1 files changed, 355 insertions, 144 deletions

diff --git a/drivers/spi/spi-nxp-fspi.c b/drivers/spi/spi-nxp-fspi.c
index 1c1991a26c15..50a7e4916a60 100644
--- a/drivers/spi/spi-nxp-fspi.c
+++ b/drivers/spi/spi-nxp-fspi.c
@@ -47,8 +47,9 @@
 #include <linux/module.h>
 #include <linux/mutex.h>
 #include <linux/of.h>
-#include <linux/of_device.h>
 #include <linux/platform_device.h>
+#include <linux/pinctrl/consumer.h>
+#include <linux/pm_runtime.h>
 #include <linux/pm_qos.h>
 #include <linux/regmap.h>
 #include <linux/sizes.h>
@@ -58,12 +59,8 @@
 #include <linux/spi/spi.h>
 #include <linux/spi/spi-mem.h>
 
-/*
- * The driver only uses one single LUT entry, that is updated on
- * each call of exec_op(). Index 0 is preset at boot with a basic
- * read operation, so let's use the last entry (31).
- */
-#define	SEQID_LUT			31
+/* runtime pm timeout */
+#define FSPI_RPM_TIMEOUT 50	/* 50ms */
 
 /* Registers used by the driver */
 #define FSPI_MCR0			0x00
@@ -214,9 +211,15 @@
 
 #define FSPI_DLLACR			0xC0
 #define FSPI_DLLACR_OVRDEN		BIT(8)
+#define FSPI_DLLACR_SLVDLY(x)		((x) << 3)
+#define FSPI_DLLACR_DLLRESET		BIT(1)
+#define FSPI_DLLACR_DLLEN		BIT(0)
 
 #define FSPI_DLLBCR			0xC4
 #define FSPI_DLLBCR_OVRDEN		BIT(8)
+#define FSPI_DLLBCR_SLVDLY(x)		((x) << 3)
+#define FSPI_DLLBCR_DLLRESET		BIT(1)
+#define FSPI_DLLBCR_DLLEN		BIT(0)
 
 #define FSPI_STS0			0xE0
 #define FSPI_STS0_DLPHB(x)		((x) << 8)
@@ -231,6 +234,16 @@
 #define FSPI_STS1_AHB_ERRCD(x)		((x) << 8)
 #define FSPI_STS1_AHB_ERRID(x)		(x)
 
+#define FSPI_STS2			0xE8
+#define FSPI_STS2_BREFLOCK		BIT(17)
+#define FSPI_STS2_BSLVLOCK		BIT(16)
+#define FSPI_STS2_AREFLOCK		BIT(1)
+#define FSPI_STS2_ASLVLOCK		BIT(0)
+#define FSPI_STS2_AB_LOCK		(FSPI_STS2_BREFLOCK | \
+					 FSPI_STS2_BSLVLOCK | \
+					 FSPI_STS2_AREFLOCK | \
+					 FSPI_STS2_ASLVLOCK)
+
 #define FSPI_AHBSPNST			0xEC
 #define FSPI_AHBSPNST_DATLFT(x)		((x) << 16)
 #define FSPI_AHBSPNST_BUFID(x)		((x) << 1)
@@ -248,9 +261,6 @@
 #define FSPI_TFDR			0x180
 
 #define FSPI_LUT_BASE			0x200
-#define FSPI_LUT_OFFSET			(SEQID_LUT * 4 * 4)
-#define FSPI_LUT_REG(idx) \
-	(FSPI_LUT_BASE + FSPI_LUT_OFFSET + (idx) * 4)
 
 /* register map end */
 
@@ -320,12 +330,15 @@
 
 /* Access flash memory using IP bus only */
 #define FSPI_QUIRK_USE_IP_ONLY	BIT(0)
+/* Disable DTR */
+#define FSPI_QUIRK_DISABLE_DTR	BIT(1)
 
 struct nxp_fspi_devtype_data {
 	unsigned int rxfifo;
 	unsigned int txfifo;
 	unsigned int ahb_buf_size;
 	unsigned int quirks;
+	unsigned int lut_num;
 	bool little_endian;
 };
 
@@ -333,7 +346,8 @@ static struct nxp_fspi_devtype_data lx2160a_data = {
 	.rxfifo = SZ_512,       /* (64  * 64 bits)  */
 	.txfifo = SZ_1K,        /* (128 * 64 bits)  */
 	.ahb_buf_size = SZ_2K,  /* (256 * 64 bits)  */
-	.quirks = 0,
+	.quirks = FSPI_QUIRK_DISABLE_DTR,
+	.lut_num = 32,
 	.little_endian = true,  /* little-endian    */
 };
 
@@ -342,6 +356,7 @@ static struct nxp_fspi_devtype_data imx8mm_data = {
 	.txfifo = SZ_1K,        /* (128 * 64 bits)  */
 	.ahb_buf_size = SZ_2K,  /* (256 * 64 bits)  */
 	.quirks = 0,
+	.lut_num = 32,
 	.little_endian = true,  /* little-endian    */
 };
 
