Merge tag 'spi-v4.19' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi

Pull spi updates from Mark Brown:
 "Quite an active release for the SPI subsystem, lots of small updates
  and fixes scattered about with highlights including:

   - 3-wire support in the GPIO driver.

   - support for setting a custom memory name in the memory mapped flash
     drivers.

   - support for extended mode in the Freescale DSPI controller.

   - support for the non-standard integration with the Microsemi Ocelot
     platform in the DesignWare driver.

   - new driver for the SocioNext UniPhier"

* tag 'spi-v4.19' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi: (47 commits)
  spi: davinci: fix a NULL pointer dereference
  spi: spi-mem: Constify spi_mem->name
  mtd: m25p80: Call spi_mem_get_name() to let controller set a custom name
  spi: spi-mem: Extend the SPI mem interface to set a custom memory name
  spi: spi-mem: Fix a typo in the documentation of struct spi_mem
  spi: uniphier: remove unnecessary include headers
  spi: spi-gpio: add SPI_3WIRE support
  spi: add flags parameter to txrx_word function pointers
  spi: add SPI controller driver for UniPhier SoC
  spi: add DT bindings for UniPhier SPI controller
  spi: dw: document Microsemi integration
  spi: img-spfi: Set device select bits for SPFI port state
  spi: omap2-mcspi: remove several redundant variables
  spi: dw-mmio: add MSCC Ocelot support
  spi: dw: export dw_spi_set_cs
  spi: spi-fsl-espi: Log fifo counters on error
  spi: imx: Use the longuest possible burst size when in dynamic_burst
  spi: imx: remove unnecessary check in spi_imx_can_dma
  spi: imx: Use correct number of bytes per words
  spi: imx: Use dynamic bursts only when bits_per_word is 8, 16 or 32
  ...

25 files changed, 1210 insertions, 415 deletions

diff --git a/drivers/mtd/devices/m25p80.c b/drivers/mtd/devices/m25p80.c
index fe260ccb2d7d..cbfafc453274 100644
--- a/drivers/mtd/devices/m25p80.c
+++ b/drivers/mtd/devices/m25p80.c
@@ -199,6 +199,9 @@ static int m25p_probe(struct spi_mem *spimem)
 	if (data && data->name)
 		nor->mtd.name = data->name;
 
+	if (!nor->mtd.name)
+		nor->mtd.name = spi_mem_get_name(spimem);
+
 	/* For some (historical?) reason many platforms provide two different
 	 * names in flash_platform_data: "name" and "type". Quite often name is
 	 * set to "m25p80" and then "type" provides a real chip name.
diff --git a/drivers/spi/Kconfig b/drivers/spi/Kconfig
index ad5d68e1dab7..671d078349cc 100644
--- a/drivers/spi/Kconfig
+++ b/drivers/spi/Kconfig
@@ -688,6 +688,19 @@ config SPI_TXX9
 	help
 	  SPI driver for Toshiba TXx9 MIPS SoCs
 
+config SPI_UNIPHIER
+	tristate "Socionext UniPhier SPI Controller"
+	depends on (ARCH_UNIPHIER || COMPILE_TEST) && OF
+	help
+	  This enables a driver for the Socionext UniPhier SoC SCSSI SPI controller.
+
+	  UniPhier SoCs have SCSSI and MCSSI SPI controllers.
+	  Every UniPhier SoC has SCSSI which supports single channel.
+	  Older UniPhier Pro4/Pro5 also has MCSSI which support multiple channels.
+	  This driver supports SCSSI only.
+
+	  If your SoC supports SCSSI, say Y here.
+
 config SPI_XCOMM
 	tristate "Analog Devices AD-FMCOMMS1-EBZ SPI-I2C-bridge driver"
 	depends on I2C
diff --git a/drivers/spi/Makefile b/drivers/spi/Makefile
index cb1f4378b87c..a90d55970036 100644
--- a/drivers/spi/Makefile
+++ b/drivers/spi/Makefile
@@ -101,6 +101,7 @@ spi-thunderx-objs			:= spi-cavium.o spi-cavium-thunderx.o
 obj-$(CONFIG_SPI_THUNDERX)		+= spi-thunderx.o
 obj-$(CONFIG_SPI_TOPCLIFF_PCH)		+= spi-topcliff-pch.o
 obj-$(CONFIG_SPI_TXX9)			+= spi-txx9.o
+obj-$(CONFIG_SPI_UNIPHIER)		+= spi-uniphier.o
 obj-$(CONFIG_SPI_XCOMM)		+= spi-xcomm.o
 obj-$(CONFIG_SPI_XILINX)		+= spi-xilinx.o
 obj-$(CONFIG_SPI_XLP)			+= spi-xlp.o
diff --git a/drivers/spi/spi-ath79.c b/drivers/spi/spi-ath79.c
index 0719bd484891..3f6b657394de 100644
--- a/drivers/spi/spi-ath79.c
+++ b/drivers/spi/spi-ath79.c
@@ -176,7 +176,7 @@ static void ath79_spi_cleanup(struct spi_device *spi)
 }
 
 static u32 ath79_spi_txrx_mode0(struct spi_device *spi, unsigned int nsecs,
-			       u32 word, u8 bits)
+			       u32 word, u8 bits, unsigned flags)
 {
 	struct ath79_spi *sp = ath79_spidev_to_sp(spi);
 	u32 ioc = sp->ioc_base;
diff --git a/drivers/spi/spi-bitbang.c b/drivers/spi/spi-bitbang.c
index 3aa9e6e3dac8..f29176000b8d 100644
--- a/drivers/spi/spi-bitbang.c
+++ b/drivers/spi/spi-bitbang.c
@@ -49,22 +49,26 @@
 struct spi_bitbang_cs {
 	unsigned	nsecs;	/* (clock cycle time)/2 */
 	u32		(*txrx_word)(struct spi_device *spi, unsigned nsecs,
-					u32 word, u8 bits);
+					u32 word, u8 bits, unsigned flags);
 	unsigned	(*txrx_bufs)(struct spi_device *,
 					u32 (*txrx_word)(
 						struct spi_device *spi,
 						unsigned nsecs,
-						u32 word, u8 bits),
-					unsigned, struct spi_transfer *);
+						u32 word, u8 bits,
+						unsigned flags),
+					unsigned, struct spi_transfer *,
+					unsigned);
 };
 
 static unsigned bitbang_txrx_8(
 	struct spi_device	*spi,
 	u32			(*txrx_word)(struct spi_device *spi,
 					unsigned nsecs,
-					u32 word, u8 bits),
+					u32 word, u8 bits,
+					unsigned flags),
 	unsigned		ns,
-	struct spi_transfer	*t
+	struct spi_transfer	*t,
+	unsigned flags
 ) {
 	unsigned		bits = t->bits_per_word;
 	unsigned		count = t->len;
@@ -76,7 +80,7 @@ static unsigned bitbang_txrx_8(
 
 		if (tx)
 			word = *tx++;
-		word = txrx_word(spi, ns, word, bits);
+		word = txrx_word(spi, ns, word, bits, flags);
 		if (rx)
 			*rx++ = word;
 		count -= 1;
@@ -88,9 +92,11 @@ static unsigned bitbang_txrx_16(
 	struct spi_device	*spi,
 	u32			(*txrx_word)(struct spi_device *spi,
 					unsigned nsecs,
-					u32 word, u8 bits),
+					u32 word, u8 bits,
+					unsigned flags),
 	unsigned		ns,
-	struct spi_transfer	*t
+	struct spi_transfer	*t,
+	unsigned flags
 ) {
 	unsigned		bits = t->bits_per_word;
 	unsigned		count = t->len;
@@ -102,7 +108,7 @@ static unsigned bitbang_txrx_16(
 
 		if (tx)
 			word = *tx++;
-		word = txrx_word(spi, ns, word, bits);
+		word = txrx_word(spi, ns, word, bits, flags);
 		if (rx)
 			*rx++ = word;
 		count -= 2;
@@ -114,9 +120,11 @@ static unsigned bitbang_txrx_32(
 	struct spi_device	*spi,
 	u32			(*txrx_word)(struct spi_device *spi,
 					unsigned nsecs,
-					u32 word, u8 bits),
+					u32 word, u8 bits,
+					unsigned flags),
 	unsigned		ns,
-	struct spi_transfer	*t
+	struct spi_transfer	*t,
+	unsigned flags
 ) {
 	unsigned		bits = t->bits_per_word;
 	unsigned		count = t->len;
@@ -128,7 +136,7 @@ static unsigned bitbang_txrx_32(
 
 		if (tx)
 			word = *tx++;
-		word = txrx_word(spi, ns, word, bits);
+		word = txrx_word(spi, ns, word, bits, flags);
 		if (rx)
 			*rx++ = word;
 		count -= 4;
@@ -235,8 +243,24 @@ static int spi_bitbang_bufs(struct spi_device *spi, struct spi_transfer *t)
 {
 	struct spi_bitbang_cs	*cs = spi->controller_state;
 	unsigned		nsecs = cs->nsecs;
+	struct spi_bitbang	*bitbang;
+
+	bitbang = spi_master_get_devdata(spi->master);
+	if (bitbang->set_line_direction) {
+		int err;
 
-	return cs->txrx_bufs(spi, cs->txrx_word, nsecs, t);
+		err = bitbang->set_line_direction(spi, !!(t->tx_buf));
+		if (err < 0)
+			return err;
+	}
+
+	if (spi->mode & SPI_3WIRE) {
+		unsigned flags;
+
+		flags = t->tx_buf ? SPI_MASTER_NO_RX : SPI_MASTER_NO_TX;
+		return cs->txrx_bufs(spi, cs->txrx_word, nsecs, t, flags);
+	}
+	return cs->txrx_bufs(spi, cs->txrx_word, nsecs, t, 0);
 }
 
 /*----------------------------------------------------------------------*/
diff --git a/drivers/spi/spi-butterfly.c b/drivers/spi/spi-butterfly.c
index 22a31e4a1a11..1a3510215841 100644
--- a/drivers/spi/spi-butterfly.c
+++ b/drivers/spi/spi-butterfly.c
@@ -144,9 +144,9 @@ static void butterfly_chipselect(struct spi_device *spi, int value)
 
 static u32
 butterfly_txrx_word_mode0(struct spi_device *spi, unsigned nsecs, u32 word,
-			  u8 bits)
+			  u8 bits, unsigned flags)
 {
-	return bitbang_txrx_be_cpha0(spi, nsecs, 0, 0, word, bits);
+	return bitbang_txrx_be_cpha0(spi, nsecs, 0, flags, word, bits);
 }
 
 /*----------------------------------------------------------------------*/
diff --git a/drivers/spi/spi-cadence.c b/drivers/spi/spi-cadence.c
index f3dad6fcdc35..7c88f74f7f47 100644
--- a/drivers/spi/spi-cadence.c
+++ b/drivers/spi/spi-cadence.c
@@ -319,7 +319,7 @@ static void cdns_spi_fill_tx_fifo(struct cdns_spi *xspi)
 		 */
 		if (cdns_spi_read(xspi, CDNS_SPI_ISR) &
 		    CDNS_SPI_IXR_TXFULL)
-			usleep_range(10, 20);
+			udelay(10);
 
 		if (xspi->txbuf)
 			cdns_spi_write(xspi, CDNS_SPI_TXD, *xspi->txbuf++);
@@ -739,7 +739,7 @@ static int __maybe_unused cnds_runtime_resume(struct device *dev)
 	ret = clk_prepare_enable(xspi->ref_clk);
 	if (ret) {
 		dev_err(dev, "Cannot enable device clock.\n");
-		clk_disable(xspi->pclk);
+		clk_disable_unprepare(xspi->pclk);
 		return ret;
 	}
 	return 0;
diff --git a/drivers/spi/spi-davinci.c b/drivers/spi/spi-davinci.c
index 577084bb911b..a02099c90c5c 100644
--- a/drivers/spi/spi-davinci.c
+++ b/drivers/spi/spi-davinci.c
@@ -217,7 +217,7 @@ static void davinci_spi_chipselect(struct spi_device *spi, int value)
 	pdata = &dspi->pdata;
 
 	/* program delay transfers if tx_delay is non zero */
-	if (spicfg->wdelay)
+	if (spicfg && spicfg->wdelay)
 		spidat1 |= SPIDAT1_WDEL;
 
 	/*
diff --git a/drivers/spi/spi-dw-mmio.c b/drivers/spi/spi-dw-mmio.c
index d25cc4037e23..e80f60ed6fdf 100644
--- a/drivers/spi/spi-dw-mmio.c
+++ b/drivers/spi/spi-dw-mmio.c
@@ -15,11 +15,13 @@
 #include <linux/slab.h>
 #include <linux/spi/spi.h>
 #include <linux/scatterlist.h>
+#include <linux/mfd/syscon.h>
 #include <linux/module.h>
 #include <linux/of.h>
 #include <linux/of_gpio.h>
 #include <linux/of_platform.h>
 #include <linux/property.h>
+#include <linux/regmap.h>
 
 #include "spi-dw.h"
 
@@ -28,10 +30,90 @@
 struct dw_spi_mmio {
 	struct dw_spi  dws;
 	struct clk     *clk;
+	void           *priv;
 };
 
