net, qe: add DM support for QE UEC ethernet

add DM/DTS support for the UEC ethernet on QUICC Engine
Block.

Signed-off-by: Heiko Schocher <hs@denx.de>
Patch-cc: Mario Six <mario.six@gdsys.cc>
Patch-cc: Qiang Zhao <qiang.zhao@nxp.com>
Patch-cc: Holger Brunck <holger.brunck@hitachi-powergrids.com>
Patch-cc: Madalin Bucur <madalin.bucur@oss.nxp.com>

Series-changes: 3
- revert:
  commit "3374264df97b" ("drivers: net: qe: deselect QE when DM_ETH is enabled")
  as now qe works with DM and DM_ETH support.
- fix mailaddress from Holger

Series-changes: 2
- add comments from Qiang Zhao:
  - add device node documentation
  - I did not drop the dm_qe_uec_phy.c and use drivers/net/fsl_mdio.c
    because using drivers/net/fsl_mdio.c leads in none existent
    udevice mdio@3320
    instead boards with DM ETH support should use now this
    driver.
- remove RFC tag

Commit-notes:

- I let the old none DM based implementation in code
  so boards should work with old implementation.
  This Code should be removed if all boards are converted
  to DM/DTS.

- add the DM based qe uec driver under drivers/net/qe

- Therefore copied the files uccf.c uccf.h uec.h from
  drivers/qe. So there are a lot of Codingstyle problems
  currently. I fix them in next version if this RFC
  patch is OK or it needs some changes.

- The dm based driver code is now under drivers/net/qe/dm_qe_uec.c
  Used a lot of functions from drivers/qe/uec.c

- seperated the PHY specific code into seperate file
  drivers/net/qe/dm_qe_uec_phy.c

END
diff --git a/drivers/net/qe/dm_qe_uec.c b/drivers/net/qe/dm_qe_uec.c
new file mode 100644
index 0000000..3482b3f
--- /dev/null
+++ b/drivers/net/qe/dm_qe_uec.c
@@ -0,0 +1,1167 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * QE UEC ethernet controller driver
+ *
+ * based on drivers/qe/uec.c from NXP
+ *
+ * Copyright (C) 2020 Heiko Schocher <hs@denx.de>
+ */
+
+#include <common.h>
+#include <dm.h>
+#include <errno.h>
+#include <memalign.h>
+#include <miiphy.h>
+#include <asm/io.h>
+
+#include "dm_qe_uec.h"
+
+#define QE_UEC_DRIVER_NAME	"ucc_geth"
+
+/* Default UTBIPAR SMI address */
+#ifndef CONFIG_UTBIPAR_INIT_TBIPA
+#define CONFIG_UTBIPAR_INIT_TBIPA 0x1F
+#endif
+
+static int uec_mac_enable(struct uec_priv *uec, comm_dir_e mode)
+{
+	uec_t		*uec_regs;
+	u32		maccfg1;
+
+	uec_regs = uec->uec_regs;
+	maccfg1 = in_be32(&uec_regs->maccfg1);
+
+	if (mode & COMM_DIR_TX)	{
+		maccfg1 |= MACCFG1_ENABLE_TX;
+		out_be32(&uec_regs->maccfg1, maccfg1);
+		uec->mac_tx_enabled = 1;
+	}
+
+	if (mode & COMM_DIR_RX)	{
+		maccfg1 |= MACCFG1_ENABLE_RX;
+		out_be32(&uec_regs->maccfg1, maccfg1);
+		uec->mac_rx_enabled = 1;
+	}
+
+	return 0;
+}
+
+static int uec_mac_disable(struct uec_priv *uec, comm_dir_e mode)
+{
+	uec_t		*uec_regs;
+	u32		maccfg1;
+
+	uec_regs = uec->uec_regs;
+	maccfg1 = in_be32(&uec_regs->maccfg1);
+
+	if (mode & COMM_DIR_TX)	{
+		maccfg1 &= ~MACCFG1_ENABLE_TX;
+		out_be32(&uec_regs->maccfg1, maccfg1);
+		uec->mac_tx_enabled = 0;
+	}
+
+	if (mode & COMM_DIR_RX)	{
+		maccfg1 &= ~MACCFG1_ENABLE_RX;
+		out_be32(&uec_regs->maccfg1, maccfg1);
+		uec->mac_rx_enabled = 0;
+	}
+
+	return 0;
+}
+
+static int uec_restart_tx(struct uec_priv *uec)
+{
+	struct uec_inf	*ui = uec->uec_info;
+	u32		cecr_subblock;
+
+	cecr_subblock = ucc_fast_get_qe_cr_subblock(ui->uf_info.ucc_num);
+	qe_issue_cmd(QE_RESTART_TX, cecr_subblock,
+		     (u8)QE_CR_PROTOCOL_ETHERNET, 0);
+
+	uec->grace_stopped_tx = 0;
+
+	return 0;
+}
+
+static int uec_restart_rx(struct uec_priv *uec)
+{
+	struct uec_inf	*ui = uec->uec_info;
+	u32		cecr_subblock;
+
+	cecr_subblock = ucc_fast_get_qe_cr_subblock(ui->uf_info.ucc_num);
+	qe_issue_cmd(QE_RESTART_RX, cecr_subblock,
+		     (u8)QE_CR_PROTOCOL_ETHERNET, 0);
+
+	uec->grace_stopped_rx = 0;
+
+	return 0;
+}
+
+static int uec_open(struct uec_priv *uec, comm_dir_e mode)
+{
+	struct ucc_fast_priv	*uccf;
+
+	uccf = uec->uccf;
+
+	/* check if the UCC number is in range. */
+	if (uec->uec_info->uf_info.ucc_num >= UCC_MAX_NUM) {
