Merge tag 'spi-v4.7' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi

Pull spi updates from Mark Brown:
 "Another quiet release for SPI, almost entirely driver specific changes
  with the diffstat dominated by two new drivers which are about two
  thirds of it in terms of lines of code:

   - new drivers for PIC32 standard and SQI controllers
   - the Cadence driver has had runtime PM support added and quite a few
     fixes and cleanups
   - flash-specific accelerated path support now has a feature query
     interface
   - the pxa2xx driver has been moved to use the core DMA mapping support"

* tag 'spi-v4.7' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi: (48 commits)
  spi: pic32-sqi: Fix linker error, undefined reference to `bad_dma_ops'
  spi: dw-pci: Spelling s/paltforms/platforms/g
  spi: pic32-sqi: Remove pic32_sqi_setup and pic32_sqi_cleanup
  spi: Fix simple typo s/impelment/implement
  spi: rockchip: potential NULL dereference on error
  spi: zynqmp: disable clocks in error paths
  spi: Drop unnecessary dependencies on relaxed I/O accessors
  spi: qup: Add spi_master_put in remove function
  spi: qup: Handle clocks in pm_runtime suspend and resume
  spi: st-ssc4: Fix missing spi_master_put in spi_st_probe error paths
  spi: st-ssc4: Allow compile test build
  spi: omap2-mcspi: Use dma_request_chan() for requesting DMA channel
  spi: davinci: Use dma_request_chan() for requesting DMA channel
  spi: pic32: Fix checking return value of devm_ioremap_resource
  spi: spi-fsl-dspi: Update DT binding documentation
  spi: Drop duplicate code to set master->dev.parent
  spi: pic32: Set proper bits_per_word_mask
  spi: return error if kmap'd buffers passed to spi_map_buf()
  spi: core: add hook flash_read_supported to spi_master
  spi: pic32-sqi: silence array overflow warning
  ...

23 files changed, 1975 insertions, 285 deletions

diff --git a/drivers/spi/Kconfig b/drivers/spi/Kconfig
index 9d8c84bb1544..4b931ec8d90b 100644
--- a/drivers/spi/Kconfig
+++ b/drivers/spi/Kconfig
@@ -410,7 +410,6 @@ config SPI_OMAP_UWIRE
 config SPI_OMAP24XX
 	tristate "McSPI driver for OMAP"
 	depends on HAS_DMA
-	depends on ARM || ARM64 || AVR32 || HEXAGON || MIPS || SUPERH
 	depends on ARCH_OMAP2PLUS || COMPILE_TEST
 	help
 	  SPI master controller for OMAP24XX and later Multichannel SPI
@@ -432,10 +431,23 @@ config SPI_OMAP_100K
 
 config SPI_ORION
 	tristate "Orion SPI master"
-	depends on PLAT_ORION || COMPILE_TEST
+	depends on PLAT_ORION || ARCH_MVEBU || COMPILE_TEST
 	help
 	  This enables using the SPI master controller on the Orion chips.
 
+config SPI_PIC32
+	tristate "Microchip PIC32 series SPI"
+	depends on MACH_PIC32 || COMPILE_TEST
+	help
+	  SPI driver for Microchip PIC32 SPI master controller.
+
+config SPI_PIC32_SQI
+	tristate "Microchip PIC32 Quad SPI driver"
+	depends on MACH_PIC32 || COMPILE_TEST
+	depends on HAS_DMA
+	help
+	  SPI driver for PIC32 Quad SPI controller.
+
 config SPI_PL022
 	tristate "ARM AMBA PL022 SSP controller"
 	depends on ARM_AMBA
@@ -469,7 +481,6 @@ config SPI_PXA2XX_PCI
 
 config SPI_ROCKCHIP
 	tristate "Rockchip SPI controller driver"
-	depends on ARM || ARM64 || AVR32 || HEXAGON || MIPS || SUPERH
 	help
 	  This selects a driver for Rockchip SPI controller.
 
@@ -569,7 +580,7 @@ config SPI_SIRF
 
 config SPI_ST_SSC4
 	tristate "STMicroelectronics SPI SSC-based driver"
-	depends on ARCH_STI
+	depends on ARCH_STI || COMPILE_TEST
 	help
 	  STMicroelectronics SoCs support for SPI. If you say yes to
 	  this option, support will be included for the SSC driven SPI.
@@ -656,7 +667,7 @@ config SPI_XILINX
 
 config SPI_XLP
 	tristate "Netlogic XLP SPI controller driver"
-	depends on CPU_XLP || COMPILE_TEST
+	depends on CPU_XLP || ARCH_VULCAN || COMPILE_TEST
 	help
 	  Enable support for the SPI controller on the Netlogic XLP SoCs.
 	  Currently supported XLP variants are XLP8XX, XLP3XX, XLP2XX, XLP9XX
diff --git a/drivers/spi/Makefile b/drivers/spi/Makefile
index fbb255c5a608..3c74d003535b 100644
--- a/drivers/spi/Makefile
+++ b/drivers/spi/Makefile
@@ -62,6 +62,8 @@ obj-$(CONFIG_SPI_OMAP_100K)		+= spi-omap-100k.o
 obj-$(CONFIG_SPI_OMAP24XX)		+= spi-omap2-mcspi.o
 obj-$(CONFIG_SPI_TI_QSPI)		+= spi-ti-qspi.o
 obj-$(CONFIG_SPI_ORION)			+= spi-orion.o
+obj-$(CONFIG_SPI_PIC32)			+= spi-pic32.o
+obj-$(CONFIG_SPI_PIC32_SQI)		+= spi-pic32-sqi.o
 obj-$(CONFIG_SPI_PL022)			+= spi-pl022.o
 obj-$(CONFIG_SPI_PPC4xx)		+= spi-ppc4xx.o
 spi-pxa2xx-platform-objs		:= spi-pxa2xx.o spi-pxa2xx-dma.o
diff --git a/drivers/spi/spi-axi-spi-engine.c b/drivers/spi/spi-axi-spi-engine.c
index c968ab210a51..2b1456e5e221 100644
--- a/drivers/spi/spi-axi-spi-engine.c
+++ b/drivers/spi/spi-axi-spi-engine.c
@@ -525,7 +525,6 @@ static int spi_engine_probe(struct platform_device *pdev)
 	if (ret)
 		goto err_ref_clk_disable;
 
-	master->dev.parent = &pdev->dev;
 	master->dev.of_node = pdev->dev.of_node;
 	master->mode_bits = SPI_CPOL | SPI_CPHA | SPI_3WIRE;
 	master->bits_per_word_mask = SPI_BPW_MASK(8);
diff --git a/drivers/spi/spi-bcm53xx.c b/drivers/spi/spi-bcm53xx.c
index cc3f938f0a6b..afb51699dbb5 100644
--- a/drivers/spi/spi-bcm53xx.c
+++ b/drivers/spi/spi-bcm53xx.c
@@ -10,6 +10,7 @@
 #include "spi-bcm53xx.h"
 
 #define BCM53XXSPI_MAX_SPI_BAUD	13500000	/* 216 MHz? */
+#define BCM53XXSPI_FLASH_WINDOW	SZ_32M
 
 /* The longest observed required wait was 19 ms */
 #define BCM53XXSPI_SPE_TIMEOUT_MS	80
@@ -17,8 +18,10 @@
 struct bcm53xxspi {
 	struct bcma_device *core;
 	struct spi_master *master;
+	void __iomem *mmio_base;
 
 	size_t read_offset;
+	bool bspi;				/* Boot SPI mode with memory mapping */
 };
 
 static inline u32 bcm53xxspi_read(struct bcm53xxspi *b53spi, u16 offset)
@@ -32,6 +35,50 @@ static inline void bcm53xxspi_write(struct bcm53xxspi *b53spi, u16 offset,
 	bcma_write32(b53spi->core, offset, value);
 }
 
+static void bcm53xxspi_disable_bspi(struct bcm53xxspi *b53spi)
+{
+	struct device *dev = &b53spi->core->dev;
+	unsigned long deadline;
+	u32 tmp;
+
+	if (!b53spi->bspi)
+		return;
+
+	tmp = bcm53xxspi_read(b53spi, B53SPI_BSPI_MAST_N_BOOT_CTRL);
+	if (tmp & 0x1)
+		return;
+
+	deadline = jiffies + usecs_to_jiffies(200);
+	do {
+		tmp = bcm53xxspi_read(b53spi, B53SPI_BSPI_BUSY_STATUS);
+		if (!(tmp & 0x1)) {
+			bcm53xxspi_write(b53spi, B53SPI_BSPI_MAST_N_BOOT_CTRL,
+					 0x1);
+			ndelay(200);
+			b53spi->bspi = false;
+			return;
+		}
+		udelay(1);
+	} while (!time_after_eq(jiffies, deadline));
+
+	dev_warn(dev, "Timeout disabling BSPI\n");
+}
+
+static void bcm53xxspi_enable_bspi(struct bcm53xxspi *b53spi)
+{
+	u32 tmp;
+
+	if (b53spi->bspi)
+		return;
+
+	tmp = bcm53xxspi_read(b53spi, B53SPI_BSPI_MAST_N_BOOT_CTRL);
+	if (!(tmp & 0x1))
+		return;
+
+	bcm53xxspi_write(b53spi, B53SPI_BSPI_MAST_N_BOOT_CTRL, 0x0);
+	b53spi->bspi = true;
+}
+
 static inline unsigned int bcm53xxspi_calc_timeout(size_t len)
 {
 	/* Do some magic calculation based on length and buad. Add 10% and 1. */
@@ -176,6 +223,8 @@ static int bcm53xxspi_transfer_one(struct spi_master *master,
 	u8 *buf;
 	size_t left;
 
+	bcm53xxspi_disable_bspi(b53spi);
+
 	if (t->tx_buf) {
 		buf = (u8 *)t->tx_buf;
 		left = t->len;
@@ -206,6 +255,22 @@ static int bcm53xxspi_transfer_one(struct spi_master *master,
 	return 0;
 }
 
+static int bcm53xxspi_flash_read(struct spi_device *spi,
+				 struct spi_flash_read_message *msg)
+{
+	struct bcm53xxspi *b53spi = spi_master_get_devdata(spi->master);
+	int ret = 0;
+
+	if (msg->from + msg->len > BCM53XXSPI_FLASH_WINDOW)
+		return -EINVAL;
+
+	bcm53xxspi_enable_bspi(b53spi);
+	memcpy_fromio(msg->buf, b53spi->mmio_base + msg->from, msg->len);
+	msg->retlen = msg->len;
+
+	return ret;
+}
+
 /**************************************************
  * BCMA
  **************************************************/
@@ -222,6 +287,7 @@ MODULE_DEVICE_TABLE(bcma, bcm53xxspi_bcma_tbl);
 
 static int bcm53xxspi_bcma_probe(struct bcma_device *core)
 {
+	struct device *dev = &core->dev;
 	struct bcm53xxspi *b53spi;
 	struct spi_master *master;
 	int err;
@@ -231,7 +297,7 @@ static int bcm53xxspi_bcma_probe(struct bcma_device *core)
 		return -ENOTSUPP;
 	}
 
-	master = spi_alloc_master(&core->dev, sizeof(*b53spi));
+	master = spi_alloc_master(dev, sizeof(*b53spi));
 	if (!master)
 		return -ENOMEM;
 
@@ -239,11 +305,19 @@ static int bcm53xxspi_bcma_probe(struct bcma_device *core)
 	b53spi->master = master;
 	b53spi->core = core;
 
+	if (core->addr_s[0])
+		b53spi->mmio_base = devm_ioremap(dev, core->addr_s[0],
+						 BCM53XXSPI_FLASH_WINDOW);
+	b53spi->bspi = true;
+	bcm53xxspi_disable_bspi(b53spi);
+
 	master->transfer_one = bcm53xxspi_transfer_one;
+	if (b53spi->mmio_base)
+		master->spi_flash_read = bcm53xxspi_flash_read;
 
 	bcma_set_drvdata(core, b53spi);
 
-	err = devm_spi_register_master(&core->dev, master);
+	err = devm_spi_register_master(dev, master);
 	if (err) {
 		spi_master_put(master);
 		bcma_set_drvdata(core, NULL);
diff --git a/drivers/spi/spi-cadence.c b/drivers/spi/spi-cadence.c
index 121a4135b540..1c57ce64abba 100644
--- a/drivers/spi/spi-cadence.c
+++ b/drivers/spi/spi-cadence.c
@@ -19,44 +19,46 @@
 #include <linux/of_irq.h>
 #include <linux/of_address.h>
 #include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
 #include <linux/spi/spi.h>
 
 /* Name of this driver */
 #define CDNS_SPI_NAME		"cdns-spi"
 