+#define MSCC_CPU_SYSTEM_CTRL_GENERAL_CTRL	0x24
+#define OCELOT_IF_SI_OWNER_MASK			GENMASK(5, 4)
+#define OCELOT_IF_SI_OWNER_OFFSET		4
+#define MSCC_IF_SI_OWNER_SISL			0
+#define MSCC_IF_SI_OWNER_SIBM			1
+#define MSCC_IF_SI_OWNER_SIMC			2
+
+#define MSCC_SPI_MST_SW_MODE			0x14
+#define MSCC_SPI_MST_SW_MODE_SW_PIN_CTRL_MODE	BIT(13)
+#define MSCC_SPI_MST_SW_MODE_SW_SPI_CS(x)	(x << 5)
+
+struct dw_spi_mscc {
+	struct regmap       *syscon;
+	void __iomem        *spi_mst;
+};
+
+/*
+ * The Designware SPI controller (referred to as master in the documentation)
+ * automatically deasserts chip select when the tx fifo is empty. The chip
+ * selects then needs to be either driven as GPIOs or, for the first 4 using the
+ * the SPI boot controller registers. the final chip select is an OR gate
+ * between the Designware SPI controller and the SPI boot controller.
+ */
+static void dw_spi_mscc_set_cs(struct spi_device *spi, bool enable)
+{
+	struct dw_spi *dws = spi_master_get_devdata(spi->master);
+	struct dw_spi_mmio *dwsmmio = container_of(dws, struct dw_spi_mmio, dws);
+	struct dw_spi_mscc *dwsmscc = dwsmmio->priv;
+	u32 cs = spi->chip_select;
+
+	if (cs < 4) {
+		u32 sw_mode = MSCC_SPI_MST_SW_MODE_SW_PIN_CTRL_MODE;
+
+		if (!enable)
+			sw_mode |= MSCC_SPI_MST_SW_MODE_SW_SPI_CS(BIT(cs));
+
+		writel(sw_mode, dwsmscc->spi_mst + MSCC_SPI_MST_SW_MODE);
+	}
+
+	dw_spi_set_cs(spi, enable);
+}
+
+static int dw_spi_mscc_init(struct platform_device *pdev,
+			    struct dw_spi_mmio *dwsmmio)
+{
+	struct dw_spi_mscc *dwsmscc;
+	struct resource *res;
+
+	dwsmscc = devm_kzalloc(&pdev->dev, sizeof(*dwsmscc), GFP_KERNEL);
+	if (!dwsmscc)
+		return -ENOMEM;
+
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 1);
+	dwsmscc->spi_mst = devm_ioremap_resource(&pdev->dev, res);
+	if (IS_ERR(dwsmscc->spi_mst)) {
+		dev_err(&pdev->dev, "SPI_MST region map failed\n");
+		return PTR_ERR(dwsmscc->spi_mst);
+	}
+
+	dwsmscc->syscon = syscon_regmap_lookup_by_compatible("mscc,ocelot-cpu-syscon");
+	if (IS_ERR(dwsmscc->syscon))
+		return PTR_ERR(dwsmscc->syscon);
+
+	/* Deassert all CS */
+	writel(0, dwsmscc->spi_mst + MSCC_SPI_MST_SW_MODE);
+
+	/* Select the owner of the SI interface */
+	regmap_update_bits(dwsmscc->syscon, MSCC_CPU_SYSTEM_CTRL_GENERAL_CTRL,
+			   OCELOT_IF_SI_OWNER_MASK,
+			   MSCC_IF_SI_OWNER_SIMC << OCELOT_IF_SI_OWNER_OFFSET);
+
+	dwsmmio->dws.set_cs = dw_spi_mscc_set_cs;
+	dwsmmio->priv = dwsmscc;
+
+	return 0;
+}
+
 static int dw_spi_mmio_probe(struct platform_device *pdev)
 {
+	int (*init_func)(struct platform_device *pdev,
+			 struct dw_spi_mmio *dwsmmio);
 	struct dw_spi_mmio *dwsmmio;
 	struct dw_spi *dws;
 	struct resource *mem;
@@ -99,6 +181,13 @@ static int dw_spi_mmio_probe(struct platform_device *pdev)
 		}
 	}
 
+	init_func = device_get_match_data(&pdev->dev);
+	if (init_func) {
+		ret = init_func(pdev, dwsmmio);
+		if (ret)
+			goto out;
+	}
+
 	ret = dw_spi_add_host(&pdev->dev, dws);
 	if (ret)
 		goto out;
@@ -123,6 +212,7 @@ static int dw_spi_mmio_remove(struct platform_device *pdev)
 
 static const struct of_device_id dw_spi_mmio_of_match[] = {
 	{ .compatible = "snps,dw-apb-ssi", },
+	{ .compatible = "mscc,ocelot-spi", .data = dw_spi_mscc_init},
 	{ /* end of table */}
 };
 MODULE_DEVICE_TABLE(of, dw_spi_mmio_of_match);
diff --git a/drivers/spi/spi-dw.c b/drivers/spi/spi-dw.c
index f693bfe95ab9..ac2eb89ef7a5 100644
--- a/drivers/spi/spi-dw.c
+++ b/drivers/spi/spi-dw.c
@@ -133,7 +133,7 @@ static inline void dw_spi_debugfs_remove(struct dw_spi *dws)
 }
 #endif /* CONFIG_DEBUG_FS */
 
-static void dw_spi_set_cs(struct spi_device *spi, bool enable)
+void dw_spi_set_cs(struct spi_device *spi, bool enable)
 {
 	struct dw_spi *dws = spi_controller_get_devdata(spi->controller);
 	struct chip_data *chip = spi_get_ctldata(spi);
@@ -145,6 +145,7 @@ static void dw_spi_set_cs(struct spi_device *spi, bool enable)
 	if (!enable)
 		dw_writel(dws, DW_SPI_SER, BIT(spi->chip_select));
 }
+EXPORT_SYMBOL_GPL(dw_spi_set_cs);
 
 /* Return the max entries we can fill into tx fifo */
 static inline u32 tx_max(struct dw_spi *dws)
@@ -485,6 +486,8 @@ int dw_spi_add_host(struct device *dev, struct dw_spi *dws)
 	dws->dma_inited = 0;
 	dws->dma_addr = (dma_addr_t)(dws->paddr + DW_SPI_DR);
 
+	spi_controller_set_devdata(master, dws);
+
 	ret = request_irq(dws->irq, dw_spi_irq, IRQF_SHARED, dev_name(dev),
 			  master);
 	if (ret < 0) {
@@ -505,6 +508,9 @@ int dw_spi_add_host(struct device *dev, struct dw_spi *dws)
 	master->dev.of_node = dev->of_node;
 	master->flags = SPI_MASTER_GPIO_SS;
 
+	if (dws->set_cs)
+		master->set_cs = dws->set_cs;
+
 	/* Basic HW init */
 	spi_hw_init(dev, dws);
 
@@ -518,7 +524,6 @@ int dw_spi_add_host(struct device *dev, struct dw_spi *dws)
 		}
 	}
 
-	spi_controller_set_devdata(master, dws);
 	ret = devm_spi_register_controller(dev, master);
 	if (ret) {
 		dev_err(&master->dev, "problem registering spi master\n");
diff --git a/drivers/spi/spi-dw.h b/drivers/spi/spi-dw.h
index 2cde2473b3e9..0168b08364d5 100644
--- a/drivers/spi/spi-dw.h
+++ b/drivers/spi/spi-dw.h
@@ -112,6 +112,7 @@ struct dw_spi {
 	u32			reg_io_width;	/* DR I/O width in bytes */
 	u16			bus_num;
 	u16			num_cs;		/* supported slave numbers */
+	void (*set_cs)(struct spi_device *spi, bool enable);
 
 	/* Current message transfer state info */
 	size_t			len;
@@ -244,6 +245,7 @@ struct dw_spi_chip {
 	void (*cs_control)(u32 command);
 };
 
+extern void dw_spi_set_cs(struct spi_device *spi, bool enable);
 extern int dw_spi_add_host(struct device *dev, struct dw_spi *dws);
 extern void dw_spi_remove_host(struct dw_spi *dws);
 extern int dw_spi_suspend_host(struct dw_spi *dws);
diff --git a/drivers/spi/spi-fsl-dspi.c b/drivers/spi/spi-fsl-dspi.c
index 0630962ce442..7cb3ab0a35a0 100644
--- a/drivers/spi/spi-fsl-dspi.c
+++ b/drivers/spi/spi-fsl-dspi.c
@@ -1,17 +1,9 @@
-/*
- * drivers/spi/spi-fsl-dspi.c
- *
- * Copyright 2013 Freescale Semiconductor, Inc.
- *
- * Freescale DSPI driver
- * This file contains a driver for the Freescale DSPI
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- */
+// SPDX-License-Identifier: GPL-2.0+
+//
+// Copyright 2013 Freescale Semiconductor, Inc.
+//
+// Freescale DSPI driver
+// This file contains a driver for the Freescale DSPI
 
 #include <linux/clk.h>
 #include <linux/delay.h>
@@ -38,10 +30,6 @@
 
 #define DRIVER_NAME "fsl-dspi"
 
-#define TRAN_STATE_RX_VOID		0x01
-#define TRAN_STATE_TX_VOID		0x02
-#define TRAN_STATE_WORD_ODD_NUM	0x04
-
 #define DSPI_FIFO_SIZE			4
 #define DSPI_DMA_BUFSIZE		(DSPI_FIFO_SIZE * 1024)
 
@@ -50,6 +38,7 @@
 #define SPI_MCR_PCSIS		(0x3F << 16)
 #define SPI_MCR_CLR_TXF	(1 << 11)
 #define SPI_MCR_CLR_RXF	(1 << 10)
+#define SPI_MCR_XSPI		(1 << 3)
 
 #define SPI_TCR			0x08
 #define SPI_TCR_GET_TCNT(x)	(((x) & 0xffff0000) >> 16)
@@ -86,11 +75,16 @@
 #define SPI_RSER_TCFQE		0x80000000
 
 #define SPI_PUSHR		0x34
-#define SPI_PUSHR_CONT		(1 << 31)
-#define SPI_PUSHR_CTAS(x)	(((x) & 0x00000003) << 28)
-#define SPI_PUSHR_EOQ		(1 << 27)
-#define SPI_PUSHR_CTCNT	(1 << 26)
-#define SPI_PUSHR_PCS(x)	(((1 << x) & 0x0000003f) << 16)
+#define SPI_PUSHR_CMD_CONT	(1 << 15)
+#define SPI_PUSHR_CONT		(SPI_PUSHR_CMD_CONT << 16)
+#define SPI_PUSHR_CMD_CTAS(x)	(((x) & 0x0003) << 12)
+#define SPI_PUSHR_CTAS(x)	(SPI_PUSHR_CMD_CTAS(x) << 16)
+#define SPI_PUSHR_CMD_EOQ	(1 << 11)
+#define SPI_PUSHR_EOQ		(SPI_PUSHR_CMD_EOQ << 16)
+#define SPI_PUSHR_CMD_CTCNT	(1 << 10)
+#define SPI_PUSHR_CTCNT		(SPI_PUSHR_CMD_CTCNT << 16)
+#define SPI_PUSHR_CMD_PCS(x)	((1 << x) & 0x003f)
+#define SPI_PUSHR_PCS(x)	(SPI_PUSHR_CMD_PCS(x) << 16)
 #define SPI_PUSHR_TXDATA(x)	((x) & 0x0000ffff)
 
 #define SPI_PUSHR_SLAVE	0x34
@@ -107,21 +101,31 @@
 #define SPI_RXFR2		0x84
 #define SPI_RXFR3		0x88
 
+#define SPI_CTARE(x)		(0x11c + (((x) & 0x3) * 4))
+#define SPI_CTARE_FMSZE(x)	(((x) & 0x1) << 16)
+#define SPI_CTARE_DTCP(x)	((x) & 0x7ff)
+
+#define SPI_SREX		0x13c
+
 #define SPI_FRAME_BITS(bits)	SPI_CTAR_FMSZ((bits) - 1)
 #define SPI_FRAME_BITS_MASK	SPI_CTAR_FMSZ(0xf)
 #define SPI_FRAME_BITS_16	SPI_CTAR_FMSZ(0xf)
 #define SPI_FRAME_BITS_8	SPI_CTAR_FMSZ(0x7)
 
+#define SPI_FRAME_EBITS(bits)	SPI_CTARE_FMSZE(((bits) - 1) >> 4)
+#define SPI_FRAME_EBITS_MASK	SPI_CTARE_FMSZE(1)
+
+/* Register offsets for regmap_pushr */
+#define PUSHR_CMD		0x0
+#define PUSHR_TX		0x2
+
 #define SPI_CS_INIT		0x01
 #define SPI_CS_ASSERT		0x02
 #define SPI_CS_DROP		0x04
 
-#define SPI_TCR_TCNT_MAX	0x10000
-
 #define DMA_COMPLETION_TIMEOUT	msecs_to_jiffies(3000)
 
 struct chip_data {
-	u32 mcr_val;
 	u32 ctar_val;
 	u16 void_write_data;
 };
@@ -135,6 +139,7 @@ enum dspi_trans_mode {
 struct fsl_dspi_devtype_data {
 	enum dspi_trans_mode trans_mode;
 	u8 max_clock_factor;
+	bool xspi_mode;
 };
 
 static const struct fsl_dspi_devtype_data vf610_data = {
@@ -145,6 +150,7 @@ static const struct fsl_dspi_devtype_data vf610_data = {
 static const struct fsl_dspi_devtype_data ls1021a_v1_data = {
 	.trans_mode = DSPI_TCFQ_MODE,
 	.max_clock_factor = 8,
+	.xspi_mode = true,
 };
 
 static const struct fsl_dspi_devtype_data ls2085a_data = {
@@ -179,6 +185,7 @@ struct fsl_dspi {
 	struct platform_device	*pdev;
 
 	struct regmap		*regmap;
+	struct regmap		*regmap_pushr;
 	int			irq;
 	struct clk		*clk;
 
@@ -186,32 +193,62 @@ struct fsl_dspi {
 	struct spi_message	*cur_msg;
 	struct chip_data	*cur_chip;
 	size_t			len;
-	void			*tx;
-	void			*tx_end;
+	const void		*tx;
 	void			*rx;
 	void			*rx_end;
-	char			dataflags;
-	u8			cs;
 	u16			void_write_data;
-	u32			cs_change;
+	u16			tx_cmd;
+	u8			bits_per_word;
+	u8			bytes_per_word;
 	const struct fsl_dspi_devtype_data *devtype_data;
 
 	wait_queue_head_t	waitq;
 	u32			waitflags;
 
-	u32			spi_tcnt;
 	struct fsl_dspi_dma	*dma;
 };
 
-static u32 dspi_data_to_pushr(struct fsl_dspi *dspi, int tx_word);
+static u32 dspi_pop_tx(struct fsl_dspi *dspi)
+{
+	u32 txdata = 0;
+
+	if (dspi->tx) {
+		if (dspi->bytes_per_word == 1)
+			txdata = *(u8 *)dspi->tx;
+		else if (dspi->bytes_per_word == 2)
+			txdata = *(u16 *)dspi->tx;
+		else  /* dspi->bytes_per_word == 4 */
+			txdata = *(u32 *)dspi->tx;
+		dspi->tx += dspi->bytes_per_word;
+	}
+	dspi->len -= dspi->bytes_per_word;
+	return txdata;
+}
 
-static inline int is_double_byte_mode(struct fsl_dspi *dspi)
+static u32 dspi_pop_tx_pushr(struct fsl_dspi *dspi)
 {
-	unsigned int val;
+	u16 cmd = dspi->tx_cmd, data = dspi_pop_tx(dspi);
 
-	regmap_read(dspi->regmap, SPI_CTAR(0), &val);
+	if (dspi->len > 0)
+		cmd |= SPI_PUSHR_CMD_CONT;
+	return cmd << 16 | data;
+}
 
