spi: npcm-fiu: add NPCM7xx FIU controller driver

Add Nuvoton NPCM BMC Flash Interface Unit(FIU) SPI master
controller driver using SPI-MEM interface.

The FIU supports single, dual or quad communication interface.
The FIU controller driver provides flash access in UMA(User
Mode Access) mode by using an indirect address/data mechanism.

the dts node is followed upstream kernel dts name.

Signed-off-by: Jim Liu <JJLIU0@nuvoton.com>
Signed-off-by: Stanley Chu <yschu@nuvoton.com>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
[Jagan: fixed the Kconfig, Makefile order]
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
diff --git a/drivers/spi/npcm_fiu_spi.c b/drivers/spi/npcm_fiu_spi.c
new file mode 100644
index 0000000..7000fe5
--- /dev/null
+++ b/drivers/spi/npcm_fiu_spi.c
@@ -0,0 +1,387 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (c) 2022 Nuvoton Technology Corp.
+ * NPCM Flash Interface Unit(FIU) SPI master controller driver.
+ */
+
+#include <clk.h>
+#include <dm.h>
+#include <spi.h>
+#include <spi-mem.h>
+#include <linux/bitfield.h>
+#include <linux/log2.h>
+#include <linux/iopoll.h>
+
+#define DW_SIZE			4
+#define CHUNK_SIZE		16
+#define XFER_TIMEOUT		1000000
+
+/* FIU UMA Configuration Register (UMA_CFG) */
+#define UMA_CFG_RDATSIZ_MASK	GENMASK(28, 24)
+#define UMA_CFG_DBSIZ_MASK	GENMASK(23, 21)
+#define UMA_CFG_WDATSIZ_MASK	GENMASK(20, 16)
+#define UMA_CFG_ADDSIZ_MASK	GENMASK(13, 11)
+#define UMA_CFG_RDBPCK_MASK	GENMASK(9, 8)
+#define UMA_CFG_DBPCK_MASK	GENMASK(7, 6)
+#define UMA_CFG_WDBPCK_MASK	GENMASK(5, 4)
+#define UMA_CFG_ADBPCK_MASK	GENMASK(3, 2)
+#define UMA_CFG_CMBPCK_MASK	GENMASK(1, 0)
+#define UMA_CFG_CMDSIZ_SHIFT	10
+
+/* FIU UMA Control and Status Register (UMA_CTS) */
+#define UMA_CTS_SW_CS		BIT(16)
+#define UMA_CTS_EXEC_DONE	BIT(0)
+#define UMA_CTS_RDYST		BIT(24)
+#define UMA_CTS_DEV_NUM_MASK	GENMASK(9, 8)
+
+struct npcm_fiu_regs {
+	unsigned int    drd_cfg;
+	unsigned int    dwr_cfg;
+	unsigned int    uma_cfg;
+	unsigned int    uma_cts;
+	unsigned int    uma_cmd;
+	unsigned int    uma_addr;
+	unsigned int    prt_cfg;
+	unsigned char	res1[4];
+	unsigned int    uma_dw0;
+	unsigned int    uma_dw1;
+	unsigned int    uma_dw2;
+	unsigned int    uma_dw3;
+	unsigned int    uma_dr0;
+	unsigned int    uma_dr1;
+	unsigned int    uma_dr2;
+	unsigned int    uma_dr3;
+	unsigned int    prt_cmd0;
+	unsigned int    prt_cmd1;
+	unsigned int    prt_cmd2;
+	unsigned int    prt_cmd3;
+	unsigned int    prt_cmd4;
+	unsigned int    prt_cmd5;
+	unsigned int    prt_cmd6;
+	unsigned int    prt_cmd7;
+	unsigned int    prt_cmd8;
+	unsigned int    prt_cmd9;
+	unsigned int    stuff[4];
+	unsigned int    fiu_cfg;
+};
+
+struct npcm_fiu_priv {
+	struct npcm_fiu_regs *regs;
+	struct clk clk;
+};
+
+static int npcm_fiu_spi_set_speed(struct udevice *bus, uint speed)
+{
+	struct npcm_fiu_priv *priv = dev_get_priv(bus);