 /* Register offset definitions */
-#define CDNS_SPI_CR_OFFSET	0x00 /* Configuration  Register, RW */
-#define CDNS_SPI_ISR_OFFSET	0x04 /* Interrupt Status Register, RO */
-#define CDNS_SPI_IER_OFFSET	0x08 /* Interrupt Enable Register, WO */
-#define CDNS_SPI_IDR_OFFSET	0x0c /* Interrupt Disable Register, WO */
-#define CDNS_SPI_IMR_OFFSET	0x10 /* Interrupt Enabled Mask Register, RO */
-#define CDNS_SPI_ER_OFFSET	0x14 /* Enable/Disable Register, RW */
-#define CDNS_SPI_DR_OFFSET	0x18 /* Delay Register, RW */
-#define CDNS_SPI_TXD_OFFSET	0x1C /* Data Transmit Register, WO */
-#define CDNS_SPI_RXD_OFFSET	0x20 /* Data Receive Register, RO */
-#define CDNS_SPI_SICR_OFFSET	0x24 /* Slave Idle Count Register, RW */
-#define CDNS_SPI_THLD_OFFSET	0x28 /* Transmit FIFO Watermark Register,RW */
-
+#define CDNS_SPI_CR	0x00 /* Configuration  Register, RW */
+#define CDNS_SPI_ISR	0x04 /* Interrupt Status Register, RO */
+#define CDNS_SPI_IER	0x08 /* Interrupt Enable Register, WO */
+#define CDNS_SPI_IDR	0x0c /* Interrupt Disable Register, WO */
+#define CDNS_SPI_IMR	0x10 /* Interrupt Enabled Mask Register, RO */
+#define CDNS_SPI_ER	0x14 /* Enable/Disable Register, RW */
+#define CDNS_SPI_DR	0x18 /* Delay Register, RW */
+#define CDNS_SPI_TXD	0x1C /* Data Transmit Register, WO */
+#define CDNS_SPI_RXD	0x20 /* Data Receive Register, RO */
+#define CDNS_SPI_SICR	0x24 /* Slave Idle Count Register, RW */
+#define CDNS_SPI_THLD	0x28 /* Transmit FIFO Watermark Register,RW */
+
+#define SPI_AUTOSUSPEND_TIMEOUT		3000
 /*
  * SPI Configuration Register bit Masks
  *
  * This register contains various control bits that affect the operation
  * of the SPI controller
  */
-#define CDNS_SPI_CR_MANSTRT_MASK	0x00010000 /* Manual TX Start */
-#define CDNS_SPI_CR_CPHA_MASK		0x00000004 /* Clock Phase Control */
-#define CDNS_SPI_CR_CPOL_MASK		0x00000002 /* Clock Polarity Control */
-#define CDNS_SPI_CR_SSCTRL_MASK		0x00003C00 /* Slave Select Mask */
-#define CDNS_SPI_CR_PERI_SEL_MASK	0x00000200 /* Peripheral Select Decode */
-#define CDNS_SPI_CR_BAUD_DIV_MASK	0x00000038 /* Baud Rate Divisor Mask */
-#define CDNS_SPI_CR_MSTREN_MASK		0x00000001 /* Master Enable Mask */
-#define CDNS_SPI_CR_MANSTRTEN_MASK	0x00008000 /* Manual TX Enable Mask */
-#define CDNS_SPI_CR_SSFORCE_MASK	0x00004000 /* Manual SS Enable Mask */
-#define CDNS_SPI_CR_BAUD_DIV_4_MASK	0x00000008 /* Default Baud Div Mask */
-#define CDNS_SPI_CR_DEFAULT_MASK	(CDNS_SPI_CR_MSTREN_MASK | \
-					CDNS_SPI_CR_SSCTRL_MASK | \
-					CDNS_SPI_CR_SSFORCE_MASK | \
-					CDNS_SPI_CR_BAUD_DIV_4_MASK)
+#define CDNS_SPI_CR_MANSTRT	0x00010000 /* Manual TX Start */
+#define CDNS_SPI_CR_CPHA		0x00000004 /* Clock Phase Control */
+#define CDNS_SPI_CR_CPOL		0x00000002 /* Clock Polarity Control */
+#define CDNS_SPI_CR_SSCTRL		0x00003C00 /* Slave Select Mask */
+#define CDNS_SPI_CR_PERI_SEL	0x00000200 /* Peripheral Select Decode */
+#define CDNS_SPI_CR_BAUD_DIV	0x00000038 /* Baud Rate Divisor Mask */
+#define CDNS_SPI_CR_MSTREN		0x00000001 /* Master Enable Mask */
+#define CDNS_SPI_CR_MANSTRTEN	0x00008000 /* Manual TX Enable Mask */
+#define CDNS_SPI_CR_SSFORCE	0x00004000 /* Manual SS Enable Mask */
+#define CDNS_SPI_CR_BAUD_DIV_4	0x00000008 /* Default Baud Div Mask */
+#define CDNS_SPI_CR_DEFAULT	(CDNS_SPI_CR_MSTREN | \
+					CDNS_SPI_CR_SSCTRL | \
+					CDNS_SPI_CR_SSFORCE | \
+					CDNS_SPI_CR_BAUD_DIV_4)
 
 /*
  * SPI Configuration Register - Baud rate and slave select
@@ -77,21 +79,21 @@
  * All the four interrupt registers (Status/Mask/Enable/Disable) have the same
  * bit definitions.
  */
-#define CDNS_SPI_IXR_TXOW_MASK	0x00000004 /* SPI TX FIFO Overwater */
-#define CDNS_SPI_IXR_MODF_MASK	0x00000002 /* SPI Mode Fault */
-#define CDNS_SPI_IXR_RXNEMTY_MASK 0x00000010 /* SPI RX FIFO Not Empty */
-#define CDNS_SPI_IXR_DEFAULT_MASK	(CDNS_SPI_IXR_TXOW_MASK | \
-					CDNS_SPI_IXR_MODF_MASK)
-#define CDNS_SPI_IXR_TXFULL_MASK	0x00000008 /* SPI TX Full */
-#define CDNS_SPI_IXR_ALL_MASK	0x0000007F /* SPI all interrupts */
+#define CDNS_SPI_IXR_TXOW	0x00000004 /* SPI TX FIFO Overwater */
+#define CDNS_SPI_IXR_MODF	0x00000002 /* SPI Mode Fault */
+#define CDNS_SPI_IXR_RXNEMTY 0x00000010 /* SPI RX FIFO Not Empty */
+#define CDNS_SPI_IXR_DEFAULT	(CDNS_SPI_IXR_TXOW | \
+					CDNS_SPI_IXR_MODF)
+#define CDNS_SPI_IXR_TXFULL	0x00000008 /* SPI TX Full */
+#define CDNS_SPI_IXR_ALL	0x0000007F /* SPI all interrupts */
 
 /*
  * SPI Enable Register bit Masks
  *
  * This register is used to enable or disable the SPI controller
  */
-#define CDNS_SPI_ER_ENABLE_MASK	0x00000001 /* SPI Enable Bit Mask */
-#define CDNS_SPI_ER_DISABLE_MASK	0x0 /* SPI Disable Bit Mask */
+#define CDNS_SPI_ER_ENABLE	0x00000001 /* SPI Enable Bit Mask */
+#define CDNS_SPI_ER_DISABLE	0x0 /* SPI Disable Bit Mask */
 
 /* SPI FIFO depth in bytes */
 #define CDNS_SPI_FIFO_DEPTH	128
@@ -149,56 +151,51 @@ static inline void cdns_spi_write(struct cdns_spi *xspi, u32 offset, u32 val)
  */
 static void cdns_spi_init_hw(struct cdns_spi *xspi)
 {
-	u32 ctrl_reg = CDNS_SPI_CR_DEFAULT_MASK;
+	u32 ctrl_reg = CDNS_SPI_CR_DEFAULT;
 
 	if (xspi->is_decoded_cs)
-		ctrl_reg |= CDNS_SPI_CR_PERI_SEL_MASK;
+		ctrl_reg |= CDNS_SPI_CR_PERI_SEL;
 
-	cdns_spi_write(xspi, CDNS_SPI_ER_OFFSET,
-		       CDNS_SPI_ER_DISABLE_MASK);
-	cdns_spi_write(xspi, CDNS_SPI_IDR_OFFSET,
-		       CDNS_SPI_IXR_ALL_MASK);
+	cdns_spi_write(xspi, CDNS_SPI_ER, CDNS_SPI_ER_DISABLE);
+	cdns_spi_write(xspi, CDNS_SPI_IDR, CDNS_SPI_IXR_ALL);
 
 	/* Clear the RX FIFO */
-	while (cdns_spi_read(xspi, CDNS_SPI_ISR_OFFSET) &
-	       CDNS_SPI_IXR_RXNEMTY_MASK)
-		cdns_spi_read(xspi, CDNS_SPI_RXD_OFFSET);
-
-	cdns_spi_write(xspi, CDNS_SPI_ISR_OFFSET,
-		       CDNS_SPI_IXR_ALL_MASK);
-	cdns_spi_write(xspi, CDNS_SPI_CR_OFFSET, ctrl_reg);
-	cdns_spi_write(xspi, CDNS_SPI_ER_OFFSET,
-		       CDNS_SPI_ER_ENABLE_MASK);
+	while (cdns_spi_read(xspi, CDNS_SPI_ISR) & CDNS_SPI_IXR_RXNEMTY)
+		cdns_spi_read(xspi, CDNS_SPI_RXD);
+
+	cdns_spi_write(xspi, CDNS_SPI_ISR, CDNS_SPI_IXR_ALL);
+	cdns_spi_write(xspi, CDNS_SPI_CR, ctrl_reg);
+	cdns_spi_write(xspi, CDNS_SPI_ER, CDNS_SPI_ER_ENABLE);
 }
 
 /**
  * cdns_spi_chipselect - Select or deselect the chip select line
  * @spi:	Pointer to the spi_device structure
- * @is_on:	Select(0) or deselect (1) the chip select line
+ * @is_high:	Select(0) or deselect (1) the chip select line
  */
 static void cdns_spi_chipselect(struct spi_device *spi, bool is_high)
 {
 	struct cdns_spi *xspi = spi_master_get_devdata(spi->master);
 	u32 ctrl_reg;
 
-	ctrl_reg = cdns_spi_read(xspi, CDNS_SPI_CR_OFFSET);
+	ctrl_reg = cdns_spi_read(xspi, CDNS_SPI_CR);
 
 	if (is_high) {
 		/* Deselect the slave */
-		ctrl_reg |= CDNS_SPI_CR_SSCTRL_MASK;
+		ctrl_reg |= CDNS_SPI_CR_SSCTRL;
 	} else {
 		/* Select the slave */
-		ctrl_reg &= ~CDNS_SPI_CR_SSCTRL_MASK;
+		ctrl_reg &= ~CDNS_SPI_CR_SSCTRL;
 		if (!(xspi->is_decoded_cs))
 			ctrl_reg |= ((~(CDNS_SPI_SS0 << spi->chip_select)) <<
 				     CDNS_SPI_SS_SHIFT) &
-				     CDNS_SPI_CR_SSCTRL_MASK;
+				     CDNS_SPI_CR_SSCTRL;
 		else
 			ctrl_reg |= (spi->chip_select << CDNS_SPI_SS_SHIFT) &
-				     CDNS_SPI_CR_SSCTRL_MASK;
+				     CDNS_SPI_CR_SSCTRL;
 	}
 
-	cdns_spi_write(xspi, CDNS_SPI_CR_OFFSET, ctrl_reg);
+	cdns_spi_write(xspi, CDNS_SPI_CR, ctrl_reg);
 }
 
 /**
@@ -212,14 +209,15 @@ static void cdns_spi_config_clock_mode(struct spi_device *spi)
 	struct cdns_spi *xspi = spi_master_get_devdata(spi->master);
 	u32 ctrl_reg, new_ctrl_reg;
 
-	new_ctrl_reg = ctrl_reg = cdns_spi_read(xspi, CDNS_SPI_CR_OFFSET);
+	new_ctrl_reg = cdns_spi_read(xspi, CDNS_SPI_CR);
+	ctrl_reg = new_ctrl_reg;
 
 	/* Set the SPI clock phase and clock polarity */
-	new_ctrl_reg &= ~(CDNS_SPI_CR_CPHA_MASK | CDNS_SPI_CR_CPOL_MASK);
+	new_ctrl_reg &= ~(CDNS_SPI_CR_CPHA | CDNS_SPI_CR_CPOL);
 	if (spi->mode & SPI_CPHA)
-		new_ctrl_reg |= CDNS_SPI_CR_CPHA_MASK;
+		new_ctrl_reg |= CDNS_SPI_CR_CPHA;
 	if (spi->mode & SPI_CPOL)
-		new_ctrl_reg |= CDNS_SPI_CR_CPOL_MASK;
+		new_ctrl_reg |= CDNS_SPI_CR_CPOL;
 
 	if (new_ctrl_reg != ctrl_reg) {
 		/*
@@ -228,11 +226,9 @@ static void cdns_spi_config_clock_mode(struct spi_device *spi)
 		 * polarity as it will cause the SPI slave to see spurious clock
 		 * transitions. To workaround the issue toggle the ER register.
 		 */
-		cdns_spi_write(xspi, CDNS_SPI_ER_OFFSET,
-				   CDNS_SPI_ER_DISABLE_MASK);
-		cdns_spi_write(xspi, CDNS_SPI_CR_OFFSET, new_ctrl_reg);
-		cdns_spi_write(xspi, CDNS_SPI_ER_OFFSET,
-				   CDNS_SPI_ER_ENABLE_MASK);
+		cdns_spi_write(xspi, CDNS_SPI_ER, CDNS_SPI_ER_DISABLE);
+		cdns_spi_write(xspi, CDNS_SPI_CR, new_ctrl_reg);
+		cdns_spi_write(xspi, CDNS_SPI_ER, CDNS_SPI_ER_ENABLE);
 	}
 }
 