@@ -350,6 +365,7 @@ static struct nxp_fspi_devtype_data imx8qxp_data = {
 	.txfifo = SZ_1K,        /* (128 * 64 bits)  */
 	.ahb_buf_size = SZ_2K,  /* (256 * 64 bits)  */
 	.quirks = 0,
+	.lut_num = 32,
 	.little_endian = true,  /* little-endian    */
 };
 
@@ -358,6 +374,16 @@ static struct nxp_fspi_devtype_data imx8dxl_data = {
 	.txfifo = SZ_1K,        /* (128 * 64 bits)  */
 	.ahb_buf_size = SZ_2K,  /* (256 * 64 bits)  */
 	.quirks = FSPI_QUIRK_USE_IP_ONLY,
+	.lut_num = 32,
+	.little_endian = true,  /* little-endian    */
+};
+
+static struct nxp_fspi_devtype_data imx8ulp_data = {
+	.rxfifo = SZ_512,       /* (64  * 64 bits)  */
+	.txfifo = SZ_1K,        /* (128 * 64 bits)  */
+	.ahb_buf_size = SZ_2K,  /* (256 * 64 bits)  */
+	.quirks = 0,
+	.lut_num = 16,
 	.little_endian = true,  /* little-endian    */
 };
 
@@ -375,6 +401,13 @@ struct nxp_fspi {
 	struct mutex lock;
 	struct pm_qos_request pm_qos_req;
 	int selected;
+#define FSPI_NEED_INIT		BIT(0)
+#define FSPI_DTR_MODE		BIT(1)
+	int flags;
+	/* save the previous operation clock rate */
+	unsigned long pre_op_rate;
+	/* the max clock rate fspi output to device */
+	unsigned long max_rate;
 };
 
 static inline int needs_ip_only(struct nxp_fspi *f)
@@ -436,7 +469,7 @@ static int nxp_fspi_check_buswidth(struct nxp_fspi *f, u8 width)
 static bool nxp_fspi_supports_op(struct spi_mem *mem,
 				 const struct spi_mem_op *op)
 {
-	struct nxp_fspi *f = spi_controller_get_devdata(mem->spi->master);
+	struct nxp_fspi *f = spi_controller_get_devdata(mem->spi->controller);
 	int ret;
 
 	ret = nxp_fspi_check_buswidth(f, op->cmd.buswidth);
@@ -505,7 +538,7 @@ static int fspi_readl_poll_tout(struct nxp_fspi *f, void __iomem *base,
 }
 
 /*
- * If the slave device content being changed by Write/Erase, need to
+ * If the target device content being changed by Write/Erase, need to
  * invalidate the AHB buffer. This can be achieved by doing the reset
  * of controller after setting MCR0[SWRESET] bit.
  */
@@ -529,14 +562,25 @@ static void nxp_fspi_prepare_lut(struct nxp_fspi *f,
 	void __iomem *base = f->iobase;
 	u32 lutval[4] = {};
 	int lutidx = 1, i;
+	u32 lut_offset = (f->devtype_data->lut_num - 1) * 4 * 4;
+	u32 target_lut_reg;
 
 	/* cmd */
-	lutval[0] |= LUT_DEF(0, LUT_CMD, LUT_PAD(op->cmd.buswidth),
-			     op->cmd.opcode);
+	if (op->cmd.dtr) {
+		lutval[0] |= LUT_DEF(0, LUT_CMD_DDR, LUT_PAD(op->cmd.buswidth),
+				     op->cmd.opcode >> 8);
+		lutval[lutidx / 2] |= LUT_DEF(lutidx, LUT_CMD_DDR,
+					      LUT_PAD(op->cmd.buswidth),
+					      op->cmd.opcode & 0xFF);
+		lutidx++;
+	} else {
+		lutval[0] |= LUT_DEF(0, LUT_CMD, LUT_PAD(op->cmd.buswidth),
+				     op->cmd.opcode);
+	}
 
 	/* addr bytes */
 	if (op->addr.nbytes) {
-		lutval[lutidx / 2] |= LUT_DEF(lutidx, LUT_ADDR,
+		lutval[lutidx / 2] |= LUT_DEF(lutidx, op->addr.dtr ? LUT_ADDR_DDR : LUT_ADDR,
 					      LUT_PAD(op->addr.buswidth),
 					      op->addr.nbytes * 8);
 		lutidx++;
@@ -544,7 +588,7 @@ static void nxp_fspi_prepare_lut(struct nxp_fspi *f,
 
 	/* dummy bytes, if needed */
 	if (op->dummy.nbytes) {
-		lutval[lutidx / 2] |= LUT_DEF(lutidx, LUT_DUMMY,
+		lutval[lutidx / 2] |= LUT_DEF(lutidx, op->dummy.dtr ? LUT_DUMMY_DDR : LUT_DUMMY,
 		/*
 		 * Due to FlexSPI controller limitation number of PAD for dummy
 		 * buswidth needs to be programmed as equal to data buswidth.
@@ -559,7 +603,8 @@ static void nxp_fspi_prepare_lut(struct nxp_fspi *f,
 	if (op->data.nbytes) {
 		lutval[lutidx / 2] |= LUT_DEF(lutidx,
 					      op->data.dir == SPI_MEM_DATA_IN ?
-					      LUT_NXP_READ : LUT_NXP_WRITE,
+					      (op->data.dtr ? LUT_READ_DDR : LUT_NXP_READ) :
+					      (op->data.dtr ? LUT_WRITE_DDR : LUT_NXP_WRITE),
 					      LUT_PAD(op->data.buswidth),
 					      0);
 		lutidx++;
@@ -573,10 +618,12 @@ static void nxp_fspi_prepare_lut(struct nxp_fspi *f,
 	fspi_writel(f, FSPI_LCKER_UNLOCK, f->iobase + FSPI_LCKCR);
 