-	return ((val & SPI_FRAME_BITS_MASK) == SPI_FRAME_BITS(8)) ? 0 : 1;
+static void dspi_push_rx(struct fsl_dspi *dspi, u32 rxdata)
+{
+	if (!dspi->rx)
+		return;
+
+	/* Mask of undefined bits */
+	rxdata &= (1 << dspi->bits_per_word) - 1;
+
+	if (dspi->bytes_per_word == 1)
+		*(u8 *)dspi->rx = rxdata;
+	else if (dspi->bytes_per_word == 2)
+		*(u16 *)dspi->rx = rxdata;
+	else /* dspi->bytes_per_word == 4 */
+		*(u32 *)dspi->rx = rxdata;
+	dspi->rx += dspi->bytes_per_word;
 }
 
 static void dspi_tx_dma_callback(void *arg)
@@ -226,19 +263,11 @@ static void dspi_rx_dma_callback(void *arg)
 {
 	struct fsl_dspi *dspi = arg;
 	struct fsl_dspi_dma *dma = dspi->dma;
-	int rx_word;
 	int i;
-	u16 d;
-
-	rx_word = is_double_byte_mode(dspi);
 
-	if (!(dspi->dataflags & TRAN_STATE_RX_VOID)) {
-		for (i = 0; i < dma->curr_xfer_len; i++) {
-			d = dspi->dma->rx_dma_buf[i];
-			rx_word ? (*(u16 *)dspi->rx = d) :
-						(*(u8 *)dspi->rx = d);
-			dspi->rx += rx_word + 1;
-		}
+	if (dspi->rx) {
+		for (i = 0; i < dma->curr_xfer_len; i++)
+			dspi_push_rx(dspi, dspi->dma->rx_dma_buf[i]);
 	}
 
 	complete(&dma->cmd_rx_complete);
@@ -249,16 +278,10 @@ static int dspi_next_xfer_dma_submit(struct fsl_dspi *dspi)
 	struct fsl_dspi_dma *dma = dspi->dma;
 	struct device *dev = &dspi->pdev->dev;
 	int time_left;
-	int tx_word;
 	int i;
 
-	tx_word = is_double_byte_mode(dspi);
-
-	for (i = 0; i < dma->curr_xfer_len; i++) {
-		dspi->dma->tx_dma_buf[i] = dspi_data_to_pushr(dspi, tx_word);
-		if ((dspi->cs_change) && (!dspi->len))
-			dspi->dma->tx_dma_buf[i] &= ~SPI_PUSHR_CONT;
-	}
+	for (i = 0; i < dma->curr_xfer_len; i++)
+		dspi->dma->tx_dma_buf[i] = dspi_pop_tx_pushr(dspi);
 
 	dma->tx_desc = dmaengine_prep_slave_single(dma->chan_tx,
 					dma->tx_dma_phys,
@@ -327,18 +350,17 @@ static int dspi_dma_xfer(struct fsl_dspi *dspi)
 {
 	struct fsl_dspi_dma *dma = dspi->dma;
 	struct device *dev = &dspi->pdev->dev;
+	struct spi_message *message = dspi->cur_msg;
 	int curr_remaining_bytes;
 	int bytes_per_buffer;
-	int word = 1;
 	int ret = 0;
 
-	if (is_double_byte_mode(dspi))
-		word = 2;
 	curr_remaining_bytes = dspi->len;
 	bytes_per_buffer = DSPI_DMA_BUFSIZE / DSPI_FIFO_SIZE;
 	while (curr_remaining_bytes) {
 		/* Check if current transfer fits the DMA buffer */
-		dma->curr_xfer_len = curr_remaining_bytes / word;
+		dma->curr_xfer_len = curr_remaining_bytes
+			/ dspi->bytes_per_word;
 		if (dma->curr_xfer_len > bytes_per_buffer)
 			dma->curr_xfer_len = bytes_per_buffer;
 
@@ -348,7 +370,10 @@ static int dspi_dma_xfer(struct fsl_dspi *dspi)
 			goto exit;
 
 		} else {
-			curr_remaining_bytes -= dma->curr_xfer_len * word;
+			const int len =
+				dma->curr_xfer_len * dspi->bytes_per_word;
+			curr_remaining_bytes -= len;
+			message->actual_length += len;
 			if (curr_remaining_bytes < 0)
 				curr_remaining_bytes = 0;
 		}
@@ -534,125 +559,91 @@ static void ns_delay_scale(char *psc, char *sc, int delay_ns,
 	}
 }
 
-static u32 dspi_data_to_pushr(struct fsl_dspi *dspi, int tx_word)
+static void fifo_write(struct fsl_dspi *dspi)
 {
-	u16 d16;
-
-	if (!(dspi->dataflags & TRAN_STATE_TX_VOID))
-		d16 = tx_word ? *(u16 *)dspi->tx : *(u8 *)dspi->tx;
-	else
-		d16 = dspi->void_write_data;
-
-	dspi->tx += tx_word + 1;
-	dspi->len -= tx_word + 1;
-
-	return	SPI_PUSHR_TXDATA(d16) |
-		SPI_PUSHR_PCS(dspi->cs) |
-		SPI_PUSHR_CTAS(0) |
-		SPI_PUSHR_CONT;
+	regmap_write(dspi->regmap, SPI_PUSHR, dspi_pop_tx_pushr(dspi));
 }
 
-static void dspi_data_from_popr(struct fsl_dspi *dspi, int rx_word)
+static void cmd_fifo_write(struct fsl_dspi *dspi)
 {
-	u16 d;
-	unsigned int val;
-
-	regmap_read(dspi->regmap, SPI_POPR, &val);
-	d = SPI_POPR_RXDATA(val);
+	u16 cmd = dspi->tx_cmd;
 
-	if (!(dspi->dataflags & TRAN_STATE_RX_VOID))
-		rx_word ? (*(u16 *)dspi->rx = d) : (*(u8 *)dspi->rx = d);
-
-	dspi->rx += rx_word + 1;
+	if (dspi->len > 0)
+		cmd |= SPI_PUSHR_CMD_CONT;
+	regmap_write(dspi->regmap_pushr, PUSHR_CMD, cmd);
 }
 
-static int dspi_eoq_write(struct fsl_dspi *dspi)
+static void tx_fifo_write(struct fsl_dspi *dspi, u16 txdata)
 {
-	int tx_count = 0;
-	int tx_word;
-	u32 dspi_pushr = 0;
+	regmap_write(dspi->regmap_pushr, PUSHR_TX, txdata);
+}
 
-	tx_word = is_double_byte_mode(dspi);
+static void dspi_tcfq_write(struct fsl_dspi *dspi)
+{
+	/* Clear transfer count */
+	dspi->tx_cmd |= SPI_PUSHR_CMD_CTCNT;
 
-	while (dspi->len && (tx_count < DSPI_FIFO_SIZE)) {
-		/* If we are in word mode, only have a single byte to transfer
-		 * switch to byte mode temporarily.  Will switch back at the
-		 * end of the transfer.
+	if (dspi->devtype_data->xspi_mode && dspi->bits_per_word > 16) {
+		/* Write two TX FIFO entries first, and then the corresponding
+		 * CMD FIFO entry.
 		 */
-		if (tx_word && (dspi->len == 1)) {
-			dspi->dataflags |= TRAN_STATE_WORD_ODD_NUM;
-			regmap_update_bits(dspi->regmap, SPI_CTAR(0),
-					SPI_FRAME_BITS_MASK, SPI_FRAME_BITS(8));
-			tx_word = 0;
-		}
-
-		dspi_pushr = dspi_data_to_pushr(dspi, tx_word);
-
-		if (dspi->len == 0 || tx_count == DSPI_FIFO_SIZE - 1) {
-			/* last transfer in the transfer */
-			dspi_pushr |= SPI_PUSHR_EOQ;
-			if ((dspi->cs_change) && (!dspi->len))
-				dspi_pushr &= ~SPI_PUSHR_CONT;
-		} else if (tx_word && (dspi->len == 1))
-			dspi_pushr |= SPI_PUSHR_EOQ;
+		u32 data = dspi_pop_tx(dspi);
 
-		regmap_write(dspi->regmap, SPI_PUSHR, dspi_pushr);
-
-		tx_count++;
+		if (dspi->cur_chip->ctar_val & SPI_CTAR_LSBFE(1)) {
+			/* LSB */
+			tx_fifo_write(dspi, data & 0xFFFF);
+			tx_fifo_write(dspi, data >> 16);
+		} else {
+			/* MSB */
+			tx_fifo_write(dspi, data >> 16);
+			tx_fifo_write(dspi, data & 0xFFFF);
+		}
+		cmd_fifo_write(dspi);
+	} else {
+		/* Write one entry to both TX FIFO and CMD FIFO
+		 * simultaneously.
+		 */
+		fifo_write(dspi);
 	}
-
-	return tx_count * (tx_word + 1);
 }
 
-static int dspi_eoq_read(struct fsl_dspi *dspi)
+static u32 fifo_read(struct fsl_dspi *dspi)
 {
-	int rx_count = 0;
-	int rx_word = is_double_byte_mode(dspi);
-
-	while ((dspi->rx < dspi->rx_end)
-			&& (rx_count < DSPI_FIFO_SIZE)) {
-		if (rx_word && (dspi->rx_end - dspi->rx) == 1)
-			rx_word = 0;
+	u32 rxdata = 0;
 
-		dspi_data_from_popr(dspi, rx_word);
-		rx_count++;
-	}
-
-	return rx_count;
+	regmap_read(dspi->regmap, SPI_POPR, &rxdata);
+	return rxdata;
 }
 
-static int dspi_tcfq_write(struct fsl_dspi *dspi)
+static void dspi_tcfq_read(struct fsl_dspi *dspi)
 {
-	int tx_word;
-	u32 dspi_pushr = 0;
-
-	tx_word = is_double_byte_mode(dspi);
+	dspi_push_rx(dspi, fifo_read(dspi));
+}
 
-	if (tx_word && (dspi->len == 1)) {
-		dspi->dataflags |= TRAN_STATE_WORD_ODD_NUM;
-		regmap_update_bits(dspi->regmap, SPI_CTAR(0),
-				SPI_FRAME_BITS_MASK, SPI_FRAME_BITS(8));
-		tx_word = 0;
+static void dspi_eoq_write(struct fsl_dspi *dspi)
+{
+	int fifo_size = DSPI_FIFO_SIZE;
+
+	/* Fill TX FIFO with as many transfers as possible */
+	while (dspi->len && fifo_size--) {
+		/* Request EOQF for last transfer in FIFO */
+		if (dspi->len == dspi->bytes_per_word || fifo_size == 0)
+			dspi->tx_cmd |= SPI_PUSHR_CMD_EOQ;
+		/* Clear transfer count for first transfer in FIFO */
+		if (fifo_size == (DSPI_FIFO_SIZE - 1))
+			dspi->tx_cmd |= SPI_PUSHR_CMD_CTCNT;
+		/* Write combined TX FIFO and CMD FIFO entry */
+		fifo_write(dspi);
 	}
-
-	dspi_pushr = dspi_data_to_pushr(dspi, tx_word);
-
-	if ((dspi->cs_change) && (!dspi->len))
-		dspi_pushr &= ~SPI_PUSHR_CONT;
-
-	regmap_write(dspi->regmap, SPI_PUSHR, dspi_pushr);
-
-	return tx_word + 1;
 }
 
-static void dspi_tcfq_read(struct fsl_dspi *dspi)
+static void dspi_eoq_read(struct fsl_dspi *dspi)
 {
-	int rx_word = is_double_byte_mode(dspi);
-
-	if (rx_word && (dspi->rx_end - dspi->rx) == 1)
-		rx_word = 0;
+	int fifo_size = DSPI_FIFO_SIZE;
 
-	dspi_data_from_popr(dspi, rx_word);
+	/* Read one FIFO entry at and push to rx buffer */
+	while ((dspi->rx < dspi->rx_end) && fifo_size--)
+		dspi_push_rx(dspi, fifo_read(dspi));
 }
 
 static int dspi_transfer_one_message(struct spi_master *master,
@@ -663,10 +654,6 @@ static int dspi_transfer_one_message(struct spi_master *master,
 	struct spi_transfer *transfer;
 	int status = 0;
 	enum dspi_trans_mode trans_mode;
-	u32 spi_tcr;
-
-	regmap_read(dspi->regmap, SPI_TCR, &spi_tcr);
-	dspi->spi_tcnt = SPI_TCR_GET_TCNT(spi_tcr);
 
 	message->actual_length = 0;
 
@@ -674,32 +661,51 @@ static int dspi_transfer_one_message(struct spi_master *master,
 		dspi->cur_transfer = transfer;
 		dspi->cur_msg = message;
 		dspi->cur_chip = spi_get_ctldata(spi);
-		dspi->cs = spi->chip_select;
-		dspi->cs_change = 0;
+		/* Prepare command word for CMD FIFO */
+		dspi->tx_cmd = SPI_PUSHR_CMD_CTAS(0) |
+			SPI_PUSHR_CMD_PCS(spi->chip_select);
 		if (list_is_last(&dspi->cur_transfer->transfer_list,
-				 &dspi->cur_msg->transfers) || transfer->cs_change)
-			dspi->cs_change = 1;
+				 &dspi->cur_msg->transfers)) {
+			/* Leave PCS activated after last transfer when
+			 * cs_change is set.
+			 */
+			if (transfer->cs_change)
+				dspi->tx_cmd |= SPI_PUSHR_CMD_CONT;
+		} else {
+			/* Keep PCS active between transfers in same message
+			 * when cs_change is not set, and de-activate PCS
+			 * between transfers in the same message when
+			 * cs_change is set.
+			 */
+			if (!transfer->cs_change)
+				dspi->tx_cmd |= SPI_PUSHR_CMD_CONT;
+		}
+
 		dspi->void_write_data = dspi->cur_chip->void_write_data;
 
-		dspi->dataflags = 0;
-		dspi->tx = (void *)transfer->tx_buf;
-		dspi->tx_end = dspi->tx + transfer->len;
+		dspi->tx = transfer->tx_buf;
 		dspi->rx = transfer->rx_buf;
 		dspi->rx_end = dspi->rx + transfer->len;
 		dspi->len = transfer->len;
+		/* Validated transfer specific frame size (defaults applied) */
+		dspi->bits_per_word = transfer->bits_per_word;
+		if (transfer->bits_per_word <= 8)
+			dspi->bytes_per_word = 1;
+		else if (transfer->bits_per_word <= 16)
+			dspi->bytes_per_word = 2;
+		else
+			dspi->bytes_per_word = 4;
 