@@ -251,7 +247,7 @@ static void cdns_spi_config_clock_mode(struct spi_device *spi)
  * controller.
  */
 static void cdns_spi_config_clock_freq(struct spi_device *spi,
-				  struct spi_transfer *transfer)
+				       struct spi_transfer *transfer)
 {
 	struct cdns_spi *xspi = spi_master_get_devdata(spi->master);
 	u32 ctrl_reg, baud_rate_val;
@@ -259,7 +255,7 @@ static void cdns_spi_config_clock_freq(struct spi_device *spi,
 
 	frequency = clk_get_rate(xspi->ref_clk);
 
-	ctrl_reg = cdns_spi_read(xspi, CDNS_SPI_CR_OFFSET);
+	ctrl_reg = cdns_spi_read(xspi, CDNS_SPI_CR);
 
 	/* Set the clock frequency */
 	if (xspi->speed_hz != transfer->speed_hz) {
@@ -269,12 +265,12 @@ static void cdns_spi_config_clock_freq(struct spi_device *spi,
 		       (frequency / (2 << baud_rate_val)) > transfer->speed_hz)
 			baud_rate_val++;
 
-		ctrl_reg &= ~CDNS_SPI_CR_BAUD_DIV_MASK;
+		ctrl_reg &= ~CDNS_SPI_CR_BAUD_DIV;
 		ctrl_reg |= baud_rate_val << CDNS_SPI_BAUD_DIV_SHIFT;
 
 		xspi->speed_hz = frequency / (2 << baud_rate_val);
 	}
-	cdns_spi_write(xspi, CDNS_SPI_CR_OFFSET, ctrl_reg);
+	cdns_spi_write(xspi, CDNS_SPI_CR, ctrl_reg);
 }
 
 /**
@@ -313,10 +309,9 @@ static void cdns_spi_fill_tx_fifo(struct cdns_spi *xspi)
 	while ((trans_cnt < CDNS_SPI_FIFO_DEPTH) &&
 	       (xspi->tx_bytes > 0)) {
 		if (xspi->txbuf)
-			cdns_spi_write(xspi, CDNS_SPI_TXD_OFFSET,
-				       *xspi->txbuf++);
+			cdns_spi_write(xspi, CDNS_SPI_TXD, *xspi->txbuf++);
 		else
-			cdns_spi_write(xspi, CDNS_SPI_TXD_OFFSET, 0);
+			cdns_spi_write(xspi, CDNS_SPI_TXD, 0);
 
 		xspi->tx_bytes--;
 		trans_cnt++;
@@ -344,19 +339,18 @@ static irqreturn_t cdns_spi_irq(int irq, void *dev_id)
 	u32 intr_status, status;
 
 	status = IRQ_NONE;
-	intr_status = cdns_spi_read(xspi, CDNS_SPI_ISR_OFFSET);
-	cdns_spi_write(xspi, CDNS_SPI_ISR_OFFSET, intr_status);
+	intr_status = cdns_spi_read(xspi, CDNS_SPI_ISR);
+	cdns_spi_write(xspi, CDNS_SPI_ISR, intr_status);
 
-	if (intr_status & CDNS_SPI_IXR_MODF_MASK) {
+	if (intr_status & CDNS_SPI_IXR_MODF) {
 		/* Indicate that transfer is completed, the SPI subsystem will
 		 * identify the error as the remaining bytes to be
 		 * transferred is non-zero
 		 */
-		cdns_spi_write(xspi, CDNS_SPI_IDR_OFFSET,
-			       CDNS_SPI_IXR_DEFAULT_MASK);
+		cdns_spi_write(xspi, CDNS_SPI_IDR, CDNS_SPI_IXR_DEFAULT);
 		spi_finalize_current_transfer(master);
 		status = IRQ_HANDLED;
-	} else if (intr_status & CDNS_SPI_IXR_TXOW_MASK) {
+	} else if (intr_status & CDNS_SPI_IXR_TXOW) {
 		unsigned long trans_cnt;
 
 		trans_cnt = xspi->rx_bytes - xspi->tx_bytes;
@@ -365,7 +359,7 @@ static irqreturn_t cdns_spi_irq(int irq, void *dev_id)
 		while (trans_cnt) {
 			u8 data;
 
-			data = cdns_spi_read(xspi, CDNS_SPI_RXD_OFFSET);
+			data = cdns_spi_read(xspi, CDNS_SPI_RXD);
 			if (xspi->rxbuf)
 				*xspi->rxbuf++ = data;
 
@@ -378,8 +372,8 @@ static irqreturn_t cdns_spi_irq(int irq, void *dev_id)
 			cdns_spi_fill_tx_fifo(xspi);
 		} else {
 			/* Transfer is completed */
-			cdns_spi_write(xspi, CDNS_SPI_IDR_OFFSET,
-				       CDNS_SPI_IXR_DEFAULT_MASK);
+			cdns_spi_write(xspi, CDNS_SPI_IDR,
+				       CDNS_SPI_IXR_DEFAULT);
 			spi_finalize_current_transfer(master);
 		}
 		status = IRQ_HANDLED;
@@ -387,6 +381,7 @@ static irqreturn_t cdns_spi_irq(int irq, void *dev_id)
 
 	return status;
 }
+
 static int cdns_prepare_message(struct spi_master *master,
 				struct spi_message *msg)
 {
@@ -421,8 +416,7 @@ static int cdns_transfer_one(struct spi_master *master,
 
 	cdns_spi_fill_tx_fifo(xspi);
 
-	cdns_spi_write(xspi, CDNS_SPI_IER_OFFSET,
-		       CDNS_SPI_IXR_DEFAULT_MASK);
+	cdns_spi_write(xspi, CDNS_SPI_IER, CDNS_SPI_IXR_DEFAULT);
 	return transfer->len;
 }
 
@@ -439,8 +433,7 @@ static int cdns_prepare_transfer_hardware(struct spi_master *master)
 {
 	struct cdns_spi *xspi = spi_master_get_devdata(master);
 
-	cdns_spi_write(xspi, CDNS_SPI_ER_OFFSET,
-		       CDNS_SPI_ER_ENABLE_MASK);
+	cdns_spi_write(xspi, CDNS_SPI_ER, CDNS_SPI_ER_ENABLE);
 
 	return 0;
 }
@@ -458,8 +451,7 @@ static int cdns_unprepare_transfer_hardware(struct spi_master *master)
 {
 	struct cdns_spi *xspi = spi_master_get_devdata(master);
 
-	cdns_spi_write(xspi, CDNS_SPI_ER_OFFSET,
-		       CDNS_SPI_ER_DISABLE_MASK);
+	cdns_spi_write(xspi, CDNS_SPI_ER, CDNS_SPI_ER_DISABLE);
 
 	return 0;
 }
@@ -481,7 +473,7 @@ static int cdns_spi_probe(struct platform_device *pdev)
 	u32 num_cs;
 
 	master = spi_alloc_master(&pdev->dev, sizeof(*xspi));
-	if (master == NULL)
+	if (!master)
 		return -ENOMEM;
 
 	xspi = spi_master_get_devdata(master);
@@ -521,6 +513,11 @@ static int cdns_spi_probe(struct platform_device *pdev)
 		goto clk_dis_apb;
 	}
 
+	pm_runtime_enable(&pdev->dev);
+	pm_runtime_use_autosuspend(&pdev->dev);
+	pm_runtime_set_autosuspend_delay(&pdev->dev, SPI_AUTOSUSPEND_TIMEOUT);
+	pm_runtime_set_active(&pdev->dev);
+
 	ret = of_property_read_u32(pdev->dev.of_node, "num-cs", &num_cs);
 	if (ret < 0)
 		master->num_chipselect = CDNS_SPI_DEFAULT_NUM_CS;
@@ -535,11 +532,14 @@ static int cdns_spi_probe(struct platform_device *pdev)
 	/* SPI controller initializations */
 	cdns_spi_init_hw(xspi);
 
+	pm_runtime_mark_last_busy(&pdev->dev);
+	pm_runtime_put_autosuspend(&pdev->dev);
+
 	irq = platform_get_irq(pdev, 0);
 	if (irq <= 0) {
 		ret = -ENXIO;
 		dev_err(&pdev->dev, "irq number is invalid\n");
-		goto remove_master;
+		goto clk_dis_all;
 	}
 
 	ret = devm_request_irq(&pdev->dev, irq, cdns_spi_irq,
@@ -547,7 +547,7 @@ static int cdns_spi_probe(struct platform_device *pdev)
 	if (ret != 0) {
 		ret = -ENXIO;
 		dev_err(&pdev->dev, "request_irq failed\n");
-		goto remove_master;
+		goto clk_dis_all;
 	}
 
 	master->prepare_transfer_hardware = cdns_prepare_transfer_hardware;
@@ -555,6 +555,7 @@ static int cdns_spi_probe(struct platform_device *pdev)
 	master->transfer_one = cdns_transfer_one;
 	master->unprepare_transfer_hardware = cdns_unprepare_transfer_hardware;
 	master->set_cs = cdns_spi_chipselect;
+	master->auto_runtime_pm = true;
 	master->mode_bits = SPI_CPOL | SPI_CPHA;
 
 	/* Set to default valid value */
@@ -572,6 +573,8 @@ static int cdns_spi_probe(struct platform_device *pdev)
 	return ret;
 
 clk_dis_all:
+	pm_runtime_set_suspended(&pdev->dev);
+	pm_runtime_disable(&pdev->dev);
 	clk_disable_unprepare(xspi->ref_clk);
 clk_dis_apb:
 	clk_disable_unprepare(xspi->pclk);
@@ -595,11 +598,12 @@ static int cdns_spi_remove(struct platform_device *pdev)
 	struct spi_master *master = platform_get_drvdata(pdev);
 	struct cdns_spi *xspi = spi_master_get_devdata(master);
 
-	cdns_spi_write(xspi, CDNS_SPI_ER_OFFSET,
-		       CDNS_SPI_ER_DISABLE_MASK);
+	cdns_spi_write(xspi, CDNS_SPI_ER, CDNS_SPI_ER_DISABLE);
 
 	clk_disable_unprepare(xspi->ref_clk);
 	clk_disable_unprepare(xspi->pclk);
+	pm_runtime_set_suspended(&pdev->dev);
+	pm_runtime_disable(&pdev->dev);
 
 	spi_unregister_master(master);
 
@@ -613,21 +617,14 @@ static int cdns_spi_remove(struct platform_device *pdev)
  * This function disables the SPI controller and
  * changes the driver state to "suspend"
  *
- * Return:	Always 0
+ * Return:	0 on success and error value on error
  */
 static int __maybe_unused cdns_spi_suspend(struct device *dev)
 {
 	struct platform_device *pdev = to_platform_device(dev);
 	struct spi_master *master = platform_get_drvdata(pdev);
-	struct cdns_spi *xspi = spi_master_get_devdata(master);
-
-	spi_master_suspend(master);
-
-	clk_disable_unprepare(xspi->ref_clk);
-
-	clk_disable_unprepare(xspi->pclk);
 