 	/* fill LUT */
-	for (i = 0; i < ARRAY_SIZE(lutval); i++)
-		fspi_writel(f, lutval[i], base + FSPI_LUT_REG(i));
+	for (i = 0; i < ARRAY_SIZE(lutval); i++) {
+		target_lut_reg = FSPI_LUT_BASE + lut_offset + i * 4;
+		fspi_writel(f, lutval[i], base + target_lut_reg);
+	}
 
-	dev_dbg(f->dev, "CMD[%x] lutval[0:%x \t 1:%x \t 2:%x \t 3:%x], size: 0x%08x\n",
+	dev_dbg(f->dev, "CMD[%02x] lutval[0:%08x 1:%08x 2:%08x 3:%08x], size: 0x%08x\n",
 		op->cmd.opcode, lutval[0], lutval[1], lutval[2], lutval[3], op->data.nbytes);
 
 	/* lock LUT */
@@ -604,20 +651,103 @@ static int nxp_fspi_clk_prep_enable(struct nxp_fspi *f)
 	return 0;
 }
 
-static int nxp_fspi_clk_disable_unprep(struct nxp_fspi *f)
+static void nxp_fspi_clk_disable_unprep(struct nxp_fspi *f)
 {
 	if (is_acpi_node(dev_fwnode(f->dev)))
-		return 0;
+		return;
 
 	clk_disable_unprepare(f->clk);
 	clk_disable_unprepare(f->clk_en);
 
-	return 0;
+	return;
+}
+
+/*
+ * Sample Clock source selection for Flash Reading
+ * Four modes defined by fspi:
+ * mode 0: Dummy Read strobe generated by FlexSPI Controller
+ *         and loopback internally
+ * mode 1: Dummy Read strobe generated by FlexSPI Controller
+ *         and loopback from DQS pad
+ * mode 2: Reserved
+ * mode 3: Flash provided Read strobe and input from DQS pad
+ *
+ * fspi default use mode 0 after reset
+ */
+static void nxp_fspi_select_rx_sample_clk_source(struct nxp_fspi *f,
+						 bool op_is_dtr)
+{
+	u32 reg;
+
+	/*
+	 * For 8D-8D-8D mode, need to use mode 3 (Flash provided Read
+	 * strobe and input from DQS pad), otherwise read operaton may
+	 * meet issue.
+	 * This mode require flash device connect the DQS pad on board.
+	 * For other modes, still use mode 0, keep align with before.
+	 * spi_nor_suspend will disable 8D-8D-8D mode, also need to
+	 * change the mode back to mode 0.
+	 */
+	reg = fspi_readl(f, f->iobase + FSPI_MCR0);
+	if (op_is_dtr) {
+		reg |= FSPI_MCR0_RXCLKSRC(3);
+		f->max_rate = 166000000;
+	} else {	/*select mode 0 */
+		reg &= ~FSPI_MCR0_RXCLKSRC(3);
+		f->max_rate = 66000000;
+	}
+	fspi_writel(f, reg, f->iobase + FSPI_MCR0);
+}
+
+static void nxp_fspi_dll_calibration(struct nxp_fspi *f)
+{
+	int ret;
+
+	/* Reset the DLL, set the DLLRESET to 1 and then set to 0 */
+	fspi_writel(f, FSPI_DLLACR_DLLRESET, f->iobase + FSPI_DLLACR);
+	fspi_writel(f, FSPI_DLLBCR_DLLRESET, f->iobase + FSPI_DLLBCR);
+	fspi_writel(f, 0, f->iobase + FSPI_DLLACR);
+	fspi_writel(f, 0, f->iobase + FSPI_DLLBCR);
+
+	/*
+	 * Enable the DLL calibration mode.
+	 * The delay target for slave delay line is:
+	 *   ((SLVDLYTARGET+1) * 1/32 * clock cycle of reference clock.
+	 * When clock rate > 100MHz, recommend SLVDLYTARGET is 0xF, which
+	 * means half of clock cycle of reference clock.
+	 */
+	fspi_writel(f, FSPI_DLLACR_DLLEN | FSPI_DLLACR_SLVDLY(0xF),
+		    f->iobase + FSPI_DLLACR);
+	fspi_writel(f, FSPI_DLLBCR_DLLEN | FSPI_DLLBCR_SLVDLY(0xF),
+		    f->iobase + FSPI_DLLBCR);
+
+	/* Wait to get REF/SLV lock */
+	ret = fspi_readl_poll_tout(f, f->iobase + FSPI_STS2, FSPI_STS2_AB_LOCK,
+				   0, POLL_TOUT, true);
+	if (ret)
+		dev_warn(f->dev, "DLL lock failed, please fix it!\n");
+
+	/*
+	 * For ERR050272, DLL lock status bit is not accurate,
+	 * wait for 4us more as a workaround.
+	 */
+	udelay(4);
+}
+
+/*
+ * Config the DLL register to default value, enable the target clock delay
+ * line delay cell override mode, and use 1 fixed delay cell in DLL delay
+ * chain, this is the suggested setting when clock rate < 100MHz.
+ */
+static void nxp_fspi_dll_override(struct nxp_fspi *f)
+{
+	fspi_writel(f, FSPI_DLLACR_OVRDEN, f->iobase + FSPI_DLLACR);
+	fspi_writel(f, FSPI_DLLBCR_OVRDEN, f->iobase + FSPI_DLLBCR);
 }
 