+		printf("%s: ucc_num out of range.\n", __func__);
+		return -EINVAL;
+	}
+
+	/* Enable MAC */
+	uec_mac_enable(uec, mode);
+
+	/* Enable UCC fast */
+	ucc_fast_enable(uccf, mode);
+
+	/* RISC microcode start */
+	if ((mode & COMM_DIR_TX) && uec->grace_stopped_tx)
+		uec_restart_tx(uec);
+
+	if ((mode & COMM_DIR_RX) && uec->grace_stopped_rx)
+		uec_restart_rx(uec);
+
+	return 0;
+}
+
+static int uec_set_mac_if_mode(struct uec_priv *uec)
+{
+	struct uec_inf		*uec_info = uec->uec_info;
+	phy_interface_t		enet_if_mode;
+	uec_t			*uec_regs;
+	u32			upsmr;
+	u32			maccfg2;
+
+	uec_regs = uec->uec_regs;
+	enet_if_mode = uec_info->enet_interface_type;
+
+	maccfg2 = in_be32(&uec_regs->maccfg2);
+	maccfg2 &= ~MACCFG2_INTERFACE_MODE_MASK;
+
+	upsmr = in_be32(&uec->uccf->uf_regs->upsmr);
+	upsmr &= ~(UPSMR_RPM | UPSMR_TBIM | UPSMR_R10M | UPSMR_RMM);
+
+	switch (uec_info->speed) {
+	case SPEED_10:
+		maccfg2 |= MACCFG2_INTERFACE_MODE_NIBBLE;
+		switch (enet_if_mode) {
+		case PHY_INTERFACE_MODE_MII:
+			break;
+		case PHY_INTERFACE_MODE_RGMII:
+			upsmr |= (UPSMR_RPM | UPSMR_R10M);
+			break;
+		case PHY_INTERFACE_MODE_RMII:
+			upsmr |= (UPSMR_R10M | UPSMR_RMM);
+			break;
+		default:
+			return -EINVAL;
+		}
+		break;
+	case SPEED_100:
+		maccfg2 |= MACCFG2_INTERFACE_MODE_NIBBLE;
+		switch (enet_if_mode) {
+		case PHY_INTERFACE_MODE_MII:
+			break;
+		case PHY_INTERFACE_MODE_RGMII:
+			upsmr |= UPSMR_RPM;
+			break;
+		case PHY_INTERFACE_MODE_RMII:
+			upsmr |= UPSMR_RMM;
+			break;
+		default:
+			return -EINVAL;
+	}
+	break;
+	case SPEED_1000:
+		maccfg2 |= MACCFG2_INTERFACE_MODE_BYTE;
+		switch (enet_if_mode) {
+		case PHY_INTERFACE_MODE_GMII:
+			break;
+		case PHY_INTERFACE_MODE_TBI:
+			upsmr |= UPSMR_TBIM;
+			break;
+		case PHY_INTERFACE_MODE_RTBI:
+			upsmr |= (UPSMR_RPM | UPSMR_TBIM);
+			break;
+		case PHY_INTERFACE_MODE_RGMII_RXID:
+		case PHY_INTERFACE_MODE_RGMII_TXID:
+		case PHY_INTERFACE_MODE_RGMII_ID:
+		case PHY_INTERFACE_MODE_RGMII:
+			upsmr |= UPSMR_RPM;
+			break;
+		case PHY_INTERFACE_MODE_SGMII:
+			upsmr |= UPSMR_SGMM;
+			break;
+		default:
+			return -EINVAL;
+		}
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	out_be32(&uec_regs->maccfg2, maccfg2);
+	out_be32(&uec->uccf->uf_regs->upsmr, upsmr);
+
+	return 0;
+}
+
+static int qe_uec_start(struct udevice *dev)
+{
+	struct qe_uec_priv *priv = dev_get_priv(dev);
+	struct uec_priv		*uec = priv->uec;
+	struct phy_device	*phydev = priv->phydev;
+	struct uec_inf		*uec_info = uec->uec_info;
+	int			err;
+
+	if (!phydev)
+		return -ENODEV;
+
+	/* Setup MAC interface mode */
+	genphy_update_link(phydev);
+	genphy_parse_link(phydev);
+	uec_info->speed = phydev->speed;
+	uec_set_mac_if_mode(uec);
+
+	err = uec_open(uec, COMM_DIR_RX_AND_TX);
+	if (err) {
+		printf("%s: cannot enable UEC device\n", dev->name);
+		return -EINVAL;
+	}
+
+	return (phydev->link ? 0 : -EINVAL);
+}
+
+static int qe_uec_send(struct udevice *dev, void *packet, int length)
+{
+	struct qe_uec_priv *priv = dev_get_priv(dev);
+	struct uec_priv		*uec = priv->uec;
+	struct ucc_fast_priv	*uccf = uec->uccf;
+	struct buffer_descriptor	*bd;
+	u16			status;
+	int			i;
+	int			result = 0;
+
+	uccf = uec->uccf;
+	bd = uec->tx_bd;
+
+	/* Find an empty TxBD */
+	for (i = 0; BD_STATUS(bd) & TX_BD_READY; i++) {
+		if (i > 0x100000) {
+			printf("%s: tx buffer not ready\n", dev->name);
+			return result;
+		}
+	}
+
+	/* Init TxBD */
+	BD_DATA_SET(bd, packet);
+	BD_LENGTH_SET(bd, length);
+	status = BD_STATUS(bd);
+	status &= BD_WRAP;
+	status |= (TX_BD_READY | TX_BD_LAST);
+	BD_STATUS_SET(bd, status);
+
+	/* Tell UCC to transmit the buffer */
+	ucc_fast_transmit_on_demand(uccf);
+
+	/* Wait for buffer to be transmitted */
+	for (i = 0; BD_STATUS(bd) & TX_BD_READY; i++) {
+		if (i > 0x100000) {
+			printf("%s: tx error\n", dev->name);
+			return result;
+		}
+	}
+
+	/* Ok, the buffer be transimitted */
+	BD_ADVANCE(bd, status, uec->p_tx_bd_ring);
+	uec->tx_bd = bd;
+	result = 1;
+
+	return result;
+}
+
+/*
+ * Receive frame:
+ * - wait for the next BD to get ready bit set
+ * - clean up the descriptor
+ * - move on and indicate to HW that the cleaned BD is available for Rx
+ */
+static int qe_uec_recv(struct udevice *dev, int flags, uchar **packetp)
+{
+	struct qe_uec_priv *priv = dev_get_priv(dev);
+	struct uec_priv		*uec = priv->uec;
+	struct buffer_descriptor	*bd;
+	u16			status;
+	u16			len = 0;
+	u8			*data;
+
+	*packetp = memalign(ARCH_DMA_MINALIGN, MAX_RXBUF_LEN);
+	if (*packetp == 0) {
+		printf("%s: error allocating packetp\n", __func__);
+		return -ENOMEM;
+	}
+
+	bd = uec->rx_bd;
+	status = BD_STATUS(bd);
+
+	while (!(status & RX_BD_EMPTY)) {
+		if (!(status & RX_BD_ERROR)) {
+			data = BD_DATA(bd);
+			len = BD_LENGTH(bd);
+			memcpy(*packetp, (char *)data, len);
+		} else {
+			printf("%s: Rx error\n", dev->name);
+		}
+		status &= BD_CLEAN;
+		BD_LENGTH_SET(bd, 0);
+		BD_STATUS_SET(bd, status | RX_BD_EMPTY);
+		BD_ADVANCE(bd, status, uec->p_rx_bd_ring);
+		status = BD_STATUS(bd);
+	}
+	uec->rx_bd = bd;
+
+	return len;
+}
+
+static int uec_graceful_stop_tx(struct uec_priv *uec)
+{
+	ucc_fast_t		*uf_regs;
+	u32			cecr_subblock;
+	u32			ucce;
+
+	uf_regs = uec->uccf->uf_regs;
+
+	/* Clear the grace stop event */
+	out_be32(&uf_regs->ucce, UCCE_GRA);
+
+	/* Issue host command */
+	cecr_subblock =
+		 ucc_fast_get_qe_cr_subblock(uec->uec_info->uf_info.ucc_num);
+	qe_issue_cmd(QE_GRACEFUL_STOP_TX, cecr_subblock,
+		     (u8)QE_CR_PROTOCOL_ETHERNET, 0);
+
+	/* Wait for command to complete */
+	do {
+		ucce = in_be32(&uf_regs->ucce);
+	} while (!(ucce & UCCE_GRA));
+
+	uec->grace_stopped_tx = 1;
+
+	return 0;
+}
+
+static int uec_graceful_stop_rx(struct uec_priv *uec)
+{
+	u32		cecr_subblock;
+	u8		ack;
+
+	if (!uec->p_rx_glbl_pram) {
+		printf("%s: No init rx global parameter\n", __func__);
+		return -EINVAL;
+	}
+
+	/* Clear acknowledge bit */
+	ack = uec->p_rx_glbl_pram->rxgstpack;
+	ack &= ~GRACEFUL_STOP_ACKNOWLEDGE_RX;
+	uec->p_rx_glbl_pram->rxgstpack = ack;
+
+	/* Keep issuing cmd and checking ack bit until it is asserted */
+	do {
+		/* Issue host command */
+		cecr_subblock =
+		ucc_fast_get_qe_cr_subblock(uec->uec_info->uf_info.ucc_num);
+		qe_issue_cmd(QE_GRACEFUL_STOP_RX, cecr_subblock,
+			     (u8)QE_CR_PROTOCOL_ETHERNET, 0);
+		ack = uec->p_rx_glbl_pram->rxgstpack;
+	} while (!(ack & GRACEFUL_STOP_ACKNOWLEDGE_RX));
+
+	uec->grace_stopped_rx = 1;
+
+	return 0;
+}
+
+static int uec_stop(struct uec_priv *uec, comm_dir_e mode)
+{
+	/* check if the UCC number is in range. */
+	if (uec->uec_info->uf_info.ucc_num >= UCC_MAX_NUM) {
+		printf("%s: ucc_num out of range.\n", __func__);
+		return -EINVAL;
+	}
+	/* Stop any transmissions */
+	if ((mode & COMM_DIR_TX) && !uec->grace_stopped_tx)
+		uec_graceful_stop_tx(uec);
+
+	/* Stop any receptions */
+	if ((mode & COMM_DIR_RX) && !uec->grace_stopped_rx)
+		uec_graceful_stop_rx(uec);
+
+	/* Disable the UCC fast */
+	ucc_fast_disable(uec->uccf, mode);
+
+	/* Disable the MAC */
+	uec_mac_disable(uec, mode);
+
+	return 0;
+}
+
+static void qe_uec_stop(struct udevice *dev)
+{
+	struct qe_uec_priv *priv = dev_get_priv(dev);
+	struct uec_priv		*uec = priv->uec;
+
+	uec_stop(uec, COMM_DIR_RX_AND_TX);
+}
+
+static int qe_uec_set_hwaddr(struct udevice *dev)
+{
+	struct qe_uec_priv *priv = dev_get_priv(dev);
+	struct eth_pdata *pdata = dev_get_platdata(dev);
+	struct uec_priv *uec = priv->uec;
+	uec_t *uec_regs = uec->uec_regs;
+	uchar *mac = pdata->enetaddr;
+	u32		mac_addr1;
+	u32		mac_addr2;
+
+	/*
+	 * if a station address of 0x12345678ABCD, perform a write to
+	 * MACSTNADDR1 of 0xCDAB7856,
+	 * MACSTNADDR2 of 0x34120000
+	 */
+
+	mac_addr1 = (mac[5] << 24) | (mac[4] << 16) |
+			(mac[3] << 8)  | (mac[2]);
+	out_be32(&uec_regs->macstnaddr1, mac_addr1);
+
+	mac_addr2 = ((mac[1] << 24) | (mac[0] << 16)) & 0xffff0000;
+	out_be32(&uec_regs->macstnaddr2, mac_addr2);
+
+	return 0;
+}
+
+static int qe_uec_free_pkt(struct udevice *dev, uchar *packet, int length)
+{
+	if (packet)
+		free(packet);
+
+	return 0;
+}
+