-	return 0;
+	return spi_master_suspend(master);
 }
 
 /**
@@ -642,8 +639,23 @@ static int __maybe_unused cdns_spi_resume(struct device *dev)
 {
 	struct platform_device *pdev = to_platform_device(dev);
 	struct spi_master *master = platform_get_drvdata(pdev);
+
+	return spi_master_resume(master);
+}
+
+/**
+ * cdns_spi_runtime_resume - Runtime resume method for the SPI driver
+ * @dev:	Address of the platform_device structure
+ *
+ * This function enables the clocks
+ *
+ * Return:	0 on success and error value on error
+ */
+static int __maybe_unused cnds_runtime_resume(struct device *dev)
+{
+	struct spi_master *master = dev_get_drvdata(dev);
 	struct cdns_spi *xspi = spi_master_get_devdata(master);
-	int ret = 0;
+	int ret;
 
 	ret = clk_prepare_enable(xspi->pclk);
 	if (ret) {
@@ -657,13 +669,33 @@ static int __maybe_unused cdns_spi_resume(struct device *dev)
 		clk_disable(xspi->pclk);
 		return ret;
 	}
-	spi_master_resume(master);
+	return 0;
+}
+
+/**
+ * cdns_spi_runtime_suspend - Runtime suspend method for the SPI driver
+ * @dev:	Address of the platform_device structure
+ *
+ * This function disables the clocks
+ *
+ * Return:	Always 0
+ */
+static int __maybe_unused cnds_runtime_suspend(struct device *dev)
+{
+	struct spi_master *master = dev_get_drvdata(dev);
+	struct cdns_spi *xspi = spi_master_get_devdata(master);
+
+	clk_disable_unprepare(xspi->ref_clk);
+	clk_disable_unprepare(xspi->pclk);
 
 	return 0;
 }
 
-static SIMPLE_DEV_PM_OPS(cdns_spi_dev_pm_ops, cdns_spi_suspend,
-			 cdns_spi_resume);
+static const struct dev_pm_ops cdns_spi_dev_pm_ops = {
+	SET_RUNTIME_PM_OPS(cnds_runtime_suspend,
+			   cnds_runtime_resume, NULL)
+	SET_SYSTEM_SLEEP_PM_OPS(cdns_spi_suspend, cdns_spi_resume)
+};
 
 static const struct of_device_id cdns_spi_of_match[] = {
 	{ .compatible = "xlnx,zynq-spi-r1p6" },
diff --git a/drivers/spi/spi-davinci.c b/drivers/spi/spi-davinci.c
index fddb7a3be322..d36c11b73a35 100644
--- a/drivers/spi/spi-davinci.c
+++ b/drivers/spi/spi-davinci.c
@@ -23,7 +23,6 @@
 #include <linux/clk.h>
 #include <linux/dmaengine.h>
 #include <linux/dma-mapping.h>
-#include <linux/edma.h>
 #include <linux/of.h>
 #include <linux/of_device.h>
 #include <linux/of_gpio.h>
@@ -33,8 +32,6 @@
 
 #include <linux/platform_data/spi-davinci.h>
 
-#define SPI_NO_RESOURCE		((resource_size_t)-1)
-
 #define CS_DEFAULT	0xFF
 
 #define SPIFMT_PHASE_MASK	BIT(16)
@@ -130,8 +127,6 @@ struct davinci_spi {
 
 	struct dma_chan		*dma_rx;
 	struct dma_chan		*dma_tx;
-	int			dma_rx_chnum;
-	int			dma_tx_chnum;
 
 	struct davinci_spi_platform_data pdata;
 
@@ -797,35 +792,19 @@ static irqreturn_t davinci_spi_irq(s32 irq, void *data)
 
 static int davinci_spi_request_dma(struct davinci_spi *dspi)
 {
-	dma_cap_mask_t mask;
 	struct device *sdev = dspi->bitbang.master->dev.parent;
-	int r;
-
-	dma_cap_zero(mask);
-	dma_cap_set(DMA_SLAVE, mask);
 
-	dspi->dma_rx = dma_request_channel(mask, edma_filter_fn,
-					   &dspi->dma_rx_chnum);
-	if (!dspi->dma_rx) {
-		dev_err(sdev, "request RX DMA channel failed\n");
-		r = -ENODEV;
-		goto rx_dma_failed;
-	}
+	dspi->dma_rx = dma_request_chan(sdev, "rx");
+	if (IS_ERR(dspi->dma_rx))
+		return PTR_ERR(dspi->dma_rx);
 
-	dspi->dma_tx = dma_request_channel(mask, edma_filter_fn,
-					   &dspi->dma_tx_chnum);
-	if (!dspi->dma_tx) {
-		dev_err(sdev, "request TX DMA channel failed\n");
-		r = -ENODEV;
-		goto tx_dma_failed;
+	dspi->dma_tx = dma_request_chan(sdev, "tx");
+	if (IS_ERR(dspi->dma_tx)) {
+		dma_release_channel(dspi->dma_rx);
+		return PTR_ERR(dspi->dma_tx);
 	}
 
 	return 0;
-
-tx_dma_failed:
-	dma_release_channel(dspi->dma_rx);
-rx_dma_failed:
-	return r;
 }
 
 #if defined(CONFIG_OF)
@@ -936,8 +915,6 @@ static int davinci_spi_probe(struct platform_device *pdev)
 	struct davinci_spi *dspi;
 	struct davinci_spi_platform_data *pdata;
 	struct resource *r;
-	resource_size_t dma_rx_chan = SPI_NO_RESOURCE;
-	resource_size_t	dma_tx_chan = SPI_NO_RESOURCE;
 	int ret = 0;
 	u32 spipc0;
 
@@ -1044,27 +1021,15 @@ static int davinci_spi_probe(struct platform_device *pdev)
 		}
 	}
 
-	r = platform_get_resource(pdev, IORESOURCE_DMA, 0);
-	if (r)
-		dma_rx_chan = r->start;
-	r = platform_get_resource(pdev, IORESOURCE_DMA, 1);
-	if (r)
-		dma_tx_chan = r->start;
-
 	dspi->bitbang.txrx_bufs = davinci_spi_bufs;
-	if (dma_rx_chan != SPI_NO_RESOURCE &&
-	    dma_tx_chan != SPI_NO_RESOURCE) {
-		dspi->dma_rx_chnum = dma_rx_chan;
-		dspi->dma_tx_chnum = dma_tx_chan;
-
-		ret = davinci_spi_request_dma(dspi);
-		if (ret)
-			goto free_clk;
-
-		dev_info(&pdev->dev, "DMA: supported\n");
-		dev_info(&pdev->dev, "DMA: RX channel: %pa, TX channel: %pa, event queue: %d\n",
-				&dma_rx_chan, &dma_tx_chan,
-				pdata->dma_event_q);
+
+	ret = davinci_spi_request_dma(dspi);
+	if (ret == -EPROBE_DEFER) {
+		goto free_clk;
+	} else if (ret) {
+		dev_info(&pdev->dev, "DMA is not supported (%d)\n", ret);
+		dspi->dma_rx = NULL;
+		dspi->dma_tx = NULL;
 	}
 
 	dspi->get_rx = davinci_spi_rx_buf_u8;
@@ -1102,8 +1067,10 @@ static int davinci_spi_probe(struct platform_device *pdev)
 	return ret;
 
 free_dma:
-	dma_release_channel(dspi->dma_rx);
-	dma_release_channel(dspi->dma_tx);
+	if (dspi->dma_rx) {
+		dma_release_channel(dspi->dma_rx);
+		dma_release_channel(dspi->dma_tx);
+	}
 free_clk:
 	clk_disable_unprepare(dspi->clk);
 free_master:
@@ -1134,6 +1101,11 @@ static int davinci_spi_remove(struct platform_device *pdev)
 	clk_disable_unprepare(dspi->clk);
 	spi_master_put(master);
 
+	if (dspi->dma_rx) {
+		dma_release_channel(dspi->dma_rx);
+		dma_release_channel(dspi->dma_tx);
+	}
+
 	return 0;
 }
 
diff --git a/drivers/spi/spi-dln2.c b/drivers/spi/spi-dln2.c
index 3b7d91d94fea..b62a99caacc0 100644
--- a/drivers/spi/spi-dln2.c
+++ b/drivers/spi/spi-dln2.c
@@ -683,6 +683,7 @@ static int dln2_spi_probe(struct platform_device *pdev)
 	struct spi_master *master;
 	struct dln2_spi *dln2;
 	struct dln2_platform_data *pdata = dev_get_platdata(&pdev->dev);
+	struct device *dev = &pdev->dev;
 	int ret;
 
 	master = spi_alloc_master(&pdev->dev, sizeof(*dln2));
@@ -700,6 +701,7 @@ static int dln2_spi_probe(struct platform_device *pdev)
 	}
 
 	dln2->master = master;
+	dln2->master->dev.of_node = dev->of_node;
 	dln2->pdev = pdev;
 	dln2->port = pdata->port;
 	/* cs/mode can never be 0xff, so the first transfer will set them */
diff --git a/drivers/spi/spi-dw-pci.c b/drivers/spi/spi-dw-pci.c
index 332ccb0539a7..ef7db75c92c1 100644
--- a/drivers/spi/spi-dw-pci.c
+++ b/drivers/spi/spi-dw-pci.c
@@ -67,7 +67,7 @@ static int spi_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	dws->irq = pdev->irq;
 
 	/*
-	 * Specific handling for paltforms, like dma setup,
+	 * Specific handling for platforms, like dma setup,
 	 * clock rate, FIFO depth.
 	 */
 	if (desc) {
diff --git a/drivers/spi/spi-fsl-dspi.c b/drivers/spi/spi-fsl-dspi.c
index c1a2d747b246..9e9dadb52b3d 100644
--- a/drivers/spi/spi-fsl-dspi.c
+++ b/drivers/spi/spi-fsl-dspi.c
@@ -121,18 +121,22 @@ enum dspi_trans_mode {
 
 struct fsl_dspi_devtype_data {
 	enum dspi_trans_mode trans_mode;
+	u8 max_clock_factor;
 };
 
 static const struct fsl_dspi_devtype_data vf610_data = {
 	.trans_mode = DSPI_EOQ_MODE,
+	.max_clock_factor = 2,
 };
 
 static const struct fsl_dspi_devtype_data ls1021a_v1_data = {
 	.trans_mode = DSPI_TCFQ_MODE,
+	.max_clock_factor = 8,
 };
 
 static const struct fsl_dspi_devtype_data ls2085a_data = {
 	.trans_mode = DSPI_TCFQ_MODE,
+	.max_clock_factor = 8,
 };
 
 struct fsl_dspi {
@@ -726,6 +730,9 @@ static int dspi_probe(struct platform_device *pdev)
 	}
 	clk_prepare_enable(dspi->clk);
 
+	master->max_speed_hz =
+		clk_get_rate(dspi->clk) / dspi->devtype_data->max_clock_factor;
+
 	init_waitqueue_head(&dspi->waitq);
 	platform_set_drvdata(pdev, master);
 
diff --git a/drivers/spi/spi-fsl-espi.c b/drivers/spi/spi-fsl-espi.c
index 7cb0c1921495..8d85a3c343da 100644
--- a/drivers/spi/spi-fsl-espi.c
+++ b/drivers/spi/spi-fsl-espi.c
@@ -245,7 +245,12 @@ static int fsl_espi_bufs(struct spi_device *spi, struct spi_transfer *t)
 	if (ret)
 		return ret;
 
-	wait_for_completion(&mpc8xxx_spi->done);
+	/* Won't hang up forever, SPI bus sometimes got lost interrupts... */
+	ret = wait_for_completion_timeout(&mpc8xxx_spi->done, 2 * HZ);
+	if (ret == 0)
+		dev_err(mpc8xxx_spi->dev,
+			"Transaction hanging up (left %d bytes)\n",
+			mpc8xxx_spi->count);
 
 	/* disable rx ints */
 	mpc8xxx_spi_write_reg(&reg_base->mask, 0);
@@ -539,16 +544,31 @@ void fsl_espi_cpu_irq(struct mpc8xxx_spi *mspi, u32 events)
 	if (events & SPIE_NE) {
 		u32 rx_data, tmp;
 		u8 rx_data_8;
+		int rx_nr_bytes = 4;
+		int ret;
 
 		/* Spin until RX is done */
-		while (SPIE_RXCNT(events) < min(4, mspi->len)) {
-			cpu_relax();
-			events = mpc8xxx_spi_read_reg(&reg_base->event);
+		if (SPIE_RXCNT(events) < min(4, mspi->len)) {
+			ret = spin_event_timeout(
+				!(SPIE_RXCNT(events =
+				mpc8xxx_spi_read_reg(&reg_base->event)) <
+						min(4, mspi->len)),
+						10000, 0); /* 10 msec */
+			if (!ret)
+				dev_err(mspi->dev,
+					 "tired waiting for SPIE_RXCNT\n");
 		}
 
 		if (mspi->len >= 4) {
 			rx_data = mpc8xxx_spi_read_reg(&reg_base->receive);
+		} else if (mspi->len <= 0) {
+			dev_err(mspi->dev,
+				"unexpected RX(SPIE_NE) interrupt occurred,\n"
+				"(local rxlen %d bytes, reg rxlen %d bytes)\n",
+				min(4, mspi->len), SPIE_RXCNT(events));
+			rx_nr_bytes = 0;
 		} else {
+			rx_nr_bytes = mspi->len;
 			tmp = mspi->len;
 			rx_data = 0;
 			while (tmp--) {
@@ -559,7 +579,7 @@ void fsl_espi_cpu_irq(struct mpc8xxx_spi *mspi, u32 events)
 			rx_data <<= (4 - mspi->len) * 8;
 		}
 