 /*
  * In FlexSPI controller, flash access is based on value of FSPI_FLSHXXCR0
- * register and start base address of the slave device.
+ * register and start base address of the target device.
  *
  *							    (Higher address)
  *				--------    <-- FLSHB2CR0
@@ -636,15 +766,15 @@ static int nxp_fspi_clk_disable_unprep(struct nxp_fspi *f)
  *
  *
  * Start base address defines the starting address range for given CS and
- * FSPI_FLSHXXCR0 defines the size of the slave device connected at given CS.
+ * FSPI_FLSHXXCR0 defines the size of the target device connected at given CS.
  *
  * But, different targets are having different combinations of number of CS,
  * some targets only have single CS or two CS covering controller's full
  * memory mapped space area.
  * Thus, implementation is being done as independent of the size and number
- * of the connected slave device.
+ * of the connected target device.
  * Assign controller memory mapped space size as the size to the connected
- * slave device.
+ * target device.
  * Mark FLSHxxCR0 as zero initially and then assign value only to the selected
  * chip-select Flash configuration register.
  *
@@ -653,17 +783,27 @@ static int nxp_fspi_clk_disable_unprep(struct nxp_fspi *f)
  * Value for rest of the CS FLSHxxCR0 register would be zero.
  *
  */
-static void nxp_fspi_select_mem(struct nxp_fspi *f, struct spi_device *spi)
+static void nxp_fspi_select_mem(struct nxp_fspi *f, struct spi_device *spi,
+				const struct spi_mem_op *op)
 {
-	unsigned long rate = spi->max_speed_hz;
+	/* flexspi only support one DTR mode: 8D-8D-8D */
+	bool op_is_dtr = op->cmd.dtr && op->addr.dtr && op->dummy.dtr && op->data.dtr;
+	unsigned long rate = op->max_freq;
 	int ret;
 	uint64_t size_kb;
 
 	/*
-	 * Return, if previously selected slave device is same as current
-	 * requested slave device.
+	 * Return when following condition all meet,
+	 * 1, if previously selected target device is same as current
+	 *    requested target device.
+	 * 2, the DTR or STR mode do not change.
+	 * 3, previous operation max rate equals current one.
+	 *
+	 * For other case, need to re-config.
 	 */
-	if (f->selected == spi->chip_select)
+	if ((f->selected == spi_get_chipselect(spi, 0)) &&
+	    (!!(f->flags & FSPI_DTR_MODE) == op_is_dtr) &&
+	    (f->pre_op_rate == op->max_freq))
 		return;
 
 	/* Reset FLSHxxCR0 registers */
@@ -676,9 +816,22 @@ static void nxp_fspi_select_mem(struct nxp_fspi *f, struct spi_device *spi)
 	size_kb = FSPI_FLSHXCR0_SZ(f->memmap_phy_size);
 
 	fspi_writel(f, size_kb, f->iobase + FSPI_FLSHA1CR0 +
-		    4 * spi->chip_select);
+		    4 * spi_get_chipselect(spi, 0));
+
+	dev_dbg(f->dev, "Target device [CS:%x] selected\n", spi_get_chipselect(spi, 0));
 
-	dev_dbg(f->dev, "Slave device [CS:%x] selected\n", spi->chip_select);
+	nxp_fspi_select_rx_sample_clk_source(f, op_is_dtr);
+	rate = min(f->max_rate, op->max_freq);
+
+	if (op_is_dtr) {
+		f->flags |= FSPI_DTR_MODE;
+		/* For DTR mode, flexspi will default div 2 and output to device.
+		 * so here to config the root clock to 2 * device rate.
+		 */
+		rate = rate * 2;
+	} else {
+		f->flags &= ~FSPI_DTR_MODE;
+	}
 
 	nxp_fspi_clk_disable_unprep(f);
 
@@ -690,7 +843,18 @@ static void nxp_fspi_select_mem(struct nxp_fspi *f, struct spi_device *spi)
 	if (ret)
 		return;
 
-	f->selected = spi->chip_select;
+	/*
+	 * If clock rate > 100MHz, then switch from DLL override mode to
+	 * DLL calibration mode.
+	 */
+	if (rate > 100000000)
+		nxp_fspi_dll_calibration(f);
+	else
+		nxp_fspi_dll_override(f);
+
+	f->pre_op_rate = op->max_freq;
+
+	f->selected = spi_get_chipselect(spi, 0);
 }
 
 static int nxp_fspi_read_ahb(struct nxp_fspi *f, const struct spi_mem_op *op)
@@ -705,10 +869,9 @@ static int nxp_fspi_read_ahb(struct nxp_fspi *f, const struct spi_mem_op *op)
 			iounmap(f->ahb_addr);
 
 		f->memmap_start = start;
-		f->memmap_len = len > NXP_FSPI_MIN_IOMAP ?
-				len : NXP_FSPI_MIN_IOMAP;
+		f->memmap_len = max_t(u32, len, NXP_FSPI_MIN_IOMAP);
 
