[U-Boot] [RFC/RESEND 21/22] net: add thunderx vnic drivers

4 Mar 2019

Note we move eth init to early init and disable it in later legacy in order
to make sure its initialized in time for the pci NIC.
Signed-off-by: Tim Harvey tharvey@gateworks.com
---
 arch/arm/include/asm/arch-thunderx/fdt.h      |   12 +
 .../include/asm/arch-thunderx/thunderx_vnic.h |   15 +
 arch/arm/mach-thunderx/fdt.c                  |  170 +-
 board/cavium/thunderx/thunderx.c              |   30 +-
 configs/thunderx_81xx_defconfig               |    7 +-
 drivers/net/Kconfig                           |   34 +
 drivers/net/Makefile                          |    1 +
 drivers/net/cavium/Makefile                   |    8 +
 drivers/net/cavium/nic.h                      |  569 ++++++
 drivers/net/cavium/nic_main.c                 |  795 +++++++++
 drivers/net/cavium/nic_reg.h                  |  228 +++
 drivers/net/cavium/nicvf_main.c               |  553 ++++++
 drivers/net/cavium/nicvf_queues.c             | 1123 ++++++++++++
 drivers/net/cavium/nicvf_queues.h             |  364 ++++
 drivers/net/cavium/q_struct.h                 |  692 ++++++++
 drivers/net/cavium/thunder_bgx.c              | 1529 +++++++++++++++++
 drivers/net/cavium/thunder_bgx.h              |  259 +++
 drivers/net/cavium/thunder_xcv.c              |  190 ++
 drivers/net/cavium/thunderx_smi.c             |  388 +++++
 include/configs/thunderx_81xx.h               |   11 +-
 net/eth_legacy.c                              |    3 +
 21 files changed, 6973 insertions(+), 8 deletions(-)
 create mode 100644 arch/arm/include/asm/arch-thunderx/fdt.h
 create mode 100644 arch/arm/include/asm/arch-thunderx/thunderx_vnic.h
 create mode 100644 drivers/net/cavium/Makefile
 create mode 100644 drivers/net/cavium/nic.h
 create mode 100644 drivers/net/cavium/nic_main.c
 create mode 100644 drivers/net/cavium/nic_reg.h
 create mode 100644 drivers/net/cavium/nicvf_main.c
 create mode 100644 drivers/net/cavium/nicvf_queues.c
 create mode 100644 drivers/net/cavium/nicvf_queues.h
 create mode 100644 drivers/net/cavium/q_struct.h
 create mode 100644 drivers/net/cavium/thunder_bgx.c
 create mode 100644 drivers/net/cavium/thunder_bgx.h
 create mode 100644 drivers/net/cavium/thunder_xcv.c
 create mode 100644 drivers/net/cavium/thunderx_smi.c

diff --git a/arch/arm/include/asm/arch-thunderx/fdt.h b/arch/arm/include/asm/arch-thunderx/fdt.h
new file mode 100644
index 0000000000..7f61e7694e
--- /dev/null
+++ b/arch/arm/include/asm/arch-thunderx/fdt.h
@@ -0,0 +1,12 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright 2018 (C) Cavium Inc.
+ */
+#ifndef __THUNDERX_FDT_H__
+#define __THUNDERX_FDT_H__
+
+void thunderx_parse_phy_info(void);
+void thunderx_parse_mac_addr(void);
+
+#endif
+
diff --git a/arch/arm/include/asm/arch-thunderx/thunderx_vnic.h b/arch/arm/include/asm/arch-thunderx/thunderx_vnic.h
new file mode 100644
index 0000000000..f1e372c61e
--- /dev/null
+++ b/arch/arm/include/asm/arch-thunderx/thunderx_vnic.h
@@ -0,0 +1,15 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright 2018 (C) Cavium Inc.
+ */
+#ifndef VNIC_H_
+#define VNIC_H_
+
+int nicvf_initialize(struct udevice *dev, int vf_num);
+int thunderx_bgx_initialize(unsigned int bgx_idx, unsigned int node);
+void bgx_get_count(int node, int *bgx_count);
+int bgx_get_lmac_count(int node, int bgx_idx);
+void bgx_set_board_info(int bgx_id, int *mdio_bus, int *phy_addr,
+			bool *autoneg_dis, bool *lmac_reg, bool *lmac_enable);
+
+#endif /* VNIC_H_ */
diff --git a/arch/arm/mach-thunderx/fdt.c b/arch/arm/mach-thunderx/fdt.c
index 31f1128e9f..c03b0ec939 100644
--- a/arch/arm/mach-thunderx/fdt.c
+++ b/arch/arm/mach-thunderx/fdt.c
@@ -12,6 +12,175 @@
 /* From lowlevel_init.S */
 extern unsigned long fdt_base_addr;
+#if defined(CONFIG_THUNDERX_VNIC)
+#include <asm/arch-thunderx/thunderx_vnic.h>
+
+#define MAX_LMAC_PER_BGX 4
+
+DECLARE_GLOBAL_DATA_PTR;
+
+/**
+ * Parse PHY info from device-tree and pass along via bgx_set_board_info()
+ *
+ * The Cavium ThunderX device-trees RGX/BGX nodes are child nodes of the
+ * mrml_bridge node alongside the mdio-nexus MDIO node.
+ *
+ * There can be one RGMII MAC (rgx0) and up to MAX_LMAC_PER_BGX BGX MACs
+ * (bgx<0..MAX_LMAC_PER_BGX-1>).
+ *
+ * Each MAC can contain multiple logical MAC's (LMAC), for example a QSGMII
+ * PHY has 4 LMACs. These LMAC's are children of the bgx/rgx node and can
+ * contain the following properties:
+ *   reg (required) - LMAC instance
+ *   qlm-mode (optional) - CNX81XX QLM phy mode (ie qsgmii, rgmii)
+ *   phy-handle (required) - points to PHY node
+ *   local-mac-address (optional) - MAC address which can be filled in by
+ *                                  firmware.
+ *
+ * These details will be parsed from the device-tree and passed into the
+ * VNIC driver bgx_set_board_info() to configure it.
+ */
+static int thunderx_parse_phy(int bgx_id)
+{
+	const void *fdt = gd->fdt_blob;
+	int offset, node, lmacid = 0;
+	int phy_offset, phandle;
+	int bdknode;
+	const u32 *val;
+	const char *str;
+	char name[24];
+	int len, eth_id = 0;
+	int subnode, i;
+	int addr[MAX_LMAC_PER_BGX] = { [0 ... MAX_LMAC_PER_BGX - 1] = -1};
+	int bus[MAX_LMAC_PER_BGX] = { [0 ... MAX_LMAC_PER_BGX - 1] = -1};
+	bool aneg_dis[MAX_LMAC_PER_BGX] = { [0 ... MAX_LMAC_PER_BGX - 1] = 0};
+	bool lmac_reg[MAX_LMAC_PER_BGX] = { [0 ... MAX_LMAC_PER_BGX - 1] = 0};
+	bool lmac_en[MAX_LMAC_PER_BGX] = { [0 ... MAX_LMAC_PER_BGX - 1] = 0};
+
+	bdknode = fdt_path_offset(fdt, "/cavium,bdk");
+	if (bdknode < 0) {
+		printf("%s: /cavium,bdk is missing from device tree. "
+		       "LMAC's will be disabled\n", __func__);
+	}
+
+	offset = fdt_node_offset_by_compatible(fdt, -1, "pci-bridge");
+	if (offset < 0)
+		return -ENODEV;
+
+	/* find bgx/rgx node */
+	snprintf(name, sizeof(name), "bgx%d", bgx_id);
+	node = fdt_subnode_offset(fdt, offset, name);
+	if (node < 0) {
+		snprintf(name, sizeof(name), "rgx0");
+		node = fdt_subnode_offset(fdt, offset, name);
+		if (node < 0)
+			return -ENODEV;
+	}
+	debug("%s: %s found at %d\n", __func__, name, node);
+
+	/* interate through the LMAC subnodes */
+	fdt_for_each_subnode(subnode, fdt, node) {
+		/* reg required to enable LMAC */
+		if (fdt_getprop(fdt, subnode, "reg", &len))
+			lmac_reg[lmacid] = 1;
+
+		/* get bus/addr encoded from phy 'reg' prop */
+		phy_offset = -1;
+		phandle = 0;
+		val = fdt_getprop(fdt, subnode, "phy-handle", &len);
+		if (val)
+			phandle = fdt32_to_cpu(*val);
+		if (phandle)
+			phy_offset = fdt_node_offset_by_phandle(fdt, phandle);
+		if (phy_offset > 0) {
+			val = fdt_getprop(fdt, phy_offset, "reg", NULL);
+			bus[lmacid] = (fdt32_to_cpu(*val) & (1 << 7)) ? 1 : 0;
+			addr[lmacid] = fdt32_to_cpu(*val);
+		}
+
+		/* check for autonegotiation property */
+		if (fdt_getprop(fdt, subnode, "cavium,disable-autonegotiation",
+				&len))
+			aneg_dis[lmacid] = 1;
+
+		eth_id++;
+		lmacid++;
+	}
+
+	/* determine if LMAC is enabled via bdk prop */
+	if (bdknode) {
+		for (i = 0; i < MAX_LMAC_PER_BGX; i++) {
+			snprintf(name, sizeof(name), "BGX-ENABLE.N0.BGX%d.P%d",
+				 bgx_id, i);
+			str = fdt_getprop(fdt, bdknode, name, &len);
+			if (str)
+				lmac_en[i] = simple_strtol(str, NULL, 10);
+		}
+	}
+
+	bgx_set_board_info(bgx_id, bus, addr, aneg_dis, lmac_reg, lmac_en);
+
+	return 0;
+}
+
+void thunderx_parse_phy_info(void)
+{
+	int bgx_id;
+
+	for (bgx_id = 0; bgx_id < CONFIG_MAX_BGX_PER_NODE; bgx_id++)
+		thunderx_parse_phy(bgx_id);
+}
+
+/* Parse MAC addrs from device-tree
+ *
+ * The Cavium ThunderX device-trees RGX/BGX MAC's are child nodes of the
+ * mrml_bridge node alongside the mdio bus named 'rgx<n>' and 'bgx<n>'.
+ * These nodes have a phy-handle pointing to the mdio bus and a
+ * local-mac-address property set by firmware.
+ */
+void thunderx_parse_mac_addr(void)
+{
+	const uint8_t *mac_addr;
+	const void *fdt = gd->fdt_blob;
+	int subnode;
+	int len, eth_id = 0;
+	int offset = 0, node, bgx_id = 0, lmacid = 0;
+	char name[16], env[16];
+
+	debug("%s\n", __func__);
+	offset = fdt_node_offset_by_compatible(fdt, -1, "pci-bridge");
+	if (offset < 0)
+		return;
+
+	for (bgx_id = 0; bgx_id < CONFIG_MAX_BGX_PER_NODE; bgx_id++) {
+		/* find bgx/rgx node */
+		snprintf(name, sizeof(name), "bgx%d", bgx_id);
+		node = fdt_subnode_offset(fdt, offset, name);
+		if (node < 0) {
+			snprintf(name, sizeof(name), "rgx0");
+			node = fdt_subnode_offset(fdt, offset, name);
+			if (node < 0)
+				continue;
+		}
+		debug("%s found\n", name);
+
+		fdt_for_each_subnode(subnode, fdt, node) {
+			sprintf(env, eth_id ? "eth%daddr" : "ethaddr",
+				eth_id);
+			mac_addr = fdt_getprop(fdt, subnode,
+					       "local-mac-address", &len);
+			if (mac_addr) {
+				debug("got %s MAC:%pM\n", name, mac_addr);
+				eth_env_set_enetaddr(env, mac_addr);
+			}
+			eth_id++;
+			lmacid++;
+		}
+		lmacid = 0;
+	}
+}
+#endif // if defined(CONFIG_THUNDERX_VNIC)
+
 /**
  * If the firmware passed a device tree use it for U-Boot
  *
@@ -47,4 +216,3 @@ int ft_board_setup(void *blob, bd_t *bd)
return ret;
 }
-
diff --git a/board/cavium/thunderx/thunderx.c b/board/cavium/thunderx/thunderx.c
index 28cf2aee22..9f9ac7ea10 100644
--- a/board/cavium/thunderx/thunderx.c
+++ b/board/cavium/thunderx/thunderx.c
@@ -8,9 +8,12 @@
 #include <errno.h>
 #include <fdt_support.h>
 #include <malloc.h>
+#include <netdev.h>
 #include <linux/compiler.h>
#include <asm/arch-thunderx/atf.h>
+#include <asm/arch-thunderx/fdt.h>
+#include <asm/arch-thunderx/thunderx.h>
 #include <asm/armv8/mmu.h>
#if !CONFIG_IS_ENABLED(OF_CONTROL)
@@ -73,8 +76,10 @@ static struct mm_region thunderx_mem_map[] = {
 struct mm_region *mem_map = thunderx_mem_map;
#ifdef CONFIG_BOARD_EARLY_INIT_R
+extern void eth_common_init(void);
 int board_early_init_r(void)
 {
+	eth_common_init();
    pci_init();
return 0;
@@ -87,6 +92,10 @@ int board_init(void)
    ulong fdt_addr = (ulong)fdt_base_addr;
    set_working_fdt_addr(fdt_addr);
 #endif
+#if defined(CONFIG_THUNDERX_VNIC)
+	thunderx_parse_phy_info();
+#endif
+
    return 0;
 }
@@ -127,12 +136,25 @@ void reset_cpu(ulong addr)
 {
 }
-/*
+#if defined(CONFIG_THUNDERX_VNIC)
+#if defined(CONFIG_BOARD_LATE_INIT)
+/**
+ * Board late initialization routine.
+ */
+int board_late_init(void)
+{
+	thunderx_parse_mac_addr();
+
+	return 0;
+}
+#endif
+
+/**
  * Board specific ethernet initialization routine.
  */
 int board_eth_init(bd_t *bis)
 {
-	int rc = 0;
-
-	return rc;
+	/* this will init non dt, pci based enet devices that are enabled */
+	return pci_eth_init(bis);
 }
+#endif // if defined(CONFIG_THUNDERX_VNIC)
diff --git a/configs/thunderx_81xx_defconfig b/configs/thunderx_81xx_defconfig
index dbdadde5fc..7ecf658734 100644
--- a/configs/thunderx_81xx_defconfig
+++ b/configs/thunderx_81xx_defconfig
@@ -27,7 +27,10 @@ CONFIG_CMD_I2C=y
 CONFIG_CMD_PCI=y
 CONFIG_CMD_SPI=y
 CONFIG_CMD_USB=y
-# CONFIG_CMD_NET is not set
+CONFIG_CMD_DHCP=y
+CONFIG_CMD_MII=y
+CONFIG_CMD_PING=y
+CONFIG_CMD_PXE=y
 CONFIG_DEFAULT_DEVICE_TREE="thunderx-81xx"
 CONFIG_DM=y
 CONFIG_SATA=y
@@ -39,6 +42,8 @@ CONFIG_DM_I2C=y
 CONFIG_LED=y
 CONFIG_LED_GPIO=y
 # CONFIG_MMC is not set
+CONFIG_THUNDERX_VNIC=y
+CONFIG_THUNDERX_BGX=y
 CONFIG_PCI=y
 CONFIG_DM_PCI=y
 CONFIG_DM_PCI_COMPAT=y
diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index 39687431fb..68fec42f4a 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -327,6 +327,40 @@ config DRIVER_TI_EMAC
    help
       Support for davinci emac
+config THUNDERX_VNIC
+        bool "Cavium ThunderX VNIC support"
+        depends on ARCH_THUNDERX
+	select PHYLIB
+	select MISC
+        help
+          This driver supports the Cavium ThunderX VNIC ethernet MAC/PHY/MDIO.
+	  It can be found in CN80XX/CN81XX/CN88XX/CN93XX based SoCs.
+
+if THUNDERX_VNIC
+
+config MAX_BGX_PER_NODE
+	hex "Maximum number of BGX interfaces per CPU node"
+	default 3
+
+config MAX_BGX
+	hex "Maximum number of BGX interfaces across all nodes"
+	default 3
+
+config THUNDERX_XCV 
+	bool
+	default y
+
+config THUNDERX_SMI
+	bool
+	default y
+
+config THUNDERX_BGX
+	select THUNDERX_XCV
+	select THUNDERX_SMI
+	bool "Enable ThunderX BGX networking support"
+
+endif
+
 config XILINX_AXIEMAC
    depends on DM_ETH && (MICROBLAZE || ARCH_ZYNQ || ARCH_ZYNQMP)
    select PHYLIB
diff --git a/drivers/net/Makefile b/drivers/net/Makefile
index 48a2878071..922b8cdd94 100644
--- a/drivers/net/Makefile
+++ b/drivers/net/Makefile
@@ -58,6 +58,7 @@ obj-$(CONFIG_SMC91111) += smc91111.o
 obj-$(CONFIG_SMC911X) += smc911x.o
 obj-$(CONFIG_DRIVER_TI_EMAC) += davinci_emac.o
 obj-$(CONFIG_TSEC_ENET) += tsec.o fsl_mdio.o
+obj-$(CONFIG_THUNDERX_VNIC) += cavium/
 obj-$(CONFIG_DRIVER_TI_CPSW) += cpsw.o cpsw-common.o
 obj-$(CONFIG_FMAN_ENET) += fsl_mdio.o
 obj-$(CONFIG_ULI526X) += uli526x.o
diff --git a/drivers/net/cavium/Makefile b/drivers/net/cavium/Makefile
new file mode 100644
index 0000000000..af1834e387
--- /dev/null
+++ b/drivers/net/cavium/Makefile
@@ -0,0 +1,8 @@
+# SPDX-License-Identifier: GPL-2.0
+#
+# Makefile for Cavium's Thunder ethernet device
+#
+obj-y = nic_main.o nicvf_queues.o nicvf_main.o
+obj-$(CONFIG_THUNDERX_BGX) += thunder_bgx.o
+obj-$(CONFIG_THUNDERX_XCV) += thunder_xcv.o
+obj-$(CONFIG_THUNDERX_SMI) += thunderx_smi.o
diff --git a/drivers/net/cavium/nic.h b/drivers/net/cavium/nic.h
new file mode 100644
index 0000000000..7a61a0ba7a
--- /dev/null
+++ b/drivers/net/cavium/nic.h
@@ -0,0 +1,569 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2018 Cavium Inc
+ */
+#ifndef NIC_H
+#define	NIC_H
+
+#include <linux/netdevice.h>
+
+#include "thunder_bgx.h"
+
+/* Subsystem device IDs */
+#define PCI_SUBSYS_DEVID_88XX_NIC_PF		0xA11E
+#define PCI_SUBSYS_DEVID_81XX_NIC_PF		0xA21E
+#define PCI_SUBSYS_DEVID_83XX_NIC_PF		0xA31E
+
+#define PCI_SUBSYS_DEVID_88XX_PASS1_NIC_VF	0xA11E
+#define PCI_SUBSYS_DEVID_88XX_NIC_VF		0xA134
+#define PCI_SUBSYS_DEVID_81XX_NIC_VF		0xA234
+#define PCI_SUBSYS_DEVID_83XX_NIC_VF		0xA334
+
+#define	NIC_INTF_COUNT			2  /* Interfaces btw VNIC and TNS/BGX */
+#define	NIC_CHANS_PER_INF		128
+#define	NIC_MAX_CHANS			(NIC_INTF_COUNT * NIC_CHANS_PER_INF)
+
+/* PCI BAR nos */
+#define	PCI_CFG_REG_BAR_NUM		0
+#define	PCI_MSIX_REG_BAR_NUM		4
+
+/* NIC SRIOV VF count */
+#define	MAX_NUM_VFS_SUPPORTED		128
+#define	DEFAULT_NUM_VF_ENABLED		8
+
+#define	NIC_TNS_BYPASS_MODE		0
+#define	NIC_TNS_MODE			1
+
+/* NIC priv flags */
+#define	NIC_SRIOV_ENABLED		BIT(0)
+#define	NIC_TNS_ENABLED			BIT(1)
+
+/* Min/Max packet size */
+#define	NIC_HW_MIN_FRS			64
+#define	NIC_HW_MAX_FRS			9190 /* Excluding L2 header and FCS */
+
+/* Max pkinds */
+#define	NIC_MAX_PKIND			16
+
+/* Max when CPI_ALG is IP diffserv */
+#define	NIC_MAX_CPI_PER_LMAC		64
+
+/* NIC VF Interrupts */
+#define	NICVF_INTR_CQ			0
+#define	NICVF_INTR_SQ			1
+#define	NICVF_INTR_RBDR			2
+#define	NICVF_INTR_PKT_DROP		3
+#define	NICVF_INTR_TCP_TIMER		4
+#define	NICVF_INTR_MBOX			5
+#define	NICVF_INTR_QS_ERR		6
+
+#define	NICVF_INTR_CQ_SHIFT		0
+#define	NICVF_INTR_SQ_SHIFT		8
+#define	NICVF_INTR_RBDR_SHIFT		16
+#define	NICVF_INTR_PKT_DROP_SHIFT	20
+#define	NICVF_INTR_TCP_TIMER_SHIFT	21
+#define	NICVF_INTR_MBOX_SHIFT		22
+#define	NICVF_INTR_QS_ERR_SHIFT		23
+
+#define	NICVF_INTR_CQ_MASK		(0xFF << NICVF_INTR_CQ_SHIFT)
+#define	NICVF_INTR_SQ_MASK		(0xFF << NICVF_INTR_SQ_SHIFT)
+#define	NICVF_INTR_RBDR_MASK		(0x03 << NICVF_INTR_RBDR_SHIFT)
+#define	NICVF_INTR_PKT_DROP_MASK	BIT(NICVF_INTR_PKT_DROP_SHIFT)
+#define	NICVF_INTR_TCP_TIMER_MASK	BIT(NICVF_INTR_TCP_TIMER_SHIFT)
+#define	NICVF_INTR_MBOX_MASK		BIT(NICVF_INTR_MBOX_SHIFT)
+#define	NICVF_INTR_QS_ERR_MASK		BIT(NICVF_INTR_QS_ERR_SHIFT)
+
+/* MSI-X interrupts */
+#define	NIC_PF_MSIX_VECTORS		10
+#define	NIC_VF_MSIX_VECTORS		20
+
+#define NIC_PF_INTR_ID_ECC0_SBE		0
+#define NIC_PF_INTR_ID_ECC0_DBE		1
+#define NIC_PF_INTR_ID_ECC1_SBE		2
+#define NIC_PF_INTR_ID_ECC1_DBE		3
+#define NIC_PF_INTR_ID_ECC2_SBE		4
+#define NIC_PF_INTR_ID_ECC2_DBE		5
+#define NIC_PF_INTR_ID_ECC3_SBE		6
+#define NIC_PF_INTR_ID_ECC3_DBE		7
+#define NIC_PF_INTR_ID_MBOX0		8
+#define NIC_PF_INTR_ID_MBOX1		9
+
+/* Global timer for CQ timer thresh interrupts
+ * Calculated for SCLK of 700Mhz
+ * value written should be a 1/16thof what is expected
+ *
+ * 1 tick per ms
+ */
+#define NICPF_CLK_PER_INT_TICK		43750
+
+struct nicvf_cq_poll {
+	u8	cq_idx;		/* Completion queue index */
+};
+
+#define NIC_MAX_RSS_HASH_BITS		8
+#define NIC_MAX_RSS_IDR_TBL_SIZE	(1 << NIC_MAX_RSS_HASH_BITS)
+#define RSS_HASH_KEY_SIZE		5 /* 320 bit key */
+
+struct nicvf_rss_info {
+	bool enable;
+#define	RSS_L2_EXTENDED_HASH_ENA	BIT(0)
+#define	RSS_IP_HASH_ENA			BIT(1)
+#define	RSS_TCP_HASH_ENA		BIT(2)
+#define	RSS_TCP_SYN_DIS			BIT(3)
+#define	RSS_UDP_HASH_ENA		BIT(4)
+#define RSS_L4_EXTENDED_HASH_ENA	BIT(5)
+#define	RSS_ROCE_ENA			BIT(6)
+#define	RSS_L3_BI_DIRECTION_ENA		BIT(7)
+#define	RSS_L4_BI_DIRECTION_ENA		BIT(8)
+	u64 cfg;
+	u8  hash_bits;
+	u16 rss_size;
+	u8  ind_tbl[NIC_MAX_RSS_IDR_TBL_SIZE];
+	u64 key[RSS_HASH_KEY_SIZE];
+};
+
+struct nicvf_pfc {
+	u8    autoneg;
+	u8    fc_rx;
+	u8    fc_tx;
+};
+
+enum rx_stats_reg_offset {
+	RX_OCTS = 0x0,
+	RX_UCAST = 0x1,
+	RX_BCAST = 0x2,
+	RX_MCAST = 0x3,
+	RX_RED = 0x4,
+	RX_RED_OCTS = 0x5,
+	RX_ORUN = 0x6,
+	RX_ORUN_OCTS = 0x7,
+	RX_FCS = 0x8,
+	RX_L2ERR = 0x9,
+	RX_DRP_BCAST = 0xa,
+	RX_DRP_MCAST = 0xb,
+	RX_DRP_L3BCAST = 0xc,
+	RX_DRP_L3MCAST = 0xd,
+	RX_STATS_ENUM_LAST,
+};
+
+enum tx_stats_reg_offset {
+	TX_OCTS = 0x0,
+	TX_UCAST = 0x1,
+	TX_BCAST = 0x2,
+	TX_MCAST = 0x3,
+	TX_DROP = 0x4,
+	TX_STATS_ENUM_LAST,
+};
+
+struct nicvf_hw_stats {
+	u64 rx_bytes;
+	u64 rx_frames;
+	u64 rx_ucast_frames;
+	u64 rx_bcast_frames;
+	u64 rx_mcast_frames;
+	u64 rx_drops;
+	u64 rx_drop_red;
+	u64 rx_drop_red_bytes;
+	u64 rx_drop_overrun;
+	u64 rx_drop_overrun_bytes;
+	u64 rx_drop_bcast;
+	u64 rx_drop_mcast;
+	u64 rx_drop_l3_bcast;
+	u64 rx_drop_l3_mcast;
+	u64 rx_fcs_errors;
+	u64 rx_l2_errors;
+
+	u64 tx_bytes;
+	u64 tx_frames;
+	u64 tx_ucast_frames;
+	u64 tx_bcast_frames;
+	u64 tx_mcast_frames;
+	u64 tx_drops;
+};
+
+struct nicvf_drv_stats {
+	/* CQE Rx errs */
+	u64 rx_bgx_truncated_pkts;
+	u64 rx_jabber_errs;
+	u64 rx_fcs_errs;
+	u64 rx_bgx_errs;
+	u64 rx_prel2_errs;
+	u64 rx_l2_hdr_malformed;
+	u64 rx_oversize;
+	u64 rx_undersize;
+	u64 rx_l2_len_mismatch;
+	u64 rx_l2_pclp;
+	u64 rx_ip_ver_errs;
+	u64 rx_ip_csum_errs;
+	u64 rx_ip_hdr_malformed;
+	u64 rx_ip_payload_malformed;
+	u64 rx_ip_ttl_errs;
+	u64 rx_l3_pclp;
+	u64 rx_l4_malformed;
+	u64 rx_l4_csum_errs;
+	u64 rx_udp_len_errs;
+	u64 rx_l4_port_errs;
+	u64 rx_tcp_flag_errs;
+	u64 rx_tcp_offset_errs;
+	u64 rx_l4_pclp;
+	u64 rx_truncated_pkts;
+
+	/* CQE Tx errs */
+	u64 tx_desc_fault;
+	u64 tx_hdr_cons_err;
+	u64 tx_subdesc_err;
+	u64 tx_max_size_exceeded;
+	u64 tx_imm_size_oflow;
+	u64 tx_data_seq_err;
+	u64 tx_mem_seq_err;
+	u64 tx_lock_viol;
+	u64 tx_data_fault;
+	u64 tx_tstmp_conflict;
+	u64 tx_tstmp_timeout;
+	u64 tx_mem_fault;
+	u64 tx_csum_overlap;
+	u64 tx_csum_overflow;
+
+	/* driver debug stats */
+	u64 tx_tso;
+	u64 tx_timeout;
+	u64 txq_stop;
+	u64 txq_wake;
+
+	u64 rcv_buffer_alloc_failures;
+	u64 page_alloc;
+};
+
+struct hw_info {
+	u8		bgx_cnt;
+	u8		chans_per_lmac;
+	u8		chans_per_bgx; /* Rx/Tx chans */
+	u8		chans_per_rgx;
+	u8		chans_per_lbk;
+	u16		cpi_cnt;
+	u16		rssi_cnt;
+	u16		rss_ind_tbl_size;
+	u16		tl4_cnt;
+	u16		tl3_cnt;
+	u8		tl2_cnt;
+	u8		tl1_cnt;
+	bool		tl1_per_bgx; /* TL1 per BGX or per LMAC */
+	u8		model_id;
+};
+
+struct nicvf {
+	struct eth_device	*netdev;
+	u8			vf_id;
+	bool			sqs_mode:1;
+	bool			loopback_supported:1;
+	u8			tns_mode;
+	u8			node;
+	u16		mtu;
+	struct queue_set	*qs;
+#define		MAX_SQS_PER_VF_SINGLE_NODE	5
+#define		MAX_SQS_PER_VF			11
+	u8			num_qs;
+	void			*addnl_qs;
+	u16		vf_mtu;
+	void __iomem		*reg_base;
+#define	MAX_QUEUES_PER_QSET			8
+	struct nicvf_cq_poll	*napi[8];
+
+	u8			cpi_alg;
+
+	struct nicvf_hw_stats	stats;
+	struct nicvf_drv_stats	drv_stats;
+
+	struct nicpf		*nicpf;
+
+	/* VF <-> PF mailbox communication */
+	bool			pf_acked;
+	bool			pf_nacked;
+	bool			set_mac_pending;
+
+	bool			link_up;
+	u8			duplex;
+	u32		speed;
+	u8			rev_id;
+	u8			rx_queues;
+	u8			tx_queues;
+
+	bool			open;
+	bool			rb_alloc_fail;
+	void			*rcv_buf;
+	bool			hw_tso;
+};
+
+static inline int node_id(void *addr)
+{
+	return ((uintptr_t)addr >> 44) & 0x3;
+}
+
+struct nicpf {
+	struct udevice		*udev;
+	struct eth_device	*netdev;
+	struct hw_info		*hw;
+	u8			node;
+	unsigned int		flags;
+	u16		total_vf_cnt;	/* Total num of VF supported */
+	u16		num_vf_en;	/* No of VF enabled */
+	void __iomem		*reg_base;	/* Register start address */
+	u16			rss_ind_tbl_size;
+	u8			num_sqs_en;	/* Secondary qsets enabled */
+	u64			nicvf[MAX_NUM_VFS_SUPPORTED];
+	u8			vf_sqs[MAX_NUM_VFS_SUPPORTED][MAX_SQS_PER_VF];
+	u8			pqs_vf[MAX_NUM_VFS_SUPPORTED];
+	bool			sqs_used[MAX_NUM_VFS_SUPPORTED];
+	struct pkind_cfg	pkind;
+	u8			bgx_cnt;
+	u8			rev_id;
+#define	NIC_SET_VF_LMAC_MAP(bgx, lmac)	(((bgx & 0xF) << 4) | (lmac & 0xF))
+#define	NIC_GET_BGX_FROM_VF_LMAC_MAP(map)	((map >> 4) & 0xF)
+#define	NIC_GET_LMAC_FROM_VF_LMAC_MAP(map)	(map & 0xF)
+	u8			vf_lmac_map[MAX_LMAC];
+	u16		cpi_base[MAX_NUM_VFS_SUPPORTED];
+	u64		mac[MAX_NUM_VFS_SUPPORTED];
+	bool			mbx_lock[MAX_NUM_VFS_SUPPORTED];
+	u8			link[MAX_LMAC];
+	u8			duplex[MAX_LMAC];
+	u32		speed[MAX_LMAC];
+	bool			vf_enabled[MAX_NUM_VFS_SUPPORTED];
+	u16		rssi_base[MAX_NUM_VFS_SUPPORTED];
+	u8			lmac_cnt;
+};
+
+/* PF <--> VF Mailbox communication
+ * Eight 64bit registers are shared between PF and VF.
+ * Separate set for each VF.
+ * Writing '1' into last register mbx7 means end of message.
+ */
+
+/* PF <--> VF mailbox communication */
+#define	NIC_PF_VF_MAILBOX_SIZE		2
+#define	NIC_PF_VF_MBX_TIMEOUT		2000 /* ms */
+
+/* Mailbox message types */
+#define	NIC_MBOX_MSG_READY		0x01	/* Is PF ready to rcv msgs */
+#define	NIC_MBOX_MSG_ACK		0x02	/* ACK the message received */
+#define	NIC_MBOX_MSG_NACK		0x03	/* NACK the message received */
+#define	NIC_MBOX_MSG_QS_CFG		0x04	/* Configure Qset */
+#define	NIC_MBOX_MSG_RQ_CFG		0x05	/* Configure receive queue */
+#define	NIC_MBOX_MSG_SQ_CFG		0x06	/* Configure Send queue */
+#define	NIC_MBOX_MSG_RQ_DROP_CFG	0x07	/* Configure receive queue */
+#define	NIC_MBOX_MSG_SET_MAC		0x08	/* Add MAC ID to DMAC filter */
+#define	NIC_MBOX_MSG_SET_MAX_FRS	0x09	/* Set max frame size */
+#define	NIC_MBOX_MSG_CPI_CFG		0x0A	/* Config CPI, RSSI */
+#define	NIC_MBOX_MSG_RSS_SIZE		0x0B	/* Get RSS indir_tbl size */
+#define	NIC_MBOX_MSG_RSS_CFG		0x0C	/* Config RSS table */
+#define	NIC_MBOX_MSG_RSS_CFG_CONT	0x0D	/* RSS config continuation */
+#define	NIC_MBOX_MSG_RQ_BP_CFG		0x0E	/* RQ backpressure config */
+#define	NIC_MBOX_MSG_RQ_SW_SYNC		0x0F	/* Flush inflight pkts to RQ */
+#define	NIC_MBOX_MSG_BGX_STATS		0x10	/* Get stats from BGX */
+#define	NIC_MBOX_MSG_BGX_LINK_CHANGE	0x11	/* BGX:LMAC link status */
+#define	NIC_MBOX_MSG_ALLOC_SQS		0x12	/* Allocate secondary Qset */
+#define	NIC_MBOX_MSG_NICVF_PTR		0x13	/* Send nicvf ptr to PF */
+#define	NIC_MBOX_MSG_PNICVF_PTR		0x14	/* Get primary qset nicvf ptr */
+#define	NIC_MBOX_MSG_SNICVF_PTR		0x15	/* Send sqet nicvf ptr to PVF */
+#define	NIC_MBOX_MSG_LOOPBACK		0x16	/* Set interface in loopback */
+#define	NIC_MBOX_MSG_RESET_STAT_COUNTER 0x17	/* Reset statistics counters */
+#define	NIC_MBOX_MSG_PFC		0x18	/* Pause frame control */
+#define	NIC_MBOX_MSG_CFG_DONE		0xF0	/* VF configuration done */
+#define	NIC_MBOX_MSG_SHUTDOWN		0xF1	/* VF is being shutdown */
+
+struct nic_cfg_msg {
+	u8    msg;
+	u8    vf_id;
+	u8    node_id;
+	u8    tns_mode:1;
+	u8    sqs_mode:1;
+	u8    loopback_supported:1;
+	u8    mac_addr[ETH_ALEN];
+};
+
+/* Qset configuration */
+struct qs_cfg_msg {
+	u8    msg;
+	u8    num;
+	u8    sqs_count;
+	u64   cfg;
+};
+
+/* Receive queue configuration */
+struct rq_cfg_msg {
+	u8    msg;
+	u8    qs_num;
+	u8    rq_num;
+	u64   cfg;
+};
+
+/* Send queue configuration */
+struct sq_cfg_msg {
+	u8    msg;
+	u8    qs_num;
+	u8    sq_num;
+	bool  sqs_mode;
+	u64   cfg;
+};
+
+/* Set VF's MAC address */
+struct set_mac_msg {
+	u8    msg;
+	u8    vf_id;
+	u8    mac_addr[ETH_ALEN];
+};
+
+/* Set Maximum frame size */
+struct set_frs_msg {
+	u8    msg;
+	u8    vf_id;
+	u16   max_frs;
+};
+
+/* Set CPI algorithm type */
+struct cpi_cfg_msg {
+	u8    msg;
+	u8    vf_id;
+	u8    rq_cnt;
+	u8    cpi_alg;
+};
+
+/* Get RSS table size */
+struct rss_sz_msg {
+	u8    msg;
+	u8    vf_id;
+	u16   ind_tbl_size;
+};
+
+/* Set RSS configuration */
+struct rss_cfg_msg {
+	u8    msg;
+	u8    vf_id;
+	u8    hash_bits;
+	u8    tbl_len;
+	u8    tbl_offset;
+#define RSS_IND_TBL_LEN_PER_MBX_MSG	8
+	u8    ind_tbl[RSS_IND_TBL_LEN_PER_MBX_MSG];
+};
+
+struct bgx_stats_msg {
+	u8    msg;
+	u8    vf_id;
+	u8    rx;
+	u8    idx;
+	u64   stats;
+};
+
+/* Physical interface link status */
+struct bgx_link_status {
+	u8    msg;
+	u8    link_up;
+	u8    duplex;
+	u32   speed;
+};
+
+#ifdef VNIC_MULTI_QSET_SUPPORT
+/* Get Extra Qset IDs */
+struct sqs_alloc {
+	u8    msg;
+	u8    vf_id;
+	u8    qs_count;
+};
+#endif
+
+struct nicvf_ptr {
+	u8    msg;
+	u8    vf_id;
+	bool  sqs_mode;
+	u8    sqs_id;
+	u64   nicvf;
+};
+
+/* Set interface in loopback mode */
+struct set_loopback {
+	u8    msg;
+	u8    vf_id;
+	bool  enable;
+};
+
+/* Reset statistics counters */
+struct reset_stat_cfg {
+	u8    msg;
+	/* Bitmap to select NIC_PF_VNIC(vf_id)_RX_STAT(0..13) */
+	u16   rx_stat_mask;
+	/* Bitmap to select NIC_PF_VNIC(vf_id)_TX_STAT(0..4) */
+	u8    tx_stat_mask;
+	/* Bitmap to select NIC_PF_QS(0..127)_RQ(0..7)_STAT(0..1)
+	 * bit14, bit15 NIC_PF_QS(vf_id)_RQ7_STAT(0..1)
+	 * bit12, bit13 NIC_PF_QS(vf_id)_RQ6_STAT(0..1)
+	 * ..