-		mspi->len -= 4;
+		mspi->len -= rx_nr_bytes;
 
 		if (mspi->rx)
 			mspi->get_rx(rx_data, mspi);
diff --git a/drivers/spi/spi-octeon.c b/drivers/spi/spi-octeon.c
index 07e4ce8273df..3b170093989f 100644
--- a/drivers/spi/spi-octeon.c
+++ b/drivers/spi/spi-octeon.c
@@ -175,6 +175,7 @@ err:
 static int octeon_spi_probe(struct platform_device *pdev)
 {
 	struct resource *res_mem;
+	void __iomem *reg_base;
 	struct spi_master *master;
 	struct octeon_spi *p;
 	int err = -ENOENT;
@@ -186,19 +187,13 @@ static int octeon_spi_probe(struct platform_device *pdev)
 	platform_set_drvdata(pdev, master);
 
 	res_mem = platform_get_resource(pdev, IORESOURCE_MEM, 0);
-
-	if (res_mem == NULL) {
-		dev_err(&pdev->dev, "found no memory resource\n");
-		err = -ENXIO;
-		goto fail;
-	}
-	if (!devm_request_mem_region(&pdev->dev, res_mem->start,
-				     resource_size(res_mem), res_mem->name)) {
-		dev_err(&pdev->dev, "request_mem_region failed\n");
+	reg_base = devm_ioremap_resource(&pdev->dev, res_mem);
+	if (IS_ERR(reg_base)) {
+		err = PTR_ERR(reg_base);
 		goto fail;
 	}
-	p->register_base = (u64)devm_ioremap(&pdev->dev, res_mem->start,
-					     resource_size(res_mem));
+
+	p->register_base = (u64)reg_base;
 
 	master->num_chipselect = 4;
 	master->mode_bits = SPI_CPHA |
diff --git a/drivers/spi/spi-omap2-mcspi.c b/drivers/spi/spi-omap2-mcspi.c
index 0caa3c8bef46..1d237e93a289 100644
--- a/drivers/spi/spi-omap2-mcspi.c
+++ b/drivers/spi/spi-omap2-mcspi.c
@@ -23,7 +23,6 @@
 #include <linux/delay.h>
 #include <linux/dma-mapping.h>
 #include <linux/dmaengine.h>
-#include <linux/omap-dma.h>
 #include <linux/pinctrl/consumer.h>
 #include <linux/platform_device.h>
 #include <linux/err.h>
@@ -103,9 +102,6 @@ struct omap2_mcspi_dma {
 	struct dma_chan *dma_tx;
 	struct dma_chan *dma_rx;
 
-	int dma_tx_sync_dev;
-	int dma_rx_sync_dev;
-
 	struct completion dma_tx_completion;
 	struct completion dma_rx_completion;
 
@@ -964,8 +960,7 @@ static int omap2_mcspi_request_dma(struct spi_device *spi)
 	struct spi_master	*master = spi->master;
 	struct omap2_mcspi	*mcspi;
 	struct omap2_mcspi_dma	*mcspi_dma;
-	dma_cap_mask_t mask;
-	unsigned sig;
+	int ret = 0;
 
 	mcspi = spi_master_get_devdata(master);
 	mcspi_dma = mcspi->dma_channels + spi->chip_select;
@@ -973,34 +968,25 @@ static int omap2_mcspi_request_dma(struct spi_device *spi)
 	init_completion(&mcspi_dma->dma_rx_completion);
 	init_completion(&mcspi_dma->dma_tx_completion);
 
-	dma_cap_zero(mask);
-	dma_cap_set(DMA_SLAVE, mask);
-	sig = mcspi_dma->dma_rx_sync_dev;
-
-	mcspi_dma->dma_rx =
-		dma_request_slave_channel_compat(mask, omap_dma_filter_fn,
-						 &sig, &master->dev,
-						 mcspi_dma->dma_rx_ch_name);
-	if (!mcspi_dma->dma_rx)
+	mcspi_dma->dma_rx = dma_request_chan(&master->dev,
+					     mcspi_dma->dma_rx_ch_name);
+	if (IS_ERR(mcspi_dma->dma_rx)) {
+		ret = PTR_ERR(mcspi_dma->dma_rx);
+		mcspi_dma->dma_rx = NULL;
 		goto no_dma;
+	}
 
-	sig = mcspi_dma->dma_tx_sync_dev;
-	mcspi_dma->dma_tx =
-		dma_request_slave_channel_compat(mask, omap_dma_filter_fn,
-						 &sig, &master->dev,
-						 mcspi_dma->dma_tx_ch_name);
-
-	if (!mcspi_dma->dma_tx) {
+	mcspi_dma->dma_tx = dma_request_chan(&master->dev,
+					     mcspi_dma->dma_tx_ch_name);
+	if (IS_ERR(mcspi_dma->dma_tx)) {
+		ret = PTR_ERR(mcspi_dma->dma_tx);
+		mcspi_dma->dma_tx = NULL;
 		dma_release_channel(mcspi_dma->dma_rx);
 		mcspi_dma->dma_rx = NULL;
-		goto no_dma;
 	}
 
-	return 0;
-
 no_dma:
-	dev_warn(&spi->dev, "not using DMA for McSPI\n");
-	return -EAGAIN;
+	return ret;
 }
 
 static int omap2_mcspi_setup(struct spi_device *spi)
@@ -1039,8 +1025,9 @@ static int omap2_mcspi_setup(struct spi_device *spi)
 
 	if (!mcspi_dma->dma_rx || !mcspi_dma->dma_tx) {
 		ret = omap2_mcspi_request_dma(spi);
-		if (ret < 0 && ret != -EAGAIN)
-			return ret;
+		if (ret)
+			dev_warn(&spi->dev, "not using DMA for McSPI (%d)\n",
+				 ret);
 	}
 
 	ret = pm_runtime_get_sync(mcspi->dev);
@@ -1434,42 +1421,8 @@ static int omap2_mcspi_probe(struct platform_device *pdev)
 	}
 
 	for (i = 0; i < master->num_chipselect; i++) {
-		char *dma_rx_ch_name = mcspi->dma_channels[i].dma_rx_ch_name;
-		char *dma_tx_ch_name = mcspi->dma_channels[i].dma_tx_ch_name;
-		struct resource *dma_res;
-
-		sprintf(dma_rx_ch_name, "rx%d", i);
-		if (!pdev->dev.of_node) {
-			dma_res =
-				platform_get_resource_byname(pdev,
-							     IORESOURCE_DMA,
-							     dma_rx_ch_name);
-			if (!dma_res) {
-				dev_dbg(&pdev->dev,
-					"cannot get DMA RX channel\n");
-				status = -ENODEV;
-				break;
-			}
-
-			mcspi->dma_channels[i].dma_rx_sync_dev =
-				dma_res->start;
-		}
-		sprintf(dma_tx_ch_name, "tx%d", i);
-		if (!pdev->dev.of_node) {
-			dma_res =
-				platform_get_resource_byname(pdev,
-							     IORESOURCE_DMA,
-							     dma_tx_ch_name);
-			if (!dma_res) {
-				dev_dbg(&pdev->dev,
-					"cannot get DMA TX channel\n");
-				status = -ENODEV;
-				break;
-			}
-
-			mcspi->dma_channels[i].dma_tx_sync_dev =
-				dma_res->start;
-		}
+		sprintf(mcspi->dma_channels[i].dma_rx_ch_name, "rx%d", i);
+		sprintf(mcspi->dma_channels[i].dma_tx_ch_name, "tx%d", i);
 	}
 