+	int ret;
+
+	debug("%s: set speed %u\n", bus->name, speed);
+	ret = clk_set_rate(&priv->clk, speed);
+	if (ret < 0)
+		return ret;
+
+	return 0;
+}
+
+static int npcm_fiu_spi_set_mode(struct udevice *bus, uint mode)
+{
+	return 0;
+}
+
+static inline void activate_cs(struct npcm_fiu_regs *regs, int cs)
+{
+	writel(FIELD_PREP(UMA_CTS_DEV_NUM_MASK, cs), &regs->uma_cts);
+}
+
+static inline void deactivate_cs(struct npcm_fiu_regs *regs, int cs)
+{
+	writel(FIELD_PREP(UMA_CTS_DEV_NUM_MASK, cs) | UMA_CTS_SW_CS, &regs->uma_cts);
+}
+
+static int fiu_uma_read(struct udevice *bus, u8 *buf, u32 size)
+{
+	struct npcm_fiu_priv *priv = dev_get_priv(bus);
+	struct npcm_fiu_regs *regs = priv->regs;
+	u32 data_reg[4];
+	u32 val;
+	int ret;
+
+	/* Set data size */
+	writel(FIELD_PREP(UMA_CFG_RDATSIZ_MASK, size), &regs->uma_cfg);
+
+	/* Initiate the read */
+	writel(readl(&regs->uma_cts) | UMA_CTS_EXEC_DONE, &regs->uma_cts);
+
+	/* Wait for completion */
+	ret = readl_poll_timeout(&regs->uma_cts, val,
+				 !(val & UMA_CTS_EXEC_DONE), XFER_TIMEOUT);
+	if (ret) {
+		printf("npcm_fiu: read timeout\n");
+		return ret;
+	}
+
+	/* Copy data from data registers */
+	if (size)
+		data_reg[0] = readl(&regs->uma_dr0);
+	if (size > DW_SIZE)
+		data_reg[1] = readl(&regs->uma_dr1);
+	if (size > DW_SIZE * 2)
+		data_reg[2] = readl(&regs->uma_dr2);
+	if (size > DW_SIZE * 3)
+		data_reg[3] = readl(&regs->uma_dr3);
+	memcpy(buf, data_reg, size);
+
+	return 0;
+}
+
+static int fiu_uma_write(struct udevice *bus, const u8 *buf, u32 size)
+{
+	struct npcm_fiu_priv *priv = dev_get_priv(bus);
+	struct npcm_fiu_regs *regs = priv->regs;
+	u32 data_reg[4];
+	u32 val;
+	int ret;
+
+	/* Set data size */
+	writel(FIELD_PREP(UMA_CFG_WDATSIZ_MASK, size), &regs->uma_cfg);
+
+	/* Write data to data registers */
+	memcpy(data_reg, buf, size);
+	if (size)
+		writel(data_reg[0], &regs->uma_dw0);
+	if (size > DW_SIZE)
+		writel(data_reg[1], &regs->uma_dw1);
+	if (size > DW_SIZE * 2)
+		writel(data_reg[2], &regs->uma_dw2);
+	if (size > DW_SIZE * 3)
+		writel(data_reg[3], &regs->uma_dw3);
+
+	/* Initiate the transaction */
+	writel(readl(&regs->uma_cts) | UMA_CTS_EXEC_DONE, &regs->uma_cts);
+
+	/* Wait for completion */
+	ret = readl_poll_timeout(&regs->uma_cts, val,
+				 !(val & UMA_CTS_EXEC_DONE), XFER_TIMEOUT);
+	if (ret)
+		printf("npcm_fiu: write timeout\n");
+
+	return ret;
+}
+
+static int npcm_fiu_spi_xfer(struct udevice *dev, unsigned int bitlen,
+			     const void *dout, void *din, unsigned long flags)
+{
+	struct udevice *bus = dev->parent;
+	struct npcm_fiu_priv *priv = dev_get_priv(bus);
+	struct npcm_fiu_regs *regs = priv->regs;
+	struct dm_spi_slave_plat *slave_plat =
+			dev_get_parent_plat(dev);
+	const u8 *tx = dout;
+	u8 *rx = din;
+	int bytes = bitlen / 8;
+	int ret = 0;
+	int len;
+
+	if (flags & SPI_XFER_BEGIN)