-		if (!dspi->rx)
-			dspi->dataflags |= TRAN_STATE_RX_VOID;
-
-		if (!dspi->tx)
-			dspi->dataflags |= TRAN_STATE_TX_VOID;
-
-		regmap_write(dspi->regmap, SPI_MCR, dspi->cur_chip->mcr_val);
 		regmap_update_bits(dspi->regmap, SPI_MCR,
-				SPI_MCR_CLR_TXF | SPI_MCR_CLR_RXF,
-				SPI_MCR_CLR_TXF | SPI_MCR_CLR_RXF);
+				   SPI_MCR_CLR_TXF | SPI_MCR_CLR_RXF,
+				   SPI_MCR_CLR_TXF | SPI_MCR_CLR_RXF);
 		regmap_write(dspi->regmap, SPI_CTAR(0),
-				dspi->cur_chip->ctar_val);
+			     dspi->cur_chip->ctar_val |
+			     SPI_FRAME_BITS(transfer->bits_per_word));
+		if (dspi->devtype_data->xspi_mode)
+			regmap_write(dspi->regmap, SPI_CTARE(0),
+				     SPI_FRAME_EBITS(transfer->bits_per_word)
+				     | SPI_CTARE_DTCP(1));
 
 		trans_mode = dspi->devtype_data->trans_mode;
 		switch (trans_mode) {
@@ -750,16 +756,9 @@ static int dspi_setup(struct spi_device *spi)
 	struct fsl_dspi_platform_data *pdata;
 	u32 cs_sck_delay = 0, sck_cs_delay = 0;
 	unsigned char br = 0, pbr = 0, pcssck = 0, cssck = 0;
-	unsigned char pasc = 0, asc = 0, fmsz = 0;
+	unsigned char pasc = 0, asc = 0;
 	unsigned long clkrate;
 
-	if ((spi->bits_per_word >= 4) && (spi->bits_per_word <= 16)) {
-		fmsz = spi->bits_per_word - 1;
-	} else {
-		pr_err("Invalid wordsize\n");
-		return -ENODEV;
-	}
-
 	/* Only alloc on first setup */
 	chip = spi_get_ctldata(spi);
 	if (chip == NULL) {
@@ -781,9 +780,6 @@ static int dspi_setup(struct spi_device *spi)
 		sck_cs_delay = pdata->sck_cs_delay;
 	}
 
-	chip->mcr_val = SPI_MCR_MASTER | SPI_MCR_PCSIS |
-		SPI_MCR_CLR_TXF | SPI_MCR_CLR_RXF;
-
 	chip->void_write_data = 0;
 
 	clkrate = clk_get_rate(dspi->clk);
@@ -795,8 +791,7 @@ static int dspi_setup(struct spi_device *spi)
 	/* Set After SCK delay scale values */
 	ns_delay_scale(&pasc, &asc, sck_cs_delay, clkrate);
 
-	chip->ctar_val =  SPI_CTAR_FMSZ(fmsz)
-		| SPI_CTAR_CPOL(spi->mode & SPI_CPOL ? 1 : 0)
+	chip->ctar_val = SPI_CTAR_CPOL(spi->mode & SPI_CPOL ? 1 : 0)
 		| SPI_CTAR_CPHA(spi->mode & SPI_CPHA ? 1 : 0)
 		| SPI_CTAR_LSBFE(spi->mode & SPI_LSB_FIRST ? 1 : 0)
 		| SPI_CTAR_PCSSCK(pcssck)
@@ -827,36 +822,20 @@ static irqreturn_t dspi_interrupt(int irq, void *dev_id)
 	struct spi_message *msg = dspi->cur_msg;
 	enum dspi_trans_mode trans_mode;
 	u32 spi_sr, spi_tcr;
-	u32 spi_tcnt, tcnt_diff;
-	int tx_word;
+	u16 spi_tcnt;
 
 	regmap_read(dspi->regmap, SPI_SR, &spi_sr);
 	regmap_write(dspi->regmap, SPI_SR, spi_sr);
 
 
 	if (spi_sr & (SPI_SR_EOQF | SPI_SR_TCFQF)) {
-		tx_word = is_double_byte_mode(dspi);
-
+		/* Get transfer counter (in number of SPI transfers). It was
+		 * reset to 0 when transfer(s) were started.
+		 */
 		regmap_read(dspi->regmap, SPI_TCR, &spi_tcr);
 		spi_tcnt = SPI_TCR_GET_TCNT(spi_tcr);
-		/*
-		 * The width of SPI Transfer Counter in SPI_TCR is 16bits,
-		 * so the max couner is 65535. When the counter reach 65535,
-		 * it will wrap around, counter reset to zero.
-		 * spi_tcnt my be less than dspi->spi_tcnt, it means the
-		 * counter already wrapped around.
-		 * SPI Transfer Counter is a counter of transmitted frames.
-		 * The size of frame maybe two bytes.
-		 */
-		tcnt_diff = ((spi_tcnt + SPI_TCR_TCNT_MAX) - dspi->spi_tcnt)
-			% SPI_TCR_TCNT_MAX;
-		tcnt_diff *= (tx_word + 1);
-		if (dspi->dataflags & TRAN_STATE_WORD_ODD_NUM)
-			tcnt_diff--;
-
-		msg->actual_length += tcnt_diff;
-
-		dspi->spi_tcnt = spi_tcnt;
+		/* Update total number of bytes that were transferred */
+		msg->actual_length += spi_tcnt * dspi->bytes_per_word;
 
 		trans_mode = dspi->devtype_data->trans_mode;
 		switch (trans_mode) {
@@ -873,14 +852,6 @@ static irqreturn_t dspi_interrupt(int irq, void *dev_id)
 		}
 
 		if (!dspi->len) {
-			if (dspi->dataflags & TRAN_STATE_WORD_ODD_NUM) {
-				regmap_update_bits(dspi->regmap,
-						   SPI_CTAR(0),
-						   SPI_FRAME_BITS_MASK,
-						   SPI_FRAME_BITS(16));
-				dspi->dataflags &= ~TRAN_STATE_WORD_ODD_NUM;
-			}
-
 			dspi->waitflags = 1;
 			wake_up_interruptible(&dspi->waitq);
 		} else {
@@ -943,16 +914,62 @@ static int dspi_resume(struct device *dev)
 
 static SIMPLE_DEV_PM_OPS(dspi_pm, dspi_suspend, dspi_resume);
 
+static const struct regmap_range dspi_volatile_ranges[] = {
+	regmap_reg_range(SPI_MCR, SPI_TCR),
+	regmap_reg_range(SPI_SR, SPI_SR),
+	regmap_reg_range(SPI_PUSHR, SPI_RXFR3),
+};
+
+static const struct regmap_access_table dspi_volatile_table = {
+	.yes_ranges     = dspi_volatile_ranges,
+	.n_yes_ranges   = ARRAY_SIZE(dspi_volatile_ranges),
+};
+
 static const struct regmap_config dspi_regmap_config = {
 	.reg_bits = 32,
 	.val_bits = 32,
 	.reg_stride = 4,
 	.max_register = 0x88,
+	.volatile_table = &dspi_volatile_table,
+};
+
+static const struct regmap_range dspi_xspi_volatile_ranges[] = {
+	regmap_reg_range(SPI_MCR, SPI_TCR),
+	regmap_reg_range(SPI_SR, SPI_SR),
+	regmap_reg_range(SPI_PUSHR, SPI_RXFR3),
+	regmap_reg_range(SPI_SREX, SPI_SREX),
+};
+
+static const struct regmap_access_table dspi_xspi_volatile_table = {
+	.yes_ranges     = dspi_xspi_volatile_ranges,
+	.n_yes_ranges   = ARRAY_SIZE(dspi_xspi_volatile_ranges),
+};
+
+static const struct regmap_config dspi_xspi_regmap_config[] = {
+	{
+		.reg_bits = 32,
+		.val_bits = 32,
+		.reg_stride = 4,
+		.max_register = 0x13c,
+		.volatile_table = &dspi_xspi_volatile_table,
+	},
+	{
+		.name = "pushr",
+		.reg_bits = 16,
+		.val_bits = 16,
+		.reg_stride = 2,
+		.max_register = 0x2,
+	},
 };
 
 static void dspi_init(struct fsl_dspi *dspi)
 {
+	regmap_write(dspi->regmap, SPI_MCR, SPI_MCR_MASTER | SPI_MCR_PCSIS |
+		     (dspi->devtype_data->xspi_mode ? SPI_MCR_XSPI : 0));
 	regmap_write(dspi->regmap, SPI_SR, SPI_SR_CLEAR);
+	if (dspi->devtype_data->xspi_mode)
+		regmap_write(dspi->regmap, SPI_CTARE(0),
+			     SPI_CTARE_FMSZE(0) | SPI_CTARE_DTCP(1));
 }
 
 static int dspi_probe(struct platform_device *pdev)
@@ -961,6 +978,7 @@ static int dspi_probe(struct platform_device *pdev)
 	struct spi_master *master;
 	struct fsl_dspi *dspi;
 	struct resource *res;
+	const struct regmap_config *regmap_config;
 	void __iomem *base;
 	struct fsl_dspi_platform_data *pdata;
 	int ret = 0, cs_num, bus_num;
@@ -980,8 +998,6 @@ static int dspi_probe(struct platform_device *pdev)
 
 	master->cleanup = dspi_cleanup;
 	master->mode_bits = SPI_CPOL | SPI_CPHA | SPI_LSB_FIRST;
-	master->bits_per_word_mask = SPI_BPW_MASK(4) | SPI_BPW_MASK(8) |
-					SPI_BPW_MASK(16);
 
 	pdata = dev_get_platdata(&pdev->dev);
 	if (pdata) {
@@ -1013,6 +1029,11 @@ static int dspi_probe(struct platform_device *pdev)
 		}
 	}
 
+	if (dspi->devtype_data->xspi_mode)
+		master->bits_per_word_mask = SPI_BPW_RANGE_MASK(4, 32);
+	else
+		master->bits_per_word_mask = SPI_BPW_RANGE_MASK(4, 16);
+
 	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
 	base = devm_ioremap_resource(&pdev->dev, res);
 	if (IS_ERR(base)) {
@@ -1020,8 +1041,11 @@ static int dspi_probe(struct platform_device *pdev)
 		goto out_master_put;
 	}
 
-	dspi->regmap = devm_regmap_init_mmio_clk(&pdev->dev, NULL, base,
-						&dspi_regmap_config);
+	if (dspi->devtype_data->xspi_mode)
+		regmap_config = &dspi_xspi_regmap_config[0];
+	else
+		regmap_config = &dspi_regmap_config;
+	dspi->regmap = devm_regmap_init_mmio(&pdev->dev, base, regmap_config);
 	if (IS_ERR(dspi->regmap)) {
 		dev_err(&pdev->dev, "failed to init regmap: %ld\n",
 				PTR_ERR(dspi->regmap));
@@ -1029,30 +1053,43 @@ static int dspi_probe(struct platform_device *pdev)
 		goto out_master_put;
 	}
 
+	if (dspi->devtype_data->xspi_mode) {
+		dspi->regmap_pushr = devm_regmap_init_mmio(
+			&pdev->dev, base + SPI_PUSHR,
+			&dspi_xspi_regmap_config[1]);
+		if (IS_ERR(dspi->regmap_pushr)) {
+			dev_err(&pdev->dev,
+				"failed to init pushr regmap: %ld\n",
+				PTR_ERR(dspi->regmap_pushr));
+			ret = PTR_ERR(dspi->regmap_pushr);
+			goto out_master_put;
+		}
+	}
+
+	dspi->clk = devm_clk_get(&pdev->dev, "dspi");
+	if (IS_ERR(dspi->clk)) {
+		ret = PTR_ERR(dspi->clk);
+		dev_err(&pdev->dev, "unable to get clock\n");
+		goto out_master_put;
+	}
+	ret = clk_prepare_enable(dspi->clk);
+	if (ret)
+		goto out_master_put;
+
 	dspi_init(dspi);
 	dspi->irq = platform_get_irq(pdev, 0);
 	if (dspi->irq < 0) {
 		dev_err(&pdev->dev, "can't get platform irq\n");
 		ret = dspi->irq;
-		goto out_master_put;
+		goto out_clk_put;
 	}
 
 	ret = devm_request_irq(&pdev->dev, dspi->irq, dspi_interrupt, 0,
 			pdev->name, dspi);
 	if (ret < 0) {
 		dev_err(&pdev->dev, "Unable to attach DSPI interrupt\n");
-		goto out_master_put;
-	}
-
-	dspi->clk = devm_clk_get(&pdev->dev, "dspi");
-	if (IS_ERR(dspi->clk)) {
-		ret = PTR_ERR(dspi->clk);
-		dev_err(&pdev->dev, "unable to get clock\n");
-		goto out_master_put;
+		goto out_clk_put;
 	}
-	ret = clk_prepare_enable(dspi->clk);
-	if (ret)
-		goto out_master_put;
 
 	if (dspi->devtype_data->trans_mode == DSPI_DMA_MODE) {
 		ret = dspi_request_dma(dspi, res->start);
diff --git a/drivers/spi/spi-fsl-espi.c b/drivers/spi/spi-fsl-espi.c
index 1d332e23f6ed..1e8ff6256079 100644
--- a/drivers/spi/spi-fsl-espi.c
+++ b/drivers/spi/spi-fsl-espi.c
@@ -547,8 +547,11 @@ static void fsl_espi_cpu_irq(struct fsl_espi *espi, u32 events)
 		dev_err(espi->dev,
 			"Transfer done but SPIE_DON isn't set!\n");
 
-	if (SPIE_RXCNT(events) || SPIE_TXCNT(events) != FSL_ESPI_FIFO_SIZE)
+	if (SPIE_RXCNT(events) || SPIE_TXCNT(events) != FSL_ESPI_FIFO_SIZE) {
 		dev_err(espi->dev, "Transfer done but rx/tx fifo's aren't empty!\n");
+		dev_err(espi->dev, "SPIE_RXCNT = %d, SPIE_TXCNT = %d\n",
+			SPIE_RXCNT(events), SPIE_TXCNT(events));
+	}
 
 	complete(&espi->done);
 }
diff --git a/drivers/spi/spi-gpio.c b/drivers/spi/spi-gpio.c
index 6ae92d4dca19..0626e6e3ea0c 100644
--- a/drivers/spi/spi-gpio.c
+++ b/drivers/spi/spi-gpio.c
@@ -121,7 +121,10 @@ static inline int getmiso(const struct spi_device *spi)
 {
 	struct spi_gpio *spi_gpio = spi_to_spi_gpio(spi);
 