 	if (status < 0)
diff --git a/drivers/spi/spi-pic32-sqi.c b/drivers/spi/spi-pic32-sqi.c
new file mode 100644
index 000000000000..ca3c8d94b290
--- /dev/null
+++ b/drivers/spi/spi-pic32-sqi.c
@@ -0,0 +1,727 @@
+/*
+ * PIC32 Quad SPI controller driver.
+ *
+ * Purna Chandra Mandal <purna.mandal@microchip.com>
+ * Copyright (c) 2016, Microchip Technology Inc.
+ *
+ * This program is free software; you can distribute it and/or modify it
+ * under the terms of the GNU General Public License (Version 2) as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ */
+
+#include <linux/clk.h>
+#include <linux/dma-mapping.h>
+#include <linux/interrupt.h>
+#include <linux/io.h>
+#include <linux/iopoll.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+#include <linux/spi/spi.h>
+
+/* SQI registers */
+#define PESQI_XIP_CONF1_REG	0x00
+#define PESQI_XIP_CONF2_REG	0x04
+#define PESQI_CONF_REG		0x08
+#define PESQI_CTRL_REG		0x0C
+#define PESQI_CLK_CTRL_REG	0x10
+#define PESQI_CMD_THRES_REG	0x14
+#define PESQI_INT_THRES_REG	0x18
+#define PESQI_INT_ENABLE_REG	0x1C
+#define PESQI_INT_STAT_REG	0x20
+#define PESQI_TX_DATA_REG	0x24
+#define PESQI_RX_DATA_REG	0x28
+#define PESQI_STAT1_REG		0x2C
+#define PESQI_STAT2_REG		0x30
+#define PESQI_BD_CTRL_REG	0x34
+#define PESQI_BD_CUR_ADDR_REG	0x38
+#define PESQI_BD_BASE_ADDR_REG	0x40
+#define PESQI_BD_STAT_REG	0x44
+#define PESQI_BD_POLL_CTRL_REG	0x48
+#define PESQI_BD_TX_DMA_STAT_REG	0x4C
+#define PESQI_BD_RX_DMA_STAT_REG	0x50
+#define PESQI_THRES_REG		0x54
+#define PESQI_INT_SIGEN_REG	0x58
+
+/* PESQI_CONF_REG fields */
+#define PESQI_MODE		0x7
+#define  PESQI_MODE_BOOT	0
+#define  PESQI_MODE_PIO		1
+#define  PESQI_MODE_DMA		2
+#define  PESQI_MODE_XIP		3
+#define PESQI_MODE_SHIFT	0
+#define PESQI_CPHA		BIT(3)
+#define PESQI_CPOL		BIT(4)
+#define PESQI_LSBF		BIT(5)
+#define PESQI_RXLATCH		BIT(7)
+#define PESQI_SERMODE		BIT(8)
+#define PESQI_WP_EN		BIT(9)
+#define PESQI_HOLD_EN		BIT(10)
+#define PESQI_BURST_EN		BIT(12)
+#define PESQI_CS_CTRL_HW	BIT(15)
+#define PESQI_SOFT_RESET	BIT(16)
+#define PESQI_LANES_SHIFT	20
+#define  PESQI_SINGLE_LANE	0
+#define  PESQI_DUAL_LANE	1
+#define  PESQI_QUAD_LANE	2
+#define PESQI_CSEN_SHIFT	24
+#define PESQI_EN		BIT(23)
+
+/* PESQI_CLK_CTRL_REG fields */
+#define PESQI_CLK_EN		BIT(0)
+#define PESQI_CLK_STABLE	BIT(1)
+#define PESQI_CLKDIV_SHIFT	8
+#define PESQI_CLKDIV		0xff
+
+/* PESQI_INT_THR/CMD_THR_REG */
+#define PESQI_TXTHR_MASK	0x1f
+#define PESQI_TXTHR_SHIFT	8
+#define PESQI_RXTHR_MASK	0x1f
+#define PESQI_RXTHR_SHIFT	0
+
+/* PESQI_INT_EN/INT_STAT/INT_SIG_EN_REG */
+#define PESQI_TXEMPTY		BIT(0)
+#define PESQI_TXFULL		BIT(1)
+#define PESQI_TXTHR		BIT(2)
+#define PESQI_RXEMPTY		BIT(3)
+#define PESQI_RXFULL		BIT(4)
+#define PESQI_RXTHR		BIT(5)
+#define PESQI_BDDONE		BIT(9)  /* BD processing complete */
+#define PESQI_PKTCOMP		BIT(10) /* packet processing complete */
+#define PESQI_DMAERR		BIT(11) /* error */
+
+/* PESQI_BD_CTRL_REG */
+#define PESQI_DMA_EN		BIT(0) /* enable DMA engine */
+#define PESQI_POLL_EN		BIT(1) /* enable polling */
+#define PESQI_BDP_START		BIT(2) /* start BD processor */
+
+/* PESQI controller buffer descriptor */
+struct buf_desc {
+	u32 bd_ctrl;	/* control */
+	u32 bd_status;	/* reserved */
+	u32 bd_addr;	/* DMA buffer addr */
+	u32 bd_nextp;	/* next item in chain */
+};
+
+/* bd_ctrl */
+#define BD_BUFLEN		0x1ff
+#define BD_CBD_INT_EN		BIT(16)	/* Current BD is processed */
+#define BD_PKT_INT_EN		BIT(17) /* All BDs of PKT processed */
+#define BD_LIFM			BIT(18) /* last data of pkt */
+#define BD_LAST			BIT(19) /* end of list */
+#define BD_DATA_RECV		BIT(20) /* receive data */
+#define BD_DDR			BIT(21) /* DDR mode */
+#define BD_DUAL			BIT(22)	/* Dual SPI */
+#define BD_QUAD			BIT(23) /* Quad SPI */
+#define BD_LSBF			BIT(25)	/* LSB First */
+#define BD_STAT_CHECK		BIT(27) /* Status poll */
+#define BD_DEVSEL_SHIFT		28	/* CS */
+#define BD_CS_DEASSERT		BIT(30) /* de-assert CS after current BD */
+#define BD_EN			BIT(31) /* BD owned by H/W */
+
+/**
+ * struct ring_desc - Representation of SQI ring descriptor
+ * @list:	list element to add to free or used list.
+ * @bd:		PESQI controller buffer descriptor
+ * @bd_dma:	DMA address of PESQI controller buffer descriptor
+ * @xfer_len:	transfer length
+ */
+struct ring_desc {
+	struct list_head list;
+	struct buf_desc *bd;
+	dma_addr_t bd_dma;
+	u32 xfer_len;
+};
+
+/* Global constants */
+#define PESQI_BD_BUF_LEN_MAX	256
+#define PESQI_BD_COUNT		256 /* max 64KB data per spi message */
+
+struct pic32_sqi {
+	void __iomem		*regs;
+	struct clk		*sys_clk;
+	struct clk		*base_clk; /* drives spi clock */
+	struct spi_master	*master;
+	int			irq;
+	struct completion	xfer_done;
+	struct ring_desc	*ring;
+	void			*bd;
+	dma_addr_t		bd_dma;
+	struct list_head	bd_list_free; /* free */
+	struct list_head	bd_list_used; /* allocated */
+	struct spi_device	*cur_spi;
+	u32			cur_speed;
+	u8			cur_mode;
+};
+
+static inline void pic32_setbits(void __iomem *reg, u32 set)
+{
+	writel(readl(reg) | set, reg);
+}
+
+static inline void pic32_clrbits(void __iomem *reg, u32 clr)
+{
+	writel(readl(reg) & ~clr, reg);
+}
+
+static int pic32_sqi_set_clk_rate(struct pic32_sqi *sqi, u32 sck)
+{
+	u32 val, div;
+
+	/* div = base_clk / (2 * spi_clk) */
+	div = clk_get_rate(sqi->base_clk) / (2 * sck);
+	div &= PESQI_CLKDIV;
+
+	val = readl(sqi->regs + PESQI_CLK_CTRL_REG);
+	/* apply new divider */
+	val &= ~(PESQI_CLK_STABLE | (PESQI_CLKDIV << PESQI_CLKDIV_SHIFT));
+	val |= div << PESQI_CLKDIV_SHIFT;
+	writel(val, sqi->regs + PESQI_CLK_CTRL_REG);
+
+	/* wait for stability */
+	return readl_poll_timeout(sqi->regs + PESQI_CLK_CTRL_REG, val,
+				  val & PESQI_CLK_STABLE, 1, 5000);
+}
+
+static inline void pic32_sqi_enable_int(struct pic32_sqi *sqi)
+{
+	u32 mask = PESQI_DMAERR | PESQI_BDDONE | PESQI_PKTCOMP;
+
+	writel(mask, sqi->regs + PESQI_INT_ENABLE_REG);
+	/* INT_SIGEN works as interrupt-gate to INTR line */
+	writel(mask, sqi->regs + PESQI_INT_SIGEN_REG);
+}
+
+static inline void pic32_sqi_disable_int(struct pic32_sqi *sqi)
+{
+	writel(0, sqi->regs + PESQI_INT_ENABLE_REG);
+	writel(0, sqi->regs + PESQI_INT_SIGEN_REG);
+}
+
+static irqreturn_t pic32_sqi_isr(int irq, void *dev_id)
+{
+	struct pic32_sqi *sqi = dev_id;
+	u32 enable, status;
+
+	enable = readl(sqi->regs + PESQI_INT_ENABLE_REG);
+	status = readl(sqi->regs + PESQI_INT_STAT_REG);
+
+	/* check spurious interrupt */
+	if (!status)
+		return IRQ_NONE;
+
+	if (status & PESQI_DMAERR) {
+		enable = 0;
+		goto irq_done;
+	}
+
+	if (status & PESQI_TXTHR)
+		enable &= ~(PESQI_TXTHR | PESQI_TXFULL | PESQI_TXEMPTY);
+
+	if (status & PESQI_RXTHR)
+		enable &= ~(PESQI_RXTHR | PESQI_RXFULL | PESQI_RXEMPTY);
+
+	if (status & PESQI_BDDONE)
+		enable &= ~PESQI_BDDONE;
+
+	/* packet processing completed */
+	if (status & PESQI_PKTCOMP) {
+		/* mask all interrupts */
+		enable = 0;
+		/* complete trasaction */
+		complete(&sqi->xfer_done);
+	}
+
+irq_done:
+	/* interrupts are sticky, so mask when handled */
+	writel(enable, sqi->regs + PESQI_INT_ENABLE_REG);
+
+	return IRQ_HANDLED;
+}
+
+static struct ring_desc *ring_desc_get(struct pic32_sqi *sqi)
+{
+	struct ring_desc *rdesc;
+
+	if (list_empty(&sqi->bd_list_free))
+		return NULL;
+
+	rdesc = list_first_entry(&sqi->bd_list_free, struct ring_desc, list);
+	list_del(&rdesc->list);
+	list_add_tail(&rdesc->list, &sqi->bd_list_used);
+	return rdesc;
+}
+
+static void ring_desc_put(struct pic32_sqi *sqi, struct ring_desc *rdesc)
+{
+	list_del(&rdesc->list);
+	list_add(&rdesc->list, &sqi->bd_list_free);
+}
+
+static int pic32_sqi_one_transfer(struct pic32_sqi *sqi,
+				  struct spi_message *mesg,
+				  struct spi_transfer *xfer)
+{
+	struct spi_device *spi = mesg->spi;
+	struct scatterlist *sg, *sgl;
+	struct ring_desc *rdesc;
+	struct buf_desc *bd;
+	int nents, i;
+	u32 bd_ctrl;
+	u32 nbits;
+
+	/* Device selection */
+	bd_ctrl = spi->chip_select << BD_DEVSEL_SHIFT;
+
+	/* half-duplex: select transfer buffer, direction and lane */
+	if (xfer->rx_buf) {
+		bd_ctrl |= BD_DATA_RECV;
+		nbits = xfer->rx_nbits;
+		sgl = xfer->rx_sg.sgl;
+		nents = xfer->rx_sg.nents;
+	} else {
+		nbits = xfer->tx_nbits;
+		sgl = xfer->tx_sg.sgl;
+		nents = xfer->tx_sg.nents;
+	}
+
+	if (nbits & SPI_NBITS_QUAD)
+		bd_ctrl |= BD_QUAD;
+	else if (nbits & SPI_NBITS_DUAL)
+		bd_ctrl |= BD_DUAL;
+
+	/* LSB first */
+	if (spi->mode & SPI_LSB_FIRST)
+		bd_ctrl |= BD_LSBF;
+
+	/* ownership to hardware */
+	bd_ctrl |= BD_EN;
+
+	for_each_sg(sgl, sg, nents, i) {
+		/* get ring descriptor */
+		rdesc = ring_desc_get(sqi);
+		if (!rdesc)
+			break;
+
+		bd = rdesc->bd;
+
+		/* BD CTRL: length */
+		rdesc->xfer_len = sg_dma_len(sg);
+		bd->bd_ctrl = bd_ctrl;
+		bd->bd_ctrl |= rdesc->xfer_len;
+
+		/* BD STAT */
+		bd->bd_status = 0;
+
+		/* BD BUFFER ADDRESS */
+		bd->bd_addr = sg->dma_address;
+	}
+
+	return 0;
+}
+
+static int pic32_sqi_prepare_hardware(struct spi_master *master)
+{
+	struct pic32_sqi *sqi = spi_master_get_devdata(master);
+
+	/* enable spi interface */
+	pic32_setbits(sqi->regs + PESQI_CONF_REG, PESQI_EN);
+	/* enable spi clk */
+	pic32_setbits(sqi->regs + PESQI_CLK_CTRL_REG, PESQI_CLK_EN);
+
+	return 0;
+}
+
+static bool pic32_sqi_can_dma(struct spi_master *master,
+			      struct spi_device *spi,
+			      struct spi_transfer *x)
+{
+	/* Do DMA irrespective of transfer size */
+	return true;
+}
+
+static int pic32_sqi_one_message(struct spi_master *master,
+				 struct spi_message *msg)
+{
+	struct spi_device *spi = msg->spi;
+	struct ring_desc *rdesc, *next;
+	struct spi_transfer *xfer;
+	struct pic32_sqi *sqi;
+	int ret = 0, mode;
+	u32 val;
+
+	sqi = spi_master_get_devdata(master);
+
+	reinit_completion(&sqi->xfer_done);
+	msg->actual_length = 0;
+
+	/* We can't handle spi_transfer specific "speed_hz", "bits_per_word"
+	 * and "delay_usecs". But spi_device specific speed and mode change
+	 * can be handled at best during spi chip-select switch.
+	 */
+	if (sqi->cur_spi != spi) {
+		/* set spi speed */
+		if (sqi->cur_speed != spi->max_speed_hz) {
+			sqi->cur_speed = spi->max_speed_hz;
+			ret = pic32_sqi_set_clk_rate(sqi, spi->max_speed_hz);
+			if (ret)
+				dev_warn(&spi->dev, "set_clk, %d\n", ret);