-		f->ahb_addr = ioremap_wc(f->memmap_phy + f->memmap_start,
+		f->ahb_addr = ioremap(f->memmap_phy + f->memmap_start,
 					 f->memmap_len);
 
 		if (!f->ahb_addr) {
@@ -754,14 +917,15 @@ static void nxp_fspi_fill_txfifo(struct nxp_fspi *f,
 	if (i < op->data.nbytes) {
 		u32 data = 0;
 		int j;
+		int remaining = op->data.nbytes - i;
 		/* Wait for TXFIFO empty */
 		ret = fspi_readl_poll_tout(f, f->iobase + FSPI_INTR,
 					   FSPI_INTR_IPTXWE, 0,
 					   POLL_TOUT, true);
 		WARN_ON(ret);
 
-		for (j = 0; j < ALIGN(op->data.nbytes - i, 4); j += 4) {
-			memcpy(&data, buf + i + j, 4);
+		for (j = 0; j < ALIGN(remaining, 4); j += 4) {
+			memcpy(&data, buf + i + j, min_t(int, 4, remaining - j));
 			fspi_writel(f, data, base + FSPI_TFDR + j);
 		}
 		fspi_writel(f, FSPI_INTR_IPTXWE, base + FSPI_INTR);
@@ -824,7 +988,7 @@ static int nxp_fspi_do_op(struct nxp_fspi *f, const struct spi_mem_op *op)
 	void __iomem *base = f->iobase;
 	int seqnum = 0;
 	int err = 0;
-	u32 reg;
+	u32 reg, seqid_lut;
 
 	reg = fspi_readl(f, base + FSPI_IPRXFCR);
 	/* invalid RXFIFO first */
@@ -840,8 +1004,9 @@ static int nxp_fspi_do_op(struct nxp_fspi *f, const struct spi_mem_op *op)
 	 * the LUT at each exec_op() call. And also specify the DATA
 	 * length, since it's has not been specified in the LUT.
 	 */
+	seqid_lut = f->devtype_data->lut_num - 1;
 	fspi_writel(f, op->data.nbytes |
-		 (SEQID_LUT << FSPI_IPCR1_SEQID_SHIFT) |
+		 (seqid_lut << FSPI_IPCR1_SEQID_SHIFT) |
 		 (seqnum << FSPI_IPCR1_SEQNUM_SHIFT),
 		 base + FSPI_IPCR1);
 
@@ -861,17 +1026,23 @@ static int nxp_fspi_do_op(struct nxp_fspi *f, const struct spi_mem_op *op)
 
 static int nxp_fspi_exec_op(struct spi_mem *mem, const struct spi_mem_op *op)
 {
-	struct nxp_fspi *f = spi_controller_get_devdata(mem->spi->master);
+	struct nxp_fspi *f = spi_controller_get_devdata(mem->spi->controller);
 	int err = 0;
 
-	mutex_lock(&f->lock);
+	guard(mutex)(&f->lock);
+
+	err = pm_runtime_get_sync(f->dev);
+	if (err < 0) {
+		dev_err(f->dev, "Failed to enable clock %d\n", __LINE__);
+		return err;
+	}
 
 	/* Wait for controller being ready. */
 	err = fspi_readl_poll_tout(f, f->iobase + FSPI_STS0,
 				   FSPI_STS0_ARB_IDLE, 1, POLL_TOUT, true);
 	WARN_ON(err);
 
-	nxp_fspi_select_mem(f, mem->spi);
+	nxp_fspi_select_mem(f, mem->spi, op);
 
 	nxp_fspi_prepare_lut(f, op);
 	/*
@@ -894,14 +1065,14 @@ static int nxp_fspi_exec_op(struct spi_mem *mem, const struct spi_mem_op *op)
 	/* Invalidate the data in the AHB buffer. */
 	nxp_fspi_invalid(f);
 
-	mutex_unlock(&f->lock);
+	pm_runtime_put_autosuspend(f->dev);
 
 	return err;
 }
 
 static int nxp_fspi_adjust_op_size(struct spi_mem *mem, struct spi_mem_op *op)
 {
-	struct nxp_fspi *f = spi_controller_get_devdata(mem->spi->master);
+	struct nxp_fspi *f = spi_controller_get_devdata(mem->spi->controller);
 
 	if (op->data.dir == SPI_MEM_DATA_OUT) {
 		if (op->data.nbytes > f->devtype_data->txfifo)
@@ -965,7 +1136,7 @@ static int nxp_fspi_default_setup(struct nxp_fspi *f)
 {
 	void __iomem *base = f->iobase;
 	int ret, i;
-	u32 reg;
+	u32 reg, seqid_lut;
 