+	 * bit2, bit3 NIC_PF_QS(vf_id)_RQ1_STAT(0..1)
+	 * bit0, bit1 NIC_PF_QS(vf_id)_RQ0_STAT(0..1)
+	 */
+	u16   rq_stat_mask;
+	/* Bitmap to select NIC_PF_QS(0..127)_SQ(0..7)_STAT(0..1)
+	 * bit14, bit15 NIC_PF_QS(vf_id)_SQ7_STAT(0..1)
+	 * bit12, bit13 NIC_PF_QS(vf_id)_SQ6_STAT(0..1)
+	 * ..
+	 * bit2, bit3 NIC_PF_QS(vf_id)_SQ1_STAT(0..1)
+	 * bit0, bit1 NIC_PF_QS(vf_id)_SQ0_STAT(0..1)
+	 */
+	u16   sq_stat_mask;
+};
+
+struct pfc {
+	u8    msg;
+	u8    get; /* Get or set PFC settings */
+	u8    autoneg;
+	u8    fc_rx;
+	u8    fc_tx;
+};
+
+/* 128 bit shared memory between PF and each VF */
+union nic_mbx {
+	struct { u8 msg; }	msg;
+	struct nic_cfg_msg	nic_cfg;
+	struct qs_cfg_msg	qs;
+	struct rq_cfg_msg	rq;
+	struct sq_cfg_msg	sq;
+	struct set_mac_msg	mac;
+	struct set_frs_msg	frs;
+	struct cpi_cfg_msg	cpi_cfg;
+	struct rss_sz_msg	rss_size;
+	struct rss_cfg_msg	rss_cfg;
+	struct bgx_stats_msg    bgx_stats;
+	struct bgx_link_status  link_status;
+#ifdef VNIC_MULTI_QSET_SUPPORT
+	struct sqs_alloc        sqs_alloc;
+	struct nicvf_ptr	nicvf;
+#endif
+	struct set_loopback	lbk;
+};
+
+int nicvf_set_real_num_queues(struct eth_device *netdev,
+			      int tx_queues, int rx_queues);
+int nicvf_open(struct eth_device *netdev, bd_t *bis);
+void nicvf_stop(struct eth_device *netdev);
+int nicvf_send_msg_to_pf(struct nicvf *vf, union nic_mbx *mbx);
+void nicvf_free_pkt(struct nicvf *nic, void *pkt);
+void nicvf_update_stats(struct nicvf *nic);
+
+void nic_handle_mbx_intr(struct nicpf *nic, int vf);
+
+int bgx_poll_for_link(int node, int bgx_idx, int lmacid);
+const u8 *bgx_get_lmac_mac(int node, int bgx_idx, int lmacid);
+void bgx_set_lmac_mac(int node, int bgx_idx, int lmacid, const u8 *mac);
+void bgx_lmac_rx_tx_enable(int node, int bgx_idx, int lmacid, bool enable);
+void bgx_lmac_internal_loopback(int node, int bgx_idx,
+				int lmac_idx, bool enable);
+
+static inline bool pass1_silicon(unsigned int revision, int model_id)
+{
+	return ((revision < 8) && (model_id == 0x88));
+}
+
+static inline bool pass2_silicon(unsigned int revision, int model_id)
+{
+	return ((revision >= 8) && (model_id == 0x88));
+}
+
+#endif /* NIC_H */
diff --git a/drivers/net/cavium/nic_main.c b/drivers/net/cavium/nic_main.c
new file mode 100644
index 0000000000..14639b5e11
--- /dev/null
+++ b/drivers/net/cavium/nic_main.c
@@ -0,0 +1,795 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2018 Cavium Inc
+ */
+#include <config.h>
+#include <common.h>
+#include <net.h>
+#include <netdev.h>
+#include <malloc.h>
+#include <miiphy.h>
+#include <dm.h>
+#include <misc.h>
+#include <pci.h>
+#include <asm/io.h>
+
+#include <asm/arch-thunderx/thunderx_vnic.h>
+
+#include "nic_reg.h"
+#include "nic.h"
+#include "q_struct.h"
+#include "thunder_bgx.h"
+
+unsigned long rounddown_pow_of_two(unsigned long n)
+{
+	n |= n >> 1;
+	n |= n >> 2;
+	n |= n >> 4;
+	n |= n >> 8;
+	n |= n >> 16;
+	n |= n >> 32;
+
+	return(n + 1);
+}
+
+static void nic_config_cpi(struct nicpf *nic, struct cpi_cfg_msg *cfg);
+static void nic_tx_channel_cfg(struct nicpf *nic, u8 vnic,
+			       struct sq_cfg_msg *sq);
+static int nic_update_hw_frs(struct nicpf *nic, int new_frs, int vf);
+static int nic_rcv_queue_sw_sync(struct nicpf *nic);
+
+/* Register read/write APIs */
+static void nic_reg_write(struct nicpf *nic, u64 offset, u64 val)
+{
+	writeq(val, nic->reg_base + offset);
+}
+
+static u64 nic_reg_read(struct nicpf *nic, u64 offset)
+{
+	return readq(nic->reg_base + offset);
+}
+
+static u64 nic_get_mbx_addr(int vf)
+{
+	return NIC_PF_VF_0_127_MAILBOX_0_1 + (vf << NIC_VF_NUM_SHIFT);
+}
+
+/* Send a mailbox message to VF
+ * @vf: vf to which this message to be sent
+ * @mbx: Message to be sent
+ */
+static void nic_send_msg_to_vf(struct nicpf *nic, int vf, union nic_mbx *mbx)
+{
+	void __iomem *mbx_addr = nic->reg_base + nic_get_mbx_addr(vf);
+	u64 *msg = (u64 *)mbx;
+
+	/* In first revision HW, mbox interrupt is triggerred
+	 * when PF writes to MBOX(1), in next revisions when
+	 * PF writes to MBOX(0)
+	 */
+	if (pass1_silicon(nic->rev_id, nic->hw->model_id)) {
+		/* see the comment for nic_reg_write()/nic_reg_read()
+		 * functions above
+		 */
+		writeq(msg[0], mbx_addr);
+		writeq(msg[1], mbx_addr + 8);
+	} else {
+		writeq(msg[1], mbx_addr + 8);
+		writeq(msg[0], mbx_addr);
+	}
+}
+
+/* Responds to VF's READY message with VF's
+ * ID, node, MAC address e.t.c
+ * @vf: VF which sent READY message
+ */
+static void nic_mbx_send_ready(struct nicpf *nic, int vf)
+{
+	union nic_mbx mbx = {};
+	int bgx_idx, lmac, timeout = 5, link = -1;
+	const u8 *mac;
+
+	mbx.nic_cfg.msg = NIC_MBOX_MSG_READY;
+	mbx.nic_cfg.vf_id = vf;
+
+	if (nic->flags & NIC_TNS_ENABLED)
+		mbx.nic_cfg.tns_mode = NIC_TNS_MODE;
+	else
+		mbx.nic_cfg.tns_mode = NIC_TNS_BYPASS_MODE;
+
+	if (vf < nic->num_vf_en) {
+		bgx_idx = NIC_GET_BGX_FROM_VF_LMAC_MAP(nic->vf_lmac_map[vf]);
+		lmac = NIC_GET_LMAC_FROM_VF_LMAC_MAP(nic->vf_lmac_map[vf]);
+
+		mac = bgx_get_lmac_mac(nic->node, bgx_idx, lmac);
+		if (mac)
+			memcpy((u8 *)&mbx.nic_cfg.mac_addr, mac, 6);
+
+		while (timeout-- && (link <= 0)){
+			link = bgx_poll_for_link(nic->node, bgx_idx, lmac);
+			debug("Link status: %d\n", link);
+			if (link <= 0)
+				mdelay(2000);
+		}
+	}
+	mbx.nic_cfg.sqs_mode = (vf >= nic->num_vf_en) ? true : false;
+	mbx.nic_cfg.node_id = nic->node;
+
+	mbx.nic_cfg.loopback_supported = vf < nic->num_vf_en;
+
+	nic_send_msg_to_vf(nic, vf, &mbx);
+}
+
+/* ACKs VF's mailbox message
+ * @vf: VF to which ACK to be sent
+ */
+static void nic_mbx_send_ack(struct nicpf *nic, int vf)
+{
+	union nic_mbx mbx = {};
+
+	mbx.msg.msg = NIC_MBOX_MSG_ACK;
+	nic_send_msg_to_vf(nic, vf, &mbx);
+}
+
+/* NACKs VF's mailbox message that PF is not able to
+ * complete the action
+ * @vf: VF to which ACK to be sent
+ */
+static void nic_mbx_send_nack(struct nicpf *nic, int vf)
+{
+	union nic_mbx mbx = {};
+
+	mbx.msg.msg = NIC_MBOX_MSG_NACK;
+	nic_send_msg_to_vf(nic, vf, &mbx);
+}
+
+static int nic_config_loopback(struct nicpf *nic, struct set_loopback *lbk)
+{
+	int bgx_idx, lmac_idx;
+
+	if (lbk->vf_id > nic->num_vf_en)
+		return -1;
+
+	bgx_idx = NIC_GET_BGX_FROM_VF_LMAC_MAP(nic->vf_lmac_map[lbk->vf_id]);
+	lmac_idx = NIC_GET_LMAC_FROM_VF_LMAC_MAP(nic->vf_lmac_map[lbk->vf_id]);
+
+	bgx_lmac_internal_loopback(nic->node, bgx_idx, lmac_idx, lbk->enable);
+
+	return 0;
+}
+
+/* Interrupt handler to handle mailbox messages from VFs */
+void nic_handle_mbx_intr(struct nicpf *nic, int vf)
+{
+	union nic_mbx mbx = {};
+	u64 *mbx_data;
+	u64 mbx_addr;
+	u64 reg_addr;
+	u64 cfg;
+	int bgx, lmac;
+	int i;
+	int ret = 0;
+
+	nic->mbx_lock[vf] = true;
+
+	mbx_addr = nic_get_mbx_addr(vf);
+	mbx_data = (u64 *)&mbx;
+
+	for (i = 0; i < NIC_PF_VF_MAILBOX_SIZE; i++) {
+		*mbx_data = nic_reg_read(nic, mbx_addr);
+		mbx_data++;
+		mbx_addr += sizeof(u64);
+	}
+
+	debug("%s: Mailbox msg %d from VF%d\n",
+		__func__, mbx.msg.msg, vf);
+	switch (mbx.msg.msg) {
+	case NIC_MBOX_MSG_READY:
+		nic_mbx_send_ready(nic, vf);
+		if (vf < nic->num_vf_en) {
+			nic->link[vf] = 0;
+			nic->duplex[vf] = 0;
+			nic->speed[vf] = 0;
+		}
+		ret = 1;
+		break;
+	case NIC_MBOX_MSG_QS_CFG:
+		reg_addr = NIC_PF_QSET_0_127_CFG |
+			   (mbx.qs.num << NIC_QS_ID_SHIFT);
+		cfg = mbx.qs.cfg;
+#ifdef VNIC_MULTI_QSET_SUPPORT
+		/* Check if its a secondary Qset */
+		if (vf >= nic->num_vf_en) {
+			cfg = cfg & (~0x7FULL);
+			/* Assign this Qset to primary Qset's VF */
+			cfg |= nic->pqs_vf[vf];
+		}
+#endif
+		nic_reg_write(nic, reg_addr, cfg);
+		break;
+	case NIC_MBOX_MSG_RQ_CFG:
+		reg_addr = NIC_PF_QSET_0_127_RQ_0_7_CFG |
+			   (mbx.rq.qs_num << NIC_QS_ID_SHIFT) |
+			   (mbx.rq.rq_num << NIC_Q_NUM_SHIFT);
+		nic_reg_write(nic, reg_addr, mbx.rq.cfg);
+		/* Enable CQE_RX2_S extension in CQE_RX descriptor.
+		 * This gets appended by default on 81xx/83xx chips,
+		 * for consistency enabling the same on 88xx pass2
+		 * where this is introduced.
+		 */
+		if (pass2_silicon(nic->rev_id, nic->hw->model_id))
+			nic_reg_write(nic, NIC_PF_RX_CFG, 0x01);
+		break;
+	case NIC_MBOX_MSG_RQ_BP_CFG:
+		reg_addr = NIC_PF_QSET_0_127_RQ_0_7_BP_CFG |
+			   (mbx.rq.qs_num << NIC_QS_ID_SHIFT) |
+			   (mbx.rq.rq_num << NIC_Q_NUM_SHIFT);
+		nic_reg_write(nic, reg_addr, mbx.rq.cfg);
+		break;
+	case NIC_MBOX_MSG_RQ_SW_SYNC:
+		ret = nic_rcv_queue_sw_sync(nic);
+		break;
+	case NIC_MBOX_MSG_RQ_DROP_CFG:
+		reg_addr = NIC_PF_QSET_0_127_RQ_0_7_DROP_CFG |
+			   (mbx.rq.qs_num << NIC_QS_ID_SHIFT) |
+			   (mbx.rq.rq_num << NIC_Q_NUM_SHIFT);
+		nic_reg_write(nic, reg_addr, mbx.rq.cfg);
+		break;
+	case NIC_MBOX_MSG_SQ_CFG:
+		reg_addr = NIC_PF_QSET_0_127_SQ_0_7_CFG |
+			   (mbx.sq.qs_num << NIC_QS_ID_SHIFT) |
+			   (mbx.sq.sq_num << NIC_Q_NUM_SHIFT);
+		nic_reg_write(nic, reg_addr, mbx.sq.cfg);
+		nic_tx_channel_cfg(nic, mbx.qs.num, (struct sq_cfg_msg*)&mbx.sq);
+		break;
+	case NIC_MBOX_MSG_SET_MAC:
+#ifdef VNIC_MULTI_QSET_SUPPORT
+		if (vf >= nic->num_vf_en)
+			break;
+#endif
+		lmac = mbx.mac.vf_id;
+		bgx = NIC_GET_BGX_FROM_VF_LMAC_MAP(nic->vf_lmac_map[lmac]);
+		lmac = NIC_GET_LMAC_FROM_VF_LMAC_MAP(nic->vf_lmac_map[lmac]);
+		bgx_set_lmac_mac(nic->node, bgx, lmac, mbx.mac.mac_addr);
+		break;
+	case NIC_MBOX_MSG_SET_MAX_FRS:
+		ret = nic_update_hw_frs(nic, mbx.frs.max_frs,
+					mbx.frs.vf_id);
+		break;
+	case NIC_MBOX_MSG_CPI_CFG:
+		nic_config_cpi(nic, &mbx.cpi_cfg);
+		break;
+	case NIC_MBOX_MSG_CFG_DONE:
+		/* Last message of VF config msg sequence */
+		nic->vf_enabled[vf] = true;
+		if (vf >= nic->lmac_cnt)
+			goto unlock;
+
+		bgx = NIC_GET_BGX_FROM_VF_LMAC_MAP(nic->vf_lmac_map[vf]);
+		lmac = NIC_GET_LMAC_FROM_VF_LMAC_MAP(nic->vf_lmac_map[vf]);
+
+		bgx_lmac_rx_tx_enable(nic->node, bgx, lmac, true);
+		goto unlock;
+	case NIC_MBOX_MSG_SHUTDOWN:
+		/* First msg in VF teardown sequence */
+		nic->vf_enabled[vf] = false;
+#ifdef VNIC_MULTI_QSET_SUPPORT
+		if (vf >= nic->num_vf_en)
+			nic->sqs_used[vf - nic->num_vf_en] = false;
+		nic->pqs_vf[vf] = 0;
+#endif
+		if (vf >= nic->lmac_cnt)
+			break;
+
+		bgx = NIC_GET_BGX_FROM_VF_LMAC_MAP(nic->vf_lmac_map[vf]);
+		lmac = NIC_GET_LMAC_FROM_VF_LMAC_MAP(nic->vf_lmac_map[vf]);
+
+		bgx_lmac_rx_tx_enable(nic->node, bgx, lmac, false);
+		break;
+#ifdef VNIC_MULTI_QSET_SUPPORT
+	case NIC_MBOX_MSG_ALLOC_SQS:
+		nic_alloc_sqs(nic, &mbx.sqs_alloc);
+		goto unlock;
+	case NIC_MBOX_MSG_NICVF_PTR:
+		nic->nicvf[vf] = mbx.nicvf.nicvf;
+		break;
+	case NIC_MBOX_MSG_PNICVF_PTR:
+		nic_send_pnicvf(nic, vf);
+		goto unlock;
+	case NIC_MBOX_MSG_SNICVF_PTR:
+		nic_send_snicvf(nic, &mbx.nicvf);
+		goto unlock;
+#endif
+	case NIC_MBOX_MSG_LOOPBACK:
+		ret = nic_config_loopback(nic, &mbx.lbk);
+		break;
+	default:
+		printf("Invalid msg from VF%d, msg 0x%x\n", vf, mbx.msg.msg);
+		break;
+	}
+
+	if (!ret)
+		nic_mbx_send_ack(nic, vf);
+	else if (mbx.msg.msg != NIC_MBOX_MSG_READY)
+		nic_mbx_send_nack(nic, vf);
+unlock:
+	nic->mbx_lock[vf] = false;
+}
+
+static int nic_rcv_queue_sw_sync(struct nicpf *nic)
+{
+	int timeout = 20;
+
+	nic_reg_write(nic, NIC_PF_SW_SYNC_RX, 0x01);
+	/* Wait till sync cycle is finished */
+	while (timeout) {
+		if (nic_reg_read(nic, NIC_PF_SW_SYNC_RX_DONE) & 0x1)
+			break;
+		udelay(2000);
+		timeout--;
+	}
+	nic_reg_write(nic, NIC_PF_SW_SYNC_RX, 0x00);
+	if (!timeout) {
+		printf("Receive queue software sync failed");
+		return 1;
+	}
+	return 0;
+}
+
+/* Update hardware min/max frame size */
+static int nic_update_hw_frs(struct nicpf *nic, int new_frs, int vf)
+{
+	u64 *pkind = (u64 *)&nic->pkind;
+	if ((new_frs > NIC_HW_MAX_FRS) || (new_frs < NIC_HW_MIN_FRS)) {
+		printf("Invalid MTU setting from VF%d rejected, should be between %d and %d\n",
+				vf, NIC_HW_MIN_FRS, NIC_HW_MAX_FRS);
+		return 1;
+	}
+	new_frs += ETH_HLEN;
+	if (new_frs <= nic->pkind.maxlen)
+		return 0;
+
+	nic->pkind.maxlen = new_frs;
+
+	nic_reg_write(nic, NIC_PF_PKIND_0_15_CFG, *pkind);
+	return 0;
+}
+
+/* Set minimum transmit packet size */
+static void nic_set_tx_pkt_pad(struct nicpf *nic, int size)
+{
+	int lmac;
+	u64 lmac_cfg;
+	struct hw_info *hw = nic->hw;
+	int max_lmac = nic->hw->bgx_cnt * MAX_LMAC_PER_BGX;
+
+	/* There is a issue in HW where-in while sending GSO sized
+	 * pkts as part of TSO, if pkt len falls below this size
+	 * NIC will zero PAD packet and also updates IP total length.
+	 * Hence set this value to lessthan min pkt size of MAC+IP+TCP
+	 * headers, BGX will do the padding to transmit 64 byte pkt.
+	 */
+	if (size > 52)
+		size = 52;
+
+	/* CN81XX has RGX configured as FAKE BGX, adjust mac_lmac accordingly */
+	if (hw->chans_per_rgx)
+		max_lmac = ((nic->hw->bgx_cnt - 1) * MAX_LMAC_PER_BGX) + 1;
+
+	for (lmac = 0; lmac < max_lmac; lmac++) {
+		lmac_cfg = nic_reg_read(nic, NIC_PF_LMAC_0_7_CFG | (lmac << 3));
+		lmac_cfg &= ~(0xF << 2);
+		lmac_cfg |= ((size / 4) << 2);
+		nic_reg_write(nic, NIC_PF_LMAC_0_7_CFG | (lmac << 3), lmac_cfg);
+	}
+}
+
+/* Function to check number of LMACs present and set VF::LMAC mapping.
+ * Mapping will be used while initializing channels.
+ */
+static void nic_set_lmac_vf_mapping(struct nicpf *nic)
+{
+	int bgx, bgx_count, next_bgx_lmac = 0;
+	int lmac, lmac_cnt = 0;
+	u64 lmac_credit;
+
+	nic->num_vf_en = 0;
+	if (nic->flags & NIC_TNS_ENABLED) {
+		nic->num_vf_en = DEFAULT_NUM_VF_ENABLED;
+		return;
+	}
+
+	bgx_get_count(nic->node, &bgx_count);
+	debug("bgx_count: %d\n", bgx_count);
+
+	for (bgx = 0; bgx < nic->hw->bgx_cnt; bgx++) {
+		if (!(bgx_count & (1 << bgx)))
+			continue;
+		nic->bgx_cnt++;
+		lmac_cnt = bgx_get_lmac_count(nic->node, bgx);
+		debug("lmac_cnt: %d for BGX%d\n", lmac_cnt, bgx);
+		for (lmac = 0; lmac < lmac_cnt; lmac++)
+			nic->vf_lmac_map[next_bgx_lmac++] =
+						NIC_SET_VF_LMAC_MAP(bgx, lmac);
+		nic->num_vf_en += lmac_cnt;
+
+		/* Program LMAC credits */
+		lmac_credit = (1ull << 1); /* channel credit enable */
+		lmac_credit |= (0x1ff << 2); /* Max outstanding pkt count */
+		/* 48KB BGX Tx buffer size, each unit is of size 16bytes */
+		lmac_credit |= (((((48 * 1024) / lmac_cnt) -
+				NIC_HW_MAX_FRS) / 16) << 12);
+		lmac = bgx * MAX_LMAC_PER_BGX;
+		for (; lmac < lmac_cnt + (bgx * MAX_LMAC_PER_BGX); lmac++)
+			nic_reg_write(nic,
+				NIC_PF_LMAC_0_7_CREDIT + (lmac * 8), lmac_credit);
+	}
+}
+
+static void nic_get_hw_info(struct nicpf *nic)
+{
+	u16 sdevid;
+	struct hw_info *hw = nic->hw;
+
+	dm_pci_read_config16(nic->udev, PCI_SUBSYSTEM_ID, &sdevid);
+
+	switch (sdevid) {
+	case PCI_SUBSYS_DEVID_88XX_NIC_PF:
+		hw->bgx_cnt = CONFIG_MAX_BGX_PER_NODE;
+		hw->chans_per_lmac = 16;
+		hw->chans_per_bgx = 128;
+		hw->cpi_cnt = 2048;
+		hw->rssi_cnt = 4096;
+		hw->rss_ind_tbl_size = NIC_MAX_RSS_IDR_TBL_SIZE;
+		hw->tl3_cnt = 256;
+		hw->tl2_cnt = 64;
+		hw->tl1_cnt = 2;
+		hw->tl1_per_bgx = true;
+		hw->model_id = 0x88;
+		break;
+	case PCI_SUBSYS_DEVID_81XX_NIC_PF:
+		hw->bgx_cnt = CONFIG_MAX_BGX_PER_NODE;
+		hw->chans_per_lmac = 8;
+		hw->chans_per_bgx = 32;
+		hw->chans_per_rgx = 8;
+		hw->chans_per_lbk = 24;
+		hw->cpi_cnt = 512;
+		hw->rssi_cnt = 256;
+		hw->rss_ind_tbl_size = 32; /* Max RSSI / Max interfaces */
+		hw->tl3_cnt = 64;
+		hw->tl2_cnt = 16;
+		hw->tl1_cnt = 10;
+		hw->tl1_per_bgx = false;
+		hw->model_id = 0x81;
+		break;
+	case PCI_SUBSYS_DEVID_83XX_NIC_PF:
+		hw->bgx_cnt = CONFIG_MAX_BGX_PER_NODE;
+		hw->chans_per_lmac = 8;
+		hw->chans_per_bgx = 32;
+		hw->chans_per_lbk = 64;
+		hw->cpi_cnt = 2048;
+		hw->rssi_cnt = 1024;
+		hw->rss_ind_tbl_size = 64; /* Max RSSI / Max interfaces */
+		hw->tl3_cnt = 256;
+		hw->tl2_cnt = 64;
+		hw->tl1_cnt = 18;
+		hw->tl1_per_bgx = false;
+		hw->model_id = 0x83;
+		break;
+	}
+
+	hw->tl4_cnt = MAX_QUEUES_PER_QSET * dm_pci_sriov_get_totalvfs(nic->udev);
+}
+
+static void nic_init_hw(struct nicpf *nic)
+{
+	int i;
+	u64 reg;
+	u64 *pkind = (u64 *)&nic->pkind;
+
+	/* Get HW capability info */
+	nic_get_hw_info(nic);
+
+	/* Enable NIC HW block */
+	nic_reg_write(nic, NIC_PF_CFG, 0x3);
+
+	/* Enable backpressure */
+	nic_reg_write(nic, NIC_PF_BP_CFG, (1ULL << 6) | 0x03);
+	nic_reg_write(nic, NIC_PF_INTF_0_1_BP_CFG, (1ULL << 63) | 0x08);
+	nic_reg_write(nic, NIC_PF_INTF_0_1_BP_CFG + (1 << 8), (1ULL << 63) | 0x09);
+
+	for (i = 0; i < NIC_MAX_CHANS; i++)
+		nic_reg_write(nic, NIC_PF_CHAN_0_255_TX_CFG | (i << 3), 1);
+
+	if (nic->flags & NIC_TNS_ENABLED) {
+		reg = NIC_TNS_MODE << 7;
+		reg |= 0x06;
+		nic_reg_write(nic, NIC_PF_INTF_0_1_SEND_CFG, reg);
+		reg &= ~0xFull;
+		reg |= 0x07;
+		nic_reg_write(nic, NIC_PF_INTF_0_1_SEND_CFG | (1 << 8), reg);
+	} else {
+		/* Disable TNS mode on both interfaces */
+		reg = NIC_TNS_BYPASS_MODE << 7;
+		reg |= 0x08; /* Block identifier */
+		nic_reg_write(nic, NIC_PF_INTF_0_1_SEND_CFG, reg);
+		reg &= ~0xFull;
+		reg |= 0x09;
+		nic_reg_write(nic, NIC_PF_INTF_0_1_SEND_CFG | (1 << 8), reg);
+	}
+
+	/* PKIND configuration */
+	nic->pkind.minlen = 0;
+	nic->pkind.maxlen = NIC_HW_MAX_FRS + ETH_HLEN;
+	nic->pkind.lenerr_en = 1;
+	nic->pkind.rx_hdr = 0;
+	nic->pkind.hdr_sl = 0;
+
+	for (i = 0; i < NIC_MAX_PKIND; i++)
+		nic_reg_write(nic, NIC_PF_PKIND_0_15_CFG | (i << 3), *pkind);
+
+	nic_set_tx_pkt_pad(nic, NIC_HW_MIN_FRS);
+
+	/* Timer config */
+	nic_reg_write(nic, NIC_PF_INTR_TIMER_CFG, NICPF_CLK_PER_INT_TICK);
+}
+
+/* Channel parse index configuration */
+static void nic_config_cpi(struct nicpf *nic, struct cpi_cfg_msg *cfg)
+{
+	struct hw_info *hw = nic->hw;
+	u32 vnic, bgx, lmac, chan;
+	u32 padd, cpi_count = 0;
+	u64 cpi_base, cpi, rssi_base, rssi;
+	u8  qset, rq_idx = 0;
+
+	vnic = cfg->vf_id;
+	bgx = NIC_GET_BGX_FROM_VF_LMAC_MAP(nic->vf_lmac_map[vnic]);
+	lmac = NIC_GET_LMAC_FROM_VF_LMAC_MAP(nic->vf_lmac_map[vnic]);
+
+	chan = (lmac * hw->chans_per_lmac) + (bgx * hw->chans_per_bgx);
+	cpi_base = vnic * NIC_MAX_CPI_PER_LMAC;
+	rssi_base = vnic * hw->rss_ind_tbl_size;
+
+	/* Rx channel configuration */
+	nic_reg_write(nic, NIC_PF_CHAN_0_255_RX_BP_CFG | (chan << 3),
+		      (1ull << 63) | (vnic << 0));
+	nic_reg_write(nic, NIC_PF_CHAN_0_255_RX_CFG | (chan << 3),
+		      ((u64)cfg->cpi_alg << 62) | (cpi_base << 48));
+
+	if (cfg->cpi_alg == CPI_ALG_NONE)
+		cpi_count = 1;
+	else if (cfg->cpi_alg == CPI_ALG_VLAN) /* 3 bits of PCP */
+		cpi_count = 8;
+	else if (cfg->cpi_alg == CPI_ALG_VLAN16) /* 3 bits PCP + DEI */
+		cpi_count = 16;
+	else if (cfg->cpi_alg == CPI_ALG_DIFF) /* 6bits DSCP */
+		cpi_count = NIC_MAX_CPI_PER_LMAC;
+
+	/* RSS Qset, Qidx mapping */
+	qset = cfg->vf_id;
+	rssi = rssi_base;
+	for (; rssi < (rssi_base + cfg->rq_cnt); rssi++) {
+		nic_reg_write(nic, NIC_PF_RSSI_0_4097_RQ | (rssi << 3),
+			      (qset << 3) | rq_idx);
+		rq_idx++;
+	}
+
+	rssi = 0;
+	cpi = cpi_base;
+	for (; cpi < (cpi_base + cpi_count); cpi++) {
+		/* Determine port to channel adder */
+		if (cfg->cpi_alg != CPI_ALG_DIFF)
+			padd = cpi % cpi_count;
+		else
+			padd = cpi % 8; /* 3 bits CS out of 6bits DSCP */
+
+		/* Leave RSS_SIZE as '0' to disable RSS */
+		if (pass1_silicon(nic->rev_id, nic->hw->model_id)) {
+			nic_reg_write(nic, NIC_PF_CPI_0_2047_CFG | (cpi << 3),
+				      (vnic << 24) | (padd << 16) |
+				      (rssi_base + rssi));
+		} else {
+			/* Set MPI_ALG to '0' to disable MCAM parsing */
+			nic_reg_write(nic, NIC_PF_CPI_0_2047_CFG | (cpi << 3),
+				      (padd << 16));
+			/* MPI index is same as CPI if MPI_ALG is not enabled */
+			nic_reg_write(nic, NIC_PF_MPI_0_2047_CFG | (cpi << 3),
+				      (vnic << 24) | (rssi_base + rssi));
+		}
+
+		if ((rssi + 1) >= cfg->rq_cnt)
+			continue;
+
+		if (cfg->cpi_alg == CPI_ALG_VLAN)
+			rssi++;
+		else if (cfg->cpi_alg == CPI_ALG_VLAN16)
+			rssi = ((cpi - cpi_base) & 0xe) >> 1;
+		else if (cfg->cpi_alg == CPI_ALG_DIFF)
+			rssi = ((cpi - cpi_base) & 0x38) >> 3;
+	}
+	nic->cpi_base[cfg->vf_id] = cpi_base;
+	nic->rssi_base[cfg->vf_id] = rssi_base;
+}
+
+/* 4 level transmit side scheduler configutation
+ * for TNS bypass mode
+ *
+ * Sample configuration for SQ0 on 88xx
+ * VNIC0-SQ0 -> TL4(0)   -> TL3[0]   -> TL2[0]  -> TL1[0] -> BGX0
+ * VNIC1-SQ0 -> TL4(8)   -> TL3[2]   -> TL2[0]  -> TL1[0] -> BGX0
+ * VNIC2-SQ0 -> TL4(16)  -> TL3[4]   -> TL2[1]  -> TL1[0] -> BGX0
+ * VNIC3-SQ0 -> TL4(24)  -> TL3[6]   -> TL2[1]  -> TL1[0] -> BGX0
+ * VNIC4-SQ0 -> TL4(512) -> TL3[128] -> TL2[32] -> TL1[1] -> BGX1
+ * VNIC5-SQ0 -> TL4(520) -> TL3[130] -> TL2[32] -> TL1[1] -> BGX1
+ * VNIC6-SQ0 -> TL4(528) -> TL3[132] -> TL2[33] -> TL1[1] -> BGX1
+ * VNIC7-SQ0 -> TL4(536) -> TL3[134] -> TL2[33] -> TL1[1] -> BGX1
+ */
+static void nic_tx_channel_cfg(struct nicpf *nic, u8 vnic,
+			       struct sq_cfg_msg *sq)
+{
+	struct hw_info *hw = nic->hw;
+	u32 bgx, lmac, chan;
+	u32 tl2, tl3, tl4;
+	u32 rr_quantum;
+	u8 sq_idx = sq->sq_num;
+	u8 pqs_vnic = vnic;
+	int svf;
+	u16 sdevid;
+
+	dm_pci_read_config16(nic->udev, PCI_SUBSYSTEM_ID, &sdevid);
+
+	bgx = NIC_GET_BGX_FROM_VF_LMAC_MAP(nic->vf_lmac_map[pqs_vnic]);
+	lmac = NIC_GET_LMAC_FROM_VF_LMAC_MAP(nic->vf_lmac_map[pqs_vnic]);
+
+	/* 24 bytes for FCS, IPG and preamble */
+	rr_quantum = ((NIC_HW_MAX_FRS + 24) / 4);
+
+	/* For 88xx 0-511 TL4 transmits via BGX0 and
+	 * 512-1023 TL4s transmit via BGX1.
+	 */
+	if (hw->tl1_per_bgx) {
+		tl4 = bgx * (hw->tl4_cnt / hw->bgx_cnt);
+		if (!sq->sqs_mode) {
+			tl4 += (lmac * MAX_QUEUES_PER_QSET);
+		} else {
+			for (svf = 0; svf < MAX_SQS_PER_VF_SINGLE_NODE; svf++) {
+				if (nic->vf_sqs[pqs_vnic][svf] == vnic)
+					break;
+			}
+			tl4 += (MAX_LMAC_PER_BGX * MAX_QUEUES_PER_QSET);
+			tl4 += (lmac * MAX_QUEUES_PER_QSET * MAX_SQS_PER_VF_SINGLE_NODE);
+			tl4 += (svf * MAX_QUEUES_PER_QSET);
+		}
+	} else {
+		tl4 = (vnic * MAX_QUEUES_PER_QSET);
+	}
+	tl4 += sq_idx;
+
+	tl3 = tl4 / (hw->tl4_cnt / hw->tl3_cnt);
+	nic_reg_write(nic, NIC_PF_QSET_0_127_SQ_0_7_CFG2 |
+		      ((u64)vnic << NIC_QS_ID_SHIFT) |
+		      ((u32)sq_idx << NIC_Q_NUM_SHIFT), tl4);
+	nic_reg_write(nic, NIC_PF_TL4_0_1023_CFG | (tl4 << 3),
+		      ((u64)vnic << 27) | ((u32)sq_idx << 24) | rr_quantum);
+
+	nic_reg_write(nic, NIC_PF_TL3_0_255_CFG | (tl3 << 3), rr_quantum);
+
+	/* On 88xx 0-127 channels are for BGX0 and
+	 * 127-255 channels for BGX1.
+	 *
+	 * On 81xx/83xx TL3_CHAN reg should be configured with channel
+	 * within LMAC i.e 0-7 and not the actual channel number like on 88xx
+	 */
+	chan = (lmac * hw->chans_per_lmac) + (bgx * hw->chans_per_bgx);
+	if (hw->tl1_per_bgx)
+		nic_reg_write(nic, NIC_PF_TL3_0_255_CHAN | (tl3 << 3), chan);
+	else
+		nic_reg_write(nic, NIC_PF_TL3_0_255_CHAN | (tl3 << 3), 0);
+
+	/* Enable backpressure on the channel */
+	nic_reg_write(nic, NIC_PF_CHAN_0_255_TX_CFG | (chan << 3), 1);
+
+	tl2 = tl3 >> 2;
+	nic_reg_write(nic, NIC_PF_TL3A_0_63_CFG | (tl2 << 3), tl2);
+	nic_reg_write(nic, NIC_PF_TL2_0_63_CFG | (tl2 << 3), rr_quantum);
+	/* No priorities as of now */
+	nic_reg_write(nic, NIC_PF_TL2_0_63_PRI | (tl2 << 3), 0x00);
+
+	/* Unlike 88xx where TL2s 0-31 transmits to TL1 '0' and rest to TL1 '1'
+	 * on 81xx/83xx TL2 needs to be configured to transmit to one of the
+	 * possible LMACs.
+	 *
+	 * This register doesn't exist on 88xx.