+		}
+
+		/* set spi mode */
+		mode = spi->mode & (SPI_MODE_3 | SPI_LSB_FIRST);
+		if (sqi->cur_mode != mode) {
+			val = readl(sqi->regs + PESQI_CONF_REG);
+			val &= ~(PESQI_CPOL | PESQI_CPHA | PESQI_LSBF);
+			if (mode & SPI_CPOL)
+				val |= PESQI_CPOL;
+			if (mode & SPI_LSB_FIRST)
+				val |= PESQI_LSBF;
+			val |= PESQI_CPHA;
+			writel(val, sqi->regs + PESQI_CONF_REG);
+
+			sqi->cur_mode = mode;
+		}
+		sqi->cur_spi = spi;
+	}
+
+	/* prepare hardware desc-list(BD) for transfer(s) */
+	list_for_each_entry(xfer, &msg->transfers, transfer_list) {
+		ret = pic32_sqi_one_transfer(sqi, msg, xfer);
+		if (ret) {
+			dev_err(&spi->dev, "xfer %p err\n", xfer);
+			goto xfer_out;
+		}
+	}
+
+	/* BDs are prepared and chained. Now mark LAST_BD, CS_DEASSERT at last
+	 * element of the list.
+	 */
+	rdesc = list_last_entry(&sqi->bd_list_used, struct ring_desc, list);
+	rdesc->bd->bd_ctrl |= BD_LAST | BD_CS_DEASSERT |
+			      BD_LIFM | BD_PKT_INT_EN;
+
+	/* set base address BD list for DMA engine */
+	rdesc = list_first_entry(&sqi->bd_list_used, struct ring_desc, list);
+	writel(rdesc->bd_dma, sqi->regs + PESQI_BD_BASE_ADDR_REG);
+
+	/* enable interrupt */
+	pic32_sqi_enable_int(sqi);
+
+	/* enable DMA engine */
+	val = PESQI_DMA_EN | PESQI_POLL_EN | PESQI_BDP_START;
+	writel(val, sqi->regs + PESQI_BD_CTRL_REG);
+
+	/* wait for xfer completion */
+	ret = wait_for_completion_timeout(&sqi->xfer_done, 5 * HZ);
+	if (ret <= 0) {
+		dev_err(&sqi->master->dev, "wait timedout/interrupted\n");
+		ret = -EIO;
+		msg->status = ret;
+	} else {
+		/* success */
+		msg->status = 0;
+		ret = 0;
+	}
+
+	/* disable DMA */
+	writel(0, sqi->regs + PESQI_BD_CTRL_REG);
+
+	pic32_sqi_disable_int(sqi);
+
+xfer_out:
+	list_for_each_entry_safe_reverse(rdesc, next,
+					 &sqi->bd_list_used, list) {
+		/* Update total byte transferred */
+		msg->actual_length += rdesc->xfer_len;
+		/* release ring descr */
+		ring_desc_put(sqi, rdesc);
+	}
+	spi_finalize_current_message(spi->master);
+
+	return ret;
+}
+
+static int pic32_sqi_unprepare_hardware(struct spi_master *master)
+{
+	struct pic32_sqi *sqi = spi_master_get_devdata(master);
+
+	/* disable clk */
+	pic32_clrbits(sqi->regs + PESQI_CLK_CTRL_REG, PESQI_CLK_EN);
+	/* disable spi */
+	pic32_clrbits(sqi->regs + PESQI_CONF_REG, PESQI_EN);
+
+	return 0;
+}
+
+static int ring_desc_ring_alloc(struct pic32_sqi *sqi)
+{
+	struct ring_desc *rdesc;
+	struct buf_desc *bd;
+	int i;
+
+	/* allocate coherent DMAable memory for hardware buffer descriptors. */
+	sqi->bd = dma_zalloc_coherent(&sqi->master->dev,
+				      sizeof(*bd) * PESQI_BD_COUNT,
+				      &sqi->bd_dma, GFP_DMA32);
+	if (!sqi->bd) {
+		dev_err(&sqi->master->dev, "failed allocating dma buffer\n");
+		return -ENOMEM;
+	}
+
+	/* allocate software ring descriptors */
+	sqi->ring = kcalloc(PESQI_BD_COUNT, sizeof(*rdesc), GFP_KERNEL);
+	if (!sqi->ring) {
+		dma_free_coherent(&sqi->master->dev,
+				  sizeof(*bd) * PESQI_BD_COUNT,
+				  sqi->bd, sqi->bd_dma);
+		return -ENOMEM;
+	}
+
+	bd = (struct buf_desc *)sqi->bd;
+
+	INIT_LIST_HEAD(&sqi->bd_list_free);
+	INIT_LIST_HEAD(&sqi->bd_list_used);
+
+	/* initialize ring-desc */
+	for (i = 0, rdesc = sqi->ring; i < PESQI_BD_COUNT; i++, rdesc++) {
+		INIT_LIST_HEAD(&rdesc->list);
+		rdesc->bd = &bd[i];
+		rdesc->bd_dma = sqi->bd_dma + (void *)&bd[i] - (void *)bd;
+		list_add_tail(&rdesc->list, &sqi->bd_list_free);
+	}
+
+	/* Prepare BD: chain to next BD(s) */
+	for (i = 0, rdesc = sqi->ring; i < PESQI_BD_COUNT - 1; i++)
+		bd[i].bd_nextp = rdesc[i + 1].bd_dma;
+	bd[PESQI_BD_COUNT - 1].bd_nextp = 0;
+
+	return 0;
+}
+
+static void ring_desc_ring_free(struct pic32_sqi *sqi)
+{
+	dma_free_coherent(&sqi->master->dev,
+			  sizeof(struct buf_desc) * PESQI_BD_COUNT,
+			  sqi->bd, sqi->bd_dma);
+	kfree(sqi->ring);
+}
+
+static void pic32_sqi_hw_init(struct pic32_sqi *sqi)
+{
+	unsigned long flags;
+	u32 val;
+
+	/* Soft-reset of PESQI controller triggers interrupt.
+	 * We are not yet ready to handle them so disable CPU
+	 * interrupt for the time being.
+	 */
+	local_irq_save(flags);
+
+	/* assert soft-reset */
+	writel(PESQI_SOFT_RESET, sqi->regs + PESQI_CONF_REG);
+
+	/* wait until clear */
+	readl_poll_timeout_atomic(sqi->regs + PESQI_CONF_REG, val,
+				  !(val & PESQI_SOFT_RESET), 1, 5000);
+
+	/* disable all interrupts */
+	pic32_sqi_disable_int(sqi);
+
+	/* Now it is safe to enable back CPU interrupt */
+	local_irq_restore(flags);
+
+	/* tx and rx fifo interrupt threshold */
+	val = readl(sqi->regs + PESQI_CMD_THRES_REG);
+	val &= ~(PESQI_TXTHR_MASK << PESQI_TXTHR_SHIFT);
+	val &= ~(PESQI_RXTHR_MASK << PESQI_RXTHR_SHIFT);
+	val |= (1U << PESQI_TXTHR_SHIFT) | (1U << PESQI_RXTHR_SHIFT);
+	writel(val, sqi->regs + PESQI_CMD_THRES_REG);
+
+	val = readl(sqi->regs + PESQI_INT_THRES_REG);
+	val &= ~(PESQI_TXTHR_MASK << PESQI_TXTHR_SHIFT);
+	val &= ~(PESQI_RXTHR_MASK << PESQI_RXTHR_SHIFT);
+	val |= (1U << PESQI_TXTHR_SHIFT) | (1U << PESQI_RXTHR_SHIFT);
+	writel(val, sqi->regs + PESQI_INT_THRES_REG);
+
+	/* default configuration */
+	val = readl(sqi->regs + PESQI_CONF_REG);
+
+	/* set mode: DMA */
+	val &= ~PESQI_MODE;
+	val |= PESQI_MODE_DMA << PESQI_MODE_SHIFT;
+	writel(val, sqi->regs + PESQI_CONF_REG);
+
+	/* DATAEN - SQIID0-ID3 */
+	val |= PESQI_QUAD_LANE << PESQI_LANES_SHIFT;
+
+	/* burst/INCR4 enable */
+	val |= PESQI_BURST_EN;
+
+	/* CSEN - all CS */
+	val |= 3U << PESQI_CSEN_SHIFT;
+	writel(val, sqi->regs + PESQI_CONF_REG);
+
+	/* write poll count */
+	writel(0, sqi->regs + PESQI_BD_POLL_CTRL_REG);
+
+	sqi->cur_speed = 0;
+	sqi->cur_mode = -1;
+}
+
+static int pic32_sqi_probe(struct platform_device *pdev)
+{
+	struct spi_master *master;
+	struct pic32_sqi *sqi;
+	struct resource *reg;
+	int ret;
+
+	master = spi_alloc_master(&pdev->dev, sizeof(*sqi));
+	if (!master)
+		return -ENOMEM;
+
+	sqi = spi_master_get_devdata(master);
+	sqi->master = master;
+
+	reg = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	sqi->regs = devm_ioremap_resource(&pdev->dev, reg);
+	if (IS_ERR(sqi->regs)) {
+		ret = PTR_ERR(sqi->regs);
+		goto err_free_master;
+	}
+
+	/* irq */
+	sqi->irq = platform_get_irq(pdev, 0);
+	if (sqi->irq < 0) {
+		dev_err(&pdev->dev, "no irq found\n");
+		ret = sqi->irq;
+		goto err_free_master;
+	}
+
+	/* clocks */
+	sqi->sys_clk = devm_clk_get(&pdev->dev, "reg_ck");
+	if (IS_ERR(sqi->sys_clk)) {
+		ret = PTR_ERR(sqi->sys_clk);
+		dev_err(&pdev->dev, "no sys_clk ?\n");
+		goto err_free_master;
+	}
+
+	sqi->base_clk = devm_clk_get(&pdev->dev, "spi_ck");
+	if (IS_ERR(sqi->base_clk)) {
+		ret = PTR_ERR(sqi->base_clk);
+		dev_err(&pdev->dev, "no base clk ?\n");
+		goto err_free_master;
+	}
+
+	ret = clk_prepare_enable(sqi->sys_clk);
+	if (ret) {
+		dev_err(&pdev->dev, "sys clk enable failed\n");
+		goto err_free_master;
+	}
+
+	ret = clk_prepare_enable(sqi->base_clk);
+	if (ret) {
+		dev_err(&pdev->dev, "base clk enable failed\n");
+		clk_disable_unprepare(sqi->sys_clk);
+		goto err_free_master;
+	}
+
+	init_completion(&sqi->xfer_done);
+
+	/* initialize hardware */
+	pic32_sqi_hw_init(sqi);
+
+	/* allocate buffers & descriptors */
+	ret = ring_desc_ring_alloc(sqi);
+	if (ret) {
+		dev_err(&pdev->dev, "ring alloc failed\n");
+		goto err_disable_clk;
+	}
+
+	/* install irq handlers */
+	ret = request_irq(sqi->irq, pic32_sqi_isr, 0,
+			  dev_name(&pdev->dev), sqi);
+	if (ret < 0) {
+		dev_err(&pdev->dev, "request_irq(%d), failed\n", sqi->irq);
+		goto err_free_ring;
+	}
+
+	/* register master */
+	master->num_chipselect	= 2;
+	master->max_speed_hz	= clk_get_rate(sqi->base_clk);
+	master->dma_alignment	= 32;
+	master->max_dma_len	= PESQI_BD_BUF_LEN_MAX;
+	master->dev.of_node	= of_node_get(pdev->dev.of_node);
+	master->mode_bits	= SPI_MODE_3 | SPI_MODE_0 | SPI_TX_DUAL |
+				  SPI_RX_DUAL | SPI_TX_QUAD | SPI_RX_QUAD;
+	master->flags		= SPI_MASTER_HALF_DUPLEX;
+	master->can_dma		= pic32_sqi_can_dma;
+	master->bits_per_word_mask	= SPI_BPW_RANGE_MASK(8, 32);
+	master->transfer_one_message	= pic32_sqi_one_message;
+	master->prepare_transfer_hardware	= pic32_sqi_prepare_hardware;
+	master->unprepare_transfer_hardware	= pic32_sqi_unprepare_hardware;
+
+	ret = devm_spi_register_master(&pdev->dev, master);
+	if (ret) {
+		dev_err(&master->dev, "failed registering spi master\n");
+		free_irq(sqi->irq, sqi);
+		goto err_free_ring;
+	}
+
+	platform_set_drvdata(pdev, sqi);
+
+	return 0;
+
+err_free_ring:
+	ring_desc_ring_free(sqi);
+
+err_disable_clk:
+	clk_disable_unprepare(sqi->base_clk);
+	clk_disable_unprepare(sqi->sys_clk);
+
+err_free_master:
+	spi_master_put(master);
+	return ret;
+}
+
+static int pic32_sqi_remove(struct platform_device *pdev)
+{
+	struct pic32_sqi *sqi = platform_get_drvdata(pdev);
+
+	/* release resources */
+	free_irq(sqi->irq, sqi);
+	ring_desc_ring_free(sqi);
+
+	/* disable clk */
+	clk_disable_unprepare(sqi->base_clk);
+	clk_disable_unprepare(sqi->sys_clk);
+
+	return 0;
+}
+
+static const struct of_device_id pic32_sqi_of_ids[] = {
+	{.compatible = "microchip,pic32mzda-sqi",},
+	{},
+};
+MODULE_DEVICE_TABLE(of, pic32_sqi_of_ids);
+
+static struct platform_driver pic32_sqi_driver = {
+	.driver = {
+		.name = "sqi-pic32",
+		.of_match_table = of_match_ptr(pic32_sqi_of_ids),
+	},
+	.probe = pic32_sqi_probe,
+	.remove = pic32_sqi_remove,
+};
+
+module_platform_driver(pic32_sqi_driver);
+
+MODULE_AUTHOR("Purna Chandra Mandal <purna.mandal@microchip.com>");
+MODULE_DESCRIPTION("Microchip SPI driver for PIC32 SQI controller.");
+MODULE_LICENSE("GPL v2");
diff --git a/drivers/spi/spi-pic32.c b/drivers/spi/spi-pic32.c
new file mode 100644
index 000000000000..73db87f805a1
--- /dev/null
+++ b/drivers/spi/spi-pic32.c
@@ -0,0 +1,878 @@
+/*
+ * Microchip PIC32 SPI controller driver.
+ *
+ * Purna Chandra Mandal <purna.mandal@microchip.com>
+ * Copyright (c) 2016, Microchip Technology Inc.
+ *
+ * This program is free software; you can distribute it and/or modify it
+ * under the terms of the GNU General Public License (Version 2) as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ */
+
+#include <linux/clk.h>
+#include <linux/clkdev.h>
+#include <linux/delay.h>
+#include <linux/dmaengine.h>
+#include <linux/dma-mapping.h>
+#include <linux/highmem.h>