+		activate_cs(regs, slave_plat->cs);
+
+	while (bytes) {
+		len = (bytes > CHUNK_SIZE) ? CHUNK_SIZE : bytes;
+		if (tx) {
+			ret = fiu_uma_write(bus, tx, len);
+			if (ret)
+				break;
+			tx += len;
+		} else {
+			ret = fiu_uma_read(bus, rx, len);
+			if (ret)
+				break;
+			rx += len;
+		}
+		bytes -= len;
+	}
+
+	if (flags & SPI_XFER_END)
+		deactivate_cs(regs, slave_plat->cs);
+
+	return ret;
+}
+
+static int npcm_fiu_uma_operation(struct npcm_fiu_priv *priv, const struct spi_mem_op *op,
+				  u32 addr, const u8 *tx, u8 *rx, u32 nbytes, bool started)
+{
+	struct npcm_fiu_regs *regs = priv->regs;
+	u32 uma_cfg = 0, val;
+	u32 data_reg[4];
+	int ret;
+
+	debug("fiu_uma: opcode 0x%x, dir %d, addr 0x%x, %d bytes\n",
+	      op->cmd.opcode, op->data.dir, addr, nbytes);
+	debug("         buswidth cmd:%d, addr:%d, dummy:%d, data:%d\n",
+	      op->cmd.buswidth, op->addr.buswidth, op->dummy.buswidth,
+	      op->data.buswidth);
+	debug("         size cmd:%d, addr:%d, dummy:%d, data:%d\n",
+	      1, op->addr.nbytes, op->dummy.nbytes, op->data.nbytes);
+	debug("         tx %p, rx %p\n", tx, rx);
+
+	if (!started) {
+		/* Send cmd/addr in the begin of an transaction */
+		writel(op->cmd.opcode, &regs->uma_cmd);
+
+		uma_cfg |= FIELD_PREP(UMA_CFG_CMBPCK_MASK, ilog2(op->cmd.buswidth)) |
+			   (1 << UMA_CFG_CMDSIZ_SHIFT);
+		/* Configure addr bytes */
+		if (op->addr.nbytes) {
+			uma_cfg |= FIELD_PREP(UMA_CFG_ADBPCK_MASK, ilog2(op->addr.buswidth)) |
+				   FIELD_PREP(UMA_CFG_ADDSIZ_MASK, op->addr.nbytes);
+			writel(addr, &regs->uma_addr);
+		}
+		/* Configure dummy bytes */
+		if (op->dummy.nbytes)
+			uma_cfg |= FIELD_PREP(UMA_CFG_DBPCK_MASK, ilog2(op->dummy.buswidth)) |
+				   FIELD_PREP(UMA_CFG_DBSIZ_MASK, op->dummy.nbytes);
+	}
+	/* Set data bus width and data size */
+	if (op->data.dir == SPI_MEM_DATA_IN && nbytes)
+		uma_cfg |= FIELD_PREP(UMA_CFG_RDBPCK_MASK, ilog2(op->data.buswidth)) |
+			   FIELD_PREP(UMA_CFG_RDATSIZ_MASK, nbytes);
+	else if (op->data.dir == SPI_MEM_DATA_OUT && nbytes)
+		uma_cfg |= FIELD_PREP(UMA_CFG_WDBPCK_MASK, ilog2(op->data.buswidth)) |
+			   FIELD_PREP(UMA_CFG_WDATSIZ_MASK, nbytes);
+	writel(uma_cfg, &regs->uma_cfg);
+
+	if (op->data.dir == SPI_MEM_DATA_OUT && nbytes) {
+		memcpy(data_reg, tx, nbytes);
+
+		if (nbytes)
+			writel(data_reg[0], &regs->uma_dw0);
+		if (nbytes > DW_SIZE)
+			writel(data_reg[1], &regs->uma_dw1);
+		if (nbytes > DW_SIZE * 2)
+			writel(data_reg[2], &regs->uma_dw2);
+		if (nbytes > DW_SIZE * 3)
+			writel(data_reg[3], &regs->uma_dw3);
+	}
+	/* Initiate the transaction */
+	writel(readl(&regs->uma_cts) | UMA_CTS_EXEC_DONE, &regs->uma_cts);
+
+	/* Wait for completion */
+	ret = readl_poll_timeout(&regs->uma_cts, val,
+				 !(val & UMA_CTS_EXEC_DONE), XFER_TIMEOUT);
+	if (ret) {
+		printf("npcm_fiu: UMA op timeout\n");
+		return ret;
+	}
+
+	if (op->data.dir == SPI_MEM_DATA_IN && nbytes) {