+	 */
+	if (!hw->tl1_per_bgx)
+		nic_reg_write(nic, NIC_PF_TL2_LMAC | (tl2 << 3),
+			      lmac + (bgx * MAX_LMAC_PER_BGX));
+}
+
+static int nic_initialize(struct udevice *dev)
+{
+	struct nicpf *nic = dev_get_priv(dev);
+	size_t size;
+
+	nic->udev = dev;
+	nic->hw = calloc(1, sizeof(struct hw_info));
+	if (!nic->hw) {
+		return -ENOMEM;
+	}
+
+	/* MAP PF's configuration registers */
+	nic->reg_base = dm_pci_map_bar(dev, 0, &size, PCI_REGION_MEM);
+	if (!nic->reg_base) {
+		printf("Cannot map config register space, aborting\n");
+		goto exit;
+	}
+
+	nic->node = node_id(nic->reg_base);
+	dm_pci_read_config8(dev, PCI_REVISION_ID, &nic->rev_id);
+
+	/* By default set NIC in TNS bypass mode */
+	nic->flags &= ~NIC_TNS_ENABLED;
+
+	/* Initialize hardware */
+	nic_init_hw(nic);
+
+	nic_set_lmac_vf_mapping(nic);
+
+	/* Set RSS TBL size for each VF */
+	nic->rss_ind_tbl_size = NIC_MAX_RSS_IDR_TBL_SIZE;
+
+	nic->rss_ind_tbl_size = rounddown_pow_of_two(nic->rss_ind_tbl_size);
+
+	return 0;
+exit:
+	free(nic->hw);
+	return -ENODEV;
+}
+
+int thunderx_nic_probe(struct udevice *dev)
+{
+	int vf;
+	int ret;
+	struct nicpf *nicpf;
+
+	nic_initialize(dev);
+	nicpf = dev_get_priv(dev);
+
+	ret = dm_pci_sriov_init(dev, nicpf->num_vf_en);
+
+	if (ret < 0) {
+		printf("enabling SRIOV failed for num VFs %d",nicpf->num_vf_en);
+		return ret;
+	}
+
+	for (vf = 0; vf < nicpf->num_vf_en; vf++) {
+		 nicvf_initialize(dev, vf);
+	}
+
+	return 0;
+}
+
+static const struct misc_ops thunderx_nic_ops = {
+};
+
+static const struct udevice_id thunderx_nic_ids[] = {
+	{ .compatible = "cavium,nic" },
+	{}
+};
+
+U_BOOT_DRIVER(thunderx_nic) = {
+	.name	= "thunderx_nic",
+	.id	= UCLASS_MISC,
+	.probe	= thunderx_nic_probe,
+	.of_match = thunderx_nic_ids,
+	.ops	= &thunderx_nic_ops,
+	.priv_auto_alloc_size = sizeof(struct nicpf),
+};
+
+static struct pci_device_id thunderx_nic_supported[] = {
+	{ PCI_VDEVICE(CAVIUM, PCI_DEVICE_ID_THUNDERX_NIC_PF) },
+	{}
+};
+
+U_BOOT_PCI_DEVICE(thunderx_nic, thunderx_nic_supported);
diff --git a/drivers/net/cavium/nic_reg.h b/drivers/net/cavium/nic_reg.h
new file mode 100644
index 0000000000..5a12bb90bf
--- /dev/null
+++ b/drivers/net/cavium/nic_reg.h
@@ -0,0 +1,228 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2018 Cavium Inc
+ */
+#ifndef NIC_REG_H
+#define NIC_REG_H
+
+#define   NIC_PF_REG_COUNT			29573
+#define   NIC_VF_REG_COUNT			249
+
+/* Physical function register offsets */
+#define   NIC_PF_CFG				(0x0000)
+#define   NIC_PF_STATUS				(0x0010)
+#define   NIC_PF_INTR_TIMER_CFG			(0x0030)
+#define   NIC_PF_BIST_STATUS			(0x0040)
+#define   NIC_PF_SOFT_RESET			(0x0050)
+#define   NIC_PF_TCP_TIMER			(0x0060)
+#define   NIC_PF_BP_CFG				(0x0080)
+#define   NIC_PF_RRM_CFG			(0x0088)
+#define   NIC_PF_CQM_CFG			(0x00A0)
+#define   NIC_PF_CNM_CF				(0x00A8)
+#define   NIC_PF_CNM_STATUS			(0x00B0)
+#define   NIC_PF_CQ_AVG_CFG			(0x00C0)
+#define   NIC_PF_RRM_AVG_CFG			(0x00C8)
+#define   NIC_PF_INTF_0_1_SEND_CFG		(0x0200)
+#define   NIC_PF_INTF_0_1_BP_CFG		(0x0208)
+#define   NIC_PF_INTF_0_1_BP_DIS_0_1		(0x0210)
+#define   NIC_PF_INTF_0_1_BP_SW_0_1		(0x0220)
+#define   NIC_PF_RBDR_BP_STATE_0_3		(0x0240)
+#define   NIC_PF_MAILBOX_INT			(0x0410)
+#define   NIC_PF_MAILBOX_INT_W1S		(0x0430)
+#define   NIC_PF_MAILBOX_ENA_W1C		(0x0450)
+#define   NIC_PF_MAILBOX_ENA_W1S		(0x0470)
+#define   NIC_PF_RX_ETYPE_0_7			(0x0500)
+#define   NIC_PF_RX_GENEVE_DEF			(0x0580)
+#define    UDP_GENEVE_PORT_NUM				0x17C1ULL
+#define   NIC_PF_RX_GENEVE_PROT_DEF		(0x0588)
+#define    IPV6_PROT					0x86DDULL
+#define    IPV4_PROT					0x800ULL
+#define    ET_PROT					0x6558ULL
+#define   NIC_PF_RX_NVGRE_PROT_DEF		(0x0598)
+#define   NIC_PF_RX_VXLAN_DEF_0_1		(0x05A0)
+#define    UDP_VXLAN_PORT_NUM				0x12B5
+#define   NIC_PF_RX_VXLAN_PROT_DEF		(0x05B0)
+#define    IPV6_PROT_DEF				0x2ULL
+#define    IPV4_PROT_DEF				0x1ULL
+#define    ET_PROT_DEF					0x3ULL
+#define   NIC_PF_RX_CFG				(0x05D0)
+#define   NIC_PF_PKIND_0_15_CFG			(0x0600)
+#define   NIC_PF_ECC0_FLIP0			(0x1000)
+#define   NIC_PF_ECC1_FLIP0			(0x1008)
+#define   NIC_PF_ECC2_FLIP0			(0x1010)
+#define   NIC_PF_ECC3_FLIP0			(0x1018)
+#define   NIC_PF_ECC0_FLIP1			(0x1080)
+#define   NIC_PF_ECC1_FLIP1			(0x1088)
+#define   NIC_PF_ECC2_FLIP1			(0x1090)
+#define   NIC_PF_ECC3_FLIP1			(0x1098)
+#define   NIC_PF_ECC0_CDIS			(0x1100)
+#define   NIC_PF_ECC1_CDIS			(0x1108)
+#define   NIC_PF_ECC2_CDIS			(0x1110)
+#define   NIC_PF_ECC3_CDIS			(0x1118)
+#define   NIC_PF_BIST0_STATUS			(0x1280)
+#define   NIC_PF_BIST1_STATUS			(0x1288)
+#define   NIC_PF_BIST2_STATUS			(0x1290)
+#define   NIC_PF_BIST3_STATUS			(0x1298)
+#define   NIC_PF_ECC0_SBE_INT			(0x2000)
+#define   NIC_PF_ECC0_SBE_INT_W1S		(0x2008)
+#define   NIC_PF_ECC0_SBE_ENA_W1C		(0x2010)
+#define   NIC_PF_ECC0_SBE_ENA_W1S		(0x2018)
+#define   NIC_PF_ECC0_DBE_INT			(0x2100)
+#define   NIC_PF_ECC0_DBE_INT_W1S		(0x2108)
+#define   NIC_PF_ECC0_DBE_ENA_W1C		(0x2110)
+#define   NIC_PF_ECC0_DBE_ENA_W1S		(0x2118)
+#define   NIC_PF_ECC1_SBE_INT			(0x2200)
+#define   NIC_PF_ECC1_SBE_INT_W1S		(0x2208)
+#define   NIC_PF_ECC1_SBE_ENA_W1C		(0x2210)
+#define   NIC_PF_ECC1_SBE_ENA_W1S		(0x2218)
+#define   NIC_PF_ECC1_DBE_INT			(0x2300)
+#define   NIC_PF_ECC1_DBE_INT_W1S		(0x2308)
+#define   NIC_PF_ECC1_DBE_ENA_W1C		(0x2310)
+#define   NIC_PF_ECC1_DBE_ENA_W1S		(0x2318)
+#define   NIC_PF_ECC2_SBE_INT			(0x2400)
+#define   NIC_PF_ECC2_SBE_INT_W1S		(0x2408)
+#define   NIC_PF_ECC2_SBE_ENA_W1C		(0x2410)
+#define   NIC_PF_ECC2_SBE_ENA_W1S		(0x2418)
+#define   NIC_PF_ECC2_DBE_INT			(0x2500)
+#define   NIC_PF_ECC2_DBE_INT_W1S		(0x2508)
+#define   NIC_PF_ECC2_DBE_ENA_W1C		(0x2510)
+#define   NIC_PF_ECC2_DBE_ENA_W1S		(0x2518)
+#define   NIC_PF_ECC3_SBE_INT			(0x2600)
+#define   NIC_PF_ECC3_SBE_INT_W1S		(0x2608)
+#define   NIC_PF_ECC3_SBE_ENA_W1C		(0x2610)
+#define   NIC_PF_ECC3_SBE_ENA_W1S		(0x2618)
+#define   NIC_PF_ECC3_DBE_INT			(0x2700)
+#define   NIC_PF_ECC3_DBE_INT_W1S		(0x2708)
+#define   NIC_PF_ECC3_DBE_ENA_W1C		(0x2710)
+#define   NIC_PF_ECC3_DBE_ENA_W1S		(0x2718)
+#define   NIC_PF_MCAM_0_191_ENA			(0x100000)
+#define   NIC_PF_MCAM_0_191_M_0_5_DATA		(0x110000)
+#define   NIC_PF_MCAM_CTRL			(0x120000)
+#define   NIC_PF_CPI_0_2047_CFG			(0x200000)
+#define   NIC_PF_MPI_0_2047_CFG			(0x210000)
+#define   NIC_PF_RSSI_0_4097_RQ			(0x220000)
+#define   NIC_PF_LMAC_0_7_CFG			(0x240000)
+#define   NIC_PF_LMAC_0_7_CFG2			(0x240100)
+#define   NIC_PF_LMAC_0_7_SW_XOFF		(0x242000)
+#define   NIC_PF_LMAC_0_7_CREDIT		(0x244000)
+#define   NIC_PF_CHAN_0_255_TX_CFG		(0x400000)
+#define   NIC_PF_CHAN_0_255_RX_CFG		(0x420000)
+#define   NIC_PF_CHAN_0_255_SW_XOFF		(0x440000)
+#define   NIC_PF_CHAN_0_255_CREDIT		(0x460000)
+#define   NIC_PF_CHAN_0_255_RX_BP_CFG		(0x480000)
+#define   NIC_PF_SW_SYNC_RX			(0x490000)
+#define   NIC_PF_SW_SYNC_RX_DONE		(0x490008)
+#define   NIC_PF_TL2_0_63_CFG			(0x500000)
+#define   NIC_PF_TL2_0_63_PRI			(0x520000)
+#define   NIC_PF_TL2_LMAC			(0x540000)
+#define   NIC_PF_TL2_0_63_SH_STATUS		(0x580000)
+#define   NIC_PF_TL3A_0_63_CFG			(0x5F0000)
+#define   NIC_PF_TL3_0_255_CFG			(0x600000)
+#define   NIC_PF_TL3_0_255_CHAN			(0x620000)
+#define   NIC_PF_TL3_0_255_PIR			(0x640000)
+#define   NIC_PF_TL3_0_255_SW_XOFF		(0x660000)
+#define   NIC_PF_TL3_0_255_CNM_RATE		(0x680000)
+#define   NIC_PF_TL3_0_255_SH_STATUS		(0x6A0000)
+#define   NIC_PF_TL4A_0_255_CFG			(0x6F0000)
+#define   NIC_PF_TL4_0_1023_CFG			(0x800000)
+#define   NIC_PF_TL4_0_1023_SW_XOFF		(0x820000)
+#define   NIC_PF_TL4_0_1023_SH_STATUS		(0x840000)
+#define   NIC_PF_TL4A_0_1023_CNM_RATE		(0x880000)
+#define   NIC_PF_TL4A_0_1023_CNM_STATUS		(0x8A0000)
+#define   NIC_PF_VF_0_127_MAILBOX_0_1		(0x20002030)
+#define   NIC_PF_VNIC_0_127_TX_STAT_0_4		(0x20004000)
+#define   NIC_PF_VNIC_0_127_RX_STAT_0_13	(0x20004100)
+#define   NIC_PF_QSET_0_127_LOCK_0_15		(0x20006000)
+#define   NIC_PF_QSET_0_127_CFG			(0x20010000)
+#define   NIC_PF_QSET_0_127_RQ_0_7_CFG		(0x20010400)
+#define   NIC_PF_QSET_0_127_RQ_0_7_DROP_CFG	(0x20010420)
+#define   NIC_PF_QSET_0_127_RQ_0_7_BP_CFG	(0x20010500)
+#define   NIC_PF_QSET_0_127_RQ_0_7_STAT_0_1	(0x20010600)
+#define   NIC_PF_QSET_0_127_SQ_0_7_CFG		(0x20010C00)
+#define   NIC_PF_QSET_0_127_SQ_0_7_CFG2		(0x20010C08)
+#define   NIC_PF_QSET_0_127_SQ_0_7_STAT_0_1	(0x20010D00)
+
+#define   NIC_PF_MSIX_VEC_0_18_ADDR		(0x000000)
+#define   NIC_PF_MSIX_VEC_0_CTL			(0x000008)
+#define   NIC_PF_MSIX_PBA_0			(0x0F0000)
+
+/* Virtual function register offsets */
+#define   NIC_VNIC_CFG				(0x000020)
+#define   NIC_VF_PF_MAILBOX_0_1			(0x000130)
+#define   NIC_VF_INT				(0x000200)
+#define   NIC_VF_INT_W1S			(0x000220)
+#define   NIC_VF_ENA_W1C			(0x000240)
+#define   NIC_VF_ENA_W1S			(0x000260)
+
+#define   NIC_VNIC_RSS_CFG			(0x0020E0)
+#define   NIC_VNIC_RSS_KEY_0_4			(0x002200)
+#define   NIC_VNIC_TX_STAT_0_4			(0x004000)
+#define   NIC_VNIC_RX_STAT_0_13			(0x004100)
+#define   NIC_QSET_RQ_GEN_CFG			(0x010010)
+
+#define   NIC_QSET_CQ_0_7_CFG			(0x010400)
+#define   NIC_QSET_CQ_0_7_CFG2			(0x010408)
+#define   NIC_QSET_CQ_0_7_THRESH		(0x010410)
+#define   NIC_QSET_CQ_0_7_BASE			(0x010420)
+#define   NIC_QSET_CQ_0_7_HEAD			(0x010428)
+#define   NIC_QSET_CQ_0_7_TAIL			(0x010430)
+#define   NIC_QSET_CQ_0_7_DOOR			(0x010438)
+#define   NIC_QSET_CQ_0_7_STATUS		(0x010440)
+#define   NIC_QSET_CQ_0_7_STATUS2		(0x010448)
+#define   NIC_QSET_CQ_0_7_DEBUG			(0x010450)
+
+#define   NIC_QSET_RQ_0_7_CFG			(0x010600)
+#define   NIC_QSET_RQ_0_7_STAT_0_1		(0x010700)
+
+#define   NIC_QSET_SQ_0_7_CFG			(0x010800)
+#define   NIC_QSET_SQ_0_7_THRESH		(0x010810)
+#define   NIC_QSET_SQ_0_7_BASE			(0x010820)
+#define   NIC_QSET_SQ_0_7_HEAD			(0x010828)
+#define   NIC_QSET_SQ_0_7_TAIL			(0x010830)
+#define   NIC_QSET_SQ_0_7_DOOR			(0x010838)
+#define   NIC_QSET_SQ_0_7_STATUS		(0x010840)
+#define   NIC_QSET_SQ_0_7_DEBUG			(0x010848)
+#define   NIC_QSET_SQ_0_7_STAT_0_1		(0x010900)
+
+#define   NIC_QSET_RBDR_0_1_CFG			(0x010C00)
+#define   NIC_QSET_RBDR_0_1_THRESH		(0x010C10)
+#define   NIC_QSET_RBDR_0_1_BASE		(0x010C20)
+#define   NIC_QSET_RBDR_0_1_HEAD		(0x010C28)
+#define   NIC_QSET_RBDR_0_1_TAIL		(0x010C30)
+#define   NIC_QSET_RBDR_0_1_DOOR		(0x010C38)
+#define   NIC_QSET_RBDR_0_1_STATUS0		(0x010C40)
+#define   NIC_QSET_RBDR_0_1_STATUS1		(0x010C48)
+#define   NIC_QSET_RBDR_0_1_PREFETCH_STATUS	(0x010C50)
+
+#define   NIC_VF_MSIX_VECTOR_0_19_ADDR		(0x000000)
+#define   NIC_VF_MSIX_VECTOR_0_19_CTL		(0x000008)
+#define   NIC_VF_MSIX_PBA			(0x0F0000)
+
+/* Offsets within registers */
+#define   NIC_MSIX_VEC_SHIFT			4
+#define   NIC_Q_NUM_SHIFT			18
+#define   NIC_QS_ID_SHIFT			21
+#define   NIC_VF_NUM_SHIFT			21
+
+/* Port kind configuration register */
+struct pkind_cfg {
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u64 reserved_42_63:22;
+	u64 hdr_sl:5;	/* Header skip length */
+	u64 rx_hdr:3;	/* TNS Receive header present */
+	u64 lenerr_en:1;/* L2 length error check enable */
+	u64 reserved_32_32:1;
+	u64 maxlen:16;	/* Max frame size */
+	u64 minlen:16;	/* Min frame size */
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u64 minlen:16;
+	u64 maxlen:16;
+	u64 reserved_32_32:1;
+	u64 lenerr_en:1;
+	u64 rx_hdr:3;
+	u64 hdr_sl:5;
+	u64 reserved_42_63:22;
+#endif
+};
+
+#endif /* NIC_REG_H */
diff --git a/drivers/net/cavium/nicvf_main.c b/drivers/net/cavium/nicvf_main.c
new file mode 100644
index 0000000000..eeb98b4f51
--- /dev/null
+++ b/drivers/net/cavium/nicvf_main.c
@@ -0,0 +1,553 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2018 Cavium Inc
+ */
+#include <config.h>
+#include <common.h>
+#include <dm.h>
+#include <environment.h>
+#include <pci.h>
+#include <net.h>
+#include <misc.h>
+#include <netdev.h>
+#include <malloc.h>
+#include <asm/io.h>
+#include <asm/arch-thunderx/thunderx_vnic.h>
+
+#include "nic_reg.h"
+#include "nic.h"
+#include "nicvf_queues.h"
+#include "thunder_bgx.h"
+
+/* Register read/write APIs */
+void nicvf_reg_write(struct nicvf *nic, u64 offset, u64 val)
+{
+	writeq(val, nic->reg_base + offset);
+}
+
+u64 nicvf_reg_read(struct nicvf *nic, u64 offset)
+{
+	return readq(nic->reg_base + offset);
+}
+
+void nicvf_queue_reg_write(struct nicvf *nic, u64 offset,
+			   u64 qidx, u64 val)
+{
+	void *addr = nic->reg_base + offset;
+
+	writeq(val, (void *)(addr + (qidx << NIC_Q_NUM_SHIFT)));
+}
+
+u64 nicvf_queue_reg_read(struct nicvf *nic, u64 offset, u64 qidx)
+{
+	void *addr = nic->reg_base + offset;
+
+	return readq((void *)(addr + (qidx << NIC_Q_NUM_SHIFT)));
+}
+
+static void  nicvf_handle_mbx_intr(struct nicvf *nic);
+
+/* VF -> PF mailbox communication */
+static void nicvf_write_to_mbx(struct nicvf *nic, union nic_mbx *mbx)
+{
+	u64 *msg = (u64 *)mbx;
+
+	nicvf_reg_write(nic, NIC_VF_PF_MAILBOX_0_1 + 0, msg[0]);
+	nicvf_reg_write(nic, NIC_VF_PF_MAILBOX_0_1 + 8, msg[1]);
+}
+
+int nicvf_send_msg_to_pf(struct nicvf *nic, union nic_mbx *mbx)
+{
+	int timeout = NIC_PF_VF_MBX_TIMEOUT;
+	int sleep = 10;
+
+	nic->pf_acked = false;
+	nic->pf_nacked = false;
+
+	nicvf_write_to_mbx(nic, mbx);
+
+	nic_handle_mbx_intr(nic->nicpf, nic->vf_id);
+
+	/* Wait for previous message to be acked, timeout 2sec */
+	while (!nic->pf_acked) {
+		if (nic->pf_nacked)
+			return -1;
+		mdelay(sleep);
+		nicvf_handle_mbx_intr(nic);
+
+		if (nic->pf_acked)
+			break;
+		timeout -= sleep;
+		if (!timeout) {
+			printf("PF didn't ack to mbox msg %d from VF%d\n",
+			       (mbx->msg.msg & 0xFF), nic->vf_id);
+			return -1;
+		}
+	}
+
+	return 0;
+}
+
+
+/* Checks if VF is able to comminicate with PF
+* and also gets the VNIC number this VF is associated to.
+*/
+static int nicvf_check_pf_ready(struct nicvf *nic)
+{
+	union nic_mbx mbx = {};
+
+	mbx.msg.msg = NIC_MBOX_MSG_READY;
+	if (nicvf_send_msg_to_pf(nic, &mbx)) {
+		printf("PF didn't respond to READY msg\n");
+		return 0;
+	}
+
+	return 1;
+}
+
+static void  nicvf_handle_mbx_intr(struct nicvf *nic)
+{
+	union nic_mbx mbx = {};
+	u64 *mbx_data;
+	u64 mbx_addr;
+	int i;
+
+	mbx_addr = NIC_VF_PF_MAILBOX_0_1;
+	mbx_data = (u64 *)&mbx;
+
+	for (i = 0; i < NIC_PF_VF_MAILBOX_SIZE; i++) {
+		*mbx_data = nicvf_reg_read(nic, mbx_addr);
+		mbx_data++;
+		mbx_addr += sizeof(u64);
+	}
+
+	debug("Mbox message: msg: 0x%x\n", mbx.msg.msg);
+	switch (mbx.msg.msg) {
+	case NIC_MBOX_MSG_READY:
+		nic->pf_acked = true;
+		nic->vf_id = mbx.nic_cfg.vf_id & 0x7F;
+		nic->tns_mode = mbx.nic_cfg.tns_mode & 0x7F;
+		nic->node = mbx.nic_cfg.node_id;
+		if (!nic->set_mac_pending)
+			memcpy(nic->netdev->enetaddr,
+					mbx.nic_cfg.mac_addr, 6);
+		nic->loopback_supported = mbx.nic_cfg.loopback_supported;
+		nic->link_up = false;
+		nic->duplex = 0;
+		nic->speed = 0;
+		break;
+	case NIC_MBOX_MSG_ACK:
+		nic->pf_acked = true;
+		break;
+	case NIC_MBOX_MSG_NACK:
+		nic->pf_nacked = true;
+		break;
+	case NIC_MBOX_MSG_BGX_LINK_CHANGE:
+		nic->pf_acked = true;
+		nic->link_up = mbx.link_status.link_up;
+		nic->duplex = mbx.link_status.duplex;
+		nic->speed = mbx.link_status.speed;
+		if (nic->link_up) {
+			printf("%s: Link is Up %d Mbps %s\n",
+				    nic->netdev->name, nic->speed,
+				    nic->duplex == 1 ?
+				"Full duplex" : "Half duplex");
+		} else {
+			printf("%s: Link is Down\n",
+				    nic->netdev->name);
+		}
+		break;
+	default:
+		printf("Invalid message from PF, msg 0x%x\n", mbx.msg.msg);
+		break;
+	}
+
+	nicvf_clear_intr(nic, NICVF_INTR_MBOX, 0);
+}
+
+static int nicvf_hw_set_mac_addr(struct nicvf *nic, struct eth_device *netdev)
+{
+	union nic_mbx mbx = {};
+
+	mbx.mac.msg = NIC_MBOX_MSG_SET_MAC;
+	mbx.mac.vf_id = nic->vf_id;
+	memcpy(mbx.mac.mac_addr, netdev->enetaddr, 6);
+
+	return nicvf_send_msg_to_pf(nic, &mbx);
+}
+
+static void nicvf_config_cpi(struct nicvf *nic)
+{
+	union nic_mbx mbx = {};
+
+	mbx.cpi_cfg.msg = NIC_MBOX_MSG_CPI_CFG;
+	mbx.cpi_cfg.vf_id = nic->vf_id;
+	mbx.cpi_cfg.cpi_alg = nic->cpi_alg;
+	mbx.cpi_cfg.rq_cnt = nic->qs->rq_cnt;
+
+	nicvf_send_msg_to_pf(nic, &mbx);
+}
+
+
+static int nicvf_init_resources(struct nicvf *nic)
+{
+	int err;
+
+	nic->num_qs = 1;
+
+	/* Enable Qset */
+	nicvf_qset_config(nic, true);
+
+	/* Initialize queues and HW for data transfer */
+	err = nicvf_config_data_transfer(nic, true);
+
+	if (err) {
+		printf("Failed to alloc/config VF's QSet resources\n");
+		return err;
+	}
+	return 0;
+}
+
+void nicvf_free_pkt(struct nicvf *nic, void *pkt)
+{
+	free(pkt);
+}
+
+static void nicvf_snd_pkt_handler(struct nicvf *nic,
+				  struct cmp_queue *cq,
+				  void *cq_desc, int cqe_type)
+{
+	struct cqe_send_t *cqe_tx;
+	struct snd_queue *sq;
+	struct sq_hdr_subdesc *hdr;
+
+	cqe_tx = (struct cqe_send_t *)cq_desc;
+	sq = &nic->qs->sq[cqe_tx->sq_idx];
+
+	hdr = (struct sq_hdr_subdesc *)GET_SQ_DESC(sq, cqe_tx->sqe_ptr);
+	if (hdr->subdesc_type != SQ_DESC_TYPE_HEADER)
+		return;
+
+	nicvf_check_cqe_tx_errs(nic, cq, cq_desc);
+	nicvf_put_sq_desc(sq, hdr->subdesc_cnt + 1);
+}
+
+static int nicvf_rcv_pkt_handler(struct nicvf *nic,
+				 struct cmp_queue *cq, void *cq_desc,
+				 void **ppkt, int cqe_type)
+{
+	void *pkt;
+
+	size_t pkt_len;
+	struct cqe_rx_t *cqe_rx = (struct cqe_rx_t *)cq_desc;
+	int err = 0;
+
+	/* Check for errors */
+	err = nicvf_check_cqe_rx_errs(nic, cq, cq_desc);
+	if (err && !cqe_rx->rb_cnt)
+		return -1;
+
+	pkt = nicvf_get_rcv_pkt(nic, cq_desc, &pkt_len);
+	if (!pkt) {
+		debug("Packet not received\n");
+		return -1;
+	}
+
+	if (pkt)
+		*ppkt = pkt;
+
+	return pkt_len;
+}
+
+static int nicvf_cq_handler(struct nicvf *nic, void **ppkt, int *pkt_len)
+{
+	int cq_qnum = 0;
+	int processed_sq_cqe = 0;
+	int processed_rq_cqe = 0;
+	int processed_cqe = 0;
+
+	unsigned long cqe_count, cqe_head;
+	struct queue_set *qs = nic->qs;
+	struct cmp_queue *cq = &qs->cq[cq_qnum];
+	struct cqe_rx_t *cq_desc;
+
+	/* Get num of valid CQ entries expect next one to be SQ completion */
+	cqe_count = nicvf_queue_reg_read(nic, NIC_QSET_CQ_0_7_STATUS, cq_qnum);
+	cqe_count &= 0xFFFF;
+	if (!cqe_count)
+		return 0;
+
+	/* Get head of the valid CQ entries */
+	cqe_head = nicvf_queue_reg_read(nic, NIC_QSET_CQ_0_7_HEAD, cq_qnum) >> 9;
+	cqe_head &= 0xFFFF;
+
+	if (cqe_count) {
+		/* Get the CQ descriptor */
+		cq_desc = (struct cqe_rx_t *)GET_CQ_DESC(cq, cqe_head);
+		cqe_head++;
+		cqe_head &= (cq->dmem.q_len - 1);
+		/* Initiate prefetch for next descriptor */
+		prefetch((struct cqe_rx_t *)GET_CQ_DESC(cq, cqe_head));
+
+		switch (cq_desc->cqe_type) {
+		case CQE_TYPE_RX:
+			debug("%s: Got Rx CQE\n", nic->netdev->name);
+			*pkt_len = nicvf_rcv_pkt_handler(nic, cq, cq_desc, ppkt, CQE_TYPE_RX);
+			processed_rq_cqe++;
+			break;
+		case CQE_TYPE_SEND:
+			debug("%s: Got Tx CQE\n", nic->netdev->name);
+			nicvf_snd_pkt_handler(nic, cq, cq_desc, CQE_TYPE_SEND);
+			processed_sq_cqe++;
+			break;
+		default:
+			debug("%s: Got CQ type %u\n", nic->netdev->name, cq_desc->cqe_type);
+			break;
+		}
+		processed_cqe++;
+	}
+
+
+	/* Dequeue CQE */
+	nicvf_queue_reg_write(nic, NIC_QSET_CQ_0_7_DOOR,
+			      cq_qnum, processed_cqe);
+
+	asm volatile ("dsb sy");
+
+	return (processed_sq_cqe | processed_rq_cqe)  ;
+}
+
+/* Qset error interrupt handler
+ *
+ * As of now only CQ errors are handled
+ */
+void nicvf_handle_qs_err(struct nicvf *nic)
+{
+	struct queue_set *qs = nic->qs;
+	int qidx;
+	u64 status;
+
+	/* Check if it is CQ err */
+	for (qidx = 0; qidx < qs->cq_cnt; qidx++) {
+		status = nicvf_queue_reg_read(nic, NIC_QSET_CQ_0_7_STATUS,
+					      qidx);
+		if (!(status & CQ_ERR_MASK))
+			continue;
+		/* Process already queued CQEs and reconfig CQ */
+		nicvf_sq_disable(nic, qidx);
+		nicvf_cmp_queue_config(nic, qs, qidx, true);
+		nicvf_sq_free_used_descs(nic->netdev, &qs->sq[qidx], qidx);
+		nicvf_sq_enable(nic, &qs->sq[qidx], qidx);
+	}
+}
+
+int nicvf_xmit(struct eth_device *netdev, void *pkt, int pkt_len)
+{
+	struct nicvf *nic = netdev->priv;
+	int ret = 0;
+	int rcv_len = 0;
+	unsigned int timeout = 5000;
+	void *rpkt = NULL;
+
+	if (!nicvf_sq_append_pkt(nic, pkt, pkt_len)) {
+		printf("VF%d: TX ring full\n", nic->vf_id);
+		return -1;
+	}
+
+	/* check and update CQ for pkt sent */
+	while (!ret && timeout--) {
+		ret = nicvf_cq_handler(nic, &rpkt, &rcv_len);
+		if (!ret)
+		{
+			debug("%s: %d, Not sent\n", __FUNCTION__, __LINE__);
+			udelay(10);
+		}
+	}
+
+	return 0;
+}
+
+int nicvf_recv(struct eth_device *netdev)
+{
+	struct nicvf *nic = netdev->priv;
+	void *pkt;
+	int pkt_len = 0;
+#ifdef DEBUG
+	u8 *dpkt;
+	int i, j;
+#endif
+
+	nicvf_cq_handler(nic, &pkt, &pkt_len);
+
+	if (pkt_len) {
+#ifdef DEBUG
+		dpkt = pkt;
+		printf("RX packet contents:\n");
+		for (i = 0; i < 8; i++) {
+			puts("\t");
+			for (j = 0; j < 10; j++) {
+				printf("%02x ", dpkt[i * 10 + j]);
+			}
+			puts("\n");
+		}
+#endif
+		net_process_received_packet(pkt, pkt_len);
+		nicvf_refill_rbdr(nic);
+	}
+
+	return pkt_len;
+}
+
+void nicvf_stop(struct eth_device *netdev)
+{
+	struct nicvf *nic = netdev->priv;
+
+	if (!nic->open)
+		return;
+
+	/* Free resources */
+	nicvf_config_data_transfer(nic, false);
+
+	/* Disable HW Qset */
+	nicvf_qset_config(nic, false);
+
+	nic->open = false;
+}
+
+int nicvf_open(struct eth_device *netdev, bd_t *bis)
+{
+	int err;
+	struct nicvf *nic = netdev->priv;
+
+	nicvf_hw_set_mac_addr(nic, netdev);
+
+	/* Configure CPI alorithm */
+	nic->cpi_alg = CPI_ALG_NONE;
+	nicvf_config_cpi(nic);
+
+	/* Initialize the queues */
+	err = nicvf_init_resources(nic);
+	if (err)
+		return -1;
+
+	if (!nicvf_check_pf_ready(nic)) {
+		return -1;
+	}
+
+	nic->open = true;
+
+	/* Make sure queue initialization is written */
+	asm volatile("dsb sy");
+
+	return 0;
+}
+
+int nicvf_initialize(struct udevice *pdev, int vf_num)
+{
+	struct eth_device *netdev = NULL;
+	struct nicvf *nicvf = NULL;
+	int    ret;
+	size_t size;
+
+	netdev = calloc(1, sizeof(struct eth_device));
+
+	if (!netdev) {
+		ret = -ENOMEM;
+		goto fail;
+	}
+
+	nicvf = calloc(1, sizeof(struct nicvf));
+
+	if (!nicvf) {
+		ret = -ENOMEM;
+		goto fail;
+	}
+
+	netdev->priv = nicvf;
+	nicvf->netdev = netdev;
+	nicvf->nicpf = dev_get_priv(pdev);
+	nicvf->vf_id = vf_num;
+
+
+	/* Enable TSO support */
+	nicvf->hw_tso = true;
+
+	nicvf->reg_base = dm_pci_map_bar(pdev, 9, &size, PCI_REGION_MEM);
+
+	nicvf->reg_base += size * nicvf->vf_id;
+
+	debug("nicvf->reg_base: %p\n", nicvf->reg_base);
+
+	if (!nicvf->reg_base) {
+		printf("Cannot map config register space, aborting\n");
+		ret = -1;
+		goto fail;
+	}
+
+	ret = nicvf_set_qset_resources(nicvf);
+	if (ret)
+		return -1;
+
+	snprintf(netdev->name, sizeof(netdev->name), "vnic%u", nicvf->vf_id);
+
+	netdev->halt = nicvf_stop;
+	netdev->init = nicvf_open;
+	netdev->send = nicvf_xmit;
+	netdev->recv = nicvf_recv;
+
+	if (!eth_env_get_enetaddr_by_index("eth", nicvf->vf_id, netdev->enetaddr)) {
+		eth_env_get_enetaddr("ethaddr", netdev->enetaddr);
+		netdev->enetaddr[5] += nicvf->vf_id;
+	}
+
+	ret = eth_register(netdev);
+
+	if (!ret)
+		return 0;
+
+	printf("Failed to register netdevice\n");
+
+fail:
+	if (nicvf)
+		free(nicvf);
+	if(netdev)
+		free(netdev);
+	return ret;
+}
+
+int thunderx_vnic_probe(struct udevice *dev)
+{
+	void *regs;
+	size_t size;
+
+	regs = dm_pci_map_bar(dev, 9, &size, PCI_REGION_MEM);
+
+	debug("%s: %d, regs: %p\n", __FUNCTION__, __LINE__, regs);
+
+	return 0;
+}
+
+static const struct misc_ops thunderx_vnic_ops = {
+};
+
+static const struct udevice_id thunderx_vnic_ids[] = {
+	{ .compatible = "cavium,vnic" },
+	{}
+};
+
+U_BOOT_DRIVER(thunderx_vnic) = {
+	.name	= "thunderx_vnic",
+	.id	= UCLASS_MISC,
+	.probe	= thunderx_vnic_probe,
+	.of_match = thunderx_vnic_ids,
+	.ops	= &thunderx_vnic_ops,
+	.priv_auto_alloc_size = sizeof(struct nicvf),
+};
+
+static struct pci_device_id thunderx_vnic_supported[] = {
+	{ PCI_VDEVICE(CAVIUM, PCI_DEVICE_ID_THUNDERX_NIC_VF_1) },
+	{ PCI_VDEVICE(CAVIUM, PCI_DEVICE_ID_THUNDERX_NIC_VF) },
+	{}
+};
+
+U_BOOT_PCI_DEVICE(thunderx_vnic, thunderx_vnic_supported);
diff --git a/drivers/net/cavium/nicvf_queues.c b/drivers/net/cavium/nicvf_queues.c
new file mode 100644
index 0000000000..562b528e44
--- /dev/null
+++ b/drivers/net/cavium/nicvf_queues.c
@@ -0,0 +1,1123 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2018 Cavium Inc
+ */
+#include <config.h>
+#include <common.h>
+#include <net.h>
+#include <netdev.h>
+#include <malloc.h>
+
+#include "nic_reg.h"
+#include "nic.h"
+#include "q_struct.h"
+#include "nicvf_queues.h"
+
+/* Poll a register for a specific value */
+static int nicvf_poll_reg(struct nicvf *nic, int qidx,
+			  u64 reg, int bit_pos, int bits, int val)
+{
+	u64 bit_mask;
+	u64 reg_val;
+	int timeout = 10;
+
+	bit_mask = (1ULL << bits) - 1;
+	bit_mask = (bit_mask << bit_pos);
+
+	while (timeout) {
+		reg_val = nicvf_queue_reg_read(nic, reg, qidx);
+		if (((reg_val & bit_mask) >> bit_pos) == val)
+			return 0;
+		udelay(2000);
+		timeout--;
+	}
+	printf("Poll on reg 0x%llx failed\n", reg);
+	return 1;
+}
+
+/* Allocate memory for a queue's descriptors */
+static int nicvf_alloc_q_desc_mem(struct nicvf *nic, struct q_desc_mem *dmem,
+				  int q_len, int desc_size, int align_bytes)
+{
+	dmem->q_len = q_len;
+	dmem->size = (desc_size * q_len) + align_bytes;
+	/* Save address, need it while freeing */
+	dmem->unalign_base = calloc(1, dmem->size);
+	dmem->dma = (uintptr_t)dmem->unalign_base;
+	if (!dmem->unalign_base)
+		return -ENOMEM;
+
+	/* Align memory address for 'align_bytes' */
+	dmem->phys_base = NICVF_ALIGNED_ADDR((u64)dmem->dma, align_bytes);
+	dmem->base = dmem->unalign_base + (dmem->phys_base - dmem->dma);
+
+	return 0;
+}
+
+/* Free queue's descriptor memory */
+static void nicvf_free_q_desc_mem(struct nicvf *nic, struct q_desc_mem *dmem)
+{
+	if (!dmem)
+		return;
+
+	free(dmem->unalign_base);
+	dmem->unalign_base = NULL;
+	dmem->base = NULL;
+}
+
+
+static void *nicvf_rb_ptr_to_pkt(struct nicvf *nic, uintptr_t rb_ptr)
+{
+	return (void *)rb_ptr;
+}
+
+static int nicvf_init_rbdr(struct nicvf *nic, struct rbdr *rbdr,
+			    int ring_len, int buf_size)
+{
+	int idx;
+	uintptr_t rbuf;
+	struct rbdr_entry_t *desc;
+
+	if (nicvf_alloc_q_desc_mem(nic, &rbdr->dmem, ring_len,
+				   sizeof(struct rbdr_entry_t),
+				   NICVF_RCV_BUF_ALIGN_BYTES)) {
+		printf("Unable to allocate memory for rcv buffer ring\n");
+		return -1;
+	}
+
+	rbdr->desc = rbdr->dmem.base;
+	/* Buffer size has to be in multiples of 128 bytes */
+	rbdr->dma_size = buf_size;
+	rbdr->enable = true;
+	rbdr->thresh = RBDR_THRESH;
+
+	debug("%s: %d: allocating %lld bytes for rcv buffers\n",
+	      __FUNCTION__, __LINE__,
+	      ring_len * buf_size + NICVF_RCV_BUF_ALIGN_BYTES);
+	rbdr->buf_mem = (uintptr_t)calloc(1, ring_len * buf_size
+						+ NICVF_RCV_BUF_ALIGN_BYTES);
+
+	if (!rbdr->buf_mem) {
+		printf("Unable to allocate memory for rcv buffers\n");
+		return -1;
+	}
+
+	rbdr->buffers = NICVF_ALIGNED_ADDR(rbdr->buf_mem, NICVF_RCV_BUF_ALIGN_BYTES);
+
+	debug("%s: %d: rbdr->buf_mem: %lx, rbdr->buffers: %lx\n",
+		__FUNCTION__, __LINE__, rbdr->buf_mem, rbdr->buffers);
+
+	for (idx = 0; idx < ring_len; idx++) {
+		rbuf = rbdr->buffers + DMA_BUFFER_LEN * idx;
+		desc = GET_RBDR_DESC(rbdr, idx);
+		desc->buf_addr = rbuf >> NICVF_RCV_BUF_ALIGN;
+		flush_dcache_range((uintptr_t)desc, (uintptr_t)desc + sizeof(desc));
+	}
+	return 0;
+}
+
+/* Free RBDR ring and its receive buffers */
+static void nicvf_free_rbdr(struct nicvf *nic, struct rbdr *rbdr)
+{
+	if (!rbdr)
+		return;
+
+	rbdr->enable = false;
+	if (!rbdr->dmem.base)
+		return;
+
+	debug("%s: %d: rbdr->buf_mem: %p\n", __FUNCTION__,
+	      __LINE__, (void *)rbdr->buf_mem);
+	free((void *)rbdr->buf_mem);
+
+	/* Free RBDR ring */
+	nicvf_free_q_desc_mem(nic, &rbdr->dmem);
+}
+
+/* Refill receive buffer descriptors with new buffers.
