma;
+	struct napi_struct napi ____cacheline_aligned_in_smp;
+	unsigned int cur_rx;
+	unsigned int dirty_rx;
+	u32 rx_zeroc_thresh;
+	dma_addr_t dma_rx_phy;
+	u32 rx_tail_addr;
+};
+
 struct stmmac_priv {
 	/* Frequently used values are kept adjacent for cache effect */
 	struct dma_extended_desc *dma_etx ____cacheline_aligned_in_smp;
@@ -56,28 +85,22 @@ struct stmmac_priv {
 	u32 tx_count_frames;
 	u32 tx_coal_frames;
 	u32 tx_coal_timer;
-	struct stmmac_tx_info *tx_skbuff_dma;
-	dma_addr_t dma_tx_phy;
 	int tx_coalesce;
 	int hwts_tx_en;
 	bool tx_path_in_lpi_mode;
 	struct timer_list txtimer;
 	bool tso;
 
-	struct dma_desc *dma_rx	____cacheline_aligned_in_smp;
-	struct dma_extended_desc *dma_erx;
-	struct sk_buff **rx_skbuff;
-	unsigned int cur_rx;
-	unsigned int dirty_rx;
+	/* TX Queue */
+	struct stmmac_tx_queue *tx_queue;
+
+	/* RX Queue */
+	struct stmmac_rx_queue *rx_queue;
+
 	unsigned int dma_buf_sz;
 	unsigned int rx_copybreak;
-	unsigned int rx_zeroc_thresh;
 	u32 rx_riwt;
 	int hwts_rx_en;
-	dma_addr_t *rx_skbuff_dma;
-	dma_addr_t dma_rx_phy;
-
-	struct napi_struct napi ____cacheline_aligned_in_smp;
 
 	void __iomem *ioaddr;
 	struct net_device *dev;
@@ -119,8 +142,6 @@ struct stmmac_priv {
 	spinlock_t ptp_lock;
 	void __iomem *mmcaddr;
 	void __iomem *ptpaddr;
-	u32 rx_tail_addr;
-	u32 tx_tail_addr;
 	u32 mss;
 
 #ifdef CONFIG_DEBUG_FS

drivers/net/ethernet/stmicro/stmmac/stmmac_pci.c

@@ -92,6 +92,13 @@ static void stmmac_default_data(struct plat_stmmacenet_data *plat)
 	/* Set default number of RX and TX queues to use */
 	plat->tx_queues_to_use = 1;
 	plat->rx_queues_to_use = 1;
+
+	/* Disable Priority config by default */
+	plat->tx_queues_cfg[0].use_prio = false;
+	plat->rx_queues_cfg[0].use_prio = false;
+
+	/* Disable RX queues routing by default */
+	plat->rx_queues_cfg[0].pkt_route = 0x0;
 }
 
 static int quark_default_data(struct plat_stmmacenet_data *plat,

drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c

@@ -182,6 +182,28 @@ static void stmmac_mtl_setup(struct platform_device *pdev,
 			plat->rx_queues_cfg[queue].chan = queue;
 		/* TODO: Dynamic mapping to be included in the future */
 
+		if (of_property_read_u32(q_node, "snps,priority",
+					&plat->rx_queues_cfg[queue].prio)) {
+			plat->rx_queues_cfg[queue].prio = 0;
+			plat->rx_queues_cfg[queue].use_prio = false;
+		} else {
+			plat->rx_queues_cfg[queue].use_prio = true;
+		}
+
+		/* RX queue specific packet type routing */
+		if (of_property_read_bool(q_node, "snps,route-avcp"))
+			plat->rx_queues_cfg[queue].pkt_route = PACKET_AVCPQ;
+		else if (of_property_read_bool(q_node, "snps,route-ptp"))
+			plat->rx_queues_cfg[queue].pkt_route = PACKET_PTPQ;
+		else if (of_property_read_bool(q_node, "snps,route-dcbcp"))
+			plat->rx_queues_cfg[queue].pkt_route = PACKET_DCBCPQ;
+		else if (of_property_read_bool(q_node, "snps,route-up"))
+			plat->rx_queues_cfg[queue].pkt_route = PACKET_UPQ;
+		else if (of_property_read_bool(q_node, "snps,route-multi-broad"))
+			plat->rx_queues_cfg[queue].pkt_route = PACKET_MCBCQ;
+		else
+			plat->rx_queues_cfg[queue].pkt_route = 0x0;
+
 		queue++;
 	}
 
@@ -235,6 +257,14 @@ static void stmmac_mtl_setup(struct platform_device *pdev,
 			plat->tx_queues_cfg[queue].mode_to_use = MTL_QUEUE_DCB;
 		}
 
+		if (of_property_read_u32(q_node, "snps,priority",
+					&plat->tx_queues_cfg[queue].prio)) {
+			plat->tx_queues_cfg[queue].prio = 0;
+			plat->tx_queues_cfg[queue].use_prio = false;
+		} else {
+			plat->tx_queues_cfg[queue].use_prio = true;
+		}
+
 		queue++;
 	}
 

include/linux/stmmac.h

@@ -127,6 +127,9 @@ struct stmmac_axi {
 struct stmmac_rxq_cfg {
 	u8 mode_to_use;
 	u8 chan;
+	u8 pkt_route;
+	bool use_prio;
+	u32 prio;
 };
 
 struct stmmac_txq_cfg {
@@ -137,6 +140,8 @@ struct stmmac_txq_cfg {
 	u32 idle_slope;
 	u32 high_credit;
 	u32 low_credit;
+	bool use_prio;
+	u32 prio;
 };
 
 struct plat_stmmacenet_data {