+static const struct eth_ops qe_uec_eth_ops = {
+	.start		= qe_uec_start,
+	.send		= qe_uec_send,
+	.recv		= qe_uec_recv,
+	.free_pkt	= qe_uec_free_pkt,
+	.stop		= qe_uec_stop,
+	.write_hwaddr	= qe_uec_set_hwaddr,
+};
+
+static int uec_convert_threads_num(enum uec_num_of_threads threads_num,
+				   int *threads_num_ret)
+{
+	int	num_threads_numerica;
+
+	switch (threads_num) {
+	case UEC_NUM_OF_THREADS_1:
+		num_threads_numerica = 1;
+		break;
+	case UEC_NUM_OF_THREADS_2:
+		num_threads_numerica = 2;
+		break;
+	case UEC_NUM_OF_THREADS_4:
+		num_threads_numerica = 4;
+		break;
+	case UEC_NUM_OF_THREADS_6:
+		num_threads_numerica = 6;
+		break;
+	case UEC_NUM_OF_THREADS_8:
+		num_threads_numerica = 8;
+		break;
+	default:
+		printf("%s: Bad number of threads value.",
+		       __func__);
+		return -EINVAL;
+	}
+
+	*threads_num_ret = num_threads_numerica;
+
+	return 0;
+}
+
+static void uec_init_tx_parameter(struct uec_priv *uec, int num_threads_tx)
+{
+	struct uec_inf	*uec_info;
+	u32		end_bd;
+	u8		bmrx = 0;
+	int		i;
+
+	uec_info = uec->uec_info;
+
+	/* Alloc global Tx parameter RAM page */
+	uec->tx_glbl_pram_offset =
+		qe_muram_alloc(sizeof(struct uec_tx_global_pram),
+			       UEC_TX_GLOBAL_PRAM_ALIGNMENT);
+	uec->p_tx_glbl_pram = (struct uec_tx_global_pram *)
+				qe_muram_addr(uec->tx_glbl_pram_offset);
+
+	/* Zero the global Tx prameter RAM */
+	memset(uec->p_tx_glbl_pram, 0, sizeof(struct uec_tx_global_pram));
+
+	/* Init global Tx parameter RAM */
+
+	/* TEMODER, RMON statistics disable, one Tx queue */
+	out_be16(&uec->p_tx_glbl_pram->temoder, TEMODER_INIT_VALUE);
+
+	/* SQPTR */
+	uec->send_q_mem_reg_offset =
+		qe_muram_alloc(sizeof(struct uec_send_queue_qd),
+			       UEC_SEND_QUEUE_QUEUE_DESCRIPTOR_ALIGNMENT);
+	uec->p_send_q_mem_reg = (struct uec_send_queue_mem_region *)
+				qe_muram_addr(uec->send_q_mem_reg_offset);
+	out_be32(&uec->p_tx_glbl_pram->sqptr, uec->send_q_mem_reg_offset);
+
+	/* Setup the table with TxBDs ring */
+	end_bd = (u32)uec->p_tx_bd_ring + (uec_info->tx_bd_ring_len - 1)
+					 * SIZEOFBD;
+	out_be32(&uec->p_send_q_mem_reg->sqqd[0].bd_ring_base,
+		 (u32)(uec->p_tx_bd_ring));
+	out_be32(&uec->p_send_q_mem_reg->sqqd[0].last_bd_completed_address,
+		 end_bd);
+
+	/* Scheduler Base Pointer, we have only one Tx queue, no need it */
+	out_be32(&uec->p_tx_glbl_pram->schedulerbasepointer, 0);
+
+	/* TxRMON Base Pointer, TxRMON disable, we don't need it */
+	out_be32(&uec->p_tx_glbl_pram->txrmonbaseptr, 0);
+
+	/* TSTATE, global snooping, big endian, the CSB bus selected */
+	bmrx = BMR_INIT_VALUE;
+	out_be32(&uec->p_tx_glbl_pram->tstate, ((u32)(bmrx) << BMR_SHIFT));
+
+	/* IPH_Offset */
+	for (i = 0; i < MAX_IPH_OFFSET_ENTRY; i++)
+		out_8(&uec->p_tx_glbl_pram->iphoffset[i], 0);
+
+	/* VTAG table */
+	for (i = 0; i < UEC_TX_VTAG_TABLE_ENTRY_MAX; i++)
+		out_be32(&uec->p_tx_glbl_pram->vtagtable[i], 0);
+
+	/* TQPTR */
+	uec->thread_dat_tx_offset =
+		qe_muram_alloc(num_threads_tx *
+			       sizeof(struct uec_thread_data_tx) +
+			       32 * (num_threads_tx == 1),
+			       UEC_THREAD_DATA_ALIGNMENT);
+
+	uec->p_thread_data_tx = (struct uec_thread_data_tx *)
+				qe_muram_addr(uec->thread_dat_tx_offset);
+	out_be32(&uec->p_tx_glbl_pram->tqptr, uec->thread_dat_tx_offset);
+}
+
+static void uec_init_rx_parameter(struct uec_priv *uec, int num_threads_rx)
+{
+	u8	bmrx = 0;
+	int	i;
+	struct uec_82xx_add_filtering_pram	*p_af_pram;
+
+	/* Allocate global Rx parameter RAM page */
+	uec->rx_glbl_pram_offset =
+		qe_muram_alloc(sizeof(struct uec_rx_global_pram),
+			       UEC_RX_GLOBAL_PRAM_ALIGNMENT);
+	uec->p_rx_glbl_pram = (struct uec_rx_global_pram *)
+				qe_muram_addr(uec->rx_glbl_pram_offset);
+
+	/* Zero Global Rx parameter RAM */
+	memset(uec->p_rx_glbl_pram, 0, sizeof(struct uec_rx_global_pram));
+
+	/* Init global Rx parameter RAM */
+	/*
+	 * REMODER, Extended feature mode disable, VLAN disable,
+	 * LossLess flow control disable, Receive firmware statisic disable,
+	 * Extended address parsing mode disable, One Rx queues,