+ */
+void nicvf_refill_rbdr(struct nicvf *nic)
+{
+	struct queue_set *qs = nic->qs;
+	int rbdr_idx = qs->rbdr_cnt;
+	unsigned long qcount, head, tail, rb_cnt;
+	struct rbdr *rbdr;
+
+	if (!rbdr_idx)
+		return;
+	rbdr_idx--;
+	rbdr = &qs->rbdr[rbdr_idx];
+	/* Check if it's enabled */
+	if (!rbdr->enable) {
+		printf("Receive queue %d is disabled\n", rbdr_idx);
+		return;
+	}
+
+	/* check if valid descs reached or crossed threshold level */
+	qcount = nicvf_queue_reg_read(nic, NIC_QSET_RBDR_0_1_STATUS0, rbdr_idx);
+	head = nicvf_queue_reg_read(nic, NIC_QSET_RBDR_0_1_HEAD, rbdr_idx);
+	tail = nicvf_queue_reg_read(nic, NIC_QSET_RBDR_0_1_TAIL, rbdr_idx);
+
+	qcount &= 0x7FFFF;
+
+	rb_cnt = qs->rbdr_len - qcount - 1;
+
+	debug("%s: %d: qcount: %lu, head: %lx, tail: %lx, rb_cnt: %lu\n",
+	      __FUNCTION__, __LINE__, qcount, head, tail, rb_cnt);
+
+
+	/* Notify HW */
+	nicvf_queue_reg_write(nic, NIC_QSET_RBDR_0_1_DOOR, rbdr_idx, rb_cnt);
+
+	asm volatile ("dsb sy");
+}
+
+
+/* TBD: how to handle full packets received in CQ
+ * i.e conversion of buffers into SKBs
+ */
+static int nicvf_init_cmp_queue(struct nicvf *nic,
+				struct cmp_queue *cq, int q_len)
+{
+	if (nicvf_alloc_q_desc_mem(nic, &cq->dmem, q_len,
+				   CMP_QUEUE_DESC_SIZE,
+				   NICVF_CQ_BASE_ALIGN_BYTES)) {
+		printf("Unable to allocate memory for completion queue\n");
+		return -1;
+	}
+	cq->desc = cq->dmem.base;
+	if (!pass1_silicon(nic->rev_id, nic->nicpf->hw->model_id))
+		cq->thresh = CMP_QUEUE_CQE_THRESH;
+	else
+		cq->thresh = 0;
+	cq->intr_timer_thresh = CMP_QUEUE_TIMER_THRESH;
+
+	return 0;
+}
+
+static void nicvf_free_cmp_queue(struct nicvf *nic, struct cmp_queue *cq)
+{
+	if (!cq)
+		return;
+	if (!cq->dmem.base)
+		return;
+
+	nicvf_free_q_desc_mem(nic, &cq->dmem);
+}
+
+static int nicvf_init_snd_queue(struct nicvf *nic,
+				struct snd_queue *sq, int q_len)
+{
+	if (nicvf_alloc_q_desc_mem(nic, &sq->dmem, q_len,
+				   SND_QUEUE_DESC_SIZE,
+				   NICVF_SQ_BASE_ALIGN_BYTES)) {
+		printf("Unable to allocate memory for send queue\n");
+		return -1;
+	}
+
+	sq->desc = sq->dmem.base;
+	sq->skbuff = calloc(q_len, sizeof(u64));
+	sq->head = 0;
+	sq->tail = 0;
+	sq->free_cnt = q_len - 1;
+	sq->thresh = SND_QUEUE_THRESH;
+
+	return 0;
+}
+
+static void nicvf_free_snd_queue(struct nicvf *nic, struct snd_queue *sq)
+{
+	if (!sq)
+		return;
+	if (!sq->dmem.base)
+		return;
+
+	debug("%s: %d\n", __FUNCTION__, __LINE__);
+	free(sq->skbuff);
+
+	nicvf_free_q_desc_mem(nic, &sq->dmem);
+}
+
+static void nicvf_reclaim_snd_queue(struct nicvf *nic,
+				    struct queue_set *qs, int qidx)
+{
+	/* Disable send queue */
+	nicvf_queue_reg_write(nic, NIC_QSET_SQ_0_7_CFG, qidx, 0);
+	/* Check if SQ is stopped */
+	if (nicvf_poll_reg(nic, qidx, NIC_QSET_SQ_0_7_STATUS, 21, 1, 0x01))
+		return;
+	/* Reset send queue */
+	nicvf_queue_reg_write(nic, NIC_QSET_SQ_0_7_CFG, qidx, NICVF_SQ_RESET);
+}
+
+static void nicvf_reclaim_rcv_queue(struct nicvf *nic,
+				    struct queue_set *qs, int qidx)
+{
+	union nic_mbx mbx = {};
+
+	/* Make sure all packets in the pipeline are written back into mem */
+	mbx.msg.msg = NIC_MBOX_MSG_RQ_SW_SYNC;
+	nicvf_send_msg_to_pf(nic, &mbx);
+}
+
+static void nicvf_reclaim_cmp_queue(struct nicvf *nic,
+				    struct queue_set *qs, int qidx)
+{
+	/* Disable timer threshold (doesn't get reset upon CQ reset */
+	nicvf_queue_reg_write(nic, NIC_QSET_CQ_0_7_CFG2, qidx, 0);
+	/* Disable completion queue */
+	nicvf_queue_reg_write(nic, NIC_QSET_CQ_0_7_CFG, qidx, 0);
+	/* Reset completion queue */
+	nicvf_queue_reg_write(nic, NIC_QSET_CQ_0_7_CFG, qidx, NICVF_CQ_RESET);
+}
+
+static void nicvf_reclaim_rbdr(struct nicvf *nic,
+			       struct rbdr *rbdr, int qidx)
+{
+	u64 tmp, fifo_state;
+	int timeout = 10;
+
+	/* Save head and tail pointers for feeing up buffers */
+	rbdr->head = nicvf_queue_reg_read(nic,
+					  NIC_QSET_RBDR_0_1_HEAD,
+					  qidx) >> 3;
+	rbdr->tail = nicvf_queue_reg_read(nic,
+					  NIC_QSET_RBDR_0_1_TAIL,
+					  qidx) >> 3;
+
+	/* If RBDR FIFO is in 'FAIL' state then do a reset first
+	 * before relaiming.
+	 */
+	fifo_state = nicvf_queue_reg_read(nic, NIC_QSET_RBDR_0_1_STATUS0, qidx);
+	if (((fifo_state >> 62) & 0x03) == 0x3)
+		nicvf_queue_reg_write(nic, NIC_QSET_RBDR_0_1_CFG,
+				      qidx, NICVF_RBDR_RESET);
+
+	/* Disable RBDR */
+	nicvf_queue_reg_write(nic, NIC_QSET_RBDR_0_1_CFG, qidx, 0);
+	if (nicvf_poll_reg(nic, qidx, NIC_QSET_RBDR_0_1_STATUS0, 62, 2, 0x00))
+		return;
+	while (1) {
+		tmp = nicvf_queue_reg_read(nic,
+					   NIC_QSET_RBDR_0_1_PREFETCH_STATUS,
+					   qidx);
+		if ((tmp & 0xFFFFFFFF) == ((tmp >> 32) & 0xFFFFFFFF))
+			break;
+		mdelay(2000);
+		timeout--;
+		if (!timeout) {
+			printf("Failed polling on prefetch status\n");
+			return;
+		}
+	}
+	nicvf_queue_reg_write(nic, NIC_QSET_RBDR_0_1_CFG,
+			      qidx, NICVF_RBDR_RESET);
+
+	if (nicvf_poll_reg(nic, qidx, NIC_QSET_RBDR_0_1_STATUS0, 62, 2, 0x02))
+		return;
+	nicvf_queue_reg_write(nic, NIC_QSET_RBDR_0_1_CFG, qidx, 0x00);
+	if (nicvf_poll_reg(nic, qidx, NIC_QSET_RBDR_0_1_STATUS0, 62, 2, 0x00))
+		return;
+}
+
+
+/* Configures receive queue */
+static void nicvf_rcv_queue_config(struct nicvf *nic, struct queue_set *qs,
+				   int qidx, bool enable)
+{
+	union nic_mbx mbx = {};
+	struct rcv_queue *rq;
+	union {
+		struct rq_cfg s;
+		u64    u;
+	} rq_cfg;
+
+	rq = &qs->rq[qidx];
+	rq->enable = enable;
+
+	/* Disable receive queue */
+	nicvf_queue_reg_write(nic, NIC_QSET_RQ_0_7_CFG, qidx, 0);
+
+	if (!rq->enable) {
+		nicvf_reclaim_rcv_queue(nic, qs, qidx);
+		return;
+	}
+
+	rq->cq_qs = qs->vnic_id;
+	rq->cq_idx = qidx;
+	rq->start_rbdr_qs = qs->vnic_id;
+	rq->start_qs_rbdr_idx = qs->rbdr_cnt - 1;
+	rq->cont_rbdr_qs = qs->vnic_id;
+	rq->cont_qs_rbdr_idx = qs->rbdr_cnt - 1;
+	/* all writes of RBDR data to be loaded into L2 Cache as well*/
+	rq->caching = 1;
+
+	/* Send a mailbox msg to PF to config RQ */
+	mbx.rq.msg = NIC_MBOX_MSG_RQ_CFG;
+	mbx.rq.qs_num = qs->vnic_id;
+	mbx.rq.rq_num = qidx;
+	mbx.rq.cfg = (rq->caching << 26) | (rq->cq_qs << 19) |
+			  (rq->cq_idx << 16) | (rq->cont_rbdr_qs << 9) |
+			  (rq->cont_qs_rbdr_idx << 8) |
+			  (rq->start_rbdr_qs << 1) | (rq->start_qs_rbdr_idx);
+	nicvf_send_msg_to_pf(nic, &mbx);
+
+	mbx.rq.msg = NIC_MBOX_MSG_RQ_BP_CFG;
+	mbx.rq.cfg = (1ULL << 63) | (1ULL << 62) | (qs->vnic_id << 0);
+	nicvf_send_msg_to_pf(nic, &mbx);
+
+	/* RQ drop config
+	 * Enable CQ drop to reserve sufficient CQEs for all tx packets
+	 */
+	mbx.rq.msg = NIC_MBOX_MSG_RQ_DROP_CFG;
+	mbx.rq.cfg = (1ULL << 62) | (RQ_CQ_DROP << 8);
+	nicvf_send_msg_to_pf(nic, &mbx);
+	nicvf_queue_reg_write(nic, NIC_QSET_RQ_GEN_CFG, 0, 0x00);
+
+	/* Enable Receive queue */
+	rq_cfg.s.ena = 1;
+	rq_cfg.s.tcp_ena = 0;
+	nicvf_queue_reg_write(nic, NIC_QSET_RQ_0_7_CFG, qidx, rq_cfg.u);
+}
+
+void nicvf_cmp_queue_config(struct nicvf *nic, struct queue_set *qs,
+			    int qidx, bool enable)
+{
+	struct cmp_queue *cq;
+	union {
+		u64 u;
+		struct cq_cfg s;
+	} cq_cfg;
+
+	cq = &qs->cq[qidx];
+	cq->enable = enable;
+
+	if (!cq->enable) {
+		nicvf_reclaim_cmp_queue(nic, qs, qidx);
+		return;
+	}
+
+	/* Reset completion queue */
+	nicvf_queue_reg_write(nic, NIC_QSET_CQ_0_7_CFG, qidx, NICVF_CQ_RESET);
+
+	if (!cq->enable)
+		return;
+
+	/* Set completion queue base address */
+	nicvf_queue_reg_write(nic, NIC_QSET_CQ_0_7_BASE,
+			      qidx, (u64)(cq->dmem.phys_base));
+
+	/* Enable Completion queue */
+	cq_cfg.s.ena = 1;
+	cq_cfg.s.reset = 0;
+	cq_cfg.s.caching = 0;
+	cq_cfg.s.qsize = CMP_QSIZE;
+	cq_cfg.s.avg_con = 0;
+	nicvf_queue_reg_write(nic, NIC_QSET_CQ_0_7_CFG, qidx, cq_cfg.u);
+
+	/* Set threshold value for interrupt generation */
+	nicvf_queue_reg_write(nic, NIC_QSET_CQ_0_7_THRESH, qidx, cq->thresh);
+	nicvf_queue_reg_write(nic, NIC_QSET_CQ_0_7_CFG2, qidx, cq->intr_timer_thresh);
+}
+
+/* Configures transmit queue */
+static void nicvf_snd_queue_config(struct nicvf *nic, struct queue_set *qs,
+				   int qidx, bool enable)
+{
+	union nic_mbx mbx = {};
+	struct snd_queue *sq;
+
+	union {
+		struct sq_cfg s;
+		u64 u;
+	} sq_cfg;
+
+	sq = &qs->sq[qidx];
+	sq->enable = enable;
+
+	if (!sq->enable) {
+		nicvf_reclaim_snd_queue(nic, qs, qidx);
+		return;
+	}
+
+	/* Reset send queue */
+	nicvf_queue_reg_write(nic, NIC_QSET_SQ_0_7_CFG, qidx, NICVF_SQ_RESET);
+
+	sq->cq_qs = qs->vnic_id;
+	sq->cq_idx = qidx;
+
+	/* Send a mailbox msg to PF to config SQ */
+	mbx.sq.msg = NIC_MBOX_MSG_SQ_CFG;
+	mbx.sq.qs_num = qs->vnic_id;
+	mbx.sq.sq_num = qidx;
+	mbx.sq.sqs_mode = nic->sqs_mode;
+	mbx.sq.cfg = (sq->cq_qs << 3) | sq->cq_idx;
+	nicvf_send_msg_to_pf(nic, &mbx);
+
+	/* Set queue base address */
+	nicvf_queue_reg_write(nic, NIC_QSET_SQ_0_7_BASE,
+			      qidx, (u64)(sq->dmem.phys_base));
+
+	/* Enable send queue  & set queue size */
+	sq_cfg.s.ena = 1;
+	sq_cfg.s.reset = 0;
+	sq_cfg.s.ldwb = 0;
+	sq_cfg.s.qsize = SND_QSIZE;
+	sq_cfg.s.tstmp_bgx_intf = 0;
+	nicvf_queue_reg_write(nic, NIC_QSET_SQ_0_7_CFG, qidx, sq_cfg.u);
+
+	/* Set threshold value for interrupt generation */
+	nicvf_queue_reg_write(nic, NIC_QSET_SQ_0_7_THRESH, qidx, sq->thresh);
+}
+
+/* Configures receive buffer descriptor ring */
+static void nicvf_rbdr_config(struct nicvf *nic, struct queue_set *qs,
+			      int qidx, bool enable)
+{
+	struct rbdr *rbdr;
+	union {
+		struct rbdr_cfg s;
+		u64 u;
+	} rbdr_cfg;
+
+	rbdr = &qs->rbdr[qidx];
+	nicvf_reclaim_rbdr(nic, rbdr, qidx);
+	if (!enable)
+		return;
+
+	/* Set descriptor base address */
+	nicvf_queue_reg_write(nic, NIC_QSET_RBDR_0_1_BASE,
+			      qidx, (u64)(rbdr->dmem.phys_base));
+
+	/* Enable RBDR  & set queue size */
+	/* Buffer size should be in multiples of 128 bytes */
+	rbdr_cfg.s.ena = 1;
+	rbdr_cfg.s.reset = 0;
+	rbdr_cfg.s.ldwb = 0;
+	rbdr_cfg.s.qsize = RBDR_SIZE;
+	rbdr_cfg.s.avg_con = 0;
+	rbdr_cfg.s.lines = rbdr->dma_size / 128;
+	nicvf_queue_reg_write(nic, NIC_QSET_RBDR_0_1_CFG,
+			      qidx, rbdr_cfg.u);
+
+	/* Notify HW */
+	nicvf_queue_reg_write(nic, NIC_QSET_RBDR_0_1_DOOR,
+			      qidx, qs->rbdr_len - 1);
+
+	/* Set threshold value for interrupt generation */
+	nicvf_queue_reg_write(nic, NIC_QSET_RBDR_0_1_THRESH,
+			      qidx, rbdr->thresh - 1);
+}
+
+/* Requests PF to assign and enable Qset */
+void nicvf_qset_config(struct nicvf *nic, bool enable)
+{
+	union nic_mbx mbx = {};
+	struct queue_set *qs = nic->qs;
+	struct qs_cfg *qs_cfg;
+
+	if (!qs) {
+		printf("Qset is still not allocated, don't init queues\n");
+		return;
+	}
+
+	qs->enable = enable;
+	qs->vnic_id = nic->vf_id;
+
+	/* Send a mailbox msg to PF to config Qset */
+	mbx.qs.msg = NIC_MBOX_MSG_QS_CFG;
+	mbx.qs.num = qs->vnic_id;
+#ifdef VNIC_MULTI_QSET_SUPPORT
+	mbx.qs.sqs_count = nic->sqs_count;
+#endif
+
+	mbx.qs.cfg = 0;
+	qs_cfg = (struct qs_cfg *)&mbx.qs.cfg;
+	if (qs->enable) {
+		qs_cfg->ena = 1;
+#ifdef __BIG_ENDIAN
+		qs_cfg->be = 1;
+#endif
+		qs_cfg->vnic = qs->vnic_id;
+	}
+	nicvf_send_msg_to_pf(nic, &mbx);
+}
+
+static void nicvf_free_resources(struct nicvf *nic)
+{
+	int qidx;
+	struct queue_set *qs = nic->qs;
+
+	/* Free receive buffer descriptor ring */
+	for (qidx = 0; qidx < qs->rbdr_cnt; qidx++)
+		nicvf_free_rbdr(nic, &qs->rbdr[qidx]);
+
+	/* Free completion queue */
+	for (qidx = 0; qidx < qs->cq_cnt; qidx++)
+		nicvf_free_cmp_queue(nic, &qs->cq[qidx]);
+
+	/* Free send queue */
+	for (qidx = 0; qidx < qs->sq_cnt; qidx++)
+		nicvf_free_snd_queue(nic, &qs->sq[qidx]);
+}
+
+static int nicvf_alloc_resources(struct nicvf *nic)
+{
+	int qidx;
+	struct queue_set *qs = nic->qs;
+
+	/* Alloc receive buffer descriptor ring */
+	for (qidx = 0; qidx < qs->rbdr_cnt; qidx++) {
+		if (nicvf_init_rbdr(nic, &qs->rbdr[qidx], qs->rbdr_len,
+				    DMA_BUFFER_LEN))
+			goto alloc_fail;
+	}
+
+	/* Alloc send queue */
+	for (qidx = 0; qidx < qs->sq_cnt; qidx++) {
+		if (nicvf_init_snd_queue(nic, &qs->sq[qidx], qs->sq_len))
+			goto alloc_fail;
+	}
+
+	/* Alloc completion queue */
+	for (qidx = 0; qidx < qs->cq_cnt; qidx++) {
+		if (nicvf_init_cmp_queue(nic, &qs->cq[qidx], qs->cq_len))
+			goto alloc_fail;
+	}
+
+	return 0;
+alloc_fail:
+	nicvf_free_resources(nic);
+	return -1;
+}
+
+int nicvf_set_qset_resources(struct nicvf *nic)
+{
+	struct queue_set *qs;
+
+	qs = calloc(1, sizeof(struct queue_set));
+	if (!qs)
+		return -1;
+	nic->qs = qs;
+
+	/* Set count of each queue */
+	qs->rbdr_cnt = RBDR_CNT;
+	qs->rq_cnt = 1;
+	qs->sq_cnt = SND_QUEUE_CNT;
+	qs->cq_cnt = CMP_QUEUE_CNT;
+
+	/* Set queue lengths */
+	qs->rbdr_len = RCV_BUF_COUNT;
+	qs->sq_len = SND_QUEUE_LEN;
+	qs->cq_len = CMP_QUEUE_LEN;
+
+	nic->rx_queues = qs->rq_cnt;
+	nic->tx_queues = qs->sq_cnt;
+
+	return 0;
+}
+
+int nicvf_config_data_transfer(struct nicvf *nic, bool enable)
+{
+	bool disable = false;
+	struct queue_set *qs = nic->qs;
+	int qidx;
+
+	if (!qs)
+		return 0;
+
+	if (enable) {
+		if (nicvf_alloc_resources(nic))
+			return -1;
+
+		for (qidx = 0; qidx < qs->sq_cnt; qidx++)
+			nicvf_snd_queue_config(nic, qs, qidx, enable);
+		for (qidx = 0; qidx < qs->cq_cnt; qidx++)
+			nicvf_cmp_queue_config(nic, qs, qidx, enable);
+		for (qidx = 0; qidx < qs->rbdr_cnt; qidx++)
+			nicvf_rbdr_config(nic, qs, qidx, enable);
+		for (qidx = 0; qidx < qs->rq_cnt; qidx++)
+			nicvf_rcv_queue_config(nic, qs, qidx, enable);
+	} else {
+		for (qidx = 0; qidx < qs->rq_cnt; qidx++)
+			nicvf_rcv_queue_config(nic, qs, qidx, disable);
+		for (qidx = 0; qidx < qs->rbdr_cnt; qidx++)
+			nicvf_rbdr_config(nic, qs, qidx, disable);
+		for (qidx = 0; qidx < qs->sq_cnt; qidx++)
+			nicvf_snd_queue_config(nic, qs, qidx, disable);
+		for (qidx = 0; qidx < qs->cq_cnt; qidx++)
+			nicvf_cmp_queue_config(nic, qs, qidx, disable);
+
+		nicvf_free_resources(nic);
+	}
+
+	return 0;
+}
+
+/* Get a free desc from SQ
+ * returns descriptor ponter & descriptor number
+ */
+static int nicvf_get_sq_desc(struct snd_queue *sq, int desc_cnt)
+{
+	int qentry;
+
+	qentry = sq->tail;
+	sq->free_cnt -= desc_cnt;
+	sq->tail += desc_cnt;
+	sq->tail &= (sq->dmem.q_len - 1);
+
+	return qentry;
+}
+
+/* Free descriptor back to SQ for future use */
+void nicvf_put_sq_desc(struct snd_queue *sq, int desc_cnt)
+{
+	sq->free_cnt += desc_cnt;
+	sq->head += desc_cnt;
+	sq->head &= (sq->dmem.q_len - 1);
+}
+
+static int nicvf_get_nxt_sqentry(struct snd_queue *sq, int qentry)
+{
+	qentry++;
+	qentry &= (sq->dmem.q_len - 1);
+	return qentry;
+}
+
+void nicvf_sq_enable(struct nicvf *nic, struct snd_queue *sq, int qidx)
+{
+	u64 sq_cfg;
+
+	sq_cfg = nicvf_queue_reg_read(nic, NIC_QSET_SQ_0_7_CFG, qidx);
+	sq_cfg |= NICVF_SQ_EN;
+	nicvf_queue_reg_write(nic, NIC_QSET_SQ_0_7_CFG, qidx, sq_cfg);
+	/* Ring doorbell so that H/W restarts processing SQEs */
+	nicvf_queue_reg_write(nic, NIC_QSET_SQ_0_7_DOOR, qidx, 0);
+}
+
+void nicvf_sq_disable(struct nicvf *nic, int qidx)
+{
+	u64 sq_cfg;
+
+	sq_cfg = nicvf_queue_reg_read(nic, NIC_QSET_SQ_0_7_CFG, qidx);
+	sq_cfg &= ~NICVF_SQ_EN;
+	nicvf_queue_reg_write(nic, NIC_QSET_SQ_0_7_CFG, qidx, sq_cfg);
+}
+
+void nicvf_sq_free_used_descs(struct eth_device *netdev, struct snd_queue *sq,
+								int qidx)
+{
+	u64 head;
+	struct nicvf *nic = netdev->priv;
+	struct sq_hdr_subdesc *hdr;
+
+	head = nicvf_queue_reg_read(nic, NIC_QSET_SQ_0_7_HEAD, qidx) >> 4;
+
+	while (sq->head != head) {
+		hdr = (struct sq_hdr_subdesc *)GET_SQ_DESC(sq, sq->head);
+		if (hdr->subdesc_type != SQ_DESC_TYPE_HEADER) {
+			nicvf_put_sq_desc(sq, 1);
+			continue;
+		}
+		nicvf_put_sq_desc(sq, hdr->subdesc_cnt + 1);
+	}
+}
+
+/* Get the number of SQ descriptors needed to xmit this skb */
+static int nicvf_sq_subdesc_required(struct nicvf *nic)
+{
+	int subdesc_cnt = MIN_SQ_DESC_PER_PKT_XMIT;
+
+	return subdesc_cnt;
+}
+
+/* Add SQ HEADER subdescriptor.
+ * First subdescriptor for every send descriptor.
+ */
+static inline void
+nicvf_sq_add_hdr_subdesc(struct nicvf *nic, struct snd_queue *sq, int qentry,
+			 int subdesc_cnt, void *pkt, size_t pkt_len)
+{
+	struct sq_hdr_subdesc *hdr;
+
+	hdr = (struct sq_hdr_subdesc *)GET_SQ_DESC(sq, qentry);
+	sq->skbuff[qentry] = (uintptr_t)pkt;
+
+	memset(hdr, 0, SND_QUEUE_DESC_SIZE);
+	hdr->subdesc_type = SQ_DESC_TYPE_HEADER;
+	/* Enable notification via CQE after processing SQE */
+	hdr->post_cqe = 1;
+	/* No of subdescriptors following this */
+	hdr->subdesc_cnt = subdesc_cnt;
+	hdr->tot_len = pkt_len;
+
+	flush_dcache_range((uintptr_t)hdr,
+			   (uintptr_t)hdr + sizeof(struct sq_hdr_subdesc));
+}
+
+/* SQ GATHER subdescriptor
+ * Must follow HDR descriptor
+ */
+static inline void nicvf_sq_add_gather_subdesc(struct snd_queue *sq, int qentry,
+					       size_t size, uintptr_t data)
+{
+	struct sq_gather_subdesc *gather;
+
+	qentry &= (sq->dmem.q_len - 1);
+	gather = (struct sq_gather_subdesc *)GET_SQ_DESC(sq, qentry);
+
+	memset(gather, 0, SND_QUEUE_DESC_SIZE);
+	gather->subdesc_type = SQ_DESC_TYPE_GATHER;
+	gather->ld_type = NIC_SEND_LD_TYPE_E_LDD;
+	gather->size = size;
+	gather->addr = data;
+
+	flush_dcache_range((uintptr_t)gather,
+			   (uintptr_t)gather + sizeof(struct sq_hdr_subdesc));
+}
+
+/* Append an skb to a SQ for packet transfer. */
+int nicvf_sq_append_pkt(struct nicvf *nic, void *pkt, size_t pkt_size)
+{
+	int subdesc_cnt;
+	int sq_num = 0, qentry;
+	struct queue_set *qs;
+	struct snd_queue *sq;
+
+	qs = nic->qs;
+	sq = &qs->sq[sq_num];
+
+	subdesc_cnt = nicvf_sq_subdesc_required(nic);
+	if (subdesc_cnt > sq->free_cnt)
+		goto append_fail;
+
+	qentry = nicvf_get_sq_desc(sq, subdesc_cnt);
+
+	/* Add SQ header subdesc */
+	nicvf_sq_add_hdr_subdesc(nic, sq, qentry, subdesc_cnt - 1,
+				 pkt, pkt_size);
+
+	/* Add SQ gather subdescs */
+	qentry = nicvf_get_nxt_sqentry(sq, qentry);
+	nicvf_sq_add_gather_subdesc(sq, qentry, pkt_size, (uintptr_t)(pkt));
+
+	flush_dcache_range((uintptr_t)pkt,
+			   (uintptr_t)pkt + pkt_size);
+
+	/* make sure all memory stores are done before ringing doorbell */
+	asm volatile ("dsb sy");
+
+	/* Inform HW to xmit new packet */
+	nicvf_queue_reg_write(nic, NIC_QSET_SQ_0_7_DOOR,
+			      sq_num, subdesc_cnt);
+	return 1;
+
+append_fail:
+	printf("Not enough SQ descriptors to xmit pkt\n");
+	return 0;
+}
+
+static unsigned frag_num(unsigned i)
+{
+#ifdef __BIG_ENDIAN
+	return (i & ~3) + 3 - (i & 3);
+#else
+	return i;
+#endif
+}
+
+void *nicvf_get_rcv_pkt(struct nicvf *nic, void *cq_desc, size_t *pkt_len)
+{
+	int frag;
+	int payload_len = 0;
+	void *pkt = NULL;
+	struct cqe_rx_t *cqe_rx;
+	struct rbdr *rbdr;
+	struct rcv_queue *rq;
+	struct queue_set *qs = nic->qs;
+	uint16_t *rb_lens = NULL;
+	u64 *rb_ptrs = NULL;
+
+	cqe_rx = (struct cqe_rx_t *)cq_desc;
+
+	rq = &qs->rq[cqe_rx->rq_idx];
+	rbdr = &qs->rbdr[rq->start_qs_rbdr_idx];
+	rb_lens = cq_desc + (3 * sizeof(u64)); /* Use offsetof */
+	/* Except 88xx pass1 on all other chips CQE_RX2_S is added to
+	 * CQE_RX at word6, hence buffer pointers move by word
+	 *
+	 * Use existing 'hw_tso' flag which will be set for all chips
+	 * except 88xx pass1 instead of a additional cache line
+	 * access (or miss) by using pci dev's revision.