 	/* disable and unprepare clock to avoid glitch pass to controller */
 	nxp_fspi_clk_disable_unprep(f);
@@ -997,9 +1168,7 @@ static int nxp_fspi_default_setup(struct nxp_fspi *f)
 	/* Disable the module */
 	fspi_writel(f, FSPI_MCR0_MDIS, base + FSPI_MCR0);
 
-	/* Reset the DLL register to default value */
-	fspi_writel(f, FSPI_DLLACR_OVRDEN, base + FSPI_DLLACR);
-	fspi_writel(f, FSPI_DLLBCR_OVRDEN, base + FSPI_DLLBCR);
+	nxp_fspi_dll_override(f);
 
 	/* enable module */
 	fspi_writel(f, FSPI_MCR0_AHB_TIMEOUT(0xFF) |
@@ -1007,7 +1176,7 @@ static int nxp_fspi_default_setup(struct nxp_fspi *f)
 		    base + FSPI_MCR0);
 
 	/*
-	 * Disable same device enable bit and configure all slave devices
+	 * Disable same device enable bit and configure all target devices
 	 * independently.
 	 */
 	reg = fspi_readl(f, f->iobase + FSPI_MCR2);
@@ -1029,11 +1198,24 @@ static int nxp_fspi_default_setup(struct nxp_fspi *f)
 	fspi_writel(f, FSPI_AHBCR_PREF_EN | FSPI_AHBCR_RDADDROPT,
 		 base + FSPI_AHBCR);
 
+	/* Reset the FLSHxCR1 registers. */
+	reg = FSPI_FLSHXCR1_TCSH(0x3) | FSPI_FLSHXCR1_TCSS(0x3);
+	fspi_writel(f, reg, base + FSPI_FLSHA1CR1);
+	fspi_writel(f, reg, base + FSPI_FLSHA2CR1);
+	fspi_writel(f, reg, base + FSPI_FLSHB1CR1);
+	fspi_writel(f, reg, base + FSPI_FLSHB2CR1);
+
+	/*
+	 * The driver only uses one single LUT entry, that is updated on
+	 * each call of exec_op(). Index 0 is preset at boot with a basic
+	 * read operation, so let's use the last entry.
+	 */
+	seqid_lut = f->devtype_data->lut_num - 1;
 	/* AHB Read - Set lut sequence ID for all CS. */
-	fspi_writel(f, SEQID_LUT, base + FSPI_FLSHA1CR2);
-	fspi_writel(f, SEQID_LUT, base + FSPI_FLSHA2CR2);
-	fspi_writel(f, SEQID_LUT, base + FSPI_FLSHB1CR2);
-	fspi_writel(f, SEQID_LUT, base + FSPI_FLSHB2CR2);
+	fspi_writel(f, seqid_lut, base + FSPI_FLSHA1CR2);
+	fspi_writel(f, seqid_lut, base + FSPI_FLSHA2CR2);
+	fspi_writel(f, seqid_lut, base + FSPI_FLSHB1CR2);
+	fspi_writel(f, seqid_lut, base + FSPI_FLSHB2CR2);
 
 	f->selected = -1;
 
@@ -1045,7 +1227,7 @@ static int nxp_fspi_default_setup(struct nxp_fspi *f)
 
 static const char *nxp_fspi_get_name(struct spi_mem *mem)
 {
-	struct nxp_fspi *f = spi_controller_get_devdata(mem->spi->master);
+	struct nxp_fspi *f = spi_controller_get_devdata(mem->spi->controller);
 	struct device *dev = &mem->spi->dev;
 	const char *name;
 
@@ -1055,7 +1237,7 @@ static const char *nxp_fspi_get_name(struct spi_mem *mem)
 
 	name = devm_kasprintf(dev, GFP_KERNEL,
 			      "%s-%d", dev_name(f->dev),
-			      mem->spi->chip_select);
+			      spi_get_chipselect(mem->spi, 0));
 
 	if (!name) {
 		dev_err(dev, "failed to get memory for custom flash name\n");
@@ -1072,17 +1254,46 @@ static const struct spi_controller_mem_ops nxp_fspi_mem_ops = {
 	.get_name = nxp_fspi_get_name,
 };
 
+static const struct spi_controller_mem_caps nxp_fspi_mem_caps = {
+	.dtr = true,
+	.swap16 = false,
+	.per_op_freq = true,
+};
+
+static const struct spi_controller_mem_caps nxp_fspi_mem_caps_disable_dtr = {
+	.dtr = false,
+	.per_op_freq = true,
+};
+
+static void nxp_fspi_cleanup(void *data)
+{
+	struct nxp_fspi *f = data;
+
+	/* enable clock first since there is register access */
+	pm_runtime_get_sync(f->dev);
+
+	/* disable the hardware */
+	fspi_writel(f, FSPI_MCR0_MDIS, f->iobase + FSPI_MCR0);
+
+	pm_runtime_disable(f->dev);
+	pm_runtime_put_noidle(f->dev);
+	nxp_fspi_clk_disable_unprep(f);
+
+	if (f->ahb_addr)
+		iounmap(f->ahb_addr);
+}
+
 static int nxp_fspi_probe(struct platform_device *pdev)
 {
 	struct spi_controller *ctlr;
 	struct device *dev = &pdev->dev;
-	struct device_node *np = dev->of_node;
+	struct fwnode_handle *fwnode = dev_fwnode(dev);
 	struct resource *res;
 	struct nxp_fspi *f;
-	int ret;
+	int ret, irq;
 	u32 reg;
 