-	return !!gpiod_get_value_cansleep(spi_gpio->miso);
+	if (spi->mode & SPI_3WIRE)
+		return !!gpiod_get_value_cansleep(spi_gpio->mosi);
+	else
+		return !!gpiod_get_value_cansleep(spi_gpio->miso);
 }
 
 /*
@@ -149,27 +152,27 @@ static inline int getmiso(const struct spi_device *spi)
  */
 
 static u32 spi_gpio_txrx_word_mode0(struct spi_device *spi,
-		unsigned nsecs, u32 word, u8 bits)
+		unsigned nsecs, u32 word, u8 bits, unsigned flags)
 {
-	return bitbang_txrx_be_cpha0(spi, nsecs, 0, 0, word, bits);
+	return bitbang_txrx_be_cpha0(spi, nsecs, 0, flags, word, bits);
 }
 
 static u32 spi_gpio_txrx_word_mode1(struct spi_device *spi,
-		unsigned nsecs, u32 word, u8 bits)
+		unsigned nsecs, u32 word, u8 bits, unsigned flags)
 {
-	return bitbang_txrx_be_cpha1(spi, nsecs, 0, 0, word, bits);
+	return bitbang_txrx_be_cpha1(spi, nsecs, 0, flags, word, bits);
 }
 
 static u32 spi_gpio_txrx_word_mode2(struct spi_device *spi,
-		unsigned nsecs, u32 word, u8 bits)
+		unsigned nsecs, u32 word, u8 bits, unsigned flags)
 {
-	return bitbang_txrx_be_cpha0(spi, nsecs, 1, 0, word, bits);
+	return bitbang_txrx_be_cpha0(spi, nsecs, 1, flags, word, bits);
 }
 
 static u32 spi_gpio_txrx_word_mode3(struct spi_device *spi,
-		unsigned nsecs, u32 word, u8 bits)
+		unsigned nsecs, u32 word, u8 bits, unsigned flags)
 {
-	return bitbang_txrx_be_cpha1(spi, nsecs, 1, 0, word, bits);
+	return bitbang_txrx_be_cpha1(spi, nsecs, 1, flags, word, bits);
 }
 
 /*
@@ -183,30 +186,30 @@ static u32 spi_gpio_txrx_word_mode3(struct spi_device *spi,
  */
 
 static u32 spi_gpio_spec_txrx_word_mode0(struct spi_device *spi,
-		unsigned nsecs, u32 word, u8 bits)
+		unsigned nsecs, u32 word, u8 bits, unsigned flags)
 {
-	unsigned flags = spi->master->flags;
+	flags = spi->master->flags;
 	return bitbang_txrx_be_cpha0(spi, nsecs, 0, flags, word, bits);
 }
 
 static u32 spi_gpio_spec_txrx_word_mode1(struct spi_device *spi,
-		unsigned nsecs, u32 word, u8 bits)
+		unsigned nsecs, u32 word, u8 bits, unsigned flags)
 {
-	unsigned flags = spi->master->flags;
+	flags = spi->master->flags;
 	return bitbang_txrx_be_cpha1(spi, nsecs, 0, flags, word, bits);
 }
 
 static u32 spi_gpio_spec_txrx_word_mode2(struct spi_device *spi,
-		unsigned nsecs, u32 word, u8 bits)
+		unsigned nsecs, u32 word, u8 bits, unsigned flags)
 {
-	unsigned flags = spi->master->flags;
+	flags = spi->master->flags;
 	return bitbang_txrx_be_cpha0(spi, nsecs, 1, flags, word, bits);
 }
 
 static u32 spi_gpio_spec_txrx_word_mode3(struct spi_device *spi,
-		unsigned nsecs, u32 word, u8 bits)
+		unsigned nsecs, u32 word, u8 bits, unsigned flags)
 {
-	unsigned flags = spi->master->flags;
+	flags = spi->master->flags;
 	return bitbang_txrx_be_cpha1(spi, nsecs, 1, flags, word, bits);
 }
 
@@ -250,6 +253,16 @@ static int spi_gpio_setup(struct spi_device *spi)
 	return status;
 }
 
+static int spi_gpio_set_direction(struct spi_device *spi, bool output)
+{
+	struct spi_gpio *spi_gpio = spi_to_spi_gpio(spi);
+
+	if (output)
+		return gpiod_direction_output(spi_gpio->mosi, 1);
+	else
+		return gpiod_direction_input(spi_gpio->mosi);
+}
+
 static void spi_gpio_cleanup(struct spi_device *spi)
 {
 	spi_bitbang_cleanup(spi);
@@ -395,6 +408,7 @@ static int spi_gpio_probe(struct platform_device *pdev)
 		return status;
 
 	master->bits_per_word_mask = SPI_BPW_RANGE_MASK(1, 32);
+	master->mode_bits = SPI_3WIRE | SPI_CPHA | SPI_CPOL;
 	master->flags = master_flags;
 	master->bus_num = pdev->id;
 	/* The master needs to think there is a chipselect even if not connected */
@@ -407,6 +421,7 @@ static int spi_gpio_probe(struct platform_device *pdev)
 
 	spi_gpio->bitbang.master = master;
 	spi_gpio->bitbang.chipselect = spi_gpio_chipselect;
+	spi_gpio->bitbang.set_line_direction = spi_gpio_set_direction;
 
 	if ((master_flags & (SPI_MASTER_NO_TX | SPI_MASTER_NO_RX)) == 0) {
 		spi_gpio->bitbang.txrx_word[SPI_MODE_0] = spi_gpio_txrx_word_mode0;
diff --git a/drivers/spi/spi-img-spfi.c b/drivers/spi/spi-img-spfi.c
index 7a37090dabbe..e6eb979f1b8a 100644
--- a/drivers/spi/spi-img-spfi.c
+++ b/drivers/spi/spi-img-spfi.c
@@ -419,6 +419,9 @@ static int img_spfi_prepare(struct spi_master *master, struct spi_message *msg)
 	u32 val;
 
 	val = spfi_readl(spfi, SPFI_PORT_STATE);
+	val &= ~(SPFI_PORT_STATE_DEV_SEL_MASK <<
+		 SPFI_PORT_STATE_DEV_SEL_SHIFT);
+	val |= msg->spi->chip_select << SPFI_PORT_STATE_DEV_SEL_SHIFT;
 	if (msg->spi->mode & SPI_CPHA)
 		val |= SPFI_PORT_STATE_CK_PHASE(msg->spi->chip_select);
 	else
diff --git a/drivers/spi/spi-imx.c b/drivers/spi/spi-imx.c
index d3b21faf6b1f..08dd3a31a3e5 100644
--- a/drivers/spi/spi-imx.c
+++ b/drivers/spi/spi-imx.c
@@ -94,8 +94,7 @@ struct spi_imx_data {
 	void *rx_buf;
 	const void *tx_buf;
 	unsigned int txfifo; /* number of words pushed in tx FIFO */
-	unsigned int dynamic_burst, read_u32;
-	unsigned int word_mask;
+	unsigned int dynamic_burst;
 
 	/* Slave mode */
 	bool slave_mode;
@@ -140,6 +139,8 @@ static void spi_imx_buf_rx_##type(struct spi_imx_data *spi_imx)		\
 		*(type *)spi_imx->rx_buf = val;				\
 		spi_imx->rx_buf += sizeof(type);			\
 	}								\
+									\
+	spi_imx->remainder -= sizeof(type);				\
 }
 
 #define MXC_SPI_BUF_TX(type)						\
@@ -203,7 +204,12 @@ out:
 
 static int spi_imx_bytes_per_word(const int bits_per_word)
 {
-	return DIV_ROUND_UP(bits_per_word, BITS_PER_BYTE);
+	if (bits_per_word <= 8)
+		return 1;
+	else if (bits_per_word <= 16)
+		return 2;
+	else
+		return 4;
 }
 
 static bool spi_imx_can_dma(struct spi_master *master, struct spi_device *spi,
@@ -220,17 +226,11 @@ static bool spi_imx_can_dma(struct spi_master *master, struct spi_device *spi,
 
 	bytes_per_word = spi_imx_bytes_per_word(transfer->bits_per_word);
 
-	if (bytes_per_word != 1 && bytes_per_word != 2 && bytes_per_word != 4)
-		return false;
-
 	for (i = spi_imx->devtype_data->fifo_size / 2; i > 0; i--) {
 		if (!(transfer->len % (i * bytes_per_word)))
 			break;
 	}
 
-	if (i == 0)
-		return false;
-
 	spi_imx->wml = i;
 	spi_imx->dynamic_burst = 0;
 
@@ -291,26 +291,39 @@ static void spi_imx_buf_rx_swap_u32(struct spi_imx_data *spi_imx)
 		else if (bytes_per_word == 2)
 			val = (val << 16) | (val >> 16);
 #endif
-		val &= spi_imx->word_mask;
 		*(u32 *)spi_imx->rx_buf = val;
 		spi_imx->rx_buf += sizeof(u32);
 	}
+
+	spi_imx->remainder -= sizeof(u32);
 }
 
 static void spi_imx_buf_rx_swap(struct spi_imx_data *spi_imx)
 {
-	unsigned int bytes_per_word;
+	int unaligned;
+	u32 val;
 
-	bytes_per_word = spi_imx_bytes_per_word(spi_imx->bits_per_word);
-	if (spi_imx->read_u32) {
+	unaligned = spi_imx->remainder % 4;
+
+	if (!unaligned) {
 		spi_imx_buf_rx_swap_u32(spi_imx);
 		return;
 	}
 
-	if (bytes_per_word == 1)
-		spi_imx_buf_rx_u8(spi_imx);
-	else if (bytes_per_word == 2)
+	if (spi_imx_bytes_per_word(spi_imx->bits_per_word) == 2) {
 		spi_imx_buf_rx_u16(spi_imx);
+		return;
+	}
+
+	val = readl(spi_imx->base + MXC_CSPIRXDATA);
+
+	while (unaligned--) {
+		if (spi_imx->rx_buf) {
+			*(u8 *)spi_imx->rx_buf = (val >> (8 * unaligned)) & 0xff;
+			spi_imx->rx_buf++;
+		}
+		spi_imx->remainder--;
+	}
 }
 
 static void spi_imx_buf_tx_swap_u32(struct spi_imx_data *spi_imx)
@@ -322,7 +335,6 @@ static void spi_imx_buf_tx_swap_u32(struct spi_imx_data *spi_imx)
 
 	if (spi_imx->tx_buf) {
 		val = *(u32 *)spi_imx->tx_buf;
-		val &= spi_imx->word_mask;
 		spi_imx->tx_buf += sizeof(u32);
 	}
 
@@ -340,40 +352,30 @@ static void spi_imx_buf_tx_swap_u32(struct spi_imx_data *spi_imx)
 
 static void spi_imx_buf_tx_swap(struct spi_imx_data *spi_imx)
 {
-	u32 ctrl, val;
-	unsigned int bytes_per_word;
+	int unaligned;
+	u32 val = 0;
 
-	if (spi_imx->count == spi_imx->remainder) {
-		ctrl = readl(spi_imx->base + MX51_ECSPI_CTRL);
-		ctrl &= ~MX51_ECSPI_CTRL_BL_MASK;
-		if (spi_imx->count > MX51_ECSPI_CTRL_MAX_BURST) {
-			spi_imx->remainder = spi_imx->count %
-					     MX51_ECSPI_CTRL_MAX_BURST;
-			val = MX51_ECSPI_CTRL_MAX_BURST * 8 - 1;
-		} else if (spi_imx->count >= sizeof(u32)) {
-			spi_imx->remainder = spi_imx->count % sizeof(u32);
-			val = (spi_imx->count - spi_imx->remainder) * 8 - 1;
-		} else {
-			spi_imx->remainder = 0;
-			val = spi_imx->bits_per_word - 1;
-			spi_imx->read_u32 = 0;
-		}
+	unaligned = spi_imx->count % 4;
 
-		ctrl |= (val << MX51_ECSPI_CTRL_BL_OFFSET);
-		writel(ctrl, spi_imx->base + MX51_ECSPI_CTRL);
+	if (!unaligned) {
+		spi_imx_buf_tx_swap_u32(spi_imx);
+		return;
 	}
 
-	if (spi_imx->count >= sizeof(u32)) {
-		spi_imx_buf_tx_swap_u32(spi_imx);
+	if (spi_imx_bytes_per_word(spi_imx->bits_per_word) == 2) {
+		spi_imx_buf_tx_u16(spi_imx);
 		return;
 	}
 
-	bytes_per_word = spi_imx_bytes_per_word(spi_imx->bits_per_word);
+	while (unaligned--) {
+		if (spi_imx->tx_buf) {
+			val |= *(u8 *)spi_imx->tx_buf << (8 * unaligned);
+			spi_imx->tx_buf++;
+		}
+		spi_imx->count--;
+	}
 
-	if (bytes_per_word == 1)
-		spi_imx_buf_tx_u8(spi_imx);
-	else if (bytes_per_word == 2)
-		spi_imx_buf_tx_u16(spi_imx);
+	writel(val, spi_imx->base + MXC_CSPITXDATA);
 }
 
 static void mx53_ecspi_rx_slave(struct spi_imx_data *spi_imx)
@@ -392,6 +394,8 @@ static void mx53_ecspi_rx_slave(struct spi_imx_data *spi_imx)
 		spi_imx->rx_buf += n_bytes;
 		spi_imx->slave_burst -= n_bytes;
 	}
+
+	spi_imx->remainder -= sizeof(u32);
 }
 
 static void mx53_ecspi_tx_slave(struct spi_imx_data *spi_imx)
@@ -1001,12 +1005,52 @@ static void spi_imx_chipselect(struct spi_device *spi, int is_active)
 	gpio_set_value(spi->cs_gpio, dev_is_lowactive ^ active);
 }
 