+	 */
+	if (!nic->hw_tso)
+		rb_ptrs = (void *)cqe_rx + (6 * sizeof(u64));
+	else
+		rb_ptrs = (void *)cqe_rx + (7 * sizeof(u64));
+
+	for (frag = 0; frag < cqe_rx->rb_cnt; frag++) {
+		payload_len = rb_lens[frag_num(frag)];
+
+		invalidate_dcache_range((uintptr_t)(*rb_ptrs),
+					(uintptr_t)(*rb_ptrs) + rbdr->dma_size);
+
+		/* First fragment */
+		*rb_ptrs = *rb_ptrs - cqe_rx->align_pad;
+
+		pkt = nicvf_rb_ptr_to_pkt(nic, *rb_ptrs);
+
+		invalidate_dcache_range((uintptr_t)pkt,
+					(uintptr_t)pkt + payload_len);
+
+		if (cqe_rx->align_pad) {
+			pkt += cqe_rx->align_pad;
+		}
+		/* Next buffer pointer */
+		rb_ptrs++;
+
+		*pkt_len = payload_len;
+	}
+	return pkt;
+}
+
+/* Clear interrupt */
+void nicvf_clear_intr(struct nicvf *nic, int int_type, int q_idx)
+{
+	u64 reg_val = 0;
+
+	switch (int_type) {
+	case NICVF_INTR_CQ:
+		reg_val = ((1ULL << q_idx) << NICVF_INTR_CQ_SHIFT);
+	break;
+	case NICVF_INTR_SQ:
+		reg_val = ((1ULL << q_idx) << NICVF_INTR_SQ_SHIFT);
+	break;
+	case NICVF_INTR_RBDR:
+		reg_val = ((1ULL << q_idx) << NICVF_INTR_RBDR_SHIFT);
+	break;
+	case NICVF_INTR_PKT_DROP:
+		reg_val = (1ULL << NICVF_INTR_PKT_DROP_SHIFT);
+	break;
+	case NICVF_INTR_TCP_TIMER:
+		reg_val = (1ULL << NICVF_INTR_TCP_TIMER_SHIFT);
+	break;
+	case NICVF_INTR_MBOX:
+		reg_val = (1ULL << NICVF_INTR_MBOX_SHIFT);
+	break;
+	case NICVF_INTR_QS_ERR:
+		reg_val |= (1ULL << NICVF_INTR_QS_ERR_SHIFT);
+	break;
+	default:
+		printf("Failed to clear interrupt: unknown type\n");
+	break;
+	}
+
+	nicvf_reg_write(nic, NIC_VF_INT, reg_val);
+}
+
+void nicvf_update_rq_stats(struct nicvf *nic, int rq_idx)
+{
+	struct rcv_queue *rq;
+
+#define GET_RQ_STATS(reg) \
+	nicvf_reg_read(nic, NIC_QSET_RQ_0_7_STAT_0_1 |\
+			    (rq_idx << NIC_Q_NUM_SHIFT) | (reg << 3))
+
+	rq = &nic->qs->rq[rq_idx];
+	rq->stats.bytes = GET_RQ_STATS(RQ_SQ_STATS_OCTS);
+	rq->stats.pkts = GET_RQ_STATS(RQ_SQ_STATS_PKTS);
+}
+
+void nicvf_update_sq_stats(struct nicvf *nic, int sq_idx)
+{
+	struct snd_queue *sq;
+
+#define GET_SQ_STATS(reg) \
+	nicvf_reg_read(nic, NIC_QSET_SQ_0_7_STAT_0_1 |\
+			    (sq_idx << NIC_Q_NUM_SHIFT) | (reg << 3))
+
+	sq = &nic->qs->sq[sq_idx];
+	sq->stats.bytes = GET_SQ_STATS(RQ_SQ_STATS_OCTS);
+	sq->stats.pkts = GET_SQ_STATS(RQ_SQ_STATS_PKTS);
+}
+
+/* Check for errors in the receive cmp.queue entry */
+int nicvf_check_cqe_rx_errs(struct nicvf *nic,
+			    struct cmp_queue *cq, void *cq_desc)
+{
+	struct cqe_rx_t *cqe_rx;
+	struct cmp_queue_stats *stats = &cq->stats;
+
+	cqe_rx = (struct cqe_rx_t *)cq_desc;
+	if (!cqe_rx->err_level && !cqe_rx->err_opcode) {
+		stats->rx.errop.good++;
+		return 0;
+	}
+
+	switch (cqe_rx->err_level) {
+	case CQ_ERRLVL_MAC:
+		stats->rx.errlvl.mac_errs++;
+	break;
+	case CQ_ERRLVL_L2:
+		stats->rx.errlvl.l2_errs++;
+	break;
+	case CQ_ERRLVL_L3:
+		stats->rx.errlvl.l3_errs++;
+	break;
+	case CQ_ERRLVL_L4:
+		stats->rx.errlvl.l4_errs++;
+	break;
+	}
+
+	switch (cqe_rx->err_opcode) {
+	case CQ_RX_ERROP_RE_PARTIAL:
+		stats->rx.errop.partial_pkts++;
+	break;
+	case CQ_RX_ERROP_RE_JABBER:
+		stats->rx.errop.jabber_errs++;
+	break;
+	case CQ_RX_ERROP_RE_FCS:
+		stats->rx.errop.fcs_errs++;
+	break;
+	case CQ_RX_ERROP_RE_TERMINATE:
+		stats->rx.errop.terminate_errs++;
+	break;
+	case CQ_RX_ERROP_RE_RX_CTL:
+		stats->rx.errop.bgx_rx_errs++;
+	break;
+	case CQ_RX_ERROP_PREL2_ERR:
+		stats->rx.errop.prel2_errs++;
+	break;
+	case CQ_RX_ERROP_L2_FRAGMENT:
+		stats->rx.errop.l2_frags++;
+	break;
+	case CQ_RX_ERROP_L2_OVERRUN:
+		stats->rx.errop.l2_overruns++;
+	break;
+	case CQ_RX_ERROP_L2_PFCS:
+		stats->rx.errop.l2_pfcs++;
+	break;
+	case CQ_RX_ERROP_L2_PUNY:
+		stats->rx.errop.l2_puny++;
+	break;
+	case CQ_RX_ERROP_L2_MAL:
+		stats->rx.errop.l2_hdr_malformed++;
+	break;
+	case CQ_RX_ERROP_L2_OVERSIZE:
+		stats->rx.errop.l2_oversize++;
+	break;
+	case CQ_RX_ERROP_L2_UNDERSIZE:
+		stats->rx.errop.l2_undersize++;
+	break;
+	case CQ_RX_ERROP_L2_LENMISM:
+		stats->rx.errop.l2_len_mismatch++;
+	break;
+	case CQ_RX_ERROP_L2_PCLP:
+		stats->rx.errop.l2_pclp++;
+	break;
+	case CQ_RX_ERROP_IP_NOT:
+		stats->rx.errop.non_ip++;
+	break;
+	case CQ_RX_ERROP_IP_CSUM_ERR:
+		stats->rx.errop.ip_csum_err++;
+	break;
+	case CQ_RX_ERROP_IP_MAL:
+		stats->rx.errop.ip_hdr_malformed++;
+	break;
+	case CQ_RX_ERROP_IP_MALD:
+		stats->rx.errop.ip_payload_malformed++;
+	break;
+	case CQ_RX_ERROP_IP_HOP:
+		stats->rx.errop.ip_hop_errs++;
+	break;
+	case CQ_RX_ERROP_L3_ICRC:
+		stats->rx.errop.l3_icrc_errs++;
+	break;
+	case CQ_RX_ERROP_L3_PCLP:
+		stats->rx.errop.l3_pclp++;
+	break;
+	case CQ_RX_ERROP_L4_MAL:
+		stats->rx.errop.l4_malformed++;
+	break;
+	case CQ_RX_ERROP_L4_CHK:
+		stats->rx.errop.l4_csum_errs++;
+	break;
+	case CQ_RX_ERROP_UDP_LEN:
+		stats->rx.errop.udp_len_err++;
+	break;
+	case CQ_RX_ERROP_L4_PORT:
+		stats->rx.errop.bad_l4_port++;
+	break;
+	case CQ_RX_ERROP_TCP_FLAG:
+		stats->rx.errop.bad_tcp_flag++;
+	break;
+	case CQ_RX_ERROP_TCP_OFFSET:
+		stats->rx.errop.tcp_offset_errs++;
+	break;
+	case CQ_RX_ERROP_L4_PCLP:
+		stats->rx.errop.l4_pclp++;
+	break;
+	case CQ_RX_ERROP_RBDR_TRUNC:
+		stats->rx.errop.pkt_truncated++;
+	break;
+	}
+
+	return 1;
+}
+
+/* Check for errors in the send cmp.queue entry */
+int nicvf_check_cqe_tx_errs(struct nicvf *nic,
+			    struct cmp_queue *cq, void *cq_desc)
+{
+	struct cqe_send_t *cqe_tx;
+	struct cmp_queue_stats *stats = &cq->stats;
+
+	cqe_tx = (struct cqe_send_t *)cq_desc;
+	switch (cqe_tx->send_status) {
+	case CQ_TX_ERROP_GOOD:
+		stats->tx.good++;
+		return 0;
+	break;
+	case CQ_TX_ERROP_DESC_FAULT:
+		stats->tx.desc_fault++;
+	break;
+	case CQ_TX_ERROP_HDR_CONS_ERR:
+		stats->tx.hdr_cons_err++;
+	break;
+	case CQ_TX_ERROP_SUBDC_ERR:
+		stats->tx.subdesc_err++;
+	break;
+	case CQ_TX_ERROP_IMM_SIZE_OFLOW:
+		stats->tx.imm_size_oflow++;
+	break;
+	case CQ_TX_ERROP_DATA_SEQUENCE_ERR:
+		stats->tx.data_seq_err++;
+	break;
+	case CQ_TX_ERROP_MEM_SEQUENCE_ERR:
+		stats->tx.mem_seq_err++;
+	break;
+	case CQ_TX_ERROP_LOCK_VIOL:
+		stats->tx.lock_viol++;
+	break;
+	case CQ_TX_ERROP_DATA_FAULT:
+		stats->tx.data_fault++;
+	break;
+	case CQ_TX_ERROP_TSTMP_CONFLICT:
+		stats->tx.tstmp_conflict++;
+	break;
+	case CQ_TX_ERROP_TSTMP_TIMEOUT:
+		stats->tx.tstmp_timeout++;
+	break;
+	case CQ_TX_ERROP_MEM_FAULT:
+		stats->tx.mem_fault++;
+	break;
+	case CQ_TX_ERROP_CK_OVERLAP:
+		stats->tx.csum_overlap++;
+	break;
+	case CQ_TX_ERROP_CK_OFLOW:
+		stats->tx.csum_overflow++;
+	break;
+	}
+
+	return 1;
+}
diff --git a/drivers/net/cavium/nicvf_queues.h b/drivers/net/cavium/nicvf_queues.h
new file mode 100644
index 0000000000..2c476b4a74
--- /dev/null
+++ b/drivers/net/cavium/nicvf_queues.h
@@ -0,0 +1,364 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2018 Cavium Inc
+ */
+#ifndef NICVF_QUEUES_H
+#define NICVF_QUEUES_H
+
+#include "q_struct.h"
+
+#define MAX_QUEUE_SET			128
+#define MAX_RCV_QUEUES_PER_QS		8
+#define MAX_RCV_BUF_DESC_RINGS_PER_QS	2
+#define MAX_SND_QUEUES_PER_QS		8
+#define MAX_CMP_QUEUES_PER_QS		8
+
+/* VF's queue interrupt ranges */
+#define	NICVF_INTR_ID_CQ		0
+#define	NICVF_INTR_ID_SQ		8
+#define	NICVF_INTR_ID_RBDR		16
+#define	NICVF_INTR_ID_MISC		18
+#define	NICVF_INTR_ID_QS_ERR		19
+
+#define	for_each_cq_irq(irq)	\
+	for (irq = NICVF_INTR_ID_CQ; irq < NICVF_INTR_ID_SQ; irq++)
+#define	for_each_sq_irq(irq)	\
+	for (irq = NICVF_INTR_ID_SQ; irq < NICVF_INTR_ID_RBDR; irq++)
+#define	for_each_rbdr_irq(irq)	\
+	for (irq = NICVF_INTR_ID_RBDR; irq < NICVF_INTR_ID_MISC; irq++)
+
+#define RBDR_SIZE0		0ULL /* 8K entries */
+#define RBDR_SIZE1		1ULL /* 16K entries */
+#define RBDR_SIZE2		2ULL /* 32K entries */
+#define RBDR_SIZE3		3ULL /* 64K entries */
+#define RBDR_SIZE4		4ULL /* 126K entries */
+#define RBDR_SIZE5		5ULL /* 256K entries */
+#define RBDR_SIZE6		6ULL /* 512K entries */
+
+#define SND_QUEUE_SIZE0		0ULL /* 1K entries */
+#define SND_QUEUE_SIZE1		1ULL /* 2K entries */
+#define SND_QUEUE_SIZE2		2ULL /* 4K entries */
+#define SND_QUEUE_SIZE3		3ULL /* 8K entries */
+#define SND_QUEUE_SIZE4		4ULL /* 16K entries */
+#define SND_QUEUE_SIZE5		5ULL /* 32K entries */
+#define SND_QUEUE_SIZE6		6ULL /* 64K entries */
+
+#define CMP_QUEUE_SIZE0		0ULL /* 1K entries */
+#define CMP_QUEUE_SIZE1		1ULL /* 2K entries */
+#define CMP_QUEUE_SIZE2		2ULL /* 4K entries */
+#define CMP_QUEUE_SIZE3		3ULL /* 8K entries */
+#define CMP_QUEUE_SIZE4		4ULL /* 16K entries */
+#define CMP_QUEUE_SIZE5		5ULL /* 32K entries */
+#define CMP_QUEUE_SIZE6		6ULL /* 64K entries */
+
+/* Default queue count per QS, its lengths and threshold values */
+#define RBDR_CNT			1
+#define RCV_QUEUE_CNT		1
+#define SND_QUEUE_CNT		1
+#define CMP_QUEUE_CNT		1 /* Max of RCV and SND qcount */
+
+#define SND_QSIZE		SND_QUEUE_SIZE0
+#define SND_QUEUE_LEN		(1ULL << (SND_QSIZE + 10))
+#define SND_QUEUE_THRESH	2ULL
+#define MIN_SQ_DESC_PER_PKT_XMIT	2
+#define MAX_CQE_PER_PKT_XMIT		2
+
+#define CMP_QSIZE		CMP_QUEUE_SIZE0
+#define CMP_QUEUE_LEN		(1ULL << (CMP_QSIZE + 10))
+#define CMP_QUEUE_CQE_THRESH	0
+#define CMP_QUEUE_TIMER_THRESH	1 /* 1 ms */
+
+#define RBDR_SIZE		RBDR_SIZE0
+#define RCV_BUF_COUNT		(1ULL << (RBDR_SIZE + 13))
+#define RBDR_THRESH		(RCV_BUF_COUNT / 2)
+#define DMA_BUFFER_LEN		2048 /* In multiples of 128bytes */
+#define RCV_FRAG_LEN	 	DMA_BUFFER_LEN
+
+#define MAX_CQES_FOR_TX		((SND_QUEUE_LEN / MIN_SQ_DESC_PER_PKT_XMIT) *\
+				 MAX_CQE_PER_PKT_XMIT)
+#define RQ_CQ_DROP		((CMP_QUEUE_LEN - MAX_CQES_FOR_TX) / 256)
+
+/* Descriptor size */
+#define SND_QUEUE_DESC_SIZE	16   /* 128 bits */
+#define CMP_QUEUE_DESC_SIZE	512
+
+/* Buffer / descriptor alignments */
+#define NICVF_RCV_BUF_ALIGN		7
+#define NICVF_RCV_BUF_ALIGN_BYTES	(1ULL << NICVF_RCV_BUF_ALIGN)
+#define NICVF_CQ_BASE_ALIGN_BYTES	512  /* 9 bits */
+#define NICVF_SQ_BASE_ALIGN_BYTES	128  /* 7 bits */
+
+#define NICVF_ALIGNED_ADDR(ADDR, ALIGN_BYTES)	ALIGN(ADDR, ALIGN_BYTES)
+#define NICVF_ADDR_ALIGN_LEN(ADDR, BYTES)\
+	(NICVF_ALIGNED_ADDR(ADDR, BYTES) - BYTES)
+#define NICVF_RCV_BUF_ALIGN_LEN(X)\
+	(NICVF_ALIGNED_ADDR(X, NICVF_RCV_BUF_ALIGN_BYTES) - X)
+
+/* Queue enable/disable */
+#define NICVF_SQ_EN            (1ULL << 19)
+
+/* Queue reset */
+#define NICVF_CQ_RESET		(1ULL << 41)
+#define NICVF_SQ_RESET		(1ULL << 17)
+#define NICVF_RBDR_RESET	(1ULL << 43)
+
+enum CQ_RX_ERRLVL_E {
+	CQ_ERRLVL_MAC,
+	CQ_ERRLVL_L2,
+	CQ_ERRLVL_L3,
+	CQ_ERRLVL_L4,
+};
+
+enum CQ_RX_ERROP_E {
+	CQ_RX_ERROP_RE_NONE = 0x0,
+	CQ_RX_ERROP_RE_PARTIAL = 0x1,
+	CQ_RX_ERROP_RE_JABBER = 0x2,
+	CQ_RX_ERROP_RE_FCS = 0x7,
+	CQ_RX_ERROP_RE_TERMINATE = 0x9,
+	CQ_RX_ERROP_RE_RX_CTL = 0xb,
+	CQ_RX_ERROP_PREL2_ERR = 0x1f,
+	CQ_RX_ERROP_L2_FRAGMENT = 0x20,
+	CQ_RX_ERROP_L2_OVERRUN = 0x21,
+	CQ_RX_ERROP_L2_PFCS = 0x22,
+	CQ_RX_ERROP_L2_PUNY = 0x23,
+	CQ_RX_ERROP_L2_MAL = 0x24,
+	CQ_RX_ERROP_L2_OVERSIZE = 0x25,
+	CQ_RX_ERROP_L2_UNDERSIZE = 0x26,
+	CQ_RX_ERROP_L2_LENMISM = 0x27,
+	CQ_RX_ERROP_L2_PCLP = 0x28,
+	CQ_RX_ERROP_IP_NOT = 0x41,
+	CQ_RX_ERROP_IP_CSUM_ERR = 0x42,
+	CQ_RX_ERROP_IP_MAL = 0x43,
+	CQ_RX_ERROP_IP_MALD = 0x44,
+	CQ_RX_ERROP_IP_HOP = 0x45,
+	CQ_RX_ERROP_L3_ICRC = 0x46,
+	CQ_RX_ERROP_L3_PCLP = 0x47,
+	CQ_RX_ERROP_L4_MAL = 0x61,
+	CQ_RX_ERROP_L4_CHK = 0x62,
+	CQ_RX_ERROP_UDP_LEN = 0x63,
+	CQ_RX_ERROP_L4_PORT = 0x64,
+	CQ_RX_ERROP_TCP_FLAG = 0x65,
+	CQ_RX_ERROP_TCP_OFFSET = 0x66,
+	CQ_RX_ERROP_L4_PCLP = 0x67,
+	CQ_RX_ERROP_RBDR_TRUNC = 0x70,
+};
+
+enum CQ_TX_ERROP_E {
+	CQ_TX_ERROP_GOOD = 0x0,
+	CQ_TX_ERROP_DESC_FAULT = 0x10,
+	CQ_TX_ERROP_HDR_CONS_ERR = 0x11,
+	CQ_TX_ERROP_SUBDC_ERR = 0x12,
+	CQ_TX_ERROP_IMM_SIZE_OFLOW = 0x80,
+	CQ_TX_ERROP_DATA_SEQUENCE_ERR = 0x81,
+	CQ_TX_ERROP_MEM_SEQUENCE_ERR = 0x82,
+	CQ_TX_ERROP_LOCK_VIOL = 0x83,
+	CQ_TX_ERROP_DATA_FAULT = 0x84,
+	CQ_TX_ERROP_TSTMP_CONFLICT = 0x85,
+	CQ_TX_ERROP_TSTMP_TIMEOUT = 0x86,
+	CQ_TX_ERROP_MEM_FAULT = 0x87,
+	CQ_TX_ERROP_CK_OVERLAP = 0x88,
+	CQ_TX_ERROP_CK_OFLOW = 0x89,
+	CQ_TX_ERROP_ENUM_LAST = 0x8a,
+};
+
+struct cmp_queue_stats {
+	struct rx_stats {
+		struct {
+			u64 mac_errs;
+			u64 l2_errs;
+			u64 l3_errs;
+			u64 l4_errs;
+		} errlvl;
+		struct {
+			u64 good;
+			u64 partial_pkts;
+			u64 jabber_errs;
+			u64 fcs_errs;
+			u64 terminate_errs;
+			u64 bgx_rx_errs;
+			u64 prel2_errs;
+			u64 l2_frags;
+			u64 l2_overruns;
+			u64 l2_pfcs;
+			u64 l2_puny;
+			u64 l2_hdr_malformed;
+			u64 l2_oversize;
+			u64 l2_undersize;
+			u64 l2_len_mismatch;
+			u64 l2_pclp;
+			u64 non_ip;
+			u64 ip_csum_err;
+			u64 ip_hdr_malformed;
+			u64 ip_payload_malformed;
+			u64 ip_hop_errs;
+			u64 l3_icrc_errs;
+			u64 l3_pclp;
+			u64 l4_malformed;
+			u64 l4_csum_errs;
+			u64 udp_len_err;
+			u64 bad_l4_port;
+			u64 bad_tcp_flag;
+			u64 tcp_offset_errs;
+			u64 l4_pclp;
+			u64 pkt_truncated;
+		} errop;
+	} rx;
+	struct tx_stats {
+		u64 good;
+		u64 desc_fault;
+		u64 hdr_cons_err;
+		u64 subdesc_err;
+		u64 imm_size_oflow;
+		u64 data_seq_err;
+		u64 mem_seq_err;
+		u64 lock_viol;
+		u64 data_fault;
+		u64 tstmp_conflict;
+		u64 tstmp_timeout;
+		u64 mem_fault;
+		u64 csum_overlap;
+		u64 csum_overflow;
+	} tx;
+};
+
+enum RQ_SQ_STATS {
+	RQ_SQ_STATS_OCTS,
+	RQ_SQ_STATS_PKTS,
+};
+
+struct rx_tx_queue_stats {
+	u64	bytes;
+	u64	pkts;
+};
+
+struct q_desc_mem {
+	uintptr_t	dma;
+	uint64_t	size;
+	uint16_t	q_len;
+	uintptr_t	phys_base;
+	void		*base;
+	void		*unalign_base;
+	bool		allocated;
+};
+
+struct rbdr {
+	bool		enable;
+	uint32_t	dma_size;
+	uint32_t	thresh;      /* Threshold level for interrupt */
+	void		*desc;
+	uint32_t	head;
+	uint32_t	tail;
+	struct		q_desc_mem   dmem;
+	uintptr_t	buf_mem;
+	uintptr_t	buffers;
+};
+
+struct rcv_queue {
+	bool		enable;
+	struct	rbdr	*rbdr_start;
+	struct	rbdr	*rbdr_cont;
+	bool		en_tcp_reassembly;
+	uint8_t		cq_qs;  /* CQ's QS to which this RQ is assigned */
+	uint8_t		cq_idx; /* CQ index (0 to 7) in the QS */
+	uint8_t		cont_rbdr_qs;      /* Continue buffer ptrs - QS num */
+	uint8_t		cont_qs_rbdr_idx;  /* RBDR idx in the cont QS */
+	uint8_t		start_rbdr_qs;     /* First buffer ptrs - QS num */
+	uint8_t		start_qs_rbdr_idx; /* RBDR idx in the above QS */
+	uint8_t         caching;
+	struct		rx_tx_queue_stats stats;
+};
+
+struct cmp_queue {
+	bool		enable;
+	uint16_t	intr_timer_thresh;
+	uint16_t	thresh;
+	void		*desc;
+	struct q_desc_mem   dmem;
+	struct cmp_queue_stats	stats;
+};
+
+struct snd_queue {
+	bool		enable;
+	uint8_t		cq_qs;  /* CQ's QS to which this SQ is pointing */
+	uint8_t		cq_idx; /* CQ index (0 to 7) in the above QS */
+	uint16_t	thresh;
+	uint32_t	free_cnt;
+	uint32_t	head;
+	uint32_t	tail;
+	uint64_t	*skbuff;
+	void		*desc;
+	struct q_desc_mem   dmem;
+	struct rx_tx_queue_stats stats;
+};
+
+struct queue_set {
+	bool		enable;
+	bool		be_en;
+	uint8_t		vnic_id;
+	uint8_t		rq_cnt;
+	uint8_t		cq_cnt;
+	uint64_t	cq_len;
+	uint8_t		sq_cnt;
+	uint64_t	sq_len;
+	uint8_t		rbdr_cnt;
+	uint64_t	rbdr_len;
+	struct	rcv_queue	rq[MAX_RCV_QUEUES_PER_QS];
+	struct	cmp_queue	cq[MAX_CMP_QUEUES_PER_QS];
+	struct	snd_queue	sq[MAX_SND_QUEUES_PER_QS];
+	struct	rbdr		rbdr[MAX_RCV_BUF_DESC_RINGS_PER_QS];
+};
+
+#define GET_RBDR_DESC(RING, idx)\
+		(&(((struct rbdr_entry_t *)((RING)->desc))[idx]))
+#define GET_SQ_DESC(RING, idx)\
+		(&(((struct sq_hdr_subdesc *)((RING)->desc))[idx]))
+#define GET_CQ_DESC(RING, idx)\
+		(&(((union cq_desc_t *)((RING)->desc))[idx]))
+
+/* CQ status bits */
+#define	CQ_WR_FULL	(1 << 26)
+#define	CQ_WR_DISABLE	(1 << 25)
+#define	CQ_WR_FAULT	(1 << 24)
+#define	CQ_CQE_COUNT	(0xFFFF << 0)
+
+#define	CQ_ERR_MASK	(CQ_WR_FULL | CQ_WR_DISABLE | CQ_WR_FAULT)
+
+int nicvf_set_qset_resources(struct nicvf *nic);
+int nicvf_config_data_transfer(struct nicvf *nic, bool enable);
+void nicvf_qset_config(struct nicvf *nic, bool enable);
+void nicvf_cmp_queue_config(struct nicvf *nic, struct queue_set *qs,
+			    int qidx, bool enable);
+
+void nicvf_sq_enable(struct nicvf *nic, struct snd_queue *sq, int qidx);
+void nicvf_sq_disable(struct nicvf *nic, int qidx);
+void nicvf_put_sq_desc(struct snd_queue *sq, int desc_cnt);
+void nicvf_sq_free_used_descs(struct eth_device *netdev,
+								struct snd_queue *sq, int qidx);
+int nicvf_sq_append_pkt(struct nicvf *nic, void *pkt, size_t pkt_len);
+
+void *nicvf_get_rcv_pkt(struct nicvf *nic, void *cq_desc, size_t *pkt_len);
+void nicvf_refill_rbdr(struct nicvf *nic);
+
+void nicvf_enable_intr(struct nicvf *nic, int int_type, int q_idx);
+void nicvf_disable_intr(struct nicvf *nic, int int_type, int q_idx);
+void nicvf_clear_intr(struct nicvf *nic, int int_type, int q_idx);
+int nicvf_is_intr_enabled(struct nicvf *nic, int int_type, int q_idx);
+
+/* Register access APIs */
+void nicvf_reg_write(struct nicvf *nic, uint64_t offset, uint64_t val);
+uint64_t nicvf_reg_read(struct nicvf *nic, uint64_t offset);
+void nicvf_qset_reg_write(struct nicvf *nic, uint64_t offset, uint64_t val);
+uint64_t nicvf_qset_reg_read(struct nicvf *nic, uint64_t offset);
+void nicvf_queue_reg_write(struct nicvf *nic, uint64_t offset,
+			   uint64_t qidx, uint64_t val);
+uint64_t nicvf_queue_reg_read(struct nicvf *nic,
+			      uint64_t offset, uint64_t qidx);
+
+/* Stats */
+void nicvf_update_rq_stats(struct nicvf *nic, int rq_idx);
+void nicvf_update_sq_stats(struct nicvf *nic, int sq_idx);
+int nicvf_check_cqe_rx_errs(struct nicvf *nic,
+			    struct cmp_queue *cq, void *cq_desc);
+int nicvf_check_cqe_tx_errs(struct nicvf *nic,
+			    struct cmp_queue *cq, void *cq_desc);
+#endif /* NICVF_QUEUES_H */
diff --git a/drivers/net/cavium/q_struct.h b/drivers/net/cavium/q_struct.h
new file mode 100644
index 0000000000..59b609024d
--- /dev/null
+++ b/drivers/net/cavium/q_struct.h
@@ -0,0 +1,692 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2018 Cavium Inc
+ */
+#ifndef Q_STRUCT_H
+#define Q_STRUCT_H
+
+/* Load transaction types for reading segment bytes specified by
+ * NIC_SEND_GATHER_S[LD_TYPE].
+ */
+enum nic_send_ld_type_e {
+	NIC_SEND_LD_TYPE_E_LDD = 0x0,
+	NIC_SEND_LD_TYPE_E_LDT = 0x1,
+	NIC_SEND_LD_TYPE_E_LDWB = 0x2,
+	NIC_SEND_LD_TYPE_E_ENUM_LAST = 0x3,
+};
+
+enum ether_type_algorithm {
+	ETYPE_ALG_NONE = 0x0,
+	ETYPE_ALG_SKIP = 0x1,
+	ETYPE_ALG_ENDPARSE = 0x2,
+	ETYPE_ALG_VLAN = 0x3,
+	ETYPE_ALG_VLAN_STRIP = 0x4,
+};
+
+enum layer3_type {
+	L3TYPE_NONE = 0x00,
+	L3TYPE_GRH = 0x01,
+	L3TYPE_IPV4 = 0x04,
+	L3TYPE_IPV4_OPTIONS = 0x05,
+	L3TYPE_IPV6 = 0x06,
+	L3TYPE_IPV6_OPTIONS = 0x07,
+	L3TYPE_ET_STOP = 0x0D,
+	L3TYPE_OTHER = 0x0E,
+};
+
+enum layer4_type {
+	L4TYPE_NONE = 0x00,
+	L4TYPE_IPSEC_ESP = 0x01,
+	L4TYPE_IPFRAG = 0x02,
+	L4TYPE_IPCOMP = 0x03,
+	L4TYPE_TCP = 0x04,
+	L4TYPE_UDP = 0x05,
+	L4TYPE_SCTP = 0x06,
+	L4TYPE_GRE = 0x07,
+	L4TYPE_ROCE_BTH = 0x08,
+	L4TYPE_OTHER = 0x0E,
+};
+
+/* CPI and RSSI configuration */
+enum cpi_algorithm_type {
+	CPI_ALG_NONE = 0x0,
+	CPI_ALG_VLAN = 0x1,
+	CPI_ALG_VLAN16 = 0x2,
+	CPI_ALG_DIFF = 0x3,
+};
+
+enum rss_algorithm_type {
+	RSS_ALG_NONE = 0x00,
+	RSS_ALG_PORT = 0x01,
+	RSS_ALG_IP = 0x02,
+	RSS_ALG_TCP_IP = 0x03,
+	RSS_ALG_UDP_IP = 0x04,
+	RSS_ALG_SCTP_IP = 0x05,
+	RSS_ALG_GRE_IP = 0x06,
+	RSS_ALG_ROCE = 0x07,
+};
+
+enum rss_hash_cfg {
+	RSS_HASH_L2ETC = 0x00,
+	RSS_HASH_IP = 0x01,
+	RSS_HASH_TCP = 0x02,
+	RSS_HASH_TCP_SYN_DIS = 0x03,
+	RSS_HASH_UDP = 0x04,
+	RSS_HASH_L4ETC = 0x05,
+	RSS_HASH_ROCE = 0x06,
+	RSS_L3_BIDI = 0x07,
+	RSS_L4_BIDI = 0x08,
+};
+
+/* Completion queue entry types */
+enum cqe_type {
+	CQE_TYPE_INVALID = 0x0,
+	CQE_TYPE_RX = 0x2,
+	CQE_TYPE_RX_SPLIT = 0x3,
+	CQE_TYPE_RX_TCP = 0x4,
+	CQE_TYPE_SEND = 0x8,
+	CQE_TYPE_SEND_PTP = 0x9,
+};
+
+enum cqe_rx_tcp_status {
+	CQE_RX_STATUS_VALID_TCP_CNXT = 0x00,
+	CQE_RX_STATUS_INVALID_TCP_CNXT = 0x0F,
+};
+
+enum cqe_send_status {
+	CQE_SEND_STATUS_GOOD = 0x00,
+	CQE_SEND_STATUS_DESC_FAULT = 0x01,
+	CQE_SEND_STATUS_HDR_CONS_ERR = 0x11,
+	CQE_SEND_STATUS_SUBDESC_ERR = 0x12,
+	CQE_SEND_STATUS_IMM_SIZE_OFLOW = 0x80,
+	CQE_SEND_STATUS_CRC_SEQ_ERR = 0x81,
+	CQE_SEND_STATUS_DATA_SEQ_ERR = 0x82,
+	CQE_SEND_STATUS_MEM_SEQ_ERR = 0x83,
+	CQE_SEND_STATUS_LOCK_VIOL = 0x84,
+	CQE_SEND_STATUS_LOCK_UFLOW = 0x85,
+	CQE_SEND_STATUS_DATA_FAULT = 0x86,
+	CQE_SEND_STATUS_TSTMP_CONFLICT = 0x87,
+	CQE_SEND_STATUS_TSTMP_TIMEOUT = 0x88,
+	CQE_SEND_STATUS_MEM_FAULT = 0x89,
+	CQE_SEND_STATUS_CSUM_OVERLAP = 0x8A,
+	CQE_SEND_STATUS_CSUM_OVERFLOW = 0x8B,
+};
+
+enum cqe_rx_tcp_end_reason {
+	CQE_RX_TCP_END_FIN_FLAG_DET = 0,
+	CQE_RX_TCP_END_INVALID_FLAG = 1,
+	CQE_RX_TCP_END_TIMEOUT = 2,
+	CQE_RX_TCP_END_OUT_OF_SEQ = 3,
+	CQE_RX_TCP_END_PKT_ERR = 4,
+	CQE_RX_TCP_END_QS_DISABLED = 0x0F,
+};
+
+/* Packet protocol level error enumeration */
+enum cqe_rx_err_level {
+	CQE_RX_ERRLVL_RE = 0x0,
+	CQE_RX_ERRLVL_L2 = 0x1,
+	CQE_RX_ERRLVL_L3 = 0x2,
+	CQE_RX_ERRLVL_L4 = 0x3,
+};
+
+/* Packet protocol level error type enumeration */
+enum cqe_rx_err_opcode {
+	CQE_RX_ERR_RE_NONE = 0x0,
+	CQE_RX_ERR_RE_PARTIAL = 0x1,
+	CQE_RX_ERR_RE_JABBER = 0x2,
+	CQE_RX_ERR_RE_FCS = 0x7,
+	CQE_RX_ERR_RE_TERMINATE = 0x9,
+	CQE_RX_ERR_RE_RX_CTL = 0xb,
+	CQE_RX_ERR_PREL2_ERR = 0x1f,
+	CQE_RX_ERR_L2_FRAGMENT = 0x20,
+	CQE_RX_ERR_L2_OVERRUN = 0x21,
+	CQE_RX_ERR_L2_PFCS = 0x22,
+	CQE_RX_ERR_L2_PUNY = 0x23,
+	CQE_RX_ERR_L2_MAL = 0x24,
+	CQE_RX_ERR_L2_OVERSIZE = 0x25,
+	CQE_RX_ERR_L2_UNDERSIZE = 0x26,
+	CQE_RX_ERR_L2_LENMISM = 0x27,
+	CQE_RX_ERR_L2_PCLP = 0x28,
+	CQE_RX_ERR_IP_NOT = 0x41,
+	CQE_RX_ERR_IP_CHK = 0x42,
+	CQE_RX_ERR_IP_MAL = 0x43,
+	CQE_RX_ERR_IP_MALD = 0x44,
+	CQE_RX_ERR_IP_HOP = 0x45,
+	CQE_RX_ERR_L3_ICRC = 0x46,
+	CQE_RX_ERR_L3_PCLP = 0x47,
+	CQE_RX_ERR_L4_MAL = 0x61,
+	CQE_RX_ERR_L4_CHK = 0x62,
+	CQE_RX_ERR_UDP_LEN = 0x63,
+	CQE_RX_ERR_L4_PORT = 0x64,
+	CQE_RX_ERR_TCP_FLAG = 0x65,
+	CQE_RX_ERR_TCP_OFFSET = 0x66,
+	CQE_RX_ERR_L4_PCLP = 0x67,
+	CQE_RX_ERR_RBDR_TRUNC = 0x70,
+};
+
+struct cqe_rx_t {
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u64   cqe_type:4; /* W0 */
+	u64   stdn_fault:1;
+	u64   rsvd0:1;
+	u64   rq_qs:7;
+	u64   rq_idx:3;
+	u64   rsvd1:12;
+	u64   rss_alg:4;
+	u64   rsvd2:4;
+	u64   rb_cnt:4;
+	u64   vlan_found:1;
+	u64   vlan_stripped:1;
+	u64   vlan2_found:1;
+	u64   vlan2_stripped:1;
+	u64   l4_type:4;
+	u64   l3_type:4;
+	u64   l2_present:1;
+	u64   err_level:3;
+	u64   err_opcode:8;
+
+	u64   pkt_len:16; /* W1 */
+	u64   l2_ptr:8;
+	u64   l3_ptr:8;
+	u64   l4_ptr:8;
+	u64   cq_pkt_len:8;
+	u64   align_pad:3;
+	u64   rsvd3:1;
+	u64   chan:12;
+
+	u64   rss_tag:32; /* W2 */
+	u64   vlan_tci:16;
+	u64   vlan_ptr:8;
+	u64   vlan2_ptr:8;
+
+	u64   rb3_sz:16; /* W3 */
+	u64   rb2_sz:16;
+	u64   rb1_sz:16;
+	u64   rb0_sz:16;
+
+	u64   rb7_sz:16; /* W4 */
+	u64   rb6_sz:16;
+	u64   rb5_sz:16;
+	u64   rb4_sz:16;
+
+	u64   rb11_sz:16; /* W5 */
+	u64   rb10_sz:16;
+	u64   rb9_sz:16;
+	u64   rb8_sz:16;
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u64   err_opcode:8;
+	u64   err_level:3;
+	u64   l2_present:1;
+	u64   l3_type:4;
+	u64   l4_type:4;
+	u64   vlan2_stripped:1;
+	u64   vlan2_found:1;
+	u64   vlan_stripped:1;
+	u64   vlan_found:1;
+	u64   rb_cnt:4;
+	u64   rsvd2:4;
+	u64   rss_alg:4;
+	u64   rsvd1:12;
+	u64   rq_idx:3;
+	u64   rq_qs:7;
+	u64   rsvd0:1;
+	u64   stdn_fault:1;
+	u64   cqe_type:4; /* W0 */
+	u64   chan:12;
+	u64   rsvd3:1;
+	u64   align_pad:3;
+	u64   cq_pkt_len:8;
+	u64   l4_ptr:8;
+	u64   l3_ptr:8;
+	u64   l2_ptr:8;
+	u64   pkt_len:16; /* W1 */
+	u64   vlan2_ptr:8;
+	u64   vlan_ptr:8;
+	u64   vlan_tci:16;
+	u64   rss_tag:32; /* W2 */
+	u64   rb0_sz:16;
+	u64   rb1_sz:16;
+	u64   rb2_sz:16;
+	u64   rb3_sz:16; /* W3 */
+	u64   rb4_sz:16;
+	u64   rb5_sz:16;
+	u64   rb6_sz:16;
+	u64   rb7_sz:16; /* W4 */
+	u64   rb8_sz:16;
+	u64   rb9_sz:16;
+	u64   rb10_sz:16;
+	u64   rb11_sz:16; /* W5 */
+#endif
+	u64   rb0_ptr:64;
+	u64   rb1_ptr:64;
+	u64   rb2_ptr:64;
+	u64   rb3_ptr:64;
+	u64   rb4_ptr:64;
+	u64   rb5_ptr:64;
+	u64   rb6_ptr:64;
+	u64   rb7_ptr:64;
+	u64   rb8_ptr:64;
+	u64   rb9_ptr:64;
+	u64   rb10_ptr:64;
+	u64   rb11_ptr:64;
+};
+
+struct cqe_rx_tcp_err_t {
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u64   cqe_type:4; /* W0 */
+	u64   rsvd0:60;
+
+	u64   rsvd1:4; /* W1 */
+	u64   partial_first:1;
+	u64   rsvd2:27;
+	u64   rbdr_bytes:8;
+	u64   rsvd3:24;
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u64   rsvd0:60;
+	u64   cqe_type:4;
+
+	u64   rsvd3:24;
+	u64   rbdr_bytes:8;
+	u64   rsvd2:27;
+	u64   partial_first:1;
+	u64   rsvd1:4;
+#endif
+};
+
+struct cqe_rx_tcp_t {
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u64   cqe_type:4; /* W0 */
+	u64   rsvd0:52;
+	u64   cq_tcp_status:8;
+
+	u64   rsvd1:32; /* W1 */
+	u64   tcp_cntx_bytes:8;
+	u64   rsvd2:8;
+	u64   tcp_err_bytes:16;