+	 * Dynamic maximum/minimum frame length disable, IP checksum check
+	 * disable, IP address alignment disable
+	 */
+	out_be32(&uec->p_rx_glbl_pram->remoder, REMODER_INIT_VALUE);
+
+	/* RQPTR */
+	uec->thread_dat_rx_offset =
+		qe_muram_alloc(num_threads_rx *
+			       sizeof(struct uec_thread_data_rx),
+			       UEC_THREAD_DATA_ALIGNMENT);
+	uec->p_thread_data_rx = (struct uec_thread_data_rx *)
+				qe_muram_addr(uec->thread_dat_rx_offset);
+	out_be32(&uec->p_rx_glbl_pram->rqptr, uec->thread_dat_rx_offset);
+
+	/* Type_or_Len */
+	out_be16(&uec->p_rx_glbl_pram->typeorlen, 3072);
+
+	/* RxRMON base pointer, we don't need it */
+	out_be32(&uec->p_rx_glbl_pram->rxrmonbaseptr, 0);
+
+	/* IntCoalescingPTR, we don't need it, no interrupt */
+	out_be32(&uec->p_rx_glbl_pram->intcoalescingptr, 0);
+
+	/* RSTATE, global snooping, big endian, the CSB bus selected */
+	bmrx = BMR_INIT_VALUE;
+	out_8(&uec->p_rx_glbl_pram->rstate, bmrx);
+
+	/* MRBLR */
+	out_be16(&uec->p_rx_glbl_pram->mrblr, MAX_RXBUF_LEN);
+
+	/* RBDQPTR */
+	uec->rx_bd_qs_tbl_offset =
+		qe_muram_alloc(sizeof(struct uec_rx_bd_queues_entry) +
+			       sizeof(struct uec_rx_pref_bds),
+			       UEC_RX_BD_QUEUES_ALIGNMENT);
+	uec->p_rx_bd_qs_tbl = (struct uec_rx_bd_queues_entry *)
+				qe_muram_addr(uec->rx_bd_qs_tbl_offset);
+
+	/* Zero it */
+	memset(uec->p_rx_bd_qs_tbl, 0, sizeof(struct uec_rx_bd_queues_entry) +
+	       sizeof(struct uec_rx_pref_bds));
+	out_be32(&uec->p_rx_glbl_pram->rbdqptr, uec->rx_bd_qs_tbl_offset);
+	out_be32(&uec->p_rx_bd_qs_tbl->externalbdbaseptr,
+		 (u32)uec->p_rx_bd_ring);
+
+	/* MFLR */
+	out_be16(&uec->p_rx_glbl_pram->mflr, MAX_FRAME_LEN);
+	/* MINFLR */
+	out_be16(&uec->p_rx_glbl_pram->minflr, MIN_FRAME_LEN);
+	/* MAXD1 */
+	out_be16(&uec->p_rx_glbl_pram->maxd1, MAX_DMA1_LEN);
+	/* MAXD2 */
+	out_be16(&uec->p_rx_glbl_pram->maxd2, MAX_DMA2_LEN);
+	/* ECAM_PTR */
+	out_be32(&uec->p_rx_glbl_pram->ecamptr, 0);
+	/* L2QT */
+	out_be32(&uec->p_rx_glbl_pram->l2qt, 0);
+	/* L3QT */
+	for (i = 0; i < 8; i++)
+		out_be32(&uec->p_rx_glbl_pram->l3qt[i], 0);
+
+	/* VLAN_TYPE */
+	out_be16(&uec->p_rx_glbl_pram->vlantype, 0x8100);
+	/* TCI */
+	out_be16(&uec->p_rx_glbl_pram->vlantci, 0);
+
+	/* Clear PQ2 style address filtering hash table */
+	p_af_pram = (struct uec_82xx_add_filtering_pram *)
+			uec->p_rx_glbl_pram->addressfiltering;
+
+	p_af_pram->iaddr_h = 0;
+	p_af_pram->iaddr_l = 0;
+	p_af_pram->gaddr_h = 0;
+	p_af_pram->gaddr_l = 0;
+}
+
+static int uec_issue_init_enet_rxtx_cmd(struct uec_priv *uec,
+					int thread_tx, int thread_rx)
+{
+	struct uec_init_cmd_pram		*p_init_enet_param;
+	u32				init_enet_param_offset;
+	struct uec_inf			*uec_info;
+	struct ucc_fast_inf			*uf_info;
+	int				i;
+	int				snum;
+	u32				off;
+	u32				entry_val;
+	u32				command;
+	u32				cecr_subblock;
+
+	uec_info = uec->uec_info;
+	uf_info = &uec_info->uf_info;
+
+	/* Allocate init enet command parameter */
+	uec->init_enet_param_offset =
+		qe_muram_alloc(sizeof(struct uec_init_cmd_pram), 4);
+	init_enet_param_offset = uec->init_enet_param_offset;
+	uec->p_init_enet_param = (struct uec_init_cmd_pram *)
+				qe_muram_addr(uec->init_enet_param_offset);
+
+	/* Zero init enet command struct */
+	memset((void *)uec->p_init_enet_param, 0,
+	       sizeof(struct uec_init_cmd_pram));
+
+	/* Init the command struct */
+	p_init_enet_param = uec->p_init_enet_param;
+	p_init_enet_param->resinit0 = ENET_INIT_PARAM_MAGIC_RES_INIT0;
+	p_init_enet_param->resinit1 = ENET_INIT_PARAM_MAGIC_RES_INIT1;
+	p_init_enet_param->resinit2 = ENET_INIT_PARAM_MAGIC_RES_INIT2;
+	p_init_enet_param->resinit3 = ENET_INIT_PARAM_MAGIC_RES_INIT3;
+	p_init_enet_param->resinit4 = ENET_INIT_PARAM_MAGIC_RES_INIT4;
+	p_init_enet_param->largestexternallookupkeysize = 0;
+
+	p_init_enet_param->rgftgfrxglobal |= ((u32)uec_info->num_threads_rx)
+					 << ENET_INIT_PARAM_RGF_SHIFT;
+	p_init_enet_param->rgftgfrxglobal |= ((u32)uec_info->num_threads_tx)
+					 << ENET_INIT_PARAM_TGF_SHIFT;
+