-	ctlr = spi_alloc_master(&pdev->dev, sizeof(*f));
+	ctlr = devm_spi_alloc_host(&pdev->dev, sizeof(*f));
 	if (!ctlr)
 		return -ENOMEM;
 
@@ -1092,146 +1303,152 @@ static int nxp_fspi_probe(struct platform_device *pdev)
 	f = spi_controller_get_devdata(ctlr);
 	f->dev = dev;
 	f->devtype_data = (struct nxp_fspi_devtype_data *)device_get_match_data(dev);
-	if (!f->devtype_data) {
-		ret = -ENODEV;
-		goto err_put_ctrl;
-	}
+	if (!f->devtype_data)
+		return -ENODEV;
 
 	platform_set_drvdata(pdev, f);
 
 	/* find the resources - configuration register address space */
-	if (is_acpi_node(dev_fwnode(f->dev)))
-		res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	if (is_acpi_node(fwnode))
+		f->iobase = devm_platform_ioremap_resource(pdev, 0);
 	else
-		res = platform_get_resource_byname(pdev,
-				IORESOURCE_MEM, "fspi_base");
-
-	f->iobase = devm_ioremap_resource(dev, res);
-	if (IS_ERR(f->iobase)) {
-		ret = PTR_ERR(f->iobase);
-		goto err_put_ctrl;
-	}
+		f->iobase = devm_platform_ioremap_resource_byname(pdev, "fspi_base");
+	if (IS_ERR(f->iobase))
+		return PTR_ERR(f->iobase);
 
 	/* find the resources - controller memory mapped space */
-	if (is_acpi_node(dev_fwnode(f->dev)))
+	if (is_acpi_node(fwnode))
 		res = platform_get_resource(pdev, IORESOURCE_MEM, 1);
 	else
 		res = platform_get_resource_byname(pdev,
 				IORESOURCE_MEM, "fspi_mmap");
-
-	if (!res) {
-		ret = -ENODEV;
-		goto err_put_ctrl;
-	}
+	if (!res)
+		return -ENODEV;
 
 	/* assign memory mapped starting address and mapped size. */
 	f->memmap_phy = res->start;
 	f->memmap_phy_size = resource_size(res);
 
 	/* find the clocks */
-	if (dev_of_node(&pdev->dev)) {
+	if (is_of_node(fwnode)) {
 		f->clk_en = devm_clk_get(dev, "fspi_en");
-		if (IS_ERR(f->clk_en)) {
-			ret = PTR_ERR(f->clk_en);
-			goto err_put_ctrl;
-		}
+		if (IS_ERR(f->clk_en))
+			return PTR_ERR(f->clk_en);
 
 		f->clk = devm_clk_get(dev, "fspi");
-		if (IS_ERR(f->clk)) {
-			ret = PTR_ERR(f->clk);
-			goto err_put_ctrl;
-		}
-
-		ret = nxp_fspi_clk_prep_enable(f);
-		if (ret) {
-			dev_err(dev, "can not enable the clock\n");
-			goto err_put_ctrl;
-		}
+		if (IS_ERR(f->clk))
+			return PTR_ERR(f->clk);
 	}
 
+	/* find the irq */
+	irq = platform_get_irq(pdev, 0);
+	if (irq < 0)
+		return dev_err_probe(dev, irq, "Failed to get irq source");
+
+	pm_runtime_enable(dev);
+	pm_runtime_set_autosuspend_delay(dev, FSPI_RPM_TIMEOUT);
+	pm_runtime_use_autosuspend(dev);
+
+	/* enable clock */
+	ret = pm_runtime_get_sync(f->dev);
+	if (ret < 0)
+		return dev_err_probe(dev, ret, "Failed to enable clock");
+
 	/* Clear potential interrupts */
 	reg = fspi_readl(f, f->iobase + FSPI_INTR);
 	if (reg)
 		fspi_writel(f, reg, f->iobase + FSPI_INTR);
 
-	/* find the irq */
-	ret = platform_get_irq(pdev, 0);
+	nxp_fspi_default_setup(f);
+
+	ret = pm_runtime_put_sync(dev);
 	if (ret < 0)
-		goto err_disable_clk;
+		return dev_err_probe(dev, ret, "Failed to disable clock");
 
-	ret = devm_request_irq(dev, ret,
+	ret = devm_request_irq(dev, irq,
 			nxp_fspi_irq_handler, 0, pdev->name, f);
-	if (ret) {
-		dev_err(dev, "failed to request irq: %d\n", ret);
-		goto err_disable_clk;
-	}
+	if (ret)
+		return dev_err_probe(dev, ret, "Failed to request irq\n");
 
-	mutex_init(&f->lock);
+	ret = devm_mutex_init(dev, &f->lock);
+	if (ret)
+		return dev_err_probe(dev, ret, "Failed to initialize lock\n");
 
 	ctlr->bus_num = -1;
 	ctlr->num_chipselect = NXP_FSPI_MAX_CHIPSELECT;
 	ctlr->mem_ops = &nxp_fspi_mem_ops;
 