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u64   cq_tcp_status:8;
+	u64   rsvd0:52;
+	u64   cqe_type:4; /* W0 */
+
+	u64   tcp_err_bytes:16;
+	u64   rsvd2:8;
+	u64   tcp_cntx_bytes:8;
+	u64   rsvd1:32; /* W1 */
+#endif
+};
+
+struct cqe_send_t {
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u64   cqe_type:4; /* W0 */
+	u64   rsvd0:4;
+	u64   sqe_ptr:16;
+	u64   rsvd1:4;
+	u64   rsvd2:10;
+	u64   sq_qs:7;
+	u64   sq_idx:3;
+	u64   rsvd3:8;
+	u64   send_status:8;
+
+	u64   ptp_timestamp:64; /* W1 */
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u64   send_status:8;
+	u64   rsvd3:8;
+	u64   sq_idx:3;
+	u64   sq_qs:7;
+	u64   rsvd2:10;
+	u64   rsvd1:4;
+	u64   sqe_ptr:16;
+	u64   rsvd0:4;
+	u64   cqe_type:4; /* W0 */
+
+	u64   ptp_timestamp:64; /* W1 */
+#endif
+};
+
+union cq_desc_t {
+	u64    u[64];
+	struct cqe_send_t snd_hdr;
+	struct cqe_rx_t rx_hdr;
+	struct cqe_rx_tcp_t rx_tcp_hdr;
+	struct cqe_rx_tcp_err_t rx_tcp_err_hdr;
+};
+
+struct rbdr_entry_t {
+	u64   buf_addr;
+};
+
+/* TCP reassembly context */
+struct rbe_tcp_cnxt_t {
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u64   tcp_pkt_cnt:12;
+	u64   rsvd1:4;
+	u64   align_hdr_bytes:4;
+	u64   align_ptr_bytes:4;
+	u64   ptr_bytes:16;
+	u64   rsvd2:24;
+	u64   cqe_type:4;
+	u64   rsvd0:54;
+	u64   tcp_end_reason:2;
+	u64   tcp_status:4;
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u64   tcp_status:4;
+	u64   tcp_end_reason:2;
+	u64   rsvd0:54;
+	u64   cqe_type:4;
+	u64   rsvd2:24;
+	u64   ptr_bytes:16;
+	u64   align_ptr_bytes:4;
+	u64   align_hdr_bytes:4;
+	u64   rsvd1:4;
+	u64   tcp_pkt_cnt:12;
+#endif
+};
+
+/* Always Big endian */
+struct rx_hdr_t {
+	u64   opaque:32;
+	u64   rss_flow:8;
+	u64   skip_length:6;
+	u64   disable_rss:1;
+	u64   disable_tcp_reassembly:1;
+	u64   nodrop:1;
+	u64   dest_alg:2;
+	u64   rsvd0:2;
+	u64   dest_rq:11;
+};
+
+enum send_l4_csum_type {
+	SEND_L4_CSUM_DISABLE = 0x00,
+	SEND_L4_CSUM_UDP = 0x01,
+	SEND_L4_CSUM_TCP = 0x02,
+	SEND_L4_CSUM_SCTP = 0x03,
+};
+
+enum send_crc_alg {
+	SEND_CRCALG_CRC32 = 0x00,
+	SEND_CRCALG_CRC32C = 0x01,
+	SEND_CRCALG_ICRC = 0x02,
+};
+
+enum send_load_type {
+	SEND_LD_TYPE_LDD = 0x00,
+	SEND_LD_TYPE_LDT = 0x01,
+	SEND_LD_TYPE_LDWB = 0x02,
+};
+
+enum send_mem_alg_type {
+	SEND_MEMALG_SET = 0x00,
+	SEND_MEMALG_ADD = 0x08,
+	SEND_MEMALG_SUB = 0x09,
+	SEND_MEMALG_ADDLEN = 0x0A,
+	SEND_MEMALG_SUBLEN = 0x0B,
+};
+
+enum send_mem_dsz_type {
+	SEND_MEMDSZ_B64 = 0x00,
+	SEND_MEMDSZ_B32 = 0x01,
+	SEND_MEMDSZ_B8 = 0x03,
+};
+
+enum sq_subdesc_type {
+	SQ_DESC_TYPE_INVALID = 0x00,
+	SQ_DESC_TYPE_HEADER = 0x01,
+	SQ_DESC_TYPE_CRC = 0x02,
+	SQ_DESC_TYPE_IMMEDIATE = 0x03,
+	SQ_DESC_TYPE_GATHER = 0x04,
+	SQ_DESC_TYPE_MEMORY = 0x05,
+};
+
+struct sq_crc_subdesc {
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u64    rsvd1:32;
+	u64    crc_ival:32;
+	u64    subdesc_type:4;
+	u64    crc_alg:2;
+	u64    rsvd0:10;
+	u64    crc_insert_pos:16;
+	u64    hdr_start:16;
+	u64    crc_len:16;
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u64    crc_len:16;
+	u64    hdr_start:16;
+	u64    crc_insert_pos:16;
+	u64    rsvd0:10;
+	u64    crc_alg:2;
+	u64    subdesc_type:4;
+	u64    crc_ival:32;
+	u64    rsvd1:32;
+#endif
+};
+
+struct sq_gather_subdesc {
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u64    subdesc_type:4; /* W0 */
+	u64    ld_type:2;
+	u64    rsvd0:42;
+	u64    size:16;
+
+	u64    rsvd1:15; /* W1 */
+	u64    addr:49;
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u64    size:16;
+	u64    rsvd0:42;
+	u64    ld_type:2;
+	u64    subdesc_type:4; /* W0 */
+
+	u64    addr:49;
+	u64    rsvd1:15; /* W1 */
+#endif
+};
+
+/* SQ immediate subdescriptor */
+struct sq_imm_subdesc {
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u64    subdesc_type:4; /* W0 */
+	u64    rsvd0:46;
+	u64    len:14;
+
+	u64    data:64; /* W1 */
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u64    len:14;
+	u64    rsvd0:46;
+	u64    subdesc_type:4; /* W0 */
+
+	u64    data:64; /* W1 */
+#endif
+};
+
+struct sq_mem_subdesc {
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u64    subdesc_type:4; /* W0 */
+	u64    mem_alg:4;
+	u64    mem_dsz:2;
+	u64    wmem:1;
+	u64    rsvd0:21;
+	u64    offset:32;
+
+	u64    rsvd1:15; /* W1 */
+	u64    addr:49;
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u64    offset:32;
+	u64    rsvd0:21;
+	u64    wmem:1;
+	u64    mem_dsz:2;
+	u64    mem_alg:4;
+	u64    subdesc_type:4; /* W0 */
+
+	u64    addr:49;
+	u64    rsvd1:15; /* W1 */
+#endif
+};
+
+struct sq_hdr_subdesc {
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u64    subdesc_type:4;
+	u64    tso:1;
+	u64    post_cqe:1; /* Post CQE on no error also */
+	u64    dont_send:1;
+	u64    tstmp:1;
+	u64    subdesc_cnt:8;
+	u64    csum_l4:2;
+	u64    csum_l3:1;
+	u64    csum_inner_l4:2;
+	u64    csum_inner_l3:1;
+	u64    rsvd0:2;
+	u64    l4_offset:8;
+	u64    l3_offset:8;
+	u64    rsvd1:4;
+	u64    tot_len:20; /* W0 */
+
+	u64    rsvd2:24;
+	u64    inner_l4_offset:8;
+	u64    inner_l3_offset:8;
+	u64    tso_start:8;
+	u64    rsvd3:2;
+	u64    tso_max_paysize:14; /* W1 */
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u64    tot_len:20;
+	u64    rsvd1:4;
+	u64    l3_offset:8;
+	u64    l4_offset:8;
+	u64    rsvd0:2;
+	u64    csum_inner_l3:1;
+	u64    csum_inner_l4:2;
+	u64    csum_l3:1;
+	u64    csum_l4:2;
+	u64    subdesc_cnt:8;
+	u64    tstmp:1;
+	u64    dont_send:1;
+	u64    post_cqe:1; /* Post CQE on no error also */
+	u64    tso:1;
+	u64    subdesc_type:4; /* W0 */
+
+	u64    tso_max_paysize:14;
+	u64    rsvd3:2;
+	u64    tso_start:8;
+	u64    inner_l3_offset:8;
+	u64    inner_l4_offset:8;
+	u64    rsvd2:24; /* W1 */
+#endif
+};
+
+/* Queue config register formats */
+struct rq_cfg {
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u64 reserved_2_63:62;
+	u64 ena:1;
+	u64 tcp_ena:1;
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u64 tcp_ena:1;
+	u64 ena:1;
+	u64 reserved_2_63:62;
+#endif
+};
+
+struct cq_cfg {
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u64 reserved_43_63:21;
+	u64 ena:1;
+	u64 reset:1;
+	u64 caching:1;
+	u64 reserved_35_39:5;
+	u64 qsize:3;
+	u64 reserved_25_31:7;
+	u64 avg_con:9;
+	u64 reserved_0_15:16;
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u64 reserved_0_15:16;
+	u64 avg_con:9;
+	u64 reserved_25_31:7;
+	u64 qsize:3;
+	u64 reserved_35_39:5;
+	u64 caching:1;
+	u64 reset:1;
+	u64 ena:1;
+	u64 reserved_43_63:21;
+#endif
+};
+
+struct sq_cfg {
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u64 reserved_32_63:32;
+	u64 cq_limit:8;
+	u64 reserved_20_23:4;
+	u64 ena:1;
+	u64 reserved_18_18:1;
+	u64 reset:1;
+	u64 ldwb:1;
+	u64 reserved_11_15:5;
+	u64 qsize:3;
+	u64 reserved_3_7:5;
+	u64 tstmp_bgx_intf:3;
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u64 tstmp_bgx_intf:3;
+	u64 reserved_3_7:5;
+	u64 qsize:3;
+	u64 reserved_11_15:5;
+	u64 ldwb:1;
+	u64 reset:1;
+	u64 reserved_18_18:1;
+	u64 ena:1;
+	u64 reserved_20_23:4;
+	u64 cq_limit:8;
+	u64 reserved_32_63:32;
+#endif
+};
+
+struct rbdr_cfg {
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u64 reserved_45_63:19;
+	u64 ena:1;
+	u64 reset:1;
+	u64 ldwb:1;
+	u64 reserved_36_41:6;
+	u64 qsize:4;
+	u64 reserved_25_31:7;
+	u64 avg_con:9;
+	u64 reserved_12_15:4;
+	u64 lines:12;
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u64 lines:12;
+	u64 reserved_12_15:4;
+	u64 avg_con:9;
+	u64 reserved_25_31:7;
+	u64 qsize:4;
+	u64 reserved_36_41:6;
+	u64 ldwb:1;
+	u64 reset:1;
+	u64 ena: 1;
+	u64 reserved_45_63:19;
+#endif
+};
+
+struct qs_cfg {
+#if defined(__BIG_ENDIAN_BITFIELD)
+	u64 reserved_32_63:32;
+	u64 ena:1;
+	u64 reserved_27_30:4;
+	u64 sq_ins_ena:1;
+	u64 sq_ins_pos:6;
+	u64 lock_ena:1;
+	u64 lock_viol_cqe_ena:1;
+	u64 send_tstmp_ena:1;
+	u64 be:1;
+	u64 reserved_7_15:9;
+	u64 vnic:7;
+#elif defined(__LITTLE_ENDIAN_BITFIELD)
+	u64 vnic:7;
+	u64 reserved_7_15:9;
+	u64 be:1;
+	u64 send_tstmp_ena:1;
+	u64 lock_viol_cqe_ena:1;
+	u64 lock_ena:1;
+	u64 sq_ins_pos:6;
+	u64 sq_ins_ena:1;
+	u64 reserved_27_30:4;
+	u64 ena:1;
+	u64 reserved_32_63:32;
+#endif
+};
+
+#endif /* Q_STRUCT_H */
diff --git a/drivers/net/cavium/thunder_bgx.c b/drivers/net/cavium/thunder_bgx.c
new file mode 100644
index 0000000000..e296350158
--- /dev/null
+++ b/drivers/net/cavium/thunder_bgx.c
@@ -0,0 +1,1529 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2018 Cavium Inc
+ */
+#include <config.h>
+#include <common.h>
+#include <net.h>
+#include <dm.h>
+#include <pci.h>
+#include <misc.h>
+#include <netdev.h>
+#include <malloc.h>
+#include <miiphy.h>
+
+#include <asm/io.h>
+#include <asm/arch-thunderx/thunderx.h>
+#include <asm/arch-thunderx/thunderx_vnic.h>
+
+#include "nic_reg.h"
+#include "nic.h"
+#include "thunder_bgx.h"
+
+/* get the physical address of a CSR on a node */
+#define CSR_PA(node, csr) ((csr) | ((u64)(node) << 44))
+
+/* Registers */
+#define GSERX_CFG(x)		(0x000087E090000080ull + (x) * 0x1000000ull)
+#define GSERX_SCRATCH(x)	(0x000087E090000020ull + (x) * 0x1000000ull)
+#define GSERX_PHY_CTL(x)	(0x000087E090000000ull + (x) * 0x1000000ull)
+#define GSERX_CFG_BGX		BIT_ULL(2)
+#define GSER_RX_EIE_DETSTS(x)	(0x000087e090000150ull + (x) * 0x1000000ull)
+#define GSER_CDRLOCK		(8)
+#define GSER_BR_RXX_CTL(x,y)	(0x000087e090000400ull + (x) * 0x1000000ull + (y) * 0x80) 
+#define GSER_BR_RXX_CTL_RXT_SWM	BIT_ULL(2)
+#define GSER_BR_RXX_EER(x,y)	(0x000087e090000418ull + (x) * 0x1000000ull + (y) * 0x80)
+#define GSER_BR_RXX_EER_RXT_ESV BIT_ULL(14)
+#define GSER_BR_RXX_EER_RXT_EER BIT_ULL(15)
+#define EER_RXT_ESV		(14)
+
+static const phy_interface_t if_mode[] = {
+	[QLM_MODE_SGMII]  = PHY_INTERFACE_MODE_SGMII,
+	[QLM_MODE_RGMII]  = PHY_INTERFACE_MODE_RGMII,
+	[QLM_MODE_QSGMII] = PHY_INTERFACE_MODE_QSGMII,
+	[QLM_MODE_XAUI]   = PHY_INTERFACE_MODE_XAUI,
+	[QLM_MODE_RXAUI]  = PHY_INTERFACE_MODE_RXAUI,
+};
+
+struct lmac {
+	struct bgx		*bgx;
+	int			dmac;
+	u8			mac[ETH_ALEN];
+	u8                      lmac_type;
+	u8                      lane_to_sds;
+	bool                    use_training;
+	bool                    autoneg;
+	bool			link_up;
+	int			lmacid; /* ID within BGX */
+	int			phy_addr; /* ID on board */
+	struct eth_device	netdev;
+	struct mii_dev		*mii_bus;
+	struct phy_device       *phydev;
+	unsigned int            last_duplex;
+	unsigned int            last_link;
+	unsigned int            last_speed;
+	u8			qlm_mode;
+	int			qlm;
+};
+
+struct bgx {
+	u8			bgx_id;
+	int			node;
+	struct	lmac		lmac[MAX_LMAC_PER_BGX];
+	u8			lmac_count;
+	u8			max_lmac;
+	void __iomem		*reg_base;
+	struct pci_dev		*pdev;
+	bool                    is_rgx;
+};
+
+struct bgx_board_info bgx_board_info[CONFIG_MAX_BGX];
+
+struct bgx *bgx_vnic[CONFIG_MAX_BGX];
+bool is_altpkg = 0;
+extern int __cavm_if_phy_xs_init(struct mii_dev *bus, int phy_addr);
+
+/* The Cavium ThunderX network controller can *only* be found in SoCs
+ * containing the ThunderX ARM64 CPU implementation.  All accesses to the device
+ * registers on this platform are implicitly strongly ordered with respect
+ * to memory accesses. So writeq_relaxed() and readq_relaxed() are safe to use
+ * with no memory barriers in this driver.  The readq()/writeq() functions add
+ * explicit ordering operation which in this case are redundant, and only
+ * add overhead.
+ */
+
+/* Register read/write APIs */
+static u64 bgx_reg_read(struct bgx *bgx, u8 lmac, u64 offset)
+{
+	u64 addr = (uintptr_t)bgx->reg_base + ((u32)lmac << 20) + offset;
+
+	return readq((void *)addr);
+}
+
+static void bgx_reg_write(struct bgx *bgx, u8 lmac, u64 offset, u64 val)
+{
+	u64 addr = (uintptr_t)bgx->reg_base + ((u32)lmac << 20) + offset;
+
+	writeq(val, (void *)addr);
+}
+
+static void bgx_reg_modify(struct bgx *bgx, u8 lmac, u64 offset, u64 val)
+{
+	u64 addr = (uintptr_t)bgx->reg_base + ((u32)lmac << 20) + offset;
+
+	writeq(val | bgx_reg_read(bgx, lmac, offset), (void *)addr);
+}
+
+static int bgx_poll_reg(struct bgx *bgx, u8 lmac, u64 reg, u64 mask, bool zero)
+{
+	int timeout = 100;
+	u64 reg_val;
+
+	while (timeout) {
+		reg_val = bgx_reg_read(bgx, lmac, reg);
+		if (zero && !(reg_val & mask))
+			return 0;
+		if (!zero && (reg_val & mask))
+			return 0;
+		mdelay(1);
+		timeout--;
+	}
+	return 1;
+}
+
+static int gser_poll_reg(u64 reg, int bit, u64 mask, u64 expected_val, int timeout)
+{
+	u64 reg_val;
+	debug("gser_poll_reg: reg = %#llx, mask = %#llx, expected_val = %#llx, bit = %d\n",
+		reg, mask, expected_val, bit);
+	while (timeout) {
+		reg_val = readq(CSR_PA(0, reg)) >> bit;
+		if ((reg_val & mask) == (expected_val))
+			return 0;
+		mdelay(1);
+		timeout--;
+	}
+	return 1;
+}
+
+static bool is_bgx_port_valid(int bgx, int lmac)
+{
+	debug("is_bgx_port_valid bgx %d lmac %d valid %d\n",
+	      bgx, lmac, bgx_board_info[bgx].lmac_reg[lmac]);
+
+	if (bgx_board_info[bgx].lmac_reg[lmac])
+		return 1;
+	else
+		return 0;
+}
+
+struct lmac *bgx_get_lmac(int node, int bgx_idx, int lmacid)
+{
+	struct bgx *bgx = bgx_vnic[(node * CONFIG_MAX_BGX_PER_NODE) + bgx_idx];
+
+	if (bgx)
+		return &bgx->lmac[lmacid];
+
+	return NULL;
+}
+
+const u8 *bgx_get_lmac_mac(int node, int bgx_idx, int lmacid)
+{
+	struct bgx *bgx = bgx_vnic[(node * CONFIG_MAX_BGX_PER_NODE) + bgx_idx];
+
+	if (bgx)
+		return bgx->lmac[lmacid].mac;
+
+	return NULL;
+}
+
+void bgx_set_lmac_mac(int node, int bgx_idx, int lmacid, const u8 *mac)
+{
+	struct bgx *bgx = bgx_vnic[(node * CONFIG_MAX_BGX_PER_NODE) + bgx_idx];
+
+	if (!bgx)
+		return;
+
+	memcpy(bgx->lmac[lmacid].mac, mac, 6);
+}
+
+/* Return number of BGX present in HW */
+void bgx_get_count(int node, int *bgx_count)
+{
+	int i;
+	struct bgx *bgx;
+
+	*bgx_count = 0;
+	for (i = 0; i < CONFIG_MAX_BGX_PER_NODE; i++) {
+		bgx = bgx_vnic[node * CONFIG_MAX_BGX_PER_NODE + i];
+		debug("bgx_vnic[%u]: %p\n", node * CONFIG_MAX_BGX_PER_NODE + i, bgx);
+		if (bgx)
+			*bgx_count |= (1 << i);
+	}
+}
+
+/* Return number of LMAC configured for this BGX */
+int bgx_get_lmac_count(int node, int bgx_idx)
+{
+	struct bgx *bgx;
+
+	bgx = bgx_vnic[(node * CONFIG_MAX_BGX_PER_NODE) + bgx_idx];
+	if (bgx)
+		return bgx->lmac_count;
+
+	return 0;
+}
+
+void bgx_lmac_rx_tx_enable(int node, int bgx_idx, int lmacid, bool enable)
+{
+	struct bgx *bgx = bgx_vnic[(node * CONFIG_MAX_BGX_PER_NODE) + bgx_idx];
+	u64 cfg;
+
+	if (!bgx)
+		return;
+
+	cfg = bgx_reg_read(bgx, lmacid, BGX_CMRX_CFG);
+	if (enable)
+		cfg |= CMR_PKT_RX_EN | CMR_PKT_TX_EN;
+	else
+		cfg &= ~(CMR_PKT_RX_EN | CMR_PKT_TX_EN);
+	bgx_reg_write(bgx, lmacid, BGX_CMRX_CFG, cfg);
+}
+
+static void bgx_flush_dmac_addrs(struct bgx *bgx, u64 lmac)
+{
+	u64 dmac = 0x00;
+	u64 offset, addr;
+
+	while (bgx->lmac[lmac].dmac > 0) {
+		offset = ((bgx->lmac[lmac].dmac - 1) * sizeof(dmac)) +
+			(lmac * MAX_DMAC_PER_LMAC * sizeof(dmac));
+		addr = (uintptr_t)bgx->reg_base +
+				BGX_CMR_RX_DMACX_CAM + offset;
+		writeq(dmac, (void *)addr);
+		bgx->lmac[lmac].dmac--;
+	}
+}
+
+/* Configure BGX LMAC in internal loopback mode */
+void bgx_lmac_internal_loopback(int node, int bgx_idx,
+				int lmac_idx, bool enable)
+{
+	struct bgx *bgx;
+	struct lmac *lmac;
+	u64    cfg;
+
+	bgx = bgx_vnic[(node * CONFIG_MAX_BGX_PER_NODE) + bgx_idx];
+	if (!bgx)
+		return;
+
+	lmac = &bgx->lmac[lmac_idx];
+	if (lmac->qlm_mode == QLM_MODE_SGMII) {
+		cfg = bgx_reg_read(bgx, lmac_idx, BGX_GMP_PCS_MRX_CTL);
+		if (enable)
+			cfg |= PCS_MRX_CTL_LOOPBACK1;
+		else
+			cfg &= ~PCS_MRX_CTL_LOOPBACK1;
+		bgx_reg_write(bgx, lmac_idx, BGX_GMP_PCS_MRX_CTL, cfg);
+	} else {
+		cfg = bgx_reg_read(bgx, lmac_idx, BGX_SPUX_CONTROL1);
+		if (enable)
+			cfg |= SPU_CTL_LOOPBACK;
+		else
+			cfg &= ~SPU_CTL_LOOPBACK;
+		bgx_reg_write(bgx, lmac_idx, BGX_SPUX_CONTROL1, cfg);
+	}
+}
+
+/* Return the DLM used for the BGX */
+static int get_qlm_for_bgx(int node, int bgx_id, int index)
+{
+	int qlm = 0;
+	u64 cfg;
+
+	if (CAVIUM_IS_MODEL(CAVIUM_CN81XX)) {
+		qlm = (bgx_id) ? 2 : 0;
+		qlm += (index >= 2) ? 1 : 0;
+	} else if (CAVIUM_IS_MODEL(CAVIUM_CN83XX)) {
+		switch (bgx_id) {
+		case 0:
+			qlm = 2;
+			break;
+		case 1:
+			qlm = 3;
+			break;
+		case 2:
+			if (index >= 2)
+				qlm = 6;
+			else
+				qlm = 5;
+			break;
+		case 3:
+			qlm = 4;
+			break;
+		}
+	}
+
+	cfg = readq(CSR_PA(node, GSERX_CFG(qlm))) & GSERX_CFG_BGX;
+	debug("get_qlm_for_bgx:qlm%d: cfg = %lld\n", qlm, cfg);
+
+	/* Check if DLM is configured as BGX# */
+	if (cfg) {
+		if (readq(CSR_PA(node, GSERX_PHY_CTL(qlm))))
+			return -1;
+		return qlm;
+	}
+	return -1;
+}
+
+static int bgx_lmac_sgmii_init(struct bgx *bgx, int lmacid)
+{
+	u64 cfg;
+	struct lmac *lmac;
+
+	lmac = &bgx->lmac[lmacid];
+
+	debug("bgx_lmac_sgmii_init: bgx_id = %d, lmacid = %d\n", bgx->bgx_id, lmacid);
+
+	bgx_reg_modify(bgx, lmacid, BGX_GMP_GMI_TXX_THRESH, 0x30);
+	/* max packet size */
+	bgx_reg_modify(bgx, lmacid, BGX_GMP_GMI_RXX_JABBER, MAX_FRAME_SIZE);
+
+	/* Disable frame alignment if using preamble */
+	cfg = bgx_reg_read(bgx, lmacid, BGX_GMP_GMI_TXX_APPEND);
+	if (cfg & 1)
+		bgx_reg_write(bgx, lmacid, BGX_GMP_GMI_TXX_SGMII_CTL, 0);
+
+	/* Enable lmac */
+	bgx_reg_modify(bgx, lmacid, BGX_CMRX_CFG, CMR_EN);
+
+	/* PCS reset */
+	bgx_reg_modify(bgx, lmacid, BGX_GMP_PCS_MRX_CTL, PCS_MRX_CTL_RESET);
+	if (bgx_poll_reg(bgx, lmacid, BGX_GMP_PCS_MRX_CTL,
+			 PCS_MRX_CTL_RESET, true)) {
+		printf("BGX PCS reset not completed\n");
+		return -1;
+	}
+
+	/* power down, reset autoneg, autoneg enable */
+	cfg = bgx_reg_read(bgx, lmacid, BGX_GMP_PCS_MRX_CTL);
+	cfg &= ~PCS_MRX_CTL_PWR_DN;
+
+	if (bgx_board_info[bgx->bgx_id].phy_info[lmacid].autoneg_dis)
+		cfg |= (PCS_MRX_CTL_RST_AN);
+	else
+		cfg |= (PCS_MRX_CTL_RST_AN | PCS_MRX_CTL_AN_EN);
+	bgx_reg_write(bgx, lmacid, BGX_GMP_PCS_MRX_CTL, cfg);
+
+	/* Disable disparity for QSGMII mode, to prevent propogation across
+	   ports. */
+
+	if (lmac->qlm_mode == QLM_MODE_QSGMII) {
+		cfg = bgx_reg_read(bgx, lmacid, BGX_GMP_PCS_MISCX_CTL);
+		cfg &= ~PCS_MISC_CTL_DISP_EN;
+		bgx_reg_write(bgx, lmacid, BGX_GMP_PCS_MISCX_CTL, cfg);
+		return 0; /* Skip checking AN_CPT */
+	}
+
+	if (lmac->qlm_mode == QLM_MODE_SGMII) {
+		if (bgx_poll_reg(bgx, lmacid, BGX_GMP_PCS_MRX_STATUS,
+			 PCS_MRX_STATUS_AN_CPT, false)) {
+			printf("BGX AN_CPT not completed\n");
+			return -1;
+		}
+	}
+
+	return 0;
+}
+
+static int bgx_lmac_sgmii_set_link_speed(struct lmac *lmac)
+{
+	u64 prtx_cfg;
+	u64 pcs_miscx_ctl;
+	u64 cfg;
+	struct bgx *bgx = lmac->bgx;
+	unsigned int lmacid = lmac->lmacid;
+		
+	debug("bgx_lmac_sgmii_set_link_speed(): lmacid %d\n", lmac->lmacid);
+
+	/* Disable LMAC before setting up speed */
+	cfg = bgx_reg_read(bgx, lmacid, BGX_CMRX_CFG);
+	cfg &= ~CMR_EN;
+        bgx_reg_write(bgx, lmacid, BGX_CMRX_CFG, cfg);
+
+	/* Read GMX CFG */
+	prtx_cfg = bgx_reg_read(bgx, lmacid,
+				BGX_GMP_GMI_PRTX_CFG);
+ 	/* Read PCS MISCS CTL */
+	pcs_miscx_ctl = bgx_reg_read(bgx, lmacid,
+				BGX_GMP_PCS_MISCX_CTL);
+
+	/* Use GMXENO to force the link down*/
+	if (lmac->link_up) {
+		pcs_miscx_ctl &= ~PCS_MISC_CTL_GMX_ENO;
+		/* change the duplex setting if the link is up */
+		prtx_cfg |= GMI_PORT_CFG_DUPLEX;
+	} else 
+		pcs_miscx_ctl |= PCS_MISC_CTL_GMX_ENO;
+
+	/* speed based setting for GMX */
+	switch( lmac->last_speed) {
+	case 10:
+		prtx_cfg &= ~GMI_PORT_CFG_SPEED;
+		prtx_cfg |= GMI_PORT_CFG_SPEED_MSB;
+		prtx_cfg &= ~GMI_PORT_CFG_SLOT_TIME;
+		pcs_miscx_ctl |= 50; /* sampling point */
+		bgx_reg_write(bgx, lmacid, BGX_GMP_GMI_TXX_SLOT, 0x40);
+		bgx_reg_write(bgx, lmacid, BGX_GMP_GMI_TXX_BURST, 0);
+		break;
+	case 100:
+		prtx_cfg &= ~GMI_PORT_CFG_SPEED;
+		prtx_cfg &= ~GMI_PORT_CFG_SPEED_MSB;
+		prtx_cfg &= ~GMI_PORT_CFG_SLOT_TIME;
+		pcs_miscx_ctl |= 0x5; /* sampling point */
+		bgx_reg_write(bgx, lmacid, BGX_GMP_GMI_TXX_SLOT, 0x40);
+		bgx_reg_write(bgx, lmacid, BGX_GMP_GMI_TXX_BURST, 0);
+		break;
+	case 1000:
+		prtx_cfg |= GMI_PORT_CFG_SPEED;
+		prtx_cfg &= ~GMI_PORT_CFG_SPEED_MSB;
+		prtx_cfg |= GMI_PORT_CFG_SLOT_TIME;
+		pcs_miscx_ctl |= 0x1; /* sampling point */
+		bgx_reg_write(bgx, lmacid, BGX_GMP_GMI_TXX_SLOT, 0x200);
+		if (lmac->last_duplex)
+			bgx_reg_write(bgx, lmacid, BGX_GMP_GMI_TXX_BURST, 0);
+		else /* half duplex */
+			bgx_reg_write(bgx, lmacid, BGX_GMP_GMI_TXX_BURST, 0x2000);
+		break;
+	default:
+		break;
+	}
+	
+	/* write back the new PCS misc and GMX settings */
+	bgx_reg_write(bgx, lmacid, BGX_GMP_PCS_MISCX_CTL, pcs_miscx_ctl);
+	bgx_reg_write(bgx, lmacid, BGX_GMP_GMI_PRTX_CFG, prtx_cfg);
+
+	/* read back GMX CFG again to check config completion */
+	bgx_reg_read(bgx, lmacid, BGX_GMP_GMI_PRTX_CFG);
+		
+	/* enable BGX back */
+	cfg = bgx_reg_read(bgx, lmacid, BGX_CMRX_CFG);
+	cfg |= CMR_EN;
+       	bgx_reg_write(bgx, lmacid, BGX_CMRX_CFG, cfg);
+
+	return 0;
+}
+
+static int bgx_lmac_xaui_init(struct bgx *bgx, int lmacid, int lmac_type)
+{
+	u64 cfg;
+	struct lmac *lmac;
+
+	lmac = &bgx->lmac[lmacid];
+
+	/* Reset SPU */
+	bgx_reg_modify(bgx, lmacid, BGX_SPUX_CONTROL1, SPU_CTL_RESET);
+	if (bgx_poll_reg(bgx, lmacid, BGX_SPUX_CONTROL1, SPU_CTL_RESET, true)) {
+		printf("BGX SPU reset not completed\n");
+		return -1;
+	}
+
+	/* Disable LMAC */
+	cfg = bgx_reg_read(bgx, lmacid, BGX_CMRX_CFG);
+	cfg &= ~CMR_EN;
+	bgx_reg_write(bgx, lmacid, BGX_CMRX_CFG, cfg);
+
+	bgx_reg_modify(bgx, lmacid, BGX_SPUX_CONTROL1, SPU_CTL_LOW_POWER);
+	/* Set interleaved running disparity for RXAUI */
+	if (lmac->qlm_mode != QLM_MODE_RXAUI)
+		bgx_reg_modify(bgx, lmacid,
+			       BGX_SPUX_MISC_CONTROL, SPU_MISC_CTL_RX_DIS);
+	else
+		bgx_reg_modify(bgx, lmacid, BGX_SPUX_MISC_CONTROL,
+			       SPU_MISC_CTL_RX_DIS | SPU_MISC_CTL_INTLV_RDISP);
+
+	/* clear all interrupts */
+	cfg = bgx_reg_read(bgx, lmacid, BGX_SMUX_RX_INT);
+	bgx_reg_write(bgx, lmacid, BGX_SMUX_RX_INT, cfg);
+	cfg = bgx_reg_read(bgx, lmacid, BGX_SMUX_TX_INT);
+	bgx_reg_write(bgx, lmacid, BGX_SMUX_TX_INT, cfg);
+	cfg = bgx_reg_read(bgx, lmacid, BGX_SPUX_INT);
+	bgx_reg_write(bgx, lmacid, BGX_SPUX_INT, cfg);
+
+	if (lmac->use_training) {
+		bgx_reg_write(bgx, lmacid, BGX_SPUX_BR_PMD_LP_CUP, 0x00);
+		bgx_reg_write(bgx, lmacid, BGX_SPUX_BR_PMD_LD_CUP, 0x00);
+		bgx_reg_write(bgx, lmacid, BGX_SPUX_BR_PMD_LD_REP, 0x00);
+		/* training enable */
+		bgx_reg_modify(bgx, lmacid,
+			       BGX_SPUX_BR_PMD_CRTL, SPU_PMD_CRTL_TRAIN_EN);
+	}
+
+	/* Append FCS to each packet */
+	bgx_reg_modify(bgx, lmacid, BGX_SMUX_TX_APPEND, SMU_TX_APPEND_FCS_D);
+
+	/* Disable forward error correction */
+	cfg = bgx_reg_read(bgx, lmacid, BGX_SPUX_FEC_CONTROL);
+	cfg &= ~SPU_FEC_CTL_FEC_EN;
+	bgx_reg_write(bgx, lmacid, BGX_SPUX_FEC_CONTROL, cfg);
+
+	/* Disable autoneg */
+	cfg = bgx_reg_read(bgx, lmacid, BGX_SPUX_AN_CONTROL);
+	cfg = cfg & ~(SPU_AN_CTL_XNP_EN);
+	if (lmac->use_training)
+		cfg = cfg | (SPU_AN_CTL_AN_EN);
+	else
+		cfg = cfg & ~(SPU_AN_CTL_AN_EN);
+	bgx_reg_write(bgx, lmacid, BGX_SPUX_AN_CONTROL, cfg);
+
+	cfg = bgx_reg_read(bgx, lmacid, BGX_SPUX_AN_ADV);
+	/* Clear all KR bits, configure according to the mode */
+	cfg &= ~((0xfULL << 22) | (1ULL << 12));
+	if (lmac->qlm_mode == QLM_MODE_10G_KR)
+		cfg |= (1 << 23);
+	else if (lmac->qlm_mode == QLM_MODE_40G_KR4)
+		cfg |= (1 << 24);
+	bgx_reg_write(bgx, lmacid, BGX_SPUX_AN_ADV, cfg);
+
+	cfg = bgx_reg_read(bgx, 0, BGX_SPU_DBG_CONTROL);
+	if (lmac->use_training)
+		cfg |= SPU_DBG_CTL_AN_ARB_LINK_CHK_EN;
+	else
+		cfg &= ~SPU_DBG_CTL_AN_ARB_LINK_CHK_EN;
+	bgx_reg_write(bgx, 0, BGX_SPU_DBG_CONTROL, cfg);
+
+	/* Enable lmac */
+	bgx_reg_modify(bgx, lmacid, BGX_CMRX_CFG, CMR_EN);
+
+	cfg = bgx_reg_read(bgx, lmacid, BGX_SPUX_CONTROL1);
+	cfg &= ~SPU_CTL_LOW_POWER;
+	bgx_reg_write(bgx, lmacid, BGX_SPUX_CONTROL1, cfg);
+
+	cfg = bgx_reg_read(bgx, lmacid, BGX_SMUX_TX_CTL);
+	cfg &= ~SMU_TX_CTL_UNI_EN;
+	cfg |= SMU_TX_CTL_DIC_EN;
+	bgx_reg_write(bgx, lmacid, BGX_SMUX_TX_CTL, cfg);
+
+	/* take lmac_count into account */
+	bgx_reg_modify(bgx, lmacid, BGX_SMUX_TX_THRESH, (0x100 - 1));
+	/* max packet size */
+	bgx_reg_modify(bgx, lmacid, BGX_SMUX_RX_JABBER, MAX_FRAME_SIZE);
+
+	debug("xaui_init: lmacid = %d, qlm = %d, qlm_mode = %d\n",
+		lmacid, lmac->qlm, lmac->qlm_mode);
+	/* RXAUI with Marvell PHY requires some tweaking */
+	if (lmac->qlm_mode == QLM_MODE_RXAUI) {
+		char mii_name[20];
+		snprintf(mii_name, sizeof(mii_name), "smi%d",
+			 bgx_board_info[bgx->bgx_id].phy_info[lmacid].mdio_bus);
+
+		debug("mii_name: %s\n", mii_name);
+		lmac->mii_bus = miiphy_get_dev_by_name(mii_name);
+		lmac->phy_addr = bgx_board_info[bgx->bgx_id].