+	/* Init Rx global parameter pointer */
+	p_init_enet_param->rgftgfrxglobal |= uec->rx_glbl_pram_offset |
+						 (u32)uec_info->risc_rx;
+
+	/* Init Rx threads */
+	for (i = 0; i < (thread_rx + 1); i++) {
+		snum = qe_get_snum();
+		if (snum < 0) {
+			printf("%s can not get snum\n", __func__);
+			return -ENOMEM;
+		}
+
+		if (i == 0) {
+			off = 0;
+		} else {
+			off = qe_muram_alloc(sizeof(struct uec_thread_rx_pram),
+					     UEC_THREAD_RX_PRAM_ALIGNMENT);
+		}
+
+		entry_val = ((u32)snum << ENET_INIT_PARAM_SNUM_SHIFT) |
+				 off | (u32)uec_info->risc_rx;
+		p_init_enet_param->rxthread[i] = entry_val;
+	}
+
+	/* Init Tx global parameter pointer */
+	p_init_enet_param->txglobal = uec->tx_glbl_pram_offset |
+					 (u32)uec_info->risc_tx;
+
+	/* Init Tx threads */
+	for (i = 0; i < thread_tx; i++) {
+		snum = qe_get_snum();
+		if (snum  < 0)	{
+			printf("%s can not get snum\n", __func__);
+			return -ENOMEM;
+		}
+
+		off = qe_muram_alloc(sizeof(struct uec_thread_tx_pram),
+				     UEC_THREAD_TX_PRAM_ALIGNMENT);
+
+		entry_val = ((u32)snum << ENET_INIT_PARAM_SNUM_SHIFT) |
+				 off | (u32)uec_info->risc_tx;
+		p_init_enet_param->txthread[i] = entry_val;
+	}
+
+	__asm__ __volatile__("sync");
+
+	/* Issue QE command */
+	command = QE_INIT_TX_RX;
+	cecr_subblock =	ucc_fast_get_qe_cr_subblock(uf_info->ucc_num);
+	qe_issue_cmd(command, cecr_subblock, (u8)QE_CR_PROTOCOL_ETHERNET,
+		     init_enet_param_offset);
+
+	return 0;
+}
+
+static int uec_startup(struct udevice *dev)
+{
+	struct qe_uec_priv *priv = dev_get_priv(dev);
+	struct uec_priv *uec = priv->uec;
+	struct uec_inf			*uec_info;
+	struct ucc_fast_inf			*uf_info;
+	struct ucc_fast_priv		*uccf;
+	ucc_fast_t			*uf_regs;
+	uec_t				*uec_regs;
+	int				num_threads_tx;
+	int				num_threads_rx;
+	u32				utbipar;
+	u32				length;
+	u32				align;
+	struct buffer_descriptor	*bd;
+	u8				*buf;
+	int				i;
+
+	uec_info = uec->uec_info;
+	uf_info = &uec_info->uf_info;
+
+	/* Check if Rx BD ring len is illegal */
+	if (uec_info->rx_bd_ring_len < UEC_RX_BD_RING_SIZE_MIN ||
+	    uec_info->rx_bd_ring_len % UEC_RX_BD_RING_SIZE_ALIGNMENT) {
+		printf("%s: Rx BD ring len must be multiple of 4, and > 8.\n",
+		       __func__);
+		return -EINVAL;
+	}
+
+	/* Check if Tx BD ring len is illegal */
+	if (uec_info->tx_bd_ring_len < UEC_TX_BD_RING_SIZE_MIN) {
+		printf("%s: Tx BD ring length must not be smaller than 2.\n",
+		       __func__);
+		return -EINVAL;
+	}
+
+	/* Check if MRBLR is illegal */
+	if (MAX_RXBUF_LEN == 0 || (MAX_RXBUF_LEN  % UEC_MRBLR_ALIGNMENT)) {
+		printf("%s: max rx buffer length must be mutliple of 128.\n",
+		       __func__);
+		return -EINVAL;
+	}
+
+	/* Both Rx and Tx are stopped */
+	uec->grace_stopped_rx = 1;
+	uec->grace_stopped_tx = 1;
+
+	/* Init UCC fast */
+	if (ucc_fast_init(uf_info, &uccf)) {
+		printf("%s: failed to init ucc fast\n", __func__);
+		return -ENOMEM;
+	}
+
+	/* Save uccf */
+	uec->uccf = uccf;
+
+	/* Convert the Tx threads number */
+	if (uec_convert_threads_num(uec_info->num_threads_tx,
+				    &num_threads_tx))
+		return -EINVAL;
+
+	/* Convert the Rx threads number */
+	if (uec_convert_threads_num(uec_info->num_threads_rx,
+				    &num_threads_rx))
+		return -EINVAL;
+
+	uf_regs = uccf->uf_regs;
+
+	/* UEC register is following UCC fast registers */
+	uec_regs = (uec_t *)(&uf_regs->ucc_eth);
+
+	/* Save the UEC register pointer to UEC private struct */
+	uec->uec_regs = uec_regs;
+
+	/* Init UPSMR, enable hardware statistics (UCC) */
+	out_be32(&uec->uccf->uf_regs->upsmr, UPSMR_INIT_VALUE);
+
+	/* Init MACCFG1, flow control disable, disable Tx and Rx */
+	out_be32(&uec_regs->maccfg1, MACCFG1_INIT_VALUE);
+
+	/* Init MACCFG2, length check, MAC PAD and CRC enable */
+	out_be32(&uec_regs->maccfg2, MACCFG2_INIT_VALUE);
+
+	/* Setup UTBIPAR */
+	utbipar = in_be32(&uec_regs->utbipar);
+	utbipar &= ~UTBIPAR_PHY_ADDRESS_MASK;
+
+	/* Initialize UTBIPAR address to CONFIG_UTBIPAR_INIT_TBIPA for ALL UEC.
+	 * This frees up the remaining SMI addresses for use.