+static void spi_imx_set_burst_len(struct spi_imx_data *spi_imx, int n_bits)
+{
+	u32 ctrl;
+
+	ctrl = readl(spi_imx->base + MX51_ECSPI_CTRL);
+	ctrl &= ~MX51_ECSPI_CTRL_BL_MASK;
+	ctrl |= ((n_bits - 1) << MX51_ECSPI_CTRL_BL_OFFSET);
+	writel(ctrl, spi_imx->base + MX51_ECSPI_CTRL);
+}
+
 static void spi_imx_push(struct spi_imx_data *spi_imx)
 {
+	unsigned int burst_len, fifo_words;
+
+	if (spi_imx->dynamic_burst)
+		fifo_words = 4;
+	else
+		fifo_words = spi_imx_bytes_per_word(spi_imx->bits_per_word);
+	/*
+	 * Reload the FIFO when the remaining bytes to be transferred in the
+	 * current burst is 0. This only applies when bits_per_word is a
+	 * multiple of 8.
+	 */
+	if (!spi_imx->remainder) {
+		if (spi_imx->dynamic_burst) {
+
+			/* We need to deal unaligned data first */
+			burst_len = spi_imx->count % MX51_ECSPI_CTRL_MAX_BURST;
+
+			if (!burst_len)
+				burst_len = MX51_ECSPI_CTRL_MAX_BURST;
+
+			spi_imx_set_burst_len(spi_imx, burst_len * 8);
+
+			spi_imx->remainder = burst_len;
+		} else {
+			spi_imx->remainder = fifo_words;
+		}
+	}
+
 	while (spi_imx->txfifo < spi_imx->devtype_data->fifo_size) {
 		if (!spi_imx->count)
 			break;
-		if (spi_imx->txfifo && (spi_imx->count == spi_imx->remainder))
+		if (spi_imx->dynamic_burst &&
+		    spi_imx->txfifo >=  DIV_ROUND_UP(spi_imx->remainder,
+						     fifo_words))
 			break;
 		spi_imx->tx(spi_imx);
 		spi_imx->txfifo++;
@@ -1102,27 +1146,20 @@ static int spi_imx_setupxfer(struct spi_device *spi,
 	spi_imx->bits_per_word = t->bits_per_word;
 	spi_imx->speed_hz  = t->speed_hz;
 
-	/* Initialize the functions for transfer */
-	if (spi_imx->devtype_data->dynamic_burst && !spi_imx->slave_mode) {
-		u32 mask;
-
-		spi_imx->dynamic_burst = 0;
-		spi_imx->remainder = 0;
-		spi_imx->read_u32  = 1;
+	/*
+	 * Initialize the functions for transfer. To transfer non byte-aligned
+	 * words, we have to use multiple word-size bursts, we can't use
+	 * dynamic_burst in that case.
+	 */
+	if (spi_imx->devtype_data->dynamic_burst && !spi_imx->slave_mode &&
+	    (spi_imx->bits_per_word == 8 ||
+	    spi_imx->bits_per_word == 16 ||
+	    spi_imx->bits_per_word == 32)) {
 
-		mask = (1 << spi_imx->bits_per_word) - 1;
 		spi_imx->rx = spi_imx_buf_rx_swap;
 		spi_imx->tx = spi_imx_buf_tx_swap;
 		spi_imx->dynamic_burst = 1;
-		spi_imx->remainder = t->len;
-
-		if (spi_imx->bits_per_word <= 8)
-			spi_imx->word_mask = mask << 24 | mask << 16
-					     | mask << 8 | mask;
-		else if (spi_imx->bits_per_word <= 16)
-			spi_imx->word_mask = mask << 16 | mask;
-		else
-			spi_imx->word_mask = mask;
+
 	} else {
 		if (spi_imx->bits_per_word <= 8) {
 			spi_imx->rx = spi_imx_buf_rx_u8;
@@ -1134,6 +1171,7 @@ static int spi_imx_setupxfer(struct spi_device *spi,
 			spi_imx->rx = spi_imx_buf_rx_u32;
 			spi_imx->tx = spi_imx_buf_tx_u32;
 		}
+		spi_imx->dynamic_burst = 0;
 	}
 
 	if (spi_imx_can_dma(spi_imx->bitbang.master, spi, t))
@@ -1317,6 +1355,7 @@ static int spi_imx_pio_transfer(struct spi_device *spi,
 	spi_imx->rx_buf = transfer->rx_buf;
 	spi_imx->count = transfer->len;
 	spi_imx->txfifo = 0;
+	spi_imx->remainder = 0;
 
 	reinit_completion(&spi_imx->xfer_done);
 
@@ -1354,6 +1393,7 @@ static int spi_imx_pio_transfer_slave(struct spi_device *spi,
 	spi_imx->rx_buf = transfer->rx_buf;
 	spi_imx->count = transfer->len;
 	spi_imx->txfifo = 0;
+	spi_imx->remainder = 0;
 
 	reinit_completion(&spi_imx->xfer_done);
 	spi_imx->slave_aborted = false;
diff --git a/drivers/spi/spi-lm70llp.c b/drivers/spi/spi-lm70llp.c
index 61ee0f4269ae..4549efd792da 100644
--- a/drivers/spi/spi-lm70llp.c
+++ b/drivers/spi/spi-lm70llp.c
@@ -188,9 +188,10 @@ static void lm70_chipselect(struct spi_device *spi, int value)
 /*
  * Our actual bitbanger routine.
  */
-static u32 lm70_txrx(struct spi_device *spi, unsigned nsecs, u32 word, u8 bits)
+static u32 lm70_txrx(struct spi_device *spi, unsigned nsecs, u32 word, u8 bits,
+		     unsigned flags)
 {
-	return bitbang_txrx_be_cpha0(spi, nsecs, 0, 0, word, bits);
+	return bitbang_txrx_be_cpha0(spi, nsecs, 0, flags, word, bits);
 }
 
 static void spi_lm70llp_attach(struct parport *p)
diff --git a/drivers/spi/spi-mem.c b/drivers/spi/spi-mem.c
index 990770dfa5cf..e43842c7a31a 100644
--- a/drivers/spi/spi-mem.c
+++ b/drivers/spi/spi-mem.c
@@ -311,6 +311,24 @@ int spi_mem_exec_op(struct spi_mem *mem, const struct spi_mem_op *op)
 EXPORT_SYMBOL_GPL(spi_mem_exec_op);
 
 /**
+ * spi_mem_get_name() - Return the SPI mem device name to be used by the
+ *			upper layer if necessary
+ * @mem: the SPI memory
+ *
+ * This function allows SPI mem users to retrieve the SPI mem device name.
+ * It is useful if the upper layer needs to expose a custom name for
+ * compatibility reasons.
+ *
+ * Return: a string containing the name of the memory device to be used
+ *	   by the SPI mem user
+ */
+const char *spi_mem_get_name(struct spi_mem *mem)
+{
+	return mem->name;
+}
+EXPORT_SYMBOL_GPL(spi_mem_get_name);
+
+/**
  * spi_mem_adjust_op_size() - Adjust the data size of a SPI mem operation to
  *			      match controller limitations
  * @mem: the SPI memory
@@ -344,6 +362,7 @@ static inline struct spi_mem_driver *to_spi_mem_drv(struct device_driver *drv)
 static int spi_mem_probe(struct spi_device *spi)
 {
 	struct spi_mem_driver *memdrv = to_spi_mem_drv(spi->dev.driver);
+	struct spi_controller *ctlr = spi->controller;
 	struct spi_mem *mem;
 
 	mem = devm_kzalloc(&spi->dev, sizeof(*mem), GFP_KERNEL);
@@ -351,6 +370,15 @@ static int spi_mem_probe(struct spi_device *spi)
 		return -ENOMEM;
 
 	mem->spi = spi;
+
+	if (ctlr->mem_ops && ctlr->mem_ops->get_name)
+		mem->name = ctlr->mem_ops->get_name(mem);
+	else
+		mem->name = dev_name(&spi->dev);
+
+	if (IS_ERR_OR_NULL(mem->name))
+		return PTR_ERR(mem->name);
+
 	spi_set_drvdata(spi, mem);
 
 	return memdrv->probe(mem);
diff --git a/drivers/spi/spi-omap2-mcspi.c b/drivers/spi/spi-omap2-mcspi.c
index 6c628a54e946..508c61c669e7 100644
--- a/drivers/spi/spi-omap2-mcspi.c
+++ b/drivers/spi/spi-omap2-mcspi.c
@@ -398,11 +398,9 @@ static void omap2_mcspi_tx_dma(struct spi_device *spi,
 {
 	struct omap2_mcspi	*mcspi;
 	struct omap2_mcspi_dma  *mcspi_dma;
-	unsigned int		count;
 
 	mcspi = spi_master_get_devdata(spi->master);
 	mcspi_dma = &mcspi->dma_channels[spi->chip_select];
-	count = xfer->len;
 
 	if (mcspi_dma->dma_tx) {
 		struct dma_async_tx_descriptor *tx;
@@ -582,7 +580,6 @@ omap2_mcspi_txrx_dma(struct spi_device *spi, struct spi_transfer *xfer)
 	struct omap2_mcspi_cs	*cs = spi->controller_state;
 	struct omap2_mcspi_dma  *mcspi_dma;
 	unsigned int		count;
-	u32			l;
 	u8			*rx;
 	const u8		*tx;
 	struct dma_slave_config	cfg;
@@ -595,8 +592,6 @@ omap2_mcspi_txrx_dma(struct spi_device *spi, struct spi_transfer *xfer)
 
 	mcspi = spi_master_get_devdata(spi->master);
 	mcspi_dma = &mcspi->dma_channels[spi->chip_select];
-	l = mcspi_cached_chconf0(spi);
-
 
 	if (cs->word_len <= 8) {
 		width = DMA_SLAVE_BUSWIDTH_1_BYTE;
@@ -676,7 +671,6 @@ omap2_mcspi_txrx_dma(struct spi_device *spi, struct spi_transfer *xfer)
 static unsigned
 omap2_mcspi_txrx_pio(struct spi_device *spi, struct spi_transfer *xfer)
 {
-	struct omap2_mcspi	*mcspi;
 	struct omap2_mcspi_cs	*cs = spi->controller_state;
 	unsigned int		count, c;
 	u32			l;
@@ -686,7 +680,6 @@ omap2_mcspi_txrx_pio(struct spi_device *spi, struct spi_transfer *xfer)
 	void __iomem		*chstat_reg;
 	int			word_len;
 
-	mcspi = spi_master_get_devdata(spi->master);
 	count = xfer->len;
 	c = count;
 	word_len = cs->word_len;
@@ -883,13 +876,11 @@ static int omap2_mcspi_setup_transfer(struct spi_device *spi,
 {
 	struct omap2_mcspi_cs *cs = spi->controller_state;
 	struct omap2_mcspi *mcspi;
-	struct spi_master *spi_cntrl;
 	u32 l = 0, clkd = 0, div, extclk = 0, clkg = 0;
 	u8 word_len = spi->bits_per_word;
 	u32 speed_hz = spi->max_speed_hz;
 
 	mcspi = spi_master_get_devdata(spi->master);
-	spi_cntrl = mcspi->master;
 
 	if (t != NULL && t->bits_per_word)
 		word_len = t->bits_per_word;
diff --git a/drivers/spi/spi-orion.c b/drivers/spi/spi-orion.c
index d01a6adc726e..47ef6b1a2e76 100644
--- a/drivers/spi/spi-orion.c
+++ b/drivers/spi/spi-orion.c
@@ -20,6 +20,7 @@
 #include <linux/of.h>
 #include <linux/of_address.h>
 #include <linux/of_device.h>
+#include <linux/of_gpio.h>
 #include <linux/clk.h>
 #include <linux/sizes.h>
 #include <linux/gpio.h>
@@ -681,9 +682,9 @@ static int orion_spi_probe(struct platform_device *pdev)
 		goto out_rel_axi_clk;
 	}
 
-	/* Scan all SPI devices of this controller for direct mapped devices */
 	for_each_available_child_of_node(pdev->dev.of_node, np) {
 		u32 cs;
+		int cs_gpio;
 
 		/* Get chip-select number from the "reg" property */
 		status = of_property_read_u32(np, "reg", &cs);
@@ -695,6 +696,44 @@ static int orion_spi_probe(struct platform_device *pdev)
 		}
 
 		/*
+		 * Initialize the CS GPIO:
+		 * - properly request the actual GPIO signal
+		 * - de-assert the logical signal so that all GPIO CS lines
+		 *   are inactive when probing for slaves
+		 * - find an unused physical CS which will be driven for any
+		 *   slave which uses a CS GPIO
+		 */
+		cs_gpio = of_get_named_gpio(pdev->dev.of_node, "cs-gpios", cs);
+		if (cs_gpio > 0) {
+			char *gpio_name;
+			int cs_flags;
+
+			if (spi->unused_hw_gpio == -1) {
+				dev_info(&pdev->dev,
+					"Selected unused HW CS#%d for any GPIO CSes\n",
+					cs);
+				spi->unused_hw_gpio = cs;
+			}
+
+			gpio_name = devm_kasprintf(&pdev->dev, GFP_KERNEL,
+					"%s-CS%d", dev_name(&pdev->dev), cs);
+			if (!gpio_name) {
+				status = -ENOMEM;
+				goto out_rel_axi_clk;
+			}
+
+			cs_flags = of_property_read_bool(np, "spi-cs-high") ?
+				GPIOF_OUT_INIT_LOW : GPIOF_OUT_INIT_HIGH;
+			status = devm_gpio_request_one(&pdev->dev, cs_gpio,
+					cs_flags, gpio_name);
+			if (status) {
+				dev_err(&pdev->dev,
+					"Can't request GPIO for CS %d\n", cs);
+				goto out_rel_axi_clk;
+			}
+		}
+
+		/*
 		 * Check if an address is configured for this SPI device. If
 		 * not, the MBus mapping via the 'ranges' property in the 'soc'
 		 * node is not configured and this device should not use the
@@ -740,44 +779,8 @@ static int orion_spi_probe(struct platform_device *pdev)
 	if (status < 0)
 		goto out_rel_pm;
 
-	if (master->cs_gpios) {
-		int i;
-		for (i = 0; i < master->num_chipselect; ++i) {
-			char *gpio_name;
-
-			if (!gpio_is_valid(master->cs_gpios[i])) {
-				continue;
-			}
-
-			gpio_name = devm_kasprintf(&pdev->dev, GFP_KERNEL,
-					"%s-CS%d", dev_name(&pdev->dev), i);
-			if (!gpio_name) {
-				status = -ENOMEM;
-				goto out_rel_master;
-			}
-
-			status = devm_gpio_request(&pdev->dev,
-					master->cs_gpios[i], gpio_name);
-			if (status) {
-				dev_err(&pdev->dev,
-					"Can't request GPIO for CS %d\n",
-					master->cs_gpios[i]);
-				goto out_rel_master;
-			}
-			if (spi->unused_hw_gpio == -1) {
-				dev_info(&pdev->dev,
-					"Selected unused HW CS#%d for any GPIO CSes\n",
-					i);
-				spi->unused_hw_gpio = i;
-			}
-		}
-	}
-
-
 	return status;
 