+				 phy_info[lmacid].phy_addr;
+		__cavm_if_phy_xs_init(lmac->mii_bus, lmac->phy_addr);
+	}
+
+	return 0;
+}
+
+/* Get max number of lanes present in a given QLM/DLM */
+static int get_qlm_lanes(int qlm)
+{
+	if (CAVIUM_IS_MODEL(CAVIUM_CN81XX))
+		return 2;
+	else if (CAVIUM_IS_MODEL(CAVIUM_CN83XX))
+		return (qlm >= 5) ? 2 : 4;
+	else
+		return -1;
+}
+
+int __rx_equalization(int qlm, int lane)
+{
+	int max_lanes = get_qlm_lanes(qlm);
+	int l;
+	int fail = 0;
+
+	/* Before completing Rx equalization wait for GSERx_RX_EIE_DETSTS[CDRLOCK] to be set
+	   This ensures the rx data is valid */
+	if (lane == -1) {
+		if (gser_poll_reg(GSER_RX_EIE_DETSTS(qlm), GSER_CDRLOCK, 0xf, (1 << max_lanes) - 1, 100)) {
+			debug("ERROR: DLM%d: CDR Lock not detected for 2 lanes\n", qlm);
+			return -1;
+		}
+	} else {
+		if (gser_poll_reg(GSER_RX_EIE_DETSTS(qlm), GSER_CDRLOCK, (0xf & (1 << lane)), (1 << lane), 100)) {
+			debug("ERROR: DLM%d: CDR Lock not detected on %d lane\n", qlm, lane);
+			return -1;
+		}
+	}
+
+	for (l = 0; l < max_lanes; l++) {
+		u64 rctl, reer;
+
+		if ((lane != -1) && (lane != l))
+			continue;
+
+		/* Enable software control */
+		rctl = readq(CSR_PA(0, GSER_BR_RXX_CTL(qlm, l)));
+		rctl |= GSER_BR_RXX_CTL_RXT_SWM;
+		writeq(rctl, CSR_PA(0, GSER_BR_RXX_CTL(qlm, l)));
+
+		/* Clear the completion flag and initiate a new request */
+		reer = readq(CSR_PA(0, GSER_BR_RXX_EER(qlm, l)));
+		reer &= ~GSER_BR_RXX_EER_RXT_ESV;
+		reer |= GSER_BR_RXX_EER_RXT_EER;
+		writeq(reer, CSR_PA(0, GSER_BR_RXX_EER(qlm, l)));
+	}
+
+	/* Wait for RX equalization to complete */
+	for (l = 0; l < max_lanes; l++) {
+		u64 rctl, reer;
+
+		if ((lane != -1) && (lane != l))
+			continue;
+
+		gser_poll_reg(GSER_BR_RXX_EER(qlm, l), EER_RXT_ESV, 1, 1, 200);
+		reer = readq(CSR_PA(0, GSER_BR_RXX_EER(qlm, l)));
+
+		/* Switch back to hardware control */
+		rctl = readq(CSR_PA(0, GSER_BR_RXX_CTL(qlm, l)));
+		rctl &= ~GSER_BR_RXX_CTL_RXT_SWM;
+		writeq(rctl, CSR_PA(0, GSER_BR_RXX_CTL(qlm, l)));
+
+		if (reer & GSER_BR_RXX_EER_RXT_ESV) {
+			debug("Rx equalization completed on DLM%d lane%d, rxt_esm = 0x%llx\n",
+				qlm, l, (reer & 0x3fff));
+		} else {
+			debug("Rx equalization timedout on DLM%d lane%d\n", qlm, l);
+			fail = 1;
+		}
+	}
+
+	return (fail) ? -1 : 0;
+}
+
+static int bgx_xaui_check_link(struct lmac *lmac)
+{
+	struct bgx *bgx = lmac->bgx;
+	int lmacid = lmac->lmacid;
+	int lmac_type = lmac->lmac_type;
+	u64 cfg;
+
+	bgx_reg_modify(bgx, lmacid, BGX_SPUX_MISC_CONTROL, SPU_MISC_CTL_RX_DIS);
+
+	/* check if auto negotiation is complete */
+	cfg = bgx_reg_read(bgx, lmacid, BGX_SPUX_AN_CONTROL);
+	if (cfg & SPU_AN_CTL_AN_EN) {
+		cfg = bgx_reg_read(bgx, lmacid, BGX_SPUX_AN_STATUS);
+		if (!(cfg & SPU_AN_STS_AN_COMPLETE)) {
+			/* Restart autonegotiation */
+			debug("restarting auto-neg\n");
+			bgx_reg_modify(bgx, lmacid, BGX_SPUX_AN_CONTROL, SPU_AN_CTL_AN_RESTART);
+			return -1;
+		}
+	}
+
+	debug("%s link use_training %d\n",__func__, lmac->use_training);
+	if (lmac->use_training) {
+		cfg = bgx_reg_read(bgx, lmacid, BGX_SPUX_INT);
+		if (!(cfg & (1ull << 13))) {
+			debug("waiting for link training\n");
+			/* Clear the training interrupts (W1C) */
+			cfg = (1ull << 13) | (1ull << 14);
+			bgx_reg_write(bgx, lmacid, BGX_SPUX_INT, cfg);
+
+			udelay(2000);
+			/* Restart training */
+			cfg = bgx_reg_read(bgx, lmacid, BGX_SPUX_BR_PMD_CRTL);
+			cfg |= (1ull << 0);
+			bgx_reg_write(bgx, lmacid, BGX_SPUX_BR_PMD_CRTL, cfg);
+			return -1;
+		}
+	}
+
+	/* Perform RX Equalization. Applies to non-KR interfaces for speeds
+	   >= 6.25Gbps. */
+	if (!lmac->use_training) {
+		int qlm;
+		bool use_dlm = 0;
+		if (CAVIUM_IS_MODEL(CAVIUM_CN81XX)
+		    || (CAVIUM_IS_MODEL(CAVIUM_CN83XX) && (bgx->bgx_id == 2)))
+			use_dlm = 1;
+		switch (lmac->lmac_type) {
+		default:
+		case BGX_MODE_SGMII:
+		case BGX_MODE_RGMII:
+		case BGX_MODE_XAUI:
+			/* Nothing to do */
+			break;
+		case BGX_MODE_XLAUI:
+			if (use_dlm) {
+				if (__rx_equalization(lmac->qlm, -1) ||
+					__rx_equalization(lmac->qlm+1, -1)) {
+					printf("BGX%d:%d: Waiting for RX Equalization on DLM%d/DLM%d\n",
+						bgx->bgx_id, lmacid, lmac->qlm, lmac->qlm+1);
+					return -1;
+				}
+			} else {
+				if (__rx_equalization(lmac->qlm, -1)) {
+					printf("BGX%d:%d: Waiting for RX Equalization on QLM%d:\n",
+						bgx->bgx_id, lmacid, lmac->qlm);
+					return -1;
+				}
+			}
+			break;
+		case BGX_MODE_RXAUI:
+			/* RXAUI0 uses LMAC0:QLM0/QLM2 and RXAUI1 uses LMAC1:QLM1/QLM3
+			   RXAUI requires 2 lanes for each interface */
+			qlm = lmac->qlm;
+			if (__rx_equalization(qlm, 0)) {
+				printf("BGX%d:%d: Waiting for RX Equalization on QLM%d, Lane0\n",
+					bgx->bgx_id, lmacid, qlm);
+				return -1;
+			}
+			if (__rx_equalization(qlm, 1)) {
+				printf("BGX%d:%d: Waiting for RX Equalization on QLM%d, Lane1\n",
+					bgx->bgx_id, lmacid, qlm);
+				return -1;
+			}
+			break;
+		case BGX_MODE_XFI:
+			{
+				int lid;
+				if ((bgx->bgx_id == 0) && is_altpkg && lmacid)
+					lid = 0;
+				else if ((lmacid >= 2) && use_dlm)
+					lid = lmacid - 2;
+				else
+					lid = lmacid;
+
+				if (__rx_equalization(lmac->qlm, lid))
+					printf("BGX%d:%d: Waiting for RX Equalization on QLM%d\n",
+						bgx->bgx_id, lid, lmac->qlm);
+			}
+			break;
+		}
+	}
+
+	/* wait for PCS to come out of reset */
+	if (bgx_poll_reg(bgx, lmacid, BGX_SPUX_CONTROL1, SPU_CTL_RESET, true)) {
+		printf("BGX SPU reset not completed\n");
+		return -1;
+	}
+
+	if ((lmac_type == 3) || (lmac_type == 4)) {
+		if (bgx_poll_reg(bgx, lmacid, BGX_SPUX_BR_STATUS1,
+				 SPU_BR_STATUS_BLK_LOCK, false)) {
+			printf("SPU_BR_STATUS_BLK_LOCK not completed\n");
+			return -1;
+		}
+	} else {
+		if (bgx_poll_reg(bgx, lmacid, BGX_SPUX_BX_STATUS,
+				 SPU_BX_STATUS_RX_ALIGN, false)) {
+			printf("SPU_BX_STATUS_RX_ALIGN not completed\n");
+			return -1;
+		}
+	}
+
+	/* Clear rcvflt bit (latching high) and read it back */
+	bgx_reg_modify(bgx, lmacid, BGX_SPUX_STATUS2, SPU_STATUS2_RCVFLT);
+	if (bgx_reg_read(bgx, lmacid, BGX_SPUX_STATUS2) & SPU_STATUS2_RCVFLT) {
+		printf("Receive fault, retry training\n");
+		if (lmac->use_training) {
+			cfg = bgx_reg_read(bgx, lmacid, BGX_SPUX_INT);
+			if (!(cfg & (1ull << 13))) {
+				cfg = (1ull << 13) | (1ull << 14);
+				bgx_reg_write(bgx, lmacid, BGX_SPUX_INT, cfg);
+				cfg = bgx_reg_read(bgx, lmacid,
+						   BGX_SPUX_BR_PMD_CRTL);
+				cfg |= (1ull << 0);
+				bgx_reg_write(bgx, lmacid,
+					      BGX_SPUX_BR_PMD_CRTL, cfg);
+				return -1;
+			}
+		}
+		return -1;
+	}
+
+	/* Wait for MAC RX to be ready */
+	if (bgx_poll_reg(bgx, lmacid, BGX_SMUX_RX_CTL,
+			 SMU_RX_CTL_STATUS, true)) {
+		printf( "SMU RX link not okay\n");
+		return -1;
+	}
+
+	/* Wait for BGX RX to be idle */
+	if (bgx_poll_reg(bgx, lmacid, BGX_SMUX_CTL, SMU_CTL_RX_IDLE, false)) {
+		printf("SMU RX not idle\n");
+		return -1;
+	}
+
+	/* Wait for BGX TX to be idle */
+	if (bgx_poll_reg(bgx, lmacid, BGX_SMUX_CTL, SMU_CTL_TX_IDLE, false)) {
+		printf("SMU TX not idle\n");
+		return -1;
+	}
+
+	if (bgx_reg_read(bgx, lmacid, BGX_SPUX_STATUS2) & SPU_STATUS2_RCVFLT) {
+		printf("Receive fault\n");
+		return -1;
+	}
+
+	/* Receive link is latching low. Force it high and verify it */
+	if (!(bgx_reg_read(bgx, lmacid, BGX_SPUX_STATUS1) & SPU_STATUS1_RCV_LNK))
+		bgx_reg_modify(bgx, lmacid, BGX_SPUX_STATUS1, SPU_STATUS1_RCV_LNK);
+	if (bgx_poll_reg(bgx, lmacid, BGX_SPUX_STATUS1,
+			 SPU_STATUS1_RCV_LNK, false)) {
+		printf("SPU receive link down\n");
+		return -1;
+	}
+
+	cfg = bgx_reg_read(bgx, lmacid, BGX_SPUX_MISC_CONTROL);
+	cfg &= ~SPU_MISC_CTL_RX_DIS;
+	bgx_reg_write(bgx, lmacid, BGX_SPUX_MISC_CONTROL, cfg);
+	return 0;
+}
+
+int bgx_poll_for_link(int node, int bgx_idx, int lmacid)
+{
+	int ret;
+	struct lmac *lmac = bgx_get_lmac(node, bgx_idx, lmacid);
+	char mii_name[10];
+
+	if (lmac == NULL) {
+		printf("LMAC %d/%d/%d is disabled or doesn't exist\n",
+		       node, bgx_idx, lmacid);
+		return 0;
+	}
+
+	debug("%s: %d, lmac: %d/%d/%d %p\n",
+	      __FILE__, __LINE__,
+	      node, bgx_idx, lmacid, lmac);
+
+	if ((lmac->qlm_mode == QLM_MODE_SGMII) ||
+	    (lmac->qlm_mode == QLM_MODE_RGMII) ||
+	    (lmac->qlm_mode == QLM_MODE_QSGMII)) {
+
+		if (bgx_board_info[bgx_idx].phy_info[lmacid].phy_addr == -1) {
+			lmac->link_up = 1;
+			lmac->last_speed = 1000;
+			lmac->last_duplex = 1;
+			printf("BGX%d:LMAC %u link up\n", bgx_idx, lmacid);
+			return lmac->link_up;
+		}
+		snprintf(mii_name, sizeof(mii_name), "smi%d",
+			 bgx_board_info[bgx_idx].phy_info[lmacid].mdio_bus);
+
+		debug("mii_name: %s\n", mii_name);
+
+		lmac->mii_bus = miiphy_get_dev_by_name(mii_name);
+		lmac->phy_addr = bgx_board_info[bgx_idx].
+				 phy_info[lmacid].phy_addr;
+
+		debug("lmac->mii_bus: %p\n",lmac->mii_bus);
+		if (!lmac->mii_bus) {
+			printf("MDIO device %s not found\n", mii_name);
+			ret = -ENODEV;
+			return ret;
+		}
+
+		lmac->phydev = phy_connect(lmac->mii_bus, lmac->phy_addr,
+					   &lmac->netdev,
+					   if_mode[lmac->qlm_mode]);
+
+		if (!lmac->phydev) {
+			printf("%s: No PHY device\n",
+				lmac->netdev.name);
+			return -1;
+		}
+
+		ret = phy_config(lmac->phydev);
+		if (ret) {
+			printf("%s: Could not initialize PHY %s\n",
+				lmac->netdev.name, lmac->phydev->dev->name);
+			return ret;
+		}
+
+		ret = phy_startup(lmac->phydev);
+		debug("%s: %d\n", __FILE__, __LINE__);
+		if (ret) {
+			printf("%s: Could not initialize PHY %s\n",
+				lmac->netdev.name, lmac->phydev->dev->name);
+		}
+
+#ifdef CONFIG_THUNDERX_XCV
+		if (lmac->qlm_mode == QLM_MODE_RGMII)
+			xcv_setup_link(lmac->phydev->link, lmac->phydev->speed);
+#endif
+
+		lmac->link_up = lmac->phydev->link;
+		lmac->last_speed = lmac->phydev->speed;
+		lmac->last_duplex = lmac->phydev->duplex;
+
+		debug("bgx_poll_for_link(), qlm_mode %d phy link status 0x%x,"
+			"last speed 0x%x, duplex 0x%x\n",
+			lmac->qlm_mode, lmac->link_up, lmac->last_speed, lmac->last_duplex);
+
+		if (lmac->qlm_mode != QLM_MODE_RGMII)
+			bgx_lmac_sgmii_set_link_speed(lmac);
+
+	} else {
+		u64 status1;
+		u64 tx_ctl;
+		u64 rx_ctl;
+		status1 = bgx_reg_read(lmac->bgx, lmac->lmacid, BGX_SPUX_STATUS1);
+		tx_ctl = bgx_reg_read(lmac->bgx, lmac->lmacid, BGX_SMUX_TX_CTL);
+		rx_ctl = bgx_reg_read(lmac->bgx, lmac->lmacid, BGX_SMUX_RX_CTL);
+
+		debug("BGX%d LMAC%d BGX_SPUX_STATUS2: %lx\n",
+		      bgx_idx, lmacid,
+		      (unsigned long)bgx_reg_read(lmac->bgx, lmac->lmacid, BGX_SPUX_STATUS2));
+		debug("BGX%d LMAC%d BGX_SPUX_STATUS1: %lx\n",
+		      bgx_idx, lmacid,
+		      (unsigned long)bgx_reg_read(lmac->bgx, lmac->lmacid, BGX_SPUX_STATUS1));
+		debug("BGX%d LMAC%d BGX_SMUX_RX_CTL: %lx\n",
+		      bgx_idx, lmacid,
+		      (unsigned long)bgx_reg_read(lmac->bgx, lmac->lmacid, BGX_SMUX_RX_CTL));
+		debug("BGX%d LMAC%d BGX_SMUX_TX_CTL: %lx\n",
+		      bgx_idx, lmacid,
+		      (unsigned long)bgx_reg_read(lmac->bgx, lmac->lmacid, BGX_SMUX_TX_CTL));
+
+		if ((status1 & SPU_STATUS1_RCV_LNK) &&
+		    ((tx_ctl & SMU_TX_CTL_LNK_STATUS) == 0) &&
+		    ((rx_ctl & SMU_RX_CTL_STATUS) == 0)) {
+			lmac->link_up = 1;
+			if (lmac->lmac_type == 4)
+				lmac->last_speed = 40000;
+			else
+				lmac->last_speed = 10000;
+			lmac->last_duplex = 1;
+		} else {
+			lmac->link_up = 0;
+			lmac->last_speed = 0;
+			lmac->last_duplex = 0;
+			return bgx_xaui_check_link(lmac);
+		}
+
+		lmac->last_link = lmac->link_up;
+	}
+
+	printf("BGX%d:LMAC %u link %s\n", bgx_idx, lmacid,  (lmac->link_up) ? "up" : "down");
+
+	return lmac->link_up;
+}
+
+
+static int bgx_lmac_enable(struct bgx *bgx, int8_t lmacid)
+{
+	struct lmac *lmac;
+	u64 cfg;
+
+	lmac = &bgx->lmac[lmacid];
+	lmac->bgx = bgx;
+
+	debug("bgx_lmac_enable: lmac: %p, lmacid = %d\n", lmac, lmacid);
+
+	if ((lmac->qlm_mode == QLM_MODE_SGMII) ||
+	    (lmac->qlm_mode == QLM_MODE_RGMII) ||
+	    (lmac->qlm_mode == QLM_MODE_QSGMII)) {
+		if (bgx_lmac_sgmii_init(bgx, lmacid)) {
+			debug("bgx_lmac_sgmii_init failed\n");
+			return -1;
+		}
+		cfg = bgx_reg_read(bgx, lmacid, BGX_GMP_GMI_TXX_APPEND);
+		cfg |= ((1ull << 2) | (1ull << 1)); /* FCS and PAD */
+		bgx_reg_modify(bgx, lmacid, BGX_GMP_GMI_TXX_APPEND, cfg);
+		bgx_reg_write(bgx, lmacid, BGX_GMP_GMI_TXX_MIN_PKT, 60 - 1);
+	} else {
+		if (bgx_lmac_xaui_init(bgx, lmacid, lmac->lmac_type))
+			return -1;
+		cfg = bgx_reg_read(bgx, lmacid, BGX_SMUX_TX_APPEND);
+		cfg |= ((1ull << 2) | (1ull << 1)); /* FCS and PAD */
+		bgx_reg_modify(bgx, lmacid, BGX_SMUX_TX_APPEND, cfg);
+		bgx_reg_write(bgx, lmacid, BGX_SMUX_TX_MIN_PKT, 60 + 4);
+	}
+
+	/* Enable lmac */
+	bgx_reg_modify(bgx, lmacid, BGX_CMRX_CFG,
+		       CMR_EN | CMR_PKT_RX_EN | CMR_PKT_TX_EN);
+
+	return 0;
+}
+
+void bgx_lmac_disable(struct bgx *bgx, uint8_t lmacid)
+{
+	struct lmac *lmac;
+	u64 cmrx_cfg;
+
+	lmac = &bgx->lmac[lmacid];
+
+	cmrx_cfg = bgx_reg_read(bgx, lmacid, BGX_CMRX_CFG);
+	cmrx_cfg &= ~(1 << 15);
+	bgx_reg_write(bgx, lmacid, BGX_CMRX_CFG, cmrx_cfg);
+	bgx_flush_dmac_addrs(bgx, lmacid);
+
+	if (lmac->phydev)
+		phy_shutdown(lmac->phydev);
+
+	lmac->phydev = NULL;
+}
+
+/* Program BGXX_CMRX_CONFIG.{lmac_type,lane_to_sds} for each interface.
+ * And the number of LMACs used by this interface. Each lmac can be in
+ * programmed in a different mode, so parse each lmac one at a time. */
+static void bgx_init_hw(struct bgx *bgx)
+{
+	struct lmac *lmac;
+	int i, lmacid, count = 0, inc = 0;
+	char buf[40];
+	static int qsgmii_configured = 0;
+
+	for (lmacid = 0; lmacid < MAX_LMAC_PER_BGX; lmacid++) {
+		struct lmac *tlmac;
+
+		lmac = &bgx->lmac[lmacid];
+		/* If QLM is not programmed, skip */
+		if (lmac->qlm == -1)
+			continue;
+
+		switch (lmac->qlm_mode) {
+		case QLM_MODE_SGMII:
+		{
+			/* EBB8000 (alternative pkg) has only lane0 present on
+			   DLM0 and DLM1, skip configuring other lanes */
+			if ((bgx->bgx_id == 0) && is_altpkg) {
+				if (lmacid % 2)
+					continue;
+			}
+				lmac->lane_to_sds = lmacid;
+			lmac->lmac_type = 0;
+			snprintf(buf, sizeof(buf),
+				 "BGX%d QLM%d LMAC%d mode: SGMII\n",
+				 bgx->bgx_id, lmac->qlm, lmacid);
+			break;
+		}
+		case QLM_MODE_XAUI:
+			if (lmacid != 0)
+				continue;
+			lmac->lmac_type = 1;
+			lmac->lane_to_sds = 0xE4;
+			snprintf(buf, sizeof(buf),
+				 "BGX%d QLM%d LMAC%d mode: XAUI\n",
+				 bgx->bgx_id, lmac->qlm, lmacid);
+			break;
+		case QLM_MODE_RXAUI:
+			if (lmacid == 0) {
+				lmac->lmac_type = 2;
+				lmac->lane_to_sds = 0x4;
+			} else if (lmacid == 1) {
+				struct lmac *tlmac;
+				tlmac = &bgx->lmac[2];
+				if (tlmac->qlm_mode == QLM_MODE_RXAUI) {
+					lmac->lmac_type = 2;
+					lmac->lane_to_sds = 0xe;
+					lmac->qlm = tlmac->qlm;
+				}
+			} else
+				continue;
+			snprintf(buf, sizeof(buf),
+				 "BGX%d QLM%d LMAC%d mode: RXAUI\n",
+				 bgx->bgx_id, lmac->qlm, lmacid);
+			break;
+		case QLM_MODE_XFI:
+			/* EBB8000 (alternative pkg) has only lane0 present on
+			   DLM0 and DLM1, skip configuring other lanes */
+			if ((bgx->bgx_id == 0) && is_altpkg) {
+				if (lmacid % 2)
+					continue;
+			}
+				lmac->lane_to_sds = lmacid;
+			lmac->lmac_type = 3;
+			snprintf(buf, sizeof(buf),
+				 "BGX%d QLM%d LMAC%d mode: XFI\n",
+				 bgx->bgx_id, lmac->qlm, lmacid);
+			break;
+		case QLM_MODE_XLAUI:
+			if (lmacid != 0)
+				continue;
+			lmac->lmac_type = 4;
+			lmac->lane_to_sds = 0xE4;
+			snprintf(buf, sizeof(buf),
+				 "BGX%d QLM%d LMAC%d mode: XLAUI\n",
+				 bgx->bgx_id, lmac->qlm, lmacid);
+			break;
+		case QLM_MODE_10G_KR:
+			/* EBB8000 (alternative pkg) has only lane0 present on
+			   DLM0 and DLM1, skip configuring other lanes */
+			if ((bgx->bgx_id == 0) && is_altpkg) {
+				if (lmacid % 2)
+					continue;
+			}
+				lmac->lane_to_sds = lmacid;
+			lmac->lmac_type = 3;
+			lmac->use_training = 1;
+			snprintf(buf, sizeof(buf),
+				 "BGX%d QLM%d LMAC%d mode: 10G-KR\n",
+				 bgx->bgx_id, lmac->qlm, lmacid);
+			break;
+		case QLM_MODE_40G_KR4:
+			if (lmacid != 0)
+				continue;
+			lmac->lmac_type = 4;
+			lmac->lane_to_sds = 0xE4;
+			lmac->use_training = 1;
+			snprintf(buf, sizeof(buf),
+				 "BGX%d QLM%d LMAC%d mode: 40G-KR4\n",
+				 bgx->bgx_id, lmac->qlm, lmacid);
+			break;
+		case QLM_MODE_RGMII:
+			if (lmacid != 0)
+				continue;
+			lmac->lmac_type = 5;
+			lmac->lane_to_sds = 0xE4;
+			snprintf(buf, sizeof(buf),
+				 "BGX%d LMAC%d mode: RGMII\n",
+				 bgx->bgx_id, lmacid);
+			break;
+		case QLM_MODE_QSGMII:
+			if (qsgmii_configured)
+				continue;
+			if ((lmacid == 0) || (lmacid == 2)) {
+				count = 4;
+				printf("BGX%d QLM%d LMAC%d mode: QSGMII\n",
+					bgx->bgx_id, lmac->qlm, lmacid);
+				for (i = 0; i < count; i++) {
+					struct lmac *l;
+					l = &bgx->lmac[i];
+					l->lmac_type = 6;
+					l->qlm_mode = QLM_MODE_QSGMII;
+					l->lane_to_sds = lmacid + i;
+					if (is_bgx_port_valid(bgx->bgx_id, i))
+						bgx_reg_write(bgx, i, BGX_CMRX_CFG,
+							      (l->lmac_type << 8) |
+							      l->lane_to_sds);
+				}
+				qsgmii_configured = 1;
+			}
+			continue;
+		default:
+			continue;
+		}
+
+
+		/* Reset lmac to the unused slot */
+		if (is_bgx_port_valid(bgx->bgx_id, count) &&
+		    (lmac->qlm_mode != QLM_MODE_QSGMII)) {
+			int lmac_enable = 0;
+			tlmac = &bgx->lmac[count];
+			tlmac->lmac_type = lmac->lmac_type;
+			/* Adjust lane_to_sds based on BGX-ENABLE */
+			for (; (inc + count) < MAX_LMAC_PER_BGX; inc++) {
+				lmac_enable = bgx_board_info[bgx->bgx_id].lmac_enable[count + inc];
+				if (lmac_enable)
+					break;
+			}
+
+			if ((inc != 0) && (inc < MAX_LMAC_PER_BGX) && lmac_enable && (inc != count))
+				tlmac->lane_to_sds = lmac->lane_to_sds + abs(inc - count);
+			else
+				tlmac->lane_to_sds = lmac->lane_to_sds;
+			tlmac->qlm = lmac->qlm;
+			tlmac->qlm_mode = lmac->qlm_mode;
+
+			printf("%s", buf);
+			/* Initialize lmac_type and lane_to_sds */
+			bgx_reg_write(bgx, count, BGX_CMRX_CFG,
+				      (tlmac->lmac_type << 8) |
+				      tlmac->lane_to_sds);
+			count += 1;
+		}
+	}
+
+	printf("BGX%d LMACs: %d\n", bgx->bgx_id, count);
+	bgx->lmac_count = count;
+	bgx_reg_write(bgx, 0, BGX_CMR_RX_LMACS, count);
+	bgx_reg_write(bgx, 0, BGX_CMR_TX_LMACS, count);
+
+	bgx_reg_modify(bgx, 0, BGX_CMR_GLOBAL_CFG, CMR_GLOBAL_CFG_FCS_STRIP);
+	if (bgx_reg_read(bgx, 0, BGX_CMR_BIST_STATUS))
+		printf("BGX%d BIST failed\n", bgx->bgx_id);
+
+	/* Set the backpressure AND mask */
+	for (i = 0; i < bgx->lmac_count; i++)
+		bgx_reg_modify(bgx, 0, BGX_CMR_CHAN_MSK_AND,
+			       ((1ULL << MAX_BGX_CHANS_PER_LMAC) - 1) <<
+				(i * MAX_BGX_CHANS_PER_LMAC));
+
+	/* Disable all MAC filtering */
+	for (i = 0; i < RX_DMAC_COUNT; i++)
+		bgx_reg_write(bgx, 0, BGX_CMR_RX_DMACX_CAM + (i * 8), 0x00);
+
+	/* Disable MAC steering (NCSI traffic) */
+	for (i = 0; i < RX_TRAFFIC_STEER_RULE_COUNT; i++)
+		bgx_reg_write(bgx, 0, BGX_CMR_RX_STREERING + (i * 8), 0x00);
+}
+
+static void bgx_get_qlm_mode(struct bgx *bgx)
+{
+	struct lmac *lmac;
+	int lmacid;
+
+	/* Read LMACx type to figure out QLM mode
+	 * This is configured by low level firmware
+	 */
+	for (lmacid = 0; lmacid < MAX_LMAC_PER_BGX; lmacid++) {
+		int lmac_type;
+		int train_en;
+		int index = 0;
+
+		if (CAVIUM_IS_MODEL(CAVIUM_CN81XX)
+		    || (CAVIUM_IS_MODEL(CAVIUM_CN83XX) && (bgx->bgx_id == 2)))
+			index = (lmacid < 2) ? 0 : 2;
+
+		lmac = &bgx->lmac[lmacid];
+
+		/* check if QLM is programmed, if not, skip */
+		if (lmac->qlm == -1)
+			continue;
+
+		lmac_type = bgx_reg_read(bgx, index, BGX_CMRX_CFG);
+		lmac->lmac_type = (lmac_type >> 8) & 0x07;
+		debug("bgx_get_qlm_mode:%d:%d: lmac_type = %d, altpkg = %d\n", bgx->bgx_id,
+				lmacid, lmac->lmac_type, is_altpkg);
+
+		train_en = (readq(CSR_PA(0, GSERX_SCRATCH(lmac->qlm))) & 0xf);
+
+		switch(lmac->lmac_type) {
+		case BGX_MODE_SGMII:
+			if (bgx->is_rgx) {
+				if (lmacid == 0) {
+					lmac->qlm_mode = QLM_MODE_RGMII;
+					debug("BGX%d LMAC%d mode: RGMII\n",
+							bgx->bgx_id, lmacid);
+				}
+				continue;
+				} else {
+				if ((bgx->bgx_id == 0) && is_altpkg) {
+					if (lmacid % 2)
+						continue;
+				}
+				lmac->qlm_mode = QLM_MODE_SGMII;
+				debug("BGX%d QLM%d LMAC%d mode: SGMII\n",
+						bgx->bgx_id, lmac->qlm, lmacid);
+			}
+			break;
+		case BGX_MODE_XAUI:
+			if ((bgx->bgx_id == 0) && is_altpkg)
+				continue;
+			lmac->qlm_mode = QLM_MODE_XAUI;
+			if (lmacid != 0)
+				continue;
+			debug("BGX%d QLM%d LMAC%d mode: XAUI\n",
+					bgx->bgx_id, lmac->qlm, lmacid);
+			break;
+		case BGX_MODE_RXAUI:
+			if ((bgx->bgx_id == 0) && is_altpkg)
+				continue;
+			lmac->qlm_mode = QLM_MODE_RXAUI;
+			if (index == lmacid) {
+				debug("BGX%d QLM%d LMAC%d mode: RXAUI\n",
+					bgx->bgx_id, lmac->qlm, (index ? 1 : 0));
+			}
+			break;
+		case BGX_MODE_XFI:
+			if ((bgx->bgx_id == 0) && is_altpkg) {
+				if (lmacid % 2)
+					continue;
+			}
+			if (((lmacid < 2) && (train_en & (1 << lmacid)))
+			    || (train_en & (1 << (lmacid - 2)))) {
+				lmac->qlm_mode = QLM_MODE_10G_KR;
+				debug("BGX%d QLM%d LMAC%d mode: 10G_KR\n",
+					bgx->bgx_id, lmac->qlm, lmacid);
+			} else {
+				lmac->qlm_mode = QLM_MODE_XFI;
+				debug("BGX%d QLM%d LMAC%d mode: XFI\n",
+					bgx->bgx_id, lmac->qlm, lmacid);
+			}
+			break;
+		case BGX_MODE_XLAUI:
+			if ((bgx->bgx_id == 0) && is_altpkg)
+				continue;
+			if (train_en) {
+				lmac->qlm_mode = QLM_MODE_40G_KR4;
+				if (lmacid != 0)
+					break;
+				debug("BGX%d QLM%d LMAC%d mode: 40G_KR4\n",
+					bgx->bgx_id, lmac->qlm, lmacid);
+			} else {
+				lmac->qlm_mode = QLM_MODE_XLAUI;
+				if (lmacid != 0)
+					break;
+				debug("BGX%d QLM%d LMAC%d mode: XLAUI\n",
+					bgx->bgx_id, lmac->qlm, lmacid);
+			}
+		break;
+		case BGX_MODE_QSGMII:
+			/* If QLM is configured as QSGMII, use lmac0 */
+			if (CAVIUM_IS_MODEL(CAVIUM_CN83XX)
+			    && (lmacid == 2)
+			    && (bgx->bgx_id != 3)) {
+				//lmac->qlm_mode = QLM_MODE_DISABLED;
+				continue;
+			}
+
+			if ((lmacid == 0) || (lmacid == 2)) {
+				lmac->qlm_mode = QLM_MODE_QSGMII;
+				debug("BGX%d QLM%d LMAC%d mode: QSGMII\n",
+					bgx->bgx_id, lmac->qlm, lmacid);
+			}
+			break;
+		default:
+			break;
+		}
+	}
+}
+
+void bgx_set_board_info(int bgx_id, int *mdio_bus,
+			int *phy_addr, bool *autoneg_dis, bool *lmac_reg,
+			bool *lmac_enable)
+{
+	unsigned int i;
+
+	for (i = 0; i < MAX_LMAC_PER_BGX; i++) {
+		bgx_board_info[bgx_id].phy_info[i].phy_addr = phy_addr[i];
+		bgx_board_info[bgx_id].phy_info[i].mdio_bus = mdio_bus[i];
+		bgx_board_info[bgx_id].phy_info[i].autoneg_dis = autoneg_dis[i];
+		bgx_board_info[bgx_id].lmac_reg[i] = lmac_reg[i];
+		bgx_board_info[bgx_id].lmac_enable[i] = lmac_enable[i];
+		debug("bgx_set_board_info bgx_id %d lmac %d phy_addr 0x%x mdio bus %d\n"
+		      "autoneg_dis %d lmac_reg %d, lmac_enable = %d\n", bgx_id, i,
+			bgx_board_info[bgx_id].phy_info[i].phy_addr,
+			bgx_board_info[bgx_id].phy_info[i].mdio_bus,
+			bgx_board_info[bgx_id].phy_info[i].autoneg_dis,
+			bgx_board_info[bgx_id].lmac_reg[i],
+			bgx_board_info[bgx_id].lmac_enable[i]);
+	}
+}
+
+
+int thunderx_bgx_remove(struct udevice *dev)
+{
+	int lmacid;
+	u64 cfg;
+	int count = MAX_LMAC_PER_BGX;
+	struct bgx *bgx = dev_get_priv(dev);
+
+	if (bgx->reg_base == NULL)
+		return 0;
+
+	if (bgx->is_rgx)
+		count = 1;
+
+	for (lmacid = 0; lmacid < count; lmacid++) {
+		struct lmac *lmac;
+		lmac = &bgx->lmac[lmacid];
+		cfg = bgx_reg_read(bgx, lmacid, BGX_CMRX_CFG);
+		cfg &= ~(CMR_PKT_RX_EN | CMR_PKT_TX_EN);
+		bgx_reg_write(bgx, lmacid, BGX_CMRX_CFG, cfg);
+
+		/* Disable PCS for 1G interface */
+		if ((lmac->lmac_type == BGX_MODE_SGMII)
+		    || (lmac->lmac_type == BGX_MODE_QSGMII)) {
+			cfg = bgx_reg_read(bgx, lmacid, BGX_GMP_PCS_MRX_CTL);
+			cfg |= PCS_MRX_CTL_PWR_DN;
+			bgx_reg_write(bgx, lmacid, BGX_GMP_PCS_MRX_CTL, cfg);
+		}
+
+		debug("%s disabling bgx%d lmacid%d\n",
+			__func__, bgx->bgx_id, lmacid);
+		bgx_lmac_disable(bgx, lmacid);
+	}
+	return 0;
+}
+
+bool alternate_pkg(void)
+{
+	u64 val = readq(CAVM_MIO_FUS_DAT2);
+	int altpkg;
+
+	altpkg = (val >> 22) & 0x3;
+
+	/* Figure out alt pkg by reading chip_id or lmc_mode32 on 81xx */
+	if (CAVIUM_IS_MODEL(CAVIUM_CN81XX) && (altpkg || ((val >> 30) & 0x1)))
+		return 2;
+	return altpkg;
+}
+
+int thunderx_bgx_probe(struct udevice *dev)
+{
+	int err;
+	struct bgx *bgx = dev_get_priv(dev);
+	uint8_t lmac = 0;
+	int qlm[4] = {-1, -1, -1, -1};
+	int bgx_idx, node;
+	size_t size;
+	int inc = 1;
+
+	bgx->reg_base = dm_pci_map_bar(dev, 0, &size, PCI_REGION_MEM);
+	if (!bgx->reg_base) {
+		printf("%s No PCI region found\n", dev->name);
+		return -EINVAL;
+	}
+	is_altpkg = alternate_pkg();
+
+	// bgx_init_of_phy()
+#ifdef CONFIG_THUNDERX_XCV
+	/* Use FAKE BGX2 for RGX interface */
+	if ((((uintptr_t)bgx->reg_base >> 24) & 0xf) == 0x8) {
+		int np;
+		const char* phy_mode;
+		int phy_interface = -1;
+
+		bgx->bgx_id = 2;
+		bgx->is_rgx = true;
+		for (lmac = 0; lmac < MAX_LMAC_PER_BGX; lmac++) {
+			if (lmac == 0) {
+				bgx->lmac[lmac].lmacid = 0;
+				bgx->lmac[lmac].qlm = 0;
+			} else {
+				bgx->lmac[lmac].qlm = -1;
+			}
+		}
+		np = fdt_first_subnode(gd->fdt_blob, dev_of_offset(dev));
+		np = fdtdec_lookup_phandle(gd->fdt_blob, np, "phy-handle");
+		phy_mode = fdt_getprop(gd->fdt_blob, np, "phy-mode", NULL);
+		phy_interface = phy_get_interface_by_name(phy_mode);
+		if (phy_interface == -1)
+			phy_interface = PHY_INTERFACE_MODE_RGMII;
+		xcv_init_hw(phy_interface);
+		goto skip_qlm_config;
+	}
+#endif
+
+	node = node_id(bgx->reg_base);
+	bgx_idx = ((uintptr_t)bgx->reg_base >> 24) & 3;
+	bgx->bgx_id = (node * CONFIG_MAX_BGX_PER_NODE) + bgx_idx;
+
+	if (CAVIUM_IS_MODEL(CAVIUM_CN81XX))
+		inc = 2;
+	else if (CAVIUM_IS_MODEL(CAVIUM_CN83XX) && (bgx_idx == 2))
+		inc = 2;
+
+	for (lmac = 0; lmac < MAX_LMAC_PER_BGX; lmac += inc) {
+		/* BGX3 (DLM4), has only 2 lanes */
+		if (CAVIUM_IS_MODEL(CAVIUM_CN83XX) && (bgx_idx == 3) && lmac >= 2)
+			continue;
+		qlm[lmac + 0] = get_qlm_for_bgx(node, bgx_idx, lmac);
+		/* Each DLM has 2 lanes, configure both lanes with
+		   same qlm configuration */
+		if (inc == 2)
+			qlm[lmac + 1] = qlm[lmac];
+		debug("qlm[%d] = %d\n", lmac, qlm[lmac]);