+	 */
+	utbipar |= CONFIG_UTBIPAR_INIT_TBIPA << UTBIPAR_PHY_ADDRESS_SHIFT;
+	out_be32(&uec_regs->utbipar, utbipar);
+
+	/* Allocate Tx BDs */
+	length = ((uec_info->tx_bd_ring_len * SIZEOFBD) /
+		 UEC_TX_BD_RING_SIZE_MEMORY_ALIGNMENT) *
+		 UEC_TX_BD_RING_SIZE_MEMORY_ALIGNMENT;
+	if ((uec_info->tx_bd_ring_len * SIZEOFBD) %
+	    UEC_TX_BD_RING_SIZE_MEMORY_ALIGNMENT)
+		length += UEC_TX_BD_RING_SIZE_MEMORY_ALIGNMENT;
+
+	align = UEC_TX_BD_RING_ALIGNMENT;
+	uec->tx_bd_ring_offset = (u32)malloc((u32)(length + align));
+	if (uec->tx_bd_ring_offset != 0)
+		uec->p_tx_bd_ring = (u8 *)((uec->tx_bd_ring_offset + align)
+					   & ~(align - 1));
+
+	/* Zero all of Tx BDs */
+	memset((void *)(uec->tx_bd_ring_offset), 0, length + align);
+
+	/* Allocate Rx BDs */
+	length = uec_info->rx_bd_ring_len * SIZEOFBD;
+	align = UEC_RX_BD_RING_ALIGNMENT;
+	uec->rx_bd_ring_offset = (u32)(malloc((u32)(length + align)));
+	if (uec->rx_bd_ring_offset != 0)
+		uec->p_rx_bd_ring = (u8 *)((uec->rx_bd_ring_offset + align)
+					   & ~(align - 1));
+
+	/* Zero all of Rx BDs */
+	memset((void *)(uec->rx_bd_ring_offset), 0, length + align);
+
+	/* Allocate Rx buffer */
+	length = uec_info->rx_bd_ring_len * MAX_RXBUF_LEN;
+	align = UEC_RX_DATA_BUF_ALIGNMENT;
+	uec->rx_buf_offset = (u32)malloc(length + align);
+	if (uec->rx_buf_offset != 0)
+		uec->p_rx_buf = (u8 *)((uec->rx_buf_offset + align)
+				       & ~(align - 1));
+
+	/* Zero all of the Rx buffer */
+	memset((void *)(uec->rx_buf_offset), 0, length + align);
+
+	/* Init TxBD ring */
+	bd = (struct buffer_descriptor *)uec->p_tx_bd_ring;
+	uec->tx_bd = bd;
+
+	for (i = 0; i < uec_info->tx_bd_ring_len; i++) {
+		BD_DATA_CLEAR(bd);
+		BD_STATUS_SET(bd, 0);
+		BD_LENGTH_SET(bd, 0);
+		bd++;
+	}
+	BD_STATUS_SET((--bd), TX_BD_WRAP);
+
+	/* Init RxBD ring */
+	bd = (struct buffer_descriptor *)uec->p_rx_bd_ring;
+	uec->rx_bd = bd;
+	buf = uec->p_rx_buf;
+	for (i = 0; i < uec_info->rx_bd_ring_len; i++) {
+		BD_DATA_SET(bd, buf);
+		BD_LENGTH_SET(bd, 0);
+		BD_STATUS_SET(bd, RX_BD_EMPTY);
+		buf += MAX_RXBUF_LEN;
+		bd++;
+	}
+	BD_STATUS_SET((--bd), RX_BD_WRAP | RX_BD_EMPTY);
+
+	/* Init global Tx parameter RAM */
+	uec_init_tx_parameter(uec, num_threads_tx);
+
+	/* Init global Rx parameter RAM */
+	uec_init_rx_parameter(uec, num_threads_rx);
+
+	/* Init ethernet Tx and Rx parameter command */
+	if (uec_issue_init_enet_rxtx_cmd(uec, num_threads_tx,
+					 num_threads_rx)) {
+		printf("%s issue init enet cmd failed\n", __func__);
+		return -ENOMEM;
+	}
+	return 0;
+}
+
+/* Convert a string to a QE clock source enum
+ *
+ * This function takes a string, typically from a property in the device
+ * tree, and returns the corresponding "enum qe_clock" value.
+ */
+enum qe_clock qe_clock_source(const char *source)
+{
+	unsigned int i;
+
+	if (strcasecmp(source, "none") == 0)
+		return QE_CLK_NONE;
+
+	if (strncasecmp(source, "brg", 3) == 0) {
+		i = simple_strtoul(source + 3, NULL, 10);
+		if (i >= 1 && i <= 16)
+			return (QE_BRG1 - 1) + i;
+		else
+			return QE_CLK_DUMMY;
+	}
+
+	if (strncasecmp(source, "clk", 3) == 0) {
+		i = simple_strtoul(source + 3, NULL, 10);
+		if (i >= 1 && i <= 24)
+			return (QE_CLK1 - 1) + i;
+		else
+			return QE_CLK_DUMMY;
+	}
+
+	return QE_CLK_DUMMY;
+}
+
+static void qe_uec_set_eth_type(struct udevice *dev)
+{
+	struct qe_uec_priv *priv = dev_get_priv(dev);
+	struct uec_priv		*uec = priv->uec;
+	struct uec_inf *uec_info  = uec->uec_info;
+	struct ucc_fast_inf *uf_info = &uec_info->uf_info;
+
+	switch (uec_info->enet_interface_type) {
+	case PHY_INTERFACE_MODE_GMII:
+	case PHY_INTERFACE_MODE_RGMII:
+	case PHY_INTERFACE_MODE_RGMII_ID:
+	case PHY_INTERFACE_MODE_RGMII_RXID:
+	case PHY_INTERFACE_MODE_RGMII_TXID:
+	case PHY_INTERFACE_MODE_TBI:
+	case PHY_INTERFACE_MODE_RTBI:
+	case PHY_INTERFACE_MODE_SGMII:
+		uf_info->eth_type = GIGA_ETH;
+		break;
+	default:
+		uf_info->eth_type = FAST_ETH;
+		break;
+	}
+}
+
+static int qe_uec_set_uec_info(struct udevice *dev)
+{
+	struct qe_uec_priv *priv = dev_get_priv(dev);
+	struct eth_pdata *pdata = dev_get_platdata(dev);
+	struct uec_priv *uec = priv->uec;
+	struct uec_inf *uec_info;
+	struct ucc_fast_inf *uf_info;
+	const char *s;
+	int ret;
+	u32 val;
+
+	uec_info = (struct uec_inf *)malloc(sizeof(struct uec_inf));
+	if (!uec_info)
+		return -ENOMEM;
+
+	uf_info = &uec_info->uf_info;
+
+	ret = dev_read_u32(dev, "cell-index", &val);
+	if (ret) {
+		ret = dev_read_u32(dev, "device-id", &val);
+		if (ret) {
+			pr_err("no cell-index nor device-id found!");
+			goto out;
+		}
+	}
+
+	uf_info->ucc_num = val - 1;
+	if (uf_info->ucc_num < 0 || uf_info->ucc_num > 7) {
+		ret = -ENODEV;
+		goto out;
+	}
+
+	ret = dev_read_string_index(dev, "rx-clock-name", 0, &s);
+	if (!ret) {
+		uf_info->rx_clock = qe_clock_source(s);
+		if (uf_info->rx_clock < QE_CLK_NONE ||
+		    uf_info->rx_clock > QE_CLK24) {
+			pr_err("invalid rx-clock-name property\n");
+			ret = -EINVAL;
+			goto out;
+		}
+	} else {
+		ret = dev_read_u32(dev, "rx-clock", &val);
+		if (ret) {
+			/*
+			 * If both rx-clock-name and rx-clock are missing,
+			 * we want to tell people to use rx-clock-name.