+		if (nbytes)
+			data_reg[0] = readl(&regs->uma_dr0);
+		if (nbytes > DW_SIZE)
+			data_reg[1] = readl(&regs->uma_dr1);
+		if (nbytes > DW_SIZE * 2)
+			data_reg[2] = readl(&regs->uma_dr2);
+		if (nbytes > DW_SIZE * 3)
+			data_reg[3] = readl(&regs->uma_dr3);
+
+		memcpy(rx, data_reg, nbytes);
+	}
+
+	return 0;
+}
+
+static int npcm_fiu_exec_op(struct spi_slave *slave,
+			    const struct spi_mem_op *op)
+{
+	struct udevice *bus = slave->dev->parent;
+	struct npcm_fiu_priv *priv = dev_get_priv(bus);
+	struct npcm_fiu_regs *regs = priv->regs;
+	struct dm_spi_slave_plat *slave_plat = dev_get_parent_plat(slave->dev);
+	u32 bytes, len, addr;
+	const u8 *tx;
+	u8 *rx;
+	bool started = false;
+	int ret;
+
+	bytes = op->data.nbytes;
+	addr = (u32)op->addr.val;
+	if (!bytes) {
+		activate_cs(regs, slave_plat->cs);
+		ret = npcm_fiu_uma_operation(priv, op, addr, NULL, NULL, 0, false);
+		deactivate_cs(regs, slave_plat->cs);
+		return ret;
+	}
+
+	tx = op->data.buf.out;
+	rx = op->data.buf.in;
+	/*
+	 * Use SW-control CS for write to extend the transaction and
+	 *     keep the Write Enable state.
+	 * Use HW-control CS for read to avoid clock and timing issues.
+	 */
+	if (op->data.dir == SPI_MEM_DATA_OUT)
+		activate_cs(regs, slave_plat->cs);
+	else
+		writel(FIELD_PREP(UMA_CTS_DEV_NUM_MASK, slave_plat->cs) | UMA_CTS_SW_CS,
+		       &regs->uma_cts);
+	while (bytes) {
+		len = (bytes > CHUNK_SIZE) ? CHUNK_SIZE : bytes;
+		ret = npcm_fiu_uma_operation(priv, op, addr, tx, rx, len, started);
+		if (ret)
+			return ret;
+
+		/* CS is kept low for uma write, extend the transaction */
+		if (op->data.dir == SPI_MEM_DATA_OUT)
+			started = true;
+
+		bytes -= len;
+		addr += len;
+		if (tx)
+			tx += len;
+		if (rx)
+			rx += len;
+	}
+	if (op->data.dir == SPI_MEM_DATA_OUT)
+		deactivate_cs(regs, slave_plat->cs);
+
+	return 0;
+}
+
+static int npcm_fiu_spi_probe(struct udevice *bus)
+{
+	struct npcm_fiu_priv *priv = dev_get_priv(bus);
+	int ret;
+
+	priv->regs = (struct npcm_fiu_regs *)dev_read_addr_ptr(bus);
+
+	ret = clk_get_by_index(bus, 0, &priv->clk);
+	if (ret < 0)
+		return ret;
+
+	return 0;
+}
+
+static const struct spi_controller_mem_ops npcm_fiu_mem_ops = {
+	.exec_op = npcm_fiu_exec_op,
+};
+
+static const struct dm_spi_ops npcm_fiu_spi_ops = {
+	.xfer           = npcm_fiu_spi_xfer,
+	.set_speed      = npcm_fiu_spi_set_speed,
+	.set_mode       = npcm_fiu_spi_set_mode,
+	.mem_ops        = &npcm_fiu_mem_ops,
+};
+
+static const struct udevice_id npcm_fiu_spi_ids[] = {
+	{ .compatible = "nuvoton,npcm845-fiu" },
+	{ .compatible = "nuvoton,npcm750-fiu" },
+	{ }
+};
+
+U_BOOT_DRIVER(npcm_fiu_spi) = {
+	.name   = "npcm_fiu_spi",
+	.id     = UCLASS_SPI,
+	.of_match = npcm_fiu_spi_ids,
+	.ops    = &npcm_fiu_spi_ops,
+	.priv_auto = sizeof(struct npcm_fiu_priv),
+	.probe  = npcm_fiu_spi_probe,
+};