-	nxp_fspi_default_setup(f);
+	if (f->devtype_data->quirks & FSPI_QUIRK_DISABLE_DTR)
+		ctlr->mem_caps = &nxp_fspi_mem_caps_disable_dtr;
+	else
+		ctlr->mem_caps = &nxp_fspi_mem_caps;
 
-	ctlr->dev.of_node = np;
+	device_set_node(&ctlr->dev, fwnode);
 
-	ret = devm_spi_register_controller(&pdev->dev, ctlr);
+	ret = devm_add_action_or_reset(dev, nxp_fspi_cleanup, f);
 	if (ret)
-		goto err_destroy_mutex;
+		return ret;
 
-	return 0;
+	return devm_spi_register_controller(&pdev->dev, ctlr);
+}
 
-err_destroy_mutex:
-	mutex_destroy(&f->lock);
+static int nxp_fspi_runtime_suspend(struct device *dev)
+{
+	struct nxp_fspi *f = dev_get_drvdata(dev);
 
-err_disable_clk:
 	nxp_fspi_clk_disable_unprep(f);
 
-err_put_ctrl:
-	spi_controller_put(ctlr);
-
-	dev_err(dev, "NXP FSPI probe failed\n");
-	return ret;
+	return 0;
 }
 
-static int nxp_fspi_remove(struct platform_device *pdev)
+static int nxp_fspi_runtime_resume(struct device *dev)
 {
-	struct nxp_fspi *f = platform_get_drvdata(pdev);
-
-	/* disable the hardware */
-	fspi_writel(f, FSPI_MCR0_MDIS, f->iobase + FSPI_MCR0);
-
-	nxp_fspi_clk_disable_unprep(f);
+	struct nxp_fspi *f = dev_get_drvdata(dev);
+	int ret;
 
-	mutex_destroy(&f->lock);
+	ret = nxp_fspi_clk_prep_enable(f);
+	if (ret)
+		return ret;
 
-	if (f->ahb_addr)
-		iounmap(f->ahb_addr);
+	if (f->flags & FSPI_NEED_INIT) {
+		nxp_fspi_default_setup(f);
+		ret = pinctrl_pm_select_default_state(dev);
+		if (ret)
+			dev_err(dev, "select flexspi default pinctrl failed!\n");
+		f->flags &= ~FSPI_NEED_INIT;
+	}
 
-	return 0;
+	return ret;
 }
 
 static int nxp_fspi_suspend(struct device *dev)
 {
-	return 0;
-}
-
-static int nxp_fspi_resume(struct device *dev)
-{
 	struct nxp_fspi *f = dev_get_drvdata(dev);
+	int ret;
 
-	nxp_fspi_default_setup(f);
+	ret = pinctrl_pm_select_sleep_state(dev);
+	if (ret) {
+		dev_err(dev, "select flexspi sleep pinctrl failed!\n");
+		return ret;
+	}
 
-	return 0;
+	f->flags |= FSPI_NEED_INIT;
+
+	return pm_runtime_force_suspend(dev);
 }
 
+static const struct dev_pm_ops nxp_fspi_pm_ops = {
+	RUNTIME_PM_OPS(nxp_fspi_runtime_suspend, nxp_fspi_runtime_resume, NULL)
+	SYSTEM_SLEEP_PM_OPS(nxp_fspi_suspend, pm_runtime_force_resume)
+};
+
 static const struct of_device_id nxp_fspi_dt_ids[] = {
 	{ .compatible = "nxp,lx2160a-fspi", .data = (void *)&lx2160a_data, },
 	{ .compatible = "nxp,imx8mm-fspi", .data = (void *)&imx8mm_data, },
 	{ .compatible = "nxp,imx8mp-fspi", .data = (void *)&imx8mm_data, },
 	{ .compatible = "nxp,imx8qxp-fspi", .data = (void *)&imx8qxp_data, },
 	{ .compatible = "nxp,imx8dxl-fspi", .data = (void *)&imx8dxl_data, },
+	{ .compatible = "nxp,imx8ulp-fspi", .data = (void *)&imx8ulp_data, },
 	{ /* sentinel */ }
 };
 MODULE_DEVICE_TABLE(of, nxp_fspi_dt_ids);
@@ -1244,20 +1461,14 @@ static const struct acpi_device_id nxp_fspi_acpi_ids[] = {
 MODULE_DEVICE_TABLE(acpi, nxp_fspi_acpi_ids);
 #endif
 
-static const struct dev_pm_ops nxp_fspi_pm_ops = {
-	.suspend	= nxp_fspi_suspend,
-	.resume		= nxp_fspi_resume,
-};
-
 static struct platform_driver nxp_fspi_driver = {
 	.driver = {
 		.name	= "nxp-fspi",
 		.of_match_table = nxp_fspi_dt_ids,
 		.acpi_match_table = ACPI_PTR(nxp_fspi_acpi_ids),
-		.pm =   &nxp_fspi_pm_ops,
+		.pm = pm_ptr(&nxp_fspi_pm_ops),
 	},
 	.probe          = nxp_fspi_probe,
-	.remove		= nxp_fspi_remove,
 };
 module_platform_driver(nxp_fspi_driver);