+			 */
+			pr_err("missing rx-clock-name property\n");
+			goto out;
+		}
+		if (val < QE_CLK_NONE || val > QE_CLK24) {
+			pr_err("invalid rx-clock property\n");
+			ret = -EINVAL;
+			goto out;
+		}
+		uf_info->rx_clock = val;
+	}
+
+	ret = dev_read_string_index(dev, "tx-clock-name", 0, &s);
+	if (!ret) {
+		uf_info->tx_clock = qe_clock_source(s);
+		if (uf_info->tx_clock < QE_CLK_NONE ||
+		    uf_info->tx_clock > QE_CLK24) {
+			pr_err("invalid tx-clock-name property\n");
+			ret = -EINVAL;
+			goto out;
+		}
+	} else {
+		ret = dev_read_u32(dev, "tx-clock", &val);
+		if (ret) {
+			pr_err("missing tx-clock-name property\n");
+			goto out;
+		}
+		if (val < QE_CLK_NONE || val > QE_CLK24) {
+			pr_err("invalid tx-clock property\n");
+			ret = -EINVAL;
+			goto out;
+		}
+		uf_info->tx_clock = val;
+	}
+
+	uec_info->num_threads_tx = UEC_NUM_OF_THREADS_1;
+	uec_info->num_threads_rx = UEC_NUM_OF_THREADS_1;
+	uec_info->risc_tx = QE_RISC_ALLOCATION_RISC1_AND_RISC2;
+	uec_info->risc_rx = QE_RISC_ALLOCATION_RISC1_AND_RISC2;
+	uec_info->tx_bd_ring_len = 16;
+	uec_info->rx_bd_ring_len = 16;
+#if (MAX_QE_RISC == 4)
+	uec_info->risc_tx = QE_RISC_ALLOCATION_FOUR_RISCS;
+	uec_info->risc_rx = QE_RISC_ALLOCATION_FOUR_RISCS;
+#endif
+
+	uec_info->enet_interface_type = pdata->phy_interface;
+
+	uec->uec_info = uec_info;
+	qe_uec_set_eth_type(dev);
+
+	return 0;
+out:
+	free(uec_info);
+	return ret;
+}
+
+static int qe_uec_probe(struct udevice *dev)
+{
+	struct qe_uec_priv *priv = dev_get_priv(dev);
+	struct eth_pdata *pdata = dev_get_platdata(dev);
+	struct uec_priv		*uec;
+	int ret;
+
+	/* Allocate the UEC private struct */
+	uec = (struct uec_priv *)malloc(sizeof(struct uec_priv));
+	if (!uec)
+		return -ENOMEM;
+
+	memset(uec, 0, sizeof(struct uec_priv));
+	priv->uec = uec;
+	uec->uec_regs = (uec_t *)pdata->iobase;
+
+	/* setup uec info struct */
+	ret = qe_uec_set_uec_info(dev);
+	if (ret) {
+		free(uec);
+		return ret;
+	}
+
+	ret = uec_startup(dev);
+	if (ret) {
+		free(uec->uec_info);
+		free(uec);
+		return ret;
+	}
+
+	priv->phydev = dm_eth_phy_connect(dev);
+	return 0;
+}
+
+/*
+ * Remove the driver from an interface:
+ * - free up allocated memory
+ */
+static int qe_uec_remove(struct udevice *dev)
+{
+	struct qe_uec_priv *priv = dev_get_priv(dev);
+
+	free(priv->uec);
+	return 0;
+}
+
+static int qe_uec_ofdata_to_platdata(struct udevice *dev)
+{
+	struct eth_pdata *pdata = dev_get_platdata(dev);
+	const char *phy_mode;
+
+	pdata->iobase = (phys_addr_t)devfdt_get_addr(dev);
+
+	pdata->phy_interface = -1;
+	phy_mode = fdt_getprop(gd->fdt_blob, dev_of_offset(dev),
+			       "phy-connection-type", NULL);
+	if (phy_mode)
+		pdata->phy_interface = phy_get_interface_by_name(phy_mode);
+	if (pdata->phy_interface == -1) {
+		debug("%s: Invalid PHY interface '%s'\n", __func__, phy_mode);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static const struct udevice_id qe_uec_ids[] = {
+	{ .compatible = QE_UEC_DRIVER_NAME },
+	{ }
+};
+
+U_BOOT_DRIVER(eth_qe_uec) = {
+	.name	= QE_UEC_DRIVER_NAME,
+	.id	= UCLASS_ETH,
+	.of_match = qe_uec_ids,
+	.ofdata_to_platdata = qe_uec_ofdata_to_platdata,
+	.probe	= qe_uec_probe,
+	.remove = qe_uec_remove,
+	.ops	= &qe_uec_eth_ops,
+	.priv_auto_alloc_size = sizeof(struct qe_uec_priv),
+	.platdata_auto_alloc_size = sizeof(struct eth_pdata),
+};