-out_rel_master:
-	spi_unregister_master(master);
 out_rel_pm:
 	pm_runtime_disable(&pdev->dev);
 out_rel_axi_clk:
diff --git a/drivers/spi/spi-pxa2xx.c b/drivers/spi/spi-pxa2xx.c
index 0b2d60d30f69..14f4ea59caff 100644
--- a/drivers/spi/spi-pxa2xx.c
+++ b/drivers/spi/spi-pxa2xx.c
@@ -1391,6 +1391,10 @@ static const struct pci_device_id pxa2xx_spi_pci_compound_match[] = {
 	{ PCI_VDEVICE(INTEL, 0x31c2), LPSS_BXT_SSP },
 	{ PCI_VDEVICE(INTEL, 0x31c4), LPSS_BXT_SSP },
 	{ PCI_VDEVICE(INTEL, 0x31c6), LPSS_BXT_SSP },
+	/* ICL-LP */
+	{ PCI_VDEVICE(INTEL, 0x34aa), LPSS_CNL_SSP },
+	{ PCI_VDEVICE(INTEL, 0x34ab), LPSS_CNL_SSP },
+	{ PCI_VDEVICE(INTEL, 0x34fb), LPSS_CNL_SSP },
 	/* APL */
 	{ PCI_VDEVICE(INTEL, 0x5ac2), LPSS_BXT_SSP },
 	{ PCI_VDEVICE(INTEL, 0x5ac4), LPSS_BXT_SSP },
diff --git a/drivers/spi/spi-sh-msiof.c b/drivers/spi/spi-sh-msiof.c
index 0e74cbf9929d..539d6d1a277a 100644
--- a/drivers/spi/spi-sh-msiof.c
+++ b/drivers/spi/spi-sh-msiof.c
@@ -49,6 +49,7 @@ struct sh_msiof_spi_priv {
 	struct platform_device *pdev;
 	struct sh_msiof_spi_info *info;
 	struct completion done;
+	struct completion done_txdma;
 	unsigned int tx_fifo_size;
 	unsigned int rx_fifo_size;
 	unsigned int min_div_pow;
@@ -649,19 +650,21 @@ static int sh_msiof_slave_abort(struct spi_master *master)
 
 	p->slave_aborted = true;
 	complete(&p->done);
+	complete(&p->done_txdma);
 	return 0;
 }
 
-static int sh_msiof_wait_for_completion(struct sh_msiof_spi_priv *p)
+static int sh_msiof_wait_for_completion(struct sh_msiof_spi_priv *p,
+					struct completion *x)
 {
 	if (spi_controller_is_slave(p->master)) {
-		if (wait_for_completion_interruptible(&p->done) ||
+		if (wait_for_completion_interruptible(x) ||
 		    p->slave_aborted) {
 			dev_dbg(&p->pdev->dev, "interrupted\n");
 			return -EINTR;
 		}
 	} else {
-		if (!wait_for_completion_timeout(&p->done, HZ)) {
+		if (!wait_for_completion_timeout(x, HZ)) {
 			dev_err(&p->pdev->dev, "timeout\n");
 			return -ETIMEDOUT;
 		}
@@ -711,7 +714,7 @@ static int sh_msiof_spi_txrx_once(struct sh_msiof_spi_priv *p,
 	}
 
 	/* wait for tx fifo to be emptied / rx fifo to be filled */
-	ret = sh_msiof_wait_for_completion(p);
+	ret = sh_msiof_wait_for_completion(p, &p->done);
 	if (ret)
 		goto stop_reset;
 
@@ -740,10 +743,7 @@ stop_ier:
 
 static void sh_msiof_dma_complete(void *arg)
 {
-	struct sh_msiof_spi_priv *p = arg;
-
-	sh_msiof_write(p, IER, 0);
-	complete(&p->done);
+	complete(arg);
 }
 
 static int sh_msiof_dma_once(struct sh_msiof_spi_priv *p, const void *tx,
@@ -764,7 +764,7 @@ static int sh_msiof_dma_once(struct sh_msiof_spi_priv *p, const void *tx,
 			return -EAGAIN;
 
 		desc_rx->callback = sh_msiof_dma_complete;
-		desc_rx->callback_param = p;
+		desc_rx->callback_param = &p->done;
 		cookie = dmaengine_submit(desc_rx);
 		if (dma_submit_error(cookie))
 			return cookie;
@@ -782,13 +782,8 @@ static int sh_msiof_dma_once(struct sh_msiof_spi_priv *p, const void *tx,
 			goto no_dma_tx;
 		}
 
-		if (rx) {
-			/* No callback */
-			desc_tx->callback = NULL;
-		} else {
-			desc_tx->callback = sh_msiof_dma_complete;
-			desc_tx->callback_param = p;
-		}
+		desc_tx->callback = sh_msiof_dma_complete;
+		desc_tx->callback_param = &p->done_txdma;
 		cookie = dmaengine_submit(desc_tx);
 		if (dma_submit_error(cookie)) {
 			ret = cookie;
@@ -805,6 +800,8 @@ static int sh_msiof_dma_once(struct sh_msiof_spi_priv *p, const void *tx,
 	sh_msiof_write(p, IER, ier_bits);
 
 	reinit_completion(&p->done);
+	if (tx)
+		reinit_completion(&p->done_txdma);
 	p->slave_aborted = false;
 
 	/* Now start DMA */
@@ -819,17 +816,24 @@ static int sh_msiof_dma_once(struct sh_msiof_spi_priv *p, const void *tx,
 		goto stop_dma;
 	}
 
-	/* wait for tx/rx DMA completion */
-	ret = sh_msiof_wait_for_completion(p);
-	if (ret)
-		goto stop_reset;
+	if (tx) {
+		/* wait for tx DMA completion */
+		ret = sh_msiof_wait_for_completion(p, &p->done_txdma);
+		if (ret)
+			goto stop_reset;
+	}
 
-	if (!rx) {
-		reinit_completion(&p->done);
-		sh_msiof_write(p, IER, IER_TEOFE);
+	if (rx) {
+		/* wait for rx DMA completion */
+		ret = sh_msiof_wait_for_completion(p, &p->done);
+		if (ret)
+			goto stop_reset;
 
+		sh_msiof_write(p, IER, 0);
+	} else {
 		/* wait for tx fifo to be emptied */
-		ret = sh_msiof_wait_for_completion(p);
+		sh_msiof_write(p, IER, IER_TEOFE);
+		ret = sh_msiof_wait_for_completion(p, &p->done);
 		if (ret)
 			goto stop_reset;
 	}
@@ -1327,6 +1331,7 @@ static int sh_msiof_spi_probe(struct platform_device *pdev)
 	p->min_div_pow = chipdata->min_div_pow;
 
 	init_completion(&p->done);
+	init_completion(&p->done_txdma);
 
 	p->clk = devm_clk_get(&pdev->dev, NULL);
 	if (IS_ERR(p->clk)) {
diff --git a/drivers/spi/spi-sh-sci.c b/drivers/spi/spi-sh-sci.c
index a9beeeed812c..393701cfca3c 100644
--- a/drivers/spi/spi-sh-sci.c
+++ b/drivers/spi/spi-sh-sci.c
@@ -80,27 +80,31 @@ static inline u32 getmiso(struct spi_device *dev)
 #include "spi-bitbang-txrx.h"
 
 static u32 sh_sci_spi_txrx_mode0(struct spi_device *spi,
-				      unsigned nsecs, u32 word, u8 bits)
+				 unsigned nsecs, u32 word, u8 bits,
+				 unsigned flags)
 {
-	return bitbang_txrx_be_cpha0(spi, nsecs, 0, 0, word, bits);
+	return bitbang_txrx_be_cpha0(spi, nsecs, 0, flags, word, bits);
 }
 
 static u32 sh_sci_spi_txrx_mode1(struct spi_device *spi,
-				      unsigned nsecs, u32 word, u8 bits)
+				 unsigned nsecs, u32 word, u8 bits,
+				 unsigned flags)
 {
-	return bitbang_txrx_be_cpha1(spi, nsecs, 0, 0, word, bits);
+	return bitbang_txrx_be_cpha1(spi, nsecs, 0, flags, word, bits);
 }
 
 static u32 sh_sci_spi_txrx_mode2(struct spi_device *spi,
-				      unsigned nsecs, u32 word, u8 bits)
+				 unsigned nsecs, u32 word, u8 bits,
+				 unsigned flags)
 {
-	return bitbang_txrx_be_cpha0(spi, nsecs, 1, 0, word, bits);
+	return bitbang_txrx_be_cpha0(spi, nsecs, 1, flags, word, bits);
 }
 
 static u32 sh_sci_spi_txrx_mode3(struct spi_device *spi,
-				      unsigned nsecs, u32 word, u8 bits)
+				 unsigned nsecs, u32 word, u8 bits,
+				 unsigned flags)
 {
-	return bitbang_txrx_be_cpha1(spi, nsecs, 1, 0, word, bits);
+	return bitbang_txrx_be_cpha1(spi, nsecs, 1, flags, word, bits);
 }
 