+	}
+
+	/* A BGX can take 1 or 2 DLMs, if both the DLMs are not configured
+	   as BGX, then return, nothing to initialize */
+	if (CAVIUM_IS_MODEL(CAVIUM_CN81XX))
+		if ((qlm[0] == -1) && (qlm[2] == -1))
+			return -ENODEV;
+
+	/* MAP configuration registers */
+	for (lmac = 0; lmac < MAX_LMAC_PER_BGX; lmac++) {
+		bgx->lmac[lmac].qlm = qlm[lmac];
+		bgx->lmac[lmac].lmacid = lmac;
+	}
+
+#ifdef CONFIG_THUNDERX_XCV
+skip_qlm_config:
+#endif
+	bgx_vnic[bgx->bgx_id] = bgx;
+	bgx_get_qlm_mode(bgx);
+	debug("bgx_vnic[%u]: %p\n", bgx->bgx_id, bgx);
+
+	bgx_init_hw(bgx);
+
+	/* Enable all LMACs */
+	for (lmac = 0; lmac < bgx->lmac_count; lmac++) {
+		snprintf(bgx->lmac[lmac].netdev.name,
+			 sizeof(bgx->lmac[lmac].netdev.name),
+			 "lmac%d", lmac);
+
+		err = bgx_lmac_enable(bgx, lmac);
+		if (err) {
+			printf("BGX%d failed to enable lmac%d\n",
+				bgx->bgx_id, lmac);
+		}
+	}
+
+	return 0;
+}
+
+static const struct misc_ops thunderx_bgx_ops = {
+};
+
+static const struct udevice_id thunderx_bgx_ids[] = {
+	{ .compatible = "cavium,thunder-8890-bgx" },
+	{}
+};
+
+U_BOOT_DRIVER(thunderx_bgx) = {
+	.name	= "thunderx_bgx",
+	.id	= UCLASS_MISC,
+	.probe	= thunderx_bgx_probe,
+	.remove	= thunderx_bgx_remove,
+	.of_match = thunderx_bgx_ids,
+	.ops	= &thunderx_bgx_ops,
+	.priv_auto_alloc_size = sizeof(struct bgx),
+	.flags  = DM_FLAG_OS_PREPARE,
+};
+
+static struct pci_device_id thunderx_pci_bgx_supported[] = {
+        { PCI_VDEVICE(CAVIUM, PCI_DEVICE_ID_THUNDERX_BGX) },
+        { PCI_VDEVICE(CAVIUM, PCI_DEVICE_ID_THUNDERX_RGX) },
+        { },
+};
+
+U_BOOT_PCI_DEVICE(thunderx_bgx, thunderx_pci_bgx_supported);
+
diff --git a/drivers/net/cavium/thunder_bgx.h b/drivers/net/cavium/thunder_bgx.h
new file mode 100644
index 0000000000..ce1bb13f6d
--- /dev/null
+++ b/drivers/net/cavium/thunder_bgx.h
@@ -0,0 +1,259 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2018 Cavium Inc
+ */
+#ifndef THUNDER_BGX_H
+#define THUNDER_BGX_H
+
+/* PCI device ID */
+#define	PCI_DEVICE_ID_THUNDER_BGX		0xA026
+#define	PCI_DEVICE_ID_THUNDER_RGX		0xA054
+
+#define    MAX_LMAC_PER_BGX			4
+#define    MAX_BGX_CHANS_PER_LMAC		16
+#define    MAX_DMAC_PER_LMAC			8
+#define    MAX_FRAME_SIZE			9216
+
+#define    MAX_DMAC_PER_LMAC_TNS_BYPASS_MODE	2
+
+#define    MAX_LMAC	(CONFIG_MAX_BGX_PER_NODE * MAX_LMAC_PER_BGX)
+
+/* Registers */
+#define BGX_CMRX_CFG			0x00
+#define  CMR_PKT_TX_EN				BIT_ULL(13)
+#define  CMR_PKT_RX_EN				BIT_ULL(14)
+#define  CMR_EN					BIT_ULL(15)
+#define BGX_CMR_GLOBAL_CFG		0x08
+#define  CMR_GLOBAL_CFG_FCS_STRIP		BIT_ULL(6)
+#define BGX_CMRX_RX_ID_MAP		0x60
+#define BGX_CMRX_RX_STAT0		0x70
+#define BGX_CMRX_RX_STAT1		0x78
+#define BGX_CMRX_RX_STAT2		0x80
+#define BGX_CMRX_RX_STAT3		0x88
+#define BGX_CMRX_RX_STAT4		0x90
+#define BGX_CMRX_RX_STAT5		0x98
+#define BGX_CMRX_RX_STAT6		0xA0
+#define BGX_CMRX_RX_STAT7		0xA8
+#define BGX_CMRX_RX_STAT8		0xB0
+#define BGX_CMRX_RX_STAT9		0xB8
+#define BGX_CMRX_RX_STAT10		0xC0
+#define BGX_CMRX_RX_BP_DROP		0xC8
+#define BGX_CMRX_RX_DMAC_CTL		0x0E8
+#define BGX_CMRX_RX_FIFO_LEN		0x108
+#define BGX_CMR_RX_DMACX_CAM		0x200
+#define  RX_DMACX_CAM_EN			BIT_ULL(48)
+#define  RX_DMACX_CAM_LMACID(x)			(x << 49)
+#define  RX_DMAC_COUNT				32
+#define BGX_CMR_RX_STREERING		0x300
+#define  RX_TRAFFIC_STEER_RULE_COUNT		8
+#define BGX_CMR_CHAN_MSK_AND		0x450
+#define BGX_CMR_BIST_STATUS		0x460
+#define BGX_CMR_RX_LMACS		0x468
+#define BGX_CMRX_TX_FIFO_LEN		0x518
+#define BGX_CMRX_TX_STAT0		0x600
+#define BGX_CMRX_TX_STAT1		0x608
+#define BGX_CMRX_TX_STAT2		0x610
+#define BGX_CMRX_TX_STAT3		0x618
+#define BGX_CMRX_TX_STAT4		0x620
+#define BGX_CMRX_TX_STAT5		0x628
+#define BGX_CMRX_TX_STAT6		0x630
+#define BGX_CMRX_TX_STAT7		0x638
+#define BGX_CMRX_TX_STAT8		0x640
+#define BGX_CMRX_TX_STAT9		0x648
+#define BGX_CMRX_TX_STAT10		0x650
+#define BGX_CMRX_TX_STAT11		0x658
+#define BGX_CMRX_TX_STAT12		0x660
+#define BGX_CMRX_TX_STAT13		0x668
+#define BGX_CMRX_TX_STAT14		0x670
+#define BGX_CMRX_TX_STAT15		0x678
+#define BGX_CMRX_TX_STAT16		0x680
+#define BGX_CMRX_TX_STAT17		0x688
+#define BGX_CMR_TX_LMACS		0x1000
+
+#define BGX_SPUX_CONTROL1		0x10000
+#define  SPU_CTL_LOW_POWER			BIT_ULL(11)
+#define  SPU_CTL_LOOPBACK			BIT_ULL(14)
+#define  SPU_CTL_RESET				BIT_ULL(15)
+#define BGX_SPUX_STATUS1		0x10008
+#define  SPU_STATUS1_RCV_LNK			BIT_ULL(2)
+#define BGX_SPUX_STATUS2		0x10020
+#define  SPU_STATUS2_RCVFLT			BIT_ULL(10)
+#define BGX_SPUX_BX_STATUS		0x10028
+#define  SPU_BX_STATUS_RX_ALIGN			BIT_ULL(12)
+#define BGX_SPUX_BR_STATUS1		0x10030
+#define  SPU_BR_STATUS_BLK_LOCK			BIT_ULL(0)
+#define  SPU_BR_STATUS_RCV_LNK			BIT_ULL(12)
+#define BGX_SPUX_BR_PMD_CRTL		0x10068
+#define  SPU_PMD_CRTL_TRAIN_EN			BIT_ULL(1)
+#define BGX_SPUX_BR_PMD_LP_CUP		0x10078
+#define BGX_SPUX_BR_PMD_LD_CUP		0x10088
+#define BGX_SPUX_BR_PMD_LD_REP		0x10090
+#define BGX_SPUX_FEC_CONTROL		0x100A0
+#define  SPU_FEC_CTL_FEC_EN			BIT_ULL(0)
+#define  SPU_FEC_CTL_ERR_EN			BIT_ULL(1)
+#define BGX_SPUX_AN_CONTROL		0x100C8
+#define  SPU_AN_CTL_AN_EN			BIT_ULL(12)
+#define  SPU_AN_CTL_XNP_EN			BIT_ULL(13)
+#define  SPU_AN_CTL_AN_RESTART			BIT_ULL(15)
+#define BGX_SPUX_AN_STATUS		0x100D0
+#define  SPU_AN_STS_AN_COMPLETE			BIT_ULL(5)
+#define BGX_SPUX_AN_ADV			0x100D8
+#define BGX_SPUX_MISC_CONTROL		0x10218
+#define  SPU_MISC_CTL_INTLV_RDISP		BIT_ULL(10)
+#define  SPU_MISC_CTL_RX_DIS			BIT_ULL(12)
+#define BGX_SPUX_INT			0x10220	/* +(0..3) << 20 */
+#define BGX_SPUX_INT_W1S		0x10228
+#define BGX_SPUX_INT_ENA_W1C		0x10230
+#define BGX_SPUX_INT_ENA_W1S		0x10238
+#define BGX_SPU_DBG_CONTROL		0x10300
+#define  SPU_DBG_CTL_AN_ARB_LINK_CHK_EN		BIT_ULL(18)
+#define  SPU_DBG_CTL_AN_NONCE_MCT_DIS		BIT_ULL(29)
+
+#define BGX_SMUX_RX_INT			0x20000
+#define BGX_SMUX_RX_JABBER		0x20030
+#define BGX_SMUX_RX_CTL			0x20048
+#define  SMU_RX_CTL_STATUS			(3ull << 0)
+#define BGX_SMUX_TX_APPEND		0x20100
+#define  SMU_TX_APPEND_FCS_D			BIT_ULL(2)
+#define BGX_SMUX_TX_PAUSE_PKT_TIME	0x20110
+#define BGX_SMUX_TX_MIN_PKT		0x20118
+#define BGX_SMUX_TX_PAUSE_PKT_INTERVAL	0x20120
+#define BGX_SMUX_TX_PAUSE_ZERO		0x20138
+#define BGX_SMUX_TX_INT			0x20140
+#define BGX_SMUX_TX_CTL			0x20178
+#define  SMU_TX_CTL_DIC_EN			BIT_ULL(0)
+#define  SMU_TX_CTL_UNI_EN			BIT_ULL(1)
+#define  SMU_TX_CTL_LNK_STATUS			(3ull << 4)
+#define BGX_SMUX_TX_THRESH		0x20180
+#define BGX_SMUX_CTL			0x20200
+#define  SMU_CTL_RX_IDLE			BIT_ULL(0)
+#define  SMU_CTL_TX_IDLE			BIT_ULL(1)
+#define	BGX_SMUX_CBFC_CTL		0x20218
+#define	RX_EN					BIT_ULL(0)
+#define	TX_EN					BIT_ULL(1)
+#define	BCK_EN					BIT_ULL(2)
+#define	DRP_EN					BIT_ULL(3)
+
+#define BGX_GMP_PCS_MRX_CTL		0x30000
+#define	 PCS_MRX_CTL_RST_AN			BIT_ULL(9)
+#define	 PCS_MRX_CTL_PWR_DN			BIT_ULL(11)
+#define	 PCS_MRX_CTL_AN_EN			BIT_ULL(12)
+#define	 PCS_MRX_CTL_LOOPBACK1			BIT_ULL(14)
+#define	 PCS_MRX_CTL_RESET			BIT_ULL(15)
+#define BGX_GMP_PCS_MRX_STATUS		0x30008
+#define	 PCS_MRX_STATUS_LINK			BIT_ULL(2)
+#define	 PCS_MRX_STATUS_AN_CPT			BIT_ULL(5)
+#define BGX_GMP_PCS_ANX_ADV		0x30010
+#define BGX_GMP_PCS_ANX_AN_RESULTS	0x30020
+#define BGX_GMP_PCS_LINKX_TIMER		0x30040
+#define PCS_LINKX_TIMER_COUNT			0x1E84
+#define BGX_GMP_PCS_SGM_AN_ADV		0x30068
+#define BGX_GMP_PCS_MISCX_CTL		0x30078
+#define  PCS_MISC_CTL_MODE			BIT_ULL(8)
+#define  PCS_MISC_CTL_DISP_EN			BIT_ULL(13)
+#define  PCS_MISC_CTL_GMX_ENO			BIT_ULL(11)
+#define  PCS_MISC_CTL_SAMP_PT_MASK	0x7Full
+#define BGX_GMP_GMI_PRTX_CFG		0x38020
+#define  GMI_PORT_CFG_SPEED			BIT_ULL(1)
+#define  GMI_PORT_CFG_DUPLEX			BIT_ULL(2)
+#define  GMI_PORT_CFG_SLOT_TIME			BIT_ULL(3)
+#define  GMI_PORT_CFG_SPEED_MSB			BIT_ULL(8)
+#define  GMI_PORT_CFG_RX_IDLE			BIT_ULL(12)
+#define  GMI_PORT_CFG_TX_IDLE			BIT_ULL(13)
+#define BGX_GMP_GMI_RXX_JABBER		0x38038
+#define BGX_GMP_GMI_TXX_THRESH		0x38210
+#define BGX_GMP_GMI_TXX_APPEND		0x38218
+#define BGX_GMP_GMI_TXX_SLOT		0x38220
+#define BGX_GMP_GMI_TXX_BURST		0x38228
+#define BGX_GMP_GMI_TXX_MIN_PKT		0x38240
+#define BGX_GMP_GMI_TXX_SGMII_CTL	0x38300
+#define BGX_GMP_GMI_TXX_INT		0x38500
+#define BGX_GMP_GMI_TXX_INT_W1S		0x38508
+#define BGX_GMP_GMI_TXX_INT_ENA_W1C	0x38510
+#define BGX_GMP_GMI_TXX_INT_ENA_W1S	0x38518
+#define  GMI_TXX_INT_PTP_LOST			BIT_ULL(4)
+#define  GMI_TXX_INT_LATE_COL			BIT_ULL(3)
+#define  GMI_TXX_INT_XSDEF			BIT_ULL(2)
+#define  GMI_TXX_INT_XSCOL			BIT_ULL(1)
+#define  GMI_TXX_INT_UNDFLW			BIT_ULL(0)
+
+#define BGX_MSIX_VEC_0_29_ADDR		0x400000 /* +(0..29) << 4 */
+#define BGX_MSIX_VEC_0_29_CTL		0x400008
+#define BGX_MSIX_PBA_0			0x4F0000
+
+/* MSI-X interrupts */
+#define BGX_MSIX_VECTORS	30
+#define BGX_LMAC_VEC_OFFSET	7
+#define BGX_MSIX_VEC_SHIFT	4
+
+#define CMRX_INT		0
+#define SPUX_INT		1
+#define SMUX_RX_INT		2
+#define SMUX_TX_INT		3
+#define GMPX_PCS_INT		4
+#define GMPX_GMI_RX_INT		5
+#define GMPX_GMI_TX_INT		6
+#define CMR_MEM_INT		28
+#define SPU_MEM_INT		29
+
+#define LMAC_INTR_LINK_UP	BIT(0)
+#define LMAC_INTR_LINK_DOWN	BIT(1)
+
+/*  RX_DMAC_CTL configuration*/
+enum MCAST_MODE {
+		MCAST_MODE_REJECT,
+		MCAST_MODE_ACCEPT,
+		MCAST_MODE_CAM_FILTER,
+		RSVD
+};
+
+#define BCAST_ACCEPT	1
+#define CAM_ACCEPT	1
+
+int thunderx_bgx_initialize(unsigned int bgx_idx, unsigned int node);
+void bgx_add_dmac_addr(uint64_t dmac, int node, int bgx_idx, int lmac);
+void bgx_get_count(int node, int *bgx_count);
+int bgx_get_lmac_count(int node, int bgx);
+void bgx_print_stats(int bgx_idx, int lmac);
+void xcv_init_hw(int phy_mode);
+void xcv_setup_link(bool link_up, int link_speed);
+
+enum qlm_mode {
+	QLM_MODE_SGMII,         /* SGMII, each lane independent */
+	QLM_MODE_XAUI,      /* 1 XAUI or DXAUI, 4 lanes */
+	QLM_MODE_RXAUI,     /* 2 RXAUI, 2 lanes each */
+	QLM_MODE_XFI,       /* 4 XFI, 1 lane each */
+	QLM_MODE_XLAUI,     /* 1 XLAUI, 4 lanes each */
+	QLM_MODE_10G_KR,    /* 4 10GBASE-KR, 1 lane each */
+	QLM_MODE_40G_KR4,   /* 1 40GBASE-KR4, 4 lanes each */
+	QLM_MODE_QSGMII,    /* 4 QSGMII, each lane independent */
+	QLM_MODE_RGMII,     /* 1 RGX */
+};
+
+struct phy_info {
+	int mdio_bus;
+	int phy_addr;
+	bool autoneg_dis;
+};
+
+struct bgx_board_info {
+	struct phy_info phy_info[MAX_LMAC_PER_BGX];
+	bool lmac_reg[MAX_LMAC_PER_BGX];
+	bool lmac_enable[MAX_LMAC_PER_BGX];
+};
+
+enum LMAC_TYPE {
+	BGX_MODE_SGMII = 0, /* 1 lane, 1.250 Gbaud */
+	BGX_MODE_XAUI = 1,  /* 4 lanes, 3.125 Gbaud */
+	BGX_MODE_DXAUI = 1, /* 4 lanes, 6.250 Gbaud */
+	BGX_MODE_RXAUI = 2, /* 2 lanes, 6.250 Gbaud */
+	BGX_MODE_XFI = 3,   /* 1 lane, 10.3125 Gbaud */
+	BGX_MODE_XLAUI = 4, /* 4 lanes, 10.3125 Gbaud */
+	BGX_MODE_10G_KR = 3,/* 1 lane, 10.3125 Gbaud */
+	BGX_MODE_40G_KR = 4,/* 4 lanes, 10.3125 Gbaud */
+	BGX_MODE_RGMII = 5,
+	BGX_MODE_QSGMII = 6,
+	BGX_MODE_INVALID = 7,
+};
+
+#endif /* THUNDER_BGX_H */
diff --git a/drivers/net/cavium/thunder_xcv.c b/drivers/net/cavium/thunder_xcv.c
new file mode 100644
index 0000000000..b4342e16b6
--- /dev/null
+++ b/drivers/net/cavium/thunder_xcv.c
@@ -0,0 +1,190 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2018 Cavium Inc
+ */
+#include <config.h>
+#include <common.h>
+#include <net.h>
+#include <dm.h>
+#include <pci.h>
+#include <misc.h>
+#include <netdev.h>
+#include <malloc.h>
+#include <miiphy.h>
+#include <asm/io.h>
+#include <errno.h>
+
+#include <asm/arch-thunderx/thunderx.h>
+#include <asm/arch-thunderx/thunderx_vnic.h>
+
+/* Register offsets */
+#define XCV_RESET		0x00
+#define   PORT_EN		BIT_ULL(63)
+#define   CLK_RESET		BIT_ULL(15)
+#define   DLL_RESET		BIT_ULL(11)
+#define   COMP_EN		BIT_ULL(7)
+#define   TX_PKT_RESET		BIT_ULL(3)
+#define   TX_DATA_RESET		BIT_ULL(2)
+#define   RX_PKT_RESET		BIT_ULL(1)
+#define   RX_DATA_RESET		BIT_ULL(0)
+#define XCV_DLL_CTL		0x10
+#define   CLKRX_BYP		BIT_ULL(23)
+#define   CLKTX_BYP		BIT_ULL(15)
+#define XCV_COMP_CTL		0x20
+#define   DRV_BYP		BIT_ULL(63)
+#define XCV_CTL			0x30
+#define XCV_INT			0x40
+#define XCV_INT_W1S		0x48
+#define XCV_INT_ENA_W1C		0x50
+#define XCV_INT_ENA_W1S		0x58
+#define XCV_INBND_STATUS	0x80
+#define XCV_BATCH_CRD_RET	0x100
+
+struct xcv {
+	void __iomem		*reg_base;
+	struct pci_dev		*pdev;
+};
+
+static struct xcv *xcv;
+
+void xcv_init_hw(int phy_mode)
+{
+	u64  cfg;
+
+	/* Take DLL out of reset */
+	cfg = readq(CAVM_XCVX_RESET);
+	cfg &= ~DLL_RESET;
+	writeq(cfg, CAVM_XCVX_RESET + XCV_RESET);
+
+	/* Take clock tree out of reset */
+	cfg = readq(CAVM_XCVX_RESET + XCV_RESET);
+	cfg &= ~CLK_RESET;
+	writeq(cfg, CAVM_XCVX_RESET + XCV_RESET);
+	/* Wait for DLL to lock */
+	udelay(1000);
+
+	/* enable/bypass DLL providing MAC based internal TX/RX delays */
+	cfg = readq(CAVM_XCVX_RESET + XCV_DLL_CTL);
+	cfg &= ~0xffff00;
+	switch (phy_mode) {
+	/* RX and TX delays are added by the MAC */
+	case PHY_INTERFACE_MODE_RGMII:
+		break;
+	/* internal RX and TX delays provided by the PHY */
+	case PHY_INTERFACE_MODE_RGMII_ID:
+		cfg |= CLKRX_BYP;
+		cfg |= CLKTX_BYP;
+		break;
+	/* internal RX delay provided by the PHY, the MAC
+	 * should not add an RX delay in this case
+	 */
+	case PHY_INTERFACE_MODE_RGMII_RXID:
+		cfg |= CLKRX_BYP;
+		break;
+	/* internal TX delay provided by the PHY, the MAC
+	 * should not add an TX delay in this case
+	 */
+	case PHY_INTERFACE_MODE_RGMII_TXID:
+		cfg |= CLKRX_BYP;
+		break;
+	}
+	writeq(cfg, CAVM_XCVX_RESET + XCV_DLL_CTL);
+
+	/* Enable compensation controller and force the
+	 * write to be visible to HW by readig back.
+	 */
+	cfg = readq(CAVM_XCVX_RESET + XCV_RESET);
+	cfg |= COMP_EN;
+	writeq(cfg, CAVM_XCVX_RESET + XCV_RESET);
+	readq(CAVM_XCVX_RESET + XCV_RESET);
+	/* Wait for compensation state machine to lock */
+	udelay(10000);
+
+	/* enable the XCV block */
+	cfg = readq(CAVM_XCVX_RESET + XCV_RESET);
+	cfg |= PORT_EN;
+	writeq(cfg, CAVM_XCVX_RESET + XCV_RESET);
+
+	cfg = readq(CAVM_XCVX_RESET + XCV_RESET);
+	cfg |= CLK_RESET;
+	writeq(cfg, CAVM_XCVX_RESET + XCV_RESET);
+}
+
+/* 
+ * Configure XCV link based on the speed
+ * link_up   : Set to 1 when link is up otherwise 0
+ * link_speed: The speed of the link.
+ */ 
+void xcv_setup_link(bool link_up, int link_speed)
+{
+	u64  cfg;
+	int speed = 2;
+
+	if (link_speed == 100)
+		speed = 1;
+	else if (link_speed == 10)
+		speed = 0;
+
+	if (link_up) {
+		/* set operating speed */
+		cfg = readq(CAVM_XCVX_RESET + XCV_CTL);
+		cfg &= ~0x03;
+		cfg |= speed;
+		writeq(cfg, CAVM_XCVX_RESET + XCV_CTL);
+
+		/* Reset datapaths */
+		cfg = readq(CAVM_XCVX_RESET + XCV_RESET);
+		cfg |= TX_DATA_RESET | RX_DATA_RESET;
+		writeq(cfg, CAVM_XCVX_RESET + XCV_RESET);
+
+		/* Enable the packet flow */
+		cfg = readq(CAVM_XCVX_RESET + XCV_RESET);
+		cfg |= TX_PKT_RESET | RX_PKT_RESET;
+		writeq(cfg, CAVM_XCVX_RESET + XCV_RESET);
+
+		/* Return credits to RGX */
+		writeq(0x01, CAVM_XCVX_RESET + XCV_BATCH_CRD_RET);
+	} else {
+		/* Disable packet flow */
+		cfg = readq(CAVM_XCVX_RESET + XCV_RESET);
+		cfg &= ~(TX_PKT_RESET | RX_PKT_RESET);
+		writeq(cfg, CAVM_XCVX_RESET + XCV_RESET);
+		readq(CAVM_XCVX_RESET + XCV_RESET);
+	}
+}
+
+int thunderx_xcv_probe(struct udevice *dev)
+{
+	size_t size;
+
+	xcv = dev_get_priv(dev);
+	if (!xcv)
+		return -ENOMEM;
+	xcv->reg_base = dm_pci_map_bar(dev, 0, &size, PCI_REGION_MEM);
+
+	return 0;
+}
+
+static const struct misc_ops thunderx_xcv_ops = {
+};
+
+static const struct udevice_id thunderx_xcv_ids[] = {
+	{ .compatible = "cavium,xcv" },
+	{}
+};
+
+U_BOOT_DRIVER(thunderx_xcv) = {
+        .name   = "thunderx_xcv",
+        .id     = UCLASS_MISC,
+        .probe  = thunderx_xcv_probe,
+        .of_match = thunderx_xcv_ids,
+        .ops    = &thunderx_xcv_ops,
+        .priv_auto_alloc_size = sizeof(struct xcv),
+};
+
+static struct pci_device_id thunderx_pci_xcv_supported[] = {
+	{ PCI_VDEVICE(CAVIUM, PCI_DEVICE_ID_THUNDERX_NIC_XCV) },
+	{ },
+};
+
+U_BOOT_PCI_DEVICE(thunderx_xcv, thunderx_pci_xcv_supported);
diff --git a/drivers/net/cavium/thunderx_smi.c b/drivers/net/cavium/thunderx_smi.c
new file mode 100644
index 0000000000..1d0dd3c0dd
--- /dev/null
+++ b/drivers/net/cavium/thunderx_smi.c
@@ -0,0 +1,388 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2018 Cavium Inc
+ */
+#include <common.h>
+#include <dm.h>
+#include <pci.h>
+#include <phy.h>
+#include <miiphy.h>
+#include <misc.h>
+#include <malloc.h>
+#include <asm/io.h>
+#include <environment.h>
+#include <linux/ctype.h>
+
+#define PCI_DEVICE_ID_THUNDERX_SMI	0xa02b
+
+DECLARE_GLOBAL_DATA_PTR;
+
+enum thunderx_smi_mode {
+	CLAUSE22 = 0,
+	CLAUSE45 = 1,
+};
+
+enum {
+	SMI_OP_C22_WRITE = 0,
+	SMI_OP_C22_READ = 1,
+
+	SMI_OP_C45_ADDR = 0,
+	SMI_OP_C45_WRITE = 1,
+	SMI_OP_C45_PRIA = 2,
+	SMI_OP_C45_READ = 3,
+};
+
+union smi_x_clk {
+	u64 u;
+	struct smi_x_clk_s {
+		int phase:8;
+		int sample:4;
+		int preamble:1;
+		int clk_idle:1;
+		int reserved_14_14:1;
+		int sample_mode:1;
+		int sample_hi:5;
+		int reserved_21_23:3;
+		int mode:1;
+	} s;
+};
+
+union smi_x_cmd {
+	u64 u;
+	struct smi_x_cmd_s {
+		int reg_adr:5;
+		int reserved_5_7:3;
+		int phy_adr:5;
+		int reserved_13_15:3;
+		int phy_op:2;
+	} s;
+};
+
+union smi_x_wr_dat {
+	u64 u;
+	struct smi_x_wr_dat_s {
+		int dat:16;
+		int val:1;
+		int pending:1;
+	} s;
+};
+
+union smi_x_rd_dat {
+	u64 u;
+	struct smi_x_rd_dat_s {
+		int dat:16;
+		int val:1;
+		int pending:1;
+	} s;
+};
+
+union smi_x_en {
+	u64 u;
+	struct smi_x_en_s {
+		int en:1;
+	} s;
+};
+
+#define SMI_X_RD_DAT	0x10ull
+#define SMI_X_WR_DAT	0x08ull
+#define SMI_X_CMD	0x00ull
+#define SMI_X_CLK	0x18ull
+#define SMI_X_EN	0x20ull
+
+struct thunderx_smi_priv {
+	void __iomem *baseaddr;
+	enum thunderx_smi_mode mode;
+};
+
+#define MDIO_TIMEOUT 10000
+
+void thunderx_smi_setmode(struct mii_dev *bus, enum thunderx_smi_mode mode)
+{
+	struct thunderx_smi_priv *priv = bus->priv;
+	union smi_x_clk smix_clk;
+
+	smix_clk.u = readq(priv->baseaddr + SMI_X_CLK);
+	smix_clk.s.mode = mode;
+	smix_clk.s.preamble = mode == CLAUSE45;
+	writeq(smix_clk.u, priv->baseaddr + SMI_X_CLK);
+
+	priv->mode = mode;
+}
+
+int thunderx_c45_addr(struct mii_dev *bus, int addr, int devad, int regnum)
+{
+	struct thunderx_smi_priv *priv = bus->priv;
+
+	union smi_x_cmd smix_cmd;
+	union smi_x_wr_dat smix_wr_dat;
+	unsigned long timeout = MDIO_TIMEOUT;
+
+	smix_wr_dat.u = 0;
+	smix_wr_dat.s.dat = regnum;
+
+	writeq(smix_wr_dat.u, priv->baseaddr + SMI_X_WR_DAT);
+
+	smix_cmd.u = 0;
+	smix_cmd.s.phy_op = SMI_OP_C45_ADDR;
+	smix_cmd.s.phy_adr = addr;
+	smix_cmd.s.reg_adr = devad;
+
+	writeq(smix_cmd.u, priv->baseaddr + SMI_X_CMD);
+
+	do {
+		smix_wr_dat.u = readq(priv->baseaddr + SMI_X_WR_DAT);
+		udelay(100);
+		timeout--;
+	} while (smix_wr_dat.s.pending && timeout);
+
+	return timeout == 0;
+}
+
+int thunderx_phy_read(struct mii_dev *bus, int addr, int devad, int regnum)
+{
+	struct thunderx_smi_priv *priv = bus->priv;
+	union smi_x_cmd smix_cmd;
+	union smi_x_rd_dat smix_rd_dat;
+	unsigned long timeout = MDIO_TIMEOUT;
+	int ret;
+
+	enum thunderx_smi_mode mode = (devad < 0) ? CLAUSE22 : CLAUSE45;
+
+	debug("RD: Mode: %u, baseaddr: %p, addr: %d, devad: %d, reg: %d\n",
+	      mode, priv->baseaddr, addr, devad, regnum);
+
+	thunderx_smi_setmode(bus, mode);
+
+	if (mode == CLAUSE45) {
+		ret = thunderx_c45_addr(bus, addr, devad, regnum);
+
+		debug("RD: ret: %u\n", ret);
+
+		if (ret)
+			return 0;
+	}
+
+	smix_cmd.u = 0;
+	smix_cmd.s.phy_adr = addr;
+
+
+	if (mode == CLAUSE45) {
+		smix_cmd.s.reg_adr = devad;
+		smix_cmd.s.phy_op = SMI_OP_C45_READ;
+	} else {
+		smix_cmd.s.reg_adr = regnum;
+		smix_cmd.s.phy_op = SMI_OP_C22_READ;
+	}
+
+	writeq(smix_cmd.u, priv->baseaddr + SMI_X_CMD);
+
+	do {
+		smix_rd_dat.u = readq(priv->baseaddr + SMI_X_RD_DAT);
+		udelay(10);
+		timeout--;
+	} while (smix_rd_dat.s.pending && timeout);
+
+	debug("SMIX_RD_DAT: %lx\n", (unsigned long)smix_rd_dat.u);
+
+	if (smix_rd_dat.s.val)
+		return smix_rd_dat.s.dat & 0xffff;
+	return -1;
+}
+
+int thunderx_phy_write(struct mii_dev *bus, int addr, int devad, int regnum,
+		       u16 value)
+{
+	struct thunderx_smi_priv *priv = bus->priv;
+	union smi_x_cmd smix_cmd;
+	union smi_x_wr_dat smix_wr_dat;
+	unsigned long timeout = MDIO_TIMEOUT;
+	int ret;
+
+	enum thunderx_smi_mode mode = (devad < 0) ? CLAUSE22 : CLAUSE45;
+
+	debug("WR: Mode: %u, baseaddr: %p, addr: %d, devad: %d, reg: %d\n",
+	      mode, priv->baseaddr, addr, devad, regnum);
+
+	if (mode == CLAUSE45) {
+		ret = thunderx_c45_addr(bus, addr, devad, regnum);
+
+		debug("WR: ret: %u\n", ret);
+
+		if (ret)
+			return ret;
+	}
+
+	smix_wr_dat.u = 0;
+	smix_wr_dat.s.dat = value;
+
+	writeq(smix_wr_dat.u, priv->baseaddr + SMI_X_WR_DAT);
+
+	smix_cmd.u = 0;
+	smix_cmd.s.phy_adr = addr;
+
+	if (mode == CLAUSE45) {
+		smix_cmd.s.reg_adr = devad;
+		smix_cmd.s.phy_op = SMI_OP_C45_WRITE;
+	} else {
+		smix_cmd.s.reg_adr = regnum;
+		smix_cmd.s.phy_op = SMI_OP_C22_WRITE;
+	}
+
+	writeq(smix_cmd.u, priv->baseaddr + SMI_X_CMD);
+
+	do {
+		smix_wr_dat.u = readq(priv->baseaddr + SMI_X_WR_DAT);
+		udelay(10);
+		timeout--;
+	} while (smix_wr_dat.s.pending && timeout);
+
+	debug("SMIX_WR_DAT: %lx\n", (unsigned long)smix_wr_dat.u);
+
+	return timeout == 0;
+}
+
+int thunderx_smi_reset(struct mii_dev *bus)
+{
+	struct thunderx_smi_priv *priv = bus->priv;
+
+	union smi_x_en smi_en;
+
+	smi_en.s.en = 0;
+	writeq(smi_en.u, priv->baseaddr + SMI_X_EN);
+
+	smi_en.s.en = 1;
+	writeq(smi_en.u, priv->baseaddr + SMI_X_EN);
+
+	thunderx_smi_setmode(bus, CLAUSE22);
+
+	return 0;
+}
+
+/* PHY XS initialization, primarily for RXAUI
+ *
+ */
+int __cavm_if_phy_xs_init(struct mii_dev *bus, int phy_addr)
+{
+	int reg;
+	ulong start_time;
+	int phy_id1, phy_id2;
+	int oui, model_number;
+
+	phy_id1 = thunderx_phy_read(bus, phy_addr, 1, 0x2);
+	phy_id2 = thunderx_phy_read(bus, phy_addr, 1, 0x3);
+	model_number = (phy_id2 >> 4) & 0x3F;
+	debug("%s model %x\n", __func__,model_number);
+	oui = phy_id1;
+	oui <<= 6;
+	oui |= (phy_id2 >> 10) & 0x3F;
+	debug("%s oui %x\n", __func__,oui);
+	switch (oui)
+	{
+		case 0x5016:
+			if (model_number == 9)
+			{
+				debug("%s +\n", __func__);
+				/* Perform hardware reset in XGXS control */
+				reg = thunderx_phy_read(bus, phy_addr, 4, 0x0);
+				if ((reg & 0xffff) < 0)
+				       goto read_error;
+				reg |= 0x8000;
+				thunderx_phy_write(bus, phy_addr, 4, 0x0, reg);
+
+				start_time = get_timer(0);
+				do {
+				       reg = thunderx_phy_read(bus, phy_addr, 4, 0x0);
+				       if ((reg & 0xffff) < 0)
+					       goto read_error;
+				} while ((reg & 0x8000) && get_timer(start_time) < 500);
+				if (reg & 0x8000) {
+				       printf("Hardware reset for M88X3120 PHY failed, MII_BMCR: 0x%x\n", reg);
+				       return -1;
+				}
+				/* program 4.49155 with 0x5 */
+				thunderx_phy_write(bus, phy_addr, 4, 0xc003, 0x5);
+			}
+			break;
+			default:
+				break;
+	}
+
+	return 0;
+
+read_error:
+	debug("M88X3120 PHY config read failed\n");
+	return -1;
+}
+
+int thunderx_smi_probe(struct udevice *dev)
+{
+	size_t size;
+	int ret, subnode, cnt = 0, node = dev->node.of_offset;
+	struct mii_dev *bus;
+	struct thunderx_smi_priv *priv;
+	pci_dev_t bdf = dm_pci_get_bdf(dev);
+
+        debug("SMI PCI device: %x\n", bdf);
+        dev->req_seq = PCI_FUNC(bdf);
+        if (!dm_pci_map_bar(dev, 0, &size, PCI_REGION_MEM)) {
+		printf("%s: Failed to map PCI region for bdf %x\n", __func__,
+			bdf);
+		return -1;
+	}
+
+	fdt_for_each_subnode(subnode, gd->fdt_blob, node) {
+		ret = fdt_node_check_compatible(gd->fdt_blob, subnode,
+						"cavium,thunder-8890-mdio");
+		if (ret)
+			continue;
+
+		bus = mdio_alloc();
+		priv = malloc(sizeof(*priv));
+		if (!bus || !priv) {
+			printf("Failed to allocate ThunderX MDIO bus # %u\n",
+				dev->seq);
+			return -1;
+		}
+
+		bus->read = thunderx_phy_read;
+		bus->write = thunderx_phy_write;
+		bus->reset = thunderx_smi_reset;
+		bus->priv = priv;
+
+		priv->mode = CLAUSE22;
+		priv->baseaddr = (void __iomem *)fdtdec_get_addr(gd->fdt_blob,
+							 subnode, "reg");
+		debug("mdio base addr %p\n", priv->baseaddr);
+
+		/* use given name or generate its own unique name */
+		snprintf(bus->name, MDIO_NAME_LEN, "smi%d", cnt++);
+
+		ret = mdio_register(bus);
+		if (ret)
+			return ret;
+	}
+	return 0;
+}
+
+static const struct misc_ops thunderx_smi_ops = {
+};
+
+static const struct udevice_id thunderx_smi_ids[] = {
+	{ .compatible = "cavium,thunder-8890-mdio-nexus" },
+	{}
+};
+
+U_BOOT_DRIVER(thunderx_smi) = {
+	.name	= "thunderx_smi",
+	.id	= UCLASS_MISC,
+	.probe	= thunderx_smi_probe,
+	.of_match = thunderx_smi_ids,
+	.ops	= &thunderx_smi_ops,
+};
+
+static struct pci_device_id thunderx_pci_smi_supported[] = {
+	{ PCI_VDEVICE(CAVIUM, PCI_DEVICE_ID_THUNDERX_SMI) },
+	{ },
+};
+
+U_BOOT_PCI_DEVICE(thunderx_smi, thunderx_pci_smi_supported);
diff --git a/include/configs/thunderx_81xx.h b/include/configs/thunderx_81xx.h
index c32aa7844c..30c2dc404e 100644
--- a/include/configs/thunderx_81xx.h
+++ b/include/configs/thunderx_81xx.h
@@ -29,7 +29,7 @@
 #define CONFIG_SYS_MEMTEST_END		(MEM_BASE + PHYS_SDRAM_1_SIZE)
/* Size of malloc() pool */
-#define CONFIG_SYS_MALLOC_LEN		(CONFIG_ENV_SIZE + 1 * SZ_1M)
+#define CONFIG_SYS_MALLOC_LEN		(CONFIG_ENV_SIZE + 32 * SZ_1M)
/* Generic Interrupt Controller Definitions */
 #define GICD_BASE			(0x801000000000)
@@ -57,7 +57,9 @@
#define BOOT_TARGET_DEVICES(func) \
    func(USB, usb, 0) \
-	func(SCSI, scsi, 0)
+	func(SCSI, scsi, 0) \
+	func(PXE, pxe, na) \
+	func(DHCP, dhcp, na)
#include <config_distro_bootcmd.h>
@@ -72,4 +74,9 @@
#define CONFIG_SYS_I2C_SPEED		100000
+#ifdef CONFIG_THUNDERX_VNIC
+/* enable board_late_init() to set env such as mac addrs */
+#define CONFIG_BOARD_LATE_INIT
+#endif
+
 #endif /* __THUNDERX_81XX_H__ */
diff --git a/net/eth_legacy.c b/net/eth_legacy.c
index 2a9caa3509..a556dc4224 100644
--- a/net/eth_legacy.c
+++ b/net/eth_legacy.c
@@ -238,9 +238,12 @@ int eth_initialize(void)
 {
    int num_devices = 0;
+#if !defined(CONFIG_BOARD_EARLY_INIT_R)
    eth_devices = NULL;
    eth_current = NULL;
    eth_common_init();
+#endif
+
    /*
     * If board-specific initialization exists, call it.
     * If not, call a CPU-specific one
-- 
2.17.1


    