 static void sh_sci_spi_chipselect(struct spi_device *dev, int value)
diff --git a/drivers/spi/spi-uniphier.c b/drivers/spi/spi-uniphier.c
new file mode 100644
index 000000000000..5a6137fe172d
--- /dev/null
+++ b/drivers/spi/spi-uniphier.c
@@ -0,0 +1,523 @@
+// SPDX-License-Identifier: GPL-2.0
+// spi-uniphier.c - Socionext UniPhier SPI controller driver
+// Copyright 2012      Panasonic Corporation
+// Copyright 2016-2018 Socionext Inc.
+
+#include <linux/kernel.h>
+#include <linux/bitfield.h>
+#include <linux/bitops.h>
+#include <linux/clk.h>
+#include <linux/interrupt.h>
+#include <linux/io.h>
+#include <linux/module.h>
+#include <linux/platform_device.h>
+#include <linux/spi/spi.h>
+
+#include <asm/unaligned.h>
+
+#define SSI_TIMEOUT_MS		2000
+#define SSI_MAX_CLK_DIVIDER	254
+#define SSI_MIN_CLK_DIVIDER	4
+
+struct uniphier_spi_priv {
+	void __iomem *base;
+	struct clk *clk;
+	struct spi_master *master;
+	struct completion xfer_done;
+
+	int error;
+	unsigned int tx_bytes;
+	unsigned int rx_bytes;
+	const u8 *tx_buf;
+	u8 *rx_buf;
+
+	bool is_save_param;
+	u8 bits_per_word;
+	u16 mode;
+	u32 speed_hz;
+};
+
+#define SSI_CTL			0x00
+#define   SSI_CTL_EN		BIT(0)
+
+#define SSI_CKS			0x04
+#define   SSI_CKS_CKRAT_MASK	GENMASK(7, 0)
+#define   SSI_CKS_CKPHS		BIT(14)
+#define   SSI_CKS_CKINIT	BIT(13)
+#define   SSI_CKS_CKDLY		BIT(12)
+
+#define SSI_TXWDS		0x08
+#define   SSI_TXWDS_WDLEN_MASK	GENMASK(13, 8)
+#define   SSI_TXWDS_TDTF_MASK	GENMASK(7, 6)
+#define   SSI_TXWDS_DTLEN_MASK	GENMASK(5, 0)
+
+#define SSI_RXWDS		0x0c
+#define   SSI_RXWDS_DTLEN_MASK	GENMASK(5, 0)
+
+#define SSI_FPS			0x10
+#define   SSI_FPS_FSPOL		BIT(15)
+#define   SSI_FPS_FSTRT		BIT(14)
+
+#define SSI_SR			0x14
+#define   SSI_SR_RNE		BIT(0)
+
+#define SSI_IE			0x18
+#define   SSI_IE_RCIE		BIT(3)
+#define   SSI_IE_RORIE		BIT(0)
+
+#define SSI_IS			0x1c
+#define   SSI_IS_RXRS		BIT(9)
+#define   SSI_IS_RCID		BIT(3)
+#define   SSI_IS_RORID		BIT(0)
+
+#define SSI_IC			0x1c
+#define   SSI_IC_TCIC		BIT(4)
+#define   SSI_IC_RCIC		BIT(3)
+#define   SSI_IC_RORIC		BIT(0)
+
+#define SSI_FC			0x20
+#define   SSI_FC_TXFFL		BIT(12)
+#define   SSI_FC_TXFTH_MASK	GENMASK(11, 8)
+#define   SSI_FC_RXFFL		BIT(4)
+#define   SSI_FC_RXFTH_MASK	GENMASK(3, 0)
+
+#define SSI_TXDR		0x24
+#define SSI_RXDR		0x24
+
+#define SSI_FIFO_DEPTH		8U
+
+static inline unsigned int bytes_per_word(unsigned int bits)
+{
+	return bits <= 8 ? 1 : (bits <= 16 ? 2 : 4);
+}
+
+static inline void uniphier_spi_irq_enable(struct spi_device *spi, u32 mask)
+{
+	struct uniphier_spi_priv *priv = spi_master_get_devdata(spi->master);
+	u32 val;
+
+	val = readl(priv->base + SSI_IE);
+	val |= mask;
+	writel(val, priv->base + SSI_IE);
+}
+
+static inline void uniphier_spi_irq_disable(struct spi_device *spi, u32 mask)
+{
+	struct uniphier_spi_priv *priv = spi_master_get_devdata(spi->master);
+	u32 val;
+
+	val = readl(priv->base + SSI_IE);
+	val &= ~mask;
+	writel(val, priv->base + SSI_IE);
+}
+
+static void uniphier_spi_set_mode(struct spi_device *spi)
+{
+	struct uniphier_spi_priv *priv = spi_master_get_devdata(spi->master);
+	u32 val1, val2;
+
+	/*
+	 * clock setting
+	 * CKPHS    capture timing. 0:rising edge, 1:falling edge
+	 * CKINIT   clock initial level. 0:low, 1:high
+	 * CKDLY    clock delay. 0:no delay, 1:delay depending on FSTRT
+	 *          (FSTRT=0: 1 clock, FSTRT=1: 0.5 clock)
+	 *
+	 * frame setting
+	 * FSPOL    frame signal porarity. 0: low, 1: high
+	 * FSTRT    start frame timing
+	 *          0: rising edge of clock, 1: falling edge of clock
+	 */
+	switch (spi->mode & (SPI_CPOL | SPI_CPHA)) {
+	case SPI_MODE_0:
+		/* CKPHS=1, CKINIT=0, CKDLY=1, FSTRT=0 */
+		val1 = SSI_CKS_CKPHS | SSI_CKS_CKDLY;
+		val2 = 0;
+		break;
+	case SPI_MODE_1:
+		/* CKPHS=0, CKINIT=0, CKDLY=0, FSTRT=1 */
+		val1 = 0;
+		val2 = SSI_FPS_FSTRT;
+		break;
+	case SPI_MODE_2:
+		/* CKPHS=0, CKINIT=1, CKDLY=1, FSTRT=1 */
+		val1 = SSI_CKS_CKINIT | SSI_CKS_CKDLY;
+		val2 = SSI_FPS_FSTRT;
+		break;
+	case SPI_MODE_3:
+		/* CKPHS=1, CKINIT=1, CKDLY=0, FSTRT=0 */
+		val1 = SSI_CKS_CKPHS | SSI_CKS_CKINIT;
+		val2 = 0;
+		break;
+	}
+
+	if (!(spi->mode & SPI_CS_HIGH))
+		val2 |= SSI_FPS_FSPOL;
+
+	writel(val1, priv->base + SSI_CKS);
+	writel(val2, priv->base + SSI_FPS);
+
+	val1 = 0;
+	if (spi->mode & SPI_LSB_FIRST)
+		val1 |= FIELD_PREP(SSI_TXWDS_TDTF_MASK, 1);
+	writel(val1, priv->base + SSI_TXWDS);
+	writel(val1, priv->base + SSI_RXWDS);
+}
+
+static void uniphier_spi_set_transfer_size(struct spi_device *spi, int size)
+{
+	struct uniphier_spi_priv *priv = spi_master_get_devdata(spi->master);
+	u32 val;
+
+	val = readl(priv->base + SSI_TXWDS);
+	val &= ~(SSI_TXWDS_WDLEN_MASK | SSI_TXWDS_DTLEN_MASK);
+	val |= FIELD_PREP(SSI_TXWDS_WDLEN_MASK, size);
+	val |= FIELD_PREP(SSI_TXWDS_DTLEN_MASK, size);
+	writel(val, priv->base + SSI_TXWDS);
+
+	val = readl(priv->base + SSI_RXWDS);
+	val &= ~SSI_RXWDS_DTLEN_MASK;
+	val |= FIELD_PREP(SSI_RXWDS_DTLEN_MASK, size);
+	writel(val, priv->base + SSI_RXWDS);
+}
+
+static void uniphier_spi_set_baudrate(struct spi_device *spi,
+				      unsigned int speed)
+{
+	struct uniphier_spi_priv *priv = spi_master_get_devdata(spi->master);
+	u32 val, ckdiv;
+
+	/*
+	 * the supported rates are even numbers from 4 to 254. (4,6,8...254)
+	 * round up as we look for equal or less speed
+	 */
+	ckdiv = DIV_ROUND_UP(clk_get_rate(priv->clk), speed);
+	ckdiv = round_up(ckdiv, 2);
+
+	val = readl(priv->base + SSI_CKS);
+	val &= ~SSI_CKS_CKRAT_MASK;
+	val |= ckdiv & SSI_CKS_CKRAT_MASK;
+	writel(val, priv->base + SSI_CKS);
+}
+
+static void uniphier_spi_setup_transfer(struct spi_device *spi,
+				       struct spi_transfer *t)
+{
+	struct uniphier_spi_priv *priv = spi_master_get_devdata(spi->master);
+	u32 val;
+
+	priv->error = 0;
+	priv->tx_buf = t->tx_buf;
+	priv->rx_buf = t->rx_buf;
+	priv->tx_bytes = priv->rx_bytes = t->len;
+
+	if (!priv->is_save_param || priv->mode != spi->mode) {
+		uniphier_spi_set_mode(spi);
+		priv->mode = spi->mode;
+	}
+
+	if (!priv->is_save_param || priv->bits_per_word != t->bits_per_word) {
+		uniphier_spi_set_transfer_size(spi, t->bits_per_word);
+		priv->bits_per_word = t->bits_per_word;
+	}
+
+	if (!priv->is_save_param || priv->speed_hz != t->speed_hz) {
+		uniphier_spi_set_baudrate(spi, t->speed_hz);
+		priv->speed_hz = t->speed_hz;
+	}
+
+	if (!priv->is_save_param)
+		priv->is_save_param = true;
+
+	/* reset FIFOs */
+	val = SSI_FC_TXFFL | SSI_FC_RXFFL;
+	writel(val, priv->base + SSI_FC);
+}
+
+static void uniphier_spi_send(struct uniphier_spi_priv *priv)
+{
+	int wsize;
+	u32 val = 0;
+
+	wsize = min(bytes_per_word(priv->bits_per_word), priv->tx_bytes);
+	priv->tx_bytes -= wsize;
+
+	if (priv->tx_buf) {
+		switch (wsize) {
+		case 1:
+			val = *priv->tx_buf;
+			break;
+		case 2:
+			val = get_unaligned_le16(priv->tx_buf);
+			break;
+		case 4:
+			val = get_unaligned_le32(priv->tx_buf);
+			break;
+		}
+
+		priv->tx_buf += wsize;
+	}
+
+	writel(val, priv->base + SSI_TXDR);
+}
+
+static void uniphier_spi_recv(struct uniphier_spi_priv *priv)
+{
+	int rsize;
+	u32 val;
+
+	rsize = min(bytes_per_word(priv->bits_per_word), priv->rx_bytes);
+	priv->rx_bytes -= rsize;
+
+	val = readl(priv->base + SSI_RXDR);
+
+	if (priv->rx_buf) {
+		switch (rsize) {
+		case 1:
+			*priv->rx_buf = val;
+			break;
+		case 2:
+			put_unaligned_le16(val, priv->rx_buf);
+			break;
+		case 4:
+			put_unaligned_le32(val, priv->rx_buf);
+			break;
+		}
+
+		priv->rx_buf += rsize;
+	}
+}
+
+static void uniphier_spi_fill_tx_fifo(struct uniphier_spi_priv *priv)
+{
+	unsigned int tx_count;
+	u32 val;
+
+	tx_count = DIV_ROUND_UP(priv->tx_bytes,
+				bytes_per_word(priv->bits_per_word));
+	tx_count = min(tx_count, SSI_FIFO_DEPTH);
+
+	/* set fifo threshold */
+	val = readl(priv->base + SSI_FC);
+	val &= ~(SSI_FC_TXFTH_MASK | SSI_FC_RXFTH_MASK);
+	val |= FIELD_PREP(SSI_FC_TXFTH_MASK, tx_count);
+	val |= FIELD_PREP(SSI_FC_RXFTH_MASK, tx_count);
+	writel(val, priv->base + SSI_FC);
+
+	while (tx_count--)
+		uniphier_spi_send(priv);
+}
+
+static void uniphier_spi_set_cs(struct spi_device *spi, bool enable)
+{
+	struct uniphier_spi_priv *priv = spi_master_get_devdata(spi->master);
+	u32 val;
+
+	val = readl(priv->base + SSI_FPS);
+
+	if (enable)
+		val |= SSI_FPS_FSPOL;
+	else
+		val &= ~SSI_FPS_FSPOL;
+
+	writel(val, priv->base + SSI_FPS);
+}
+
+static int uniphier_spi_transfer_one(struct spi_master *master,
+				     struct spi_device *spi,
+				     struct spi_transfer *t)
+{
+	struct uniphier_spi_priv *priv = spi_master_get_devdata(master);
+	int status;
+
+	uniphier_spi_setup_transfer(spi, t);
+
+	reinit_completion(&priv->xfer_done);
+
+	uniphier_spi_fill_tx_fifo(priv);
+
+	uniphier_spi_irq_enable(spi, SSI_IE_RCIE | SSI_IE_RORIE);
+
+	status = wait_for_completion_timeout(&priv->xfer_done,
+					     msecs_to_jiffies(SSI_TIMEOUT_MS));
+
+	uniphier_spi_irq_disable(spi, SSI_IE_RCIE | SSI_IE_RORIE);
+
+	if (status < 0)
+		return status;
+
+	return priv->error;
+}
+
+static int uniphier_spi_prepare_transfer_hardware(struct spi_master *master)
+{
+	struct uniphier_spi_priv *priv = spi_master_get_devdata(master);
+
+	writel(SSI_CTL_EN, priv->base + SSI_CTL);
+
+	return 0;
+}
+
+static int uniphier_spi_unprepare_transfer_hardware(struct spi_master *master)
+{
+	struct uniphier_spi_priv *priv = spi_master_get_devdata(master);
+
+	writel(0, priv->base + SSI_CTL);
+
+	return 0;
+}
+
+static irqreturn_t uniphier_spi_handler(int irq, void *dev_id)
+{
+	struct uniphier_spi_priv *priv = dev_id;
+	u32 val, stat;
+
+	stat = readl(priv->base + SSI_IS);
+	val = SSI_IC_TCIC | SSI_IC_RCIC | SSI_IC_RORIC;
+	writel(val, priv->base + SSI_IC);
+
+	/* rx fifo overrun */
+	if (stat & SSI_IS_RORID) {
+		priv->error = -EIO;
+		goto done;
+	}
+
+	/* rx complete */
+	if ((stat & SSI_IS_RCID) && (stat & SSI_IS_RXRS)) {
+		while ((readl(priv->base + SSI_SR) & SSI_SR_RNE) &&
+				(priv->rx_bytes - priv->tx_bytes) > 0)
+			uniphier_spi_recv(priv);
+
+		if ((readl(priv->base + SSI_SR) & SSI_SR_RNE) ||
+				(priv->rx_bytes != priv->tx_bytes)) {
+			priv->error = -EIO;
+			goto done;
+		} else if (priv->rx_bytes == 0)
+			goto done;
+
+		/* next tx transfer */
+		uniphier_spi_fill_tx_fifo(priv);
+
+		return IRQ_HANDLED;
+	}
+
+	return IRQ_NONE;
+
+done:
+	complete(&priv->xfer_done);
+	return IRQ_HANDLED;
+}
+
+static int uniphier_spi_probe(struct platform_device *pdev)
+{
+	struct uniphier_spi_priv *priv;
+	struct spi_master *master;
+	struct resource *res;
+	unsigned long clk_rate;
+	int irq;
+	int ret;
+
+	master = spi_alloc_master(&pdev->dev, sizeof(*priv));
+	if (!master)
+		return -ENOMEM;
+
+	platform_set_drvdata(pdev, master);
+
+	priv = spi_master_get_devdata(master);
+	priv->master = master;
+	priv->is_save_param = false;
+
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	priv->base = devm_ioremap_resource(&pdev->dev, res);
+	if (IS_ERR(priv->base)) {
+		ret = PTR_ERR(priv->base);
+		goto out_master_put;
+	}
+
+	priv->clk = devm_clk_get(&pdev->dev, NULL);
+	if (IS_ERR(priv->clk)) {
+		dev_err(&pdev->dev, "failed to get clock\n");
+		ret = PTR_ERR(priv->clk);
+		goto out_master_put;
+	}
+
+	ret = clk_prepare_enable(priv->clk);
+	if (ret)
+		goto out_master_put;
+
+	irq = platform_get_irq(pdev, 0);
+	if (irq < 0) {
+		dev_err(&pdev->dev, "failed to get IRQ\n");
+		ret = irq;
+		goto out_disable_clk;
+	}
+
+	ret = devm_request_irq(&pdev->dev, irq, uniphier_spi_handler,
+			       0, "uniphier-spi", priv);
+	if (ret) {
+		dev_err(&pdev->dev, "failed to request IRQ\n");
+		goto out_disable_clk;
+	}
+
+	init_completion(&priv->xfer_done);
+
+	clk_rate = clk_get_rate(priv->clk);
+
+	master->max_speed_hz = DIV_ROUND_UP(clk_rate, SSI_MIN_CLK_DIVIDER);
+	master->min_speed_hz = DIV_ROUND_UP(clk_rate, SSI_MAX_CLK_DIVIDER);
+	master->mode_bits = SPI_CPOL | SPI_CPHA | SPI_CS_HIGH | SPI_LSB_FIRST;
+	master->dev.of_node = pdev->dev.of_node;
+	master->bus_num = pdev->id;
+	master->bits_per_word_mask = SPI_BPW_RANGE_MASK(1, 32);
+
+	master->set_cs = uniphier_spi_set_cs;
+	master->transfer_one = uniphier_spi_transfer_one;
+	master->prepare_transfer_hardware
+				= uniphier_spi_prepare_transfer_hardware;
+	master->unprepare_transfer_hardware
+				= uniphier_spi_unprepare_transfer_hardware;
+	master->num_chipselect = 1;
+
+	ret = devm_spi_register_master(&pdev->dev, master);
+	if (ret)
+		goto out_disable_clk;
+
+	return 0;
+
+out_disable_clk:
+	clk_disable_unprepare(priv->clk);
+
+out_master_put:
+	spi_master_put(master);
+	return ret;
+}
+
+static int uniphier_spi_remove(struct platform_device *pdev)
+{
+	struct uniphier_spi_priv *priv = platform_get_drvdata(pdev);
+
+	clk_disable_unprepare(priv->clk);
+
+	return 0;
+}
+
+static const struct of_device_id uniphier_spi_match[] = {
+	{ .compatible = "socionext,uniphier-scssi" },
+	{ /* sentinel */ }
+};
+MODULE_DEVICE_TABLE(of, uniphier_spi_match);
+
+static struct platform_driver uniphier_spi_driver = {
+	.probe = uniphier_spi_probe,
+	.remove = uniphier_spi_remove,
+	.driver = {
+		.name = "uniphier-spi",
+		.of_match_table = uniphier_spi_match,
+	},
+};
+module_platform_driver(uniphier_spi_driver);
+
+MODULE_AUTHOR("Kunihiko Hayashi <hayashi.kunihiko@socionext.com>");
+MODULE_AUTHOR("Keiji Hayashibara <hayashibara.keiji@socionext.com>");
+MODULE_DESCRIPTION("Socionext UniPhier SPI controller driver");
+MODULE_LICENSE("GPL v2");
diff --git a/drivers/spi/spi-xtensa-xtfpga.c b/drivers/spi/spi-xtensa-xtfpga.c
index be6155cba9de..8ce04f829a80 100644
--- a/drivers/spi/spi-xtensa-xtfpga.c
+++ b/drivers/spi/spi-xtensa-xtfpga.c
@@ -54,7 +54,7 @@ static inline void xtfpga_spi_wait_busy(struct xtfpga_spi *xspi)
 }
 
 static u32 xtfpga_spi_txrx_word(struct spi_device *spi, unsigned nsecs,
-				u32 v, u8 bits)
+				u32 v, u8 bits, unsigned flags)
 {
 	struct xtfpga_spi *xspi = spi_master_get_devdata(spi->master);