+#include <linux/module.h>
+#include <linux/io.h>
+#include <linux/interrupt.h>
+#include <linux/of.h>
+#include <linux/of_irq.h>
+#include <linux/of_gpio.h>
+#include <linux/of_address.h>
+#include <linux/platform_device.h>
+#include <linux/spi/spi.h>
+
+/* SPI controller registers */
+struct pic32_spi_regs {
+	u32 ctrl;
+	u32 ctrl_clr;
+	u32 ctrl_set;
+	u32 ctrl_inv;
+	u32 status;
+	u32 status_clr;
+	u32 status_set;
+	u32 status_inv;
+	u32 buf;
+	u32 dontuse[3];
+	u32 baud;
+	u32 dontuse2[3];
+	u32 ctrl2;
+	u32 ctrl2_clr;
+	u32 ctrl2_set;
+	u32 ctrl2_inv;
+};
+
+/* Bit fields of SPI Control Register */
+#define CTRL_RX_INT_SHIFT	0  /* Rx interrupt generation */
+#define  RX_FIFO_EMTPY		0
+#define  RX_FIFO_NOT_EMPTY	1 /* not empty */
+#define  RX_FIFO_HALF_FULL	2 /* full by half or more */
+#define  RX_FIFO_FULL		3 /* completely full */
+
+#define CTRL_TX_INT_SHIFT	2  /* TX interrupt generation */
+#define  TX_FIFO_ALL_EMPTY	0 /* completely empty */
+#define  TX_FIFO_EMTPY		1 /* empty */
+#define  TX_FIFO_HALF_EMPTY	2 /* empty by half or more */
+#define  TX_FIFO_NOT_FULL	3 /* atleast one empty */
+
+#define CTRL_MSTEN	BIT(5) /* enable master mode */
+#define CTRL_CKP	BIT(6) /* active low */
+#define CTRL_CKE	BIT(8) /* Tx on falling edge */
+#define CTRL_SMP	BIT(9) /* Rx at middle or end of tx */
+#define CTRL_BPW_MASK	0x03   /* bits per word/sample */
+#define CTRL_BPW_SHIFT	10
+#define  PIC32_BPW_8	0
+#define  PIC32_BPW_16	1
+#define  PIC32_BPW_32	2
+#define CTRL_SIDL	BIT(13) /* sleep when idle */
+#define CTRL_ON		BIT(15) /* enable macro */
+#define CTRL_ENHBUF	BIT(16) /* enable enhanced buffering */
+#define CTRL_MCLKSEL	BIT(23) /* select clock source */
+#define CTRL_MSSEN	BIT(28) /* macro driven /SS */
+#define CTRL_FRMEN	BIT(31) /* enable framing mode */
+
+/* Bit fields of SPI Status Register */
+#define STAT_RF_EMPTY	BIT(5) /* RX Fifo empty */
+#define STAT_RX_OV	BIT(6) /* err, s/w needs to clear */
+#define STAT_TX_UR	BIT(8) /* UR in Framed SPI modes */
+#define STAT_FRM_ERR	BIT(12) /* Multiple Frame Sync pulse */
+#define STAT_TF_LVL_MASK	0x1F
+#define STAT_TF_LVL_SHIFT	16
+#define STAT_RF_LVL_MASK	0x1F
+#define STAT_RF_LVL_SHIFT	24
+
+/* Bit fields of SPI Baud Register */
+#define BAUD_MASK		0x1ff
+
+/* Bit fields of SPI Control2 Register */
+#define CTRL2_TX_UR_EN		BIT(10) /* Enable int on Tx under-run */
+#define CTRL2_RX_OV_EN		BIT(11) /* Enable int on Rx over-run */
+#define CTRL2_FRM_ERR_EN	BIT(12) /* Enable frame err int */
+
+/* Minimum DMA transfer size */
+#define PIC32_DMA_LEN_MIN	64
+
+struct pic32_spi {
+	dma_addr_t		dma_base;
+	struct pic32_spi_regs __iomem *regs;
+	int			fault_irq;
+	int			rx_irq;
+	int			tx_irq;
+	u32			fifo_n_byte; /* FIFO depth in bytes */
+	struct clk		*clk;
+	struct spi_master	*master;
+	/* Current controller setting */
+	u32			speed_hz; /* spi-clk rate */
+	u32			mode;
+	u32			bits_per_word;
+	u32			fifo_n_elm; /* FIFO depth in words */
+#define PIC32F_DMA_PREP		0 /* DMA chnls configured */
+	unsigned long		flags;
+	/* Current transfer state */
+	struct completion	xfer_done;
+	/* PIO transfer specific */
+	const void		*tx;
+	const void		*tx_end;
+	const void		*rx;
+	const void		*rx_end;
+	int			len;
+	void (*rx_fifo)(struct pic32_spi *);
+	void (*tx_fifo)(struct pic32_spi *);
+};
+
+static inline void pic32_spi_enable(struct pic32_spi *pic32s)
+{
+	writel(CTRL_ON | CTRL_SIDL, &pic32s->regs->ctrl_set);
+}
+
+static inline void pic32_spi_disable(struct pic32_spi *pic32s)
+{
+	writel(CTRL_ON | CTRL_SIDL, &pic32s->regs->ctrl_clr);
+
+	/* avoid SPI registers read/write at immediate next CPU clock */
+	ndelay(20);
+}
+
+static void pic32_spi_set_clk_rate(struct pic32_spi *pic32s, u32 spi_ck)
+{
+	u32 div;
+
+	/* div = (clk_in / 2 * spi_ck) - 1 */
+	div = DIV_ROUND_CLOSEST(clk_get_rate(pic32s->clk), 2 * spi_ck) - 1;
+
+	writel(div & BAUD_MASK, &pic32s->regs->baud);
+}
+
+static inline u32 pic32_rx_fifo_level(struct pic32_spi *pic32s)
+{
+	u32 sr = readl(&pic32s->regs->status);
+
+	return (sr >> STAT_RF_LVL_SHIFT) & STAT_RF_LVL_MASK;
+}
+
+static inline u32 pic32_tx_fifo_level(struct pic32_spi *pic32s)
+{
+	u32 sr = readl(&pic32s->regs->status);
+
+	return (sr >> STAT_TF_LVL_SHIFT) & STAT_TF_LVL_MASK;
+}
+
+/* Return the max entries we can fill into tx fifo */
+static u32 pic32_tx_max(struct pic32_spi *pic32s, int n_bytes)
+{
+	u32 tx_left, tx_room, rxtx_gap;
+
+	tx_left = (pic32s->tx_end - pic32s->tx) / n_bytes;
+	tx_room = pic32s->fifo_n_elm - pic32_tx_fifo_level(pic32s);
+
+	/*
+	 * Another concern is about the tx/rx mismatch, we
+	 * though to use (pic32s->fifo_n_byte - rxfl - txfl) as
+	 * one maximum value for tx, but it doesn't cover the
+	 * data which is out of tx/rx fifo and inside the
+	 * shift registers. So a ctrl from sw point of
+	 * view is taken.
+	 */
+	rxtx_gap = ((pic32s->rx_end - pic32s->rx) -
+		    (pic32s->tx_end - pic32s->tx)) / n_bytes;
+	return min3(tx_left, tx_room, (u32)(pic32s->fifo_n_elm - rxtx_gap));
+}
+
+/* Return the max entries we should read out of rx fifo */
+static u32 pic32_rx_max(struct pic32_spi *pic32s, int n_bytes)
+{
+	u32 rx_left = (pic32s->rx_end - pic32s->rx) / n_bytes;
+
+	return min_t(u32, rx_left, pic32_rx_fifo_level(pic32s));
+}
+
+#define BUILD_SPI_FIFO_RW(__name, __type, __bwl)		\
+static void pic32_spi_rx_##__name(struct pic32_spi *pic32s)	\
+{								\
+	__type v;						\
+	u32 mx = pic32_rx_max(pic32s, sizeof(__type));		\
+	for (; mx; mx--) {					\
+		v = read##__bwl(&pic32s->regs->buf);		\
+		if (pic32s->rx_end - pic32s->len)		\
+			*(__type *)(pic32s->rx) = v;		\
+		pic32s->rx += sizeof(__type);			\
+	}							\
+}								\
+								\
+static void pic32_spi_tx_##__name(struct pic32_spi *pic32s)	\
+{								\
+	__type v;						\
+	u32 mx = pic32_tx_max(pic32s, sizeof(__type));		\
+	for (; mx ; mx--) {					\
+		v = (__type)~0U;				\
+		if (pic32s->tx_end - pic32s->len)		\
+			v = *(__type *)(pic32s->tx);		\
+		write##__bwl(v, &pic32s->regs->buf);		\
+		pic32s->tx += sizeof(__type);			\
+	}							\
+}
+
+BUILD_SPI_FIFO_RW(byte, u8, b);
+BUILD_SPI_FIFO_RW(word, u16, w);
+BUILD_SPI_FIFO_RW(dword, u32, l);
+
+static void pic32_err_stop(struct pic32_spi *pic32s, const char *msg)
+{
+	/* disable all interrupts */
+	disable_irq_nosync(pic32s->fault_irq);
+	disable_irq_nosync(pic32s->rx_irq);
+	disable_irq_nosync(pic32s->tx_irq);
+
+	/* Show err message and abort xfer with err */
+	dev_err(&pic32s->master->dev, "%s\n", msg);
+	if (pic32s->master->cur_msg)
+		pic32s->master->cur_msg->status = -EIO;
+	complete(&pic32s->xfer_done);
+}
+
+static irqreturn_t pic32_spi_fault_irq(int irq, void *dev_id)
+{
+	struct pic32_spi *pic32s = dev_id;
+	u32 status;
+
+	status = readl(&pic32s->regs->status);
+
+	/* Error handling */
+	if (status & (STAT_RX_OV | STAT_TX_UR)) {
+		writel(STAT_RX_OV, &pic32s->regs->status_clr);
+		writel(STAT_TX_UR, &pic32s->regs->status_clr);
+		pic32_err_stop(pic32s, "err_irq: fifo ov/ur-run\n");
+		return IRQ_HANDLED;
+	}
+
+	if (status & STAT_FRM_ERR) {
+		pic32_err_stop(pic32s, "err_irq: frame error");
+		return IRQ_HANDLED;
+	}
+
+	if (!pic32s->master->cur_msg) {
+		pic32_err_stop(pic32s, "err_irq: no mesg");
+		return IRQ_NONE;
+	}
+
+	return IRQ_NONE;
+}
+
+static irqreturn_t pic32_spi_rx_irq(int irq, void *dev_id)
+{
+	struct pic32_spi *pic32s = dev_id;
+
+	pic32s->rx_fifo(pic32s);
+
+	/* rx complete ? */
+	if (pic32s->rx_end == pic32s->rx) {
+		/* disable all interrupts */
+		disable_irq_nosync(pic32s->fault_irq);
+		disable_irq_nosync(pic32s->rx_irq);
+
+		/* complete current xfer */
+		complete(&pic32s->xfer_done);
+	}
+
+	return IRQ_HANDLED;
+}
+
+static irqreturn_t pic32_spi_tx_irq(int irq, void *dev_id)
+{
+	struct pic32_spi *pic32s = dev_id;
+
+	pic32s->tx_fifo(pic32s);
+
+	/* tx complete? disable tx interrupt */
+	if (pic32s->tx_end == pic32s->tx)
+		disable_irq_nosync(pic32s->tx_irq);
+
+	return IRQ_HANDLED;
+}
+
+static void pic32_spi_dma_rx_notify(void *data)
+{
+	struct pic32_spi *pic32s = data;
+
+	complete(&pic32s->xfer_done);
+}
+
+static int pic32_spi_dma_transfer(struct pic32_spi *pic32s,
+				  struct spi_transfer *xfer)
+{
+	struct spi_master *master = pic32s->master;
+	struct dma_async_tx_descriptor *desc_rx;
+	struct dma_async_tx_descriptor *desc_tx;
+	dma_cookie_t cookie;
+	int ret;
+
+	if (!master->dma_rx || !master->dma_tx)
+		return -ENODEV;
+
+	desc_rx = dmaengine_prep_slave_sg(master->dma_rx,
+					  xfer->rx_sg.sgl,
+					  xfer->rx_sg.nents,
+					  DMA_FROM_DEVICE,
+					  DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
+	if (!desc_rx) {
+		ret = -EINVAL;
+		goto err_dma;
+	}
+
+	desc_tx = dmaengine_prep_slave_sg(master->dma_tx,
+					  xfer->tx_sg.sgl,
+					  xfer->tx_sg.nents,
+					  DMA_TO_DEVICE,
+					  DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
+	if (!desc_tx) {
+		ret = -EINVAL;
+		goto err_dma;
+	}
+
+	/* Put callback on the RX transfer, that should finish last */
+	desc_rx->callback = pic32_spi_dma_rx_notify;
+	desc_rx->callback_param = pic32s;
+
+	cookie = dmaengine_submit(desc_rx);
+	ret = dma_submit_error(cookie);
+	if (ret)
+		goto err_dma;
+
+	cookie = dmaengine_submit(desc_tx);
+	ret = dma_submit_error(cookie);
+	if (ret)
+		goto err_dma_tx;
+
+	dma_async_issue_pending(master->dma_rx);
+	dma_async_issue_pending(master->dma_tx);
+
+	return 0;
+
+err_dma_tx:
+	dmaengine_terminate_all(master->dma_rx);
+err_dma:
+	return ret;
+}
+
+static int pic32_spi_dma_config(struct pic32_spi *pic32s, u32 dma_width)
+{
+	int buf_offset = offsetof(struct pic32_spi_regs, buf);
+	struct spi_master *master = pic32s->master;
+	struct dma_slave_config cfg;
+	int ret;
+
+	cfg.device_fc = true;
+	cfg.src_addr = pic32s->dma_base + buf_offset;
+	cfg.dst_addr = pic32s->dma_base + buf_offset;
+	cfg.src_maxburst = pic32s->fifo_n_elm / 2; /* fill one-half */
+	cfg.dst_maxburst = pic32s->fifo_n_elm / 2; /* drain one-half */
+	cfg.src_addr_width = dma_width;
+	cfg.dst_addr_width = dma_width;
+	/* tx channel */
+	cfg.slave_id = pic32s->tx_irq;
+	cfg.direction = DMA_MEM_TO_DEV;
+	ret = dmaengine_slave_config(master->dma_tx, &cfg);
+	if (ret) {
+		dev_err(&master->dev, "tx channel setup failed\n");
+		return ret;
+	}
+	/* rx channel */
+	cfg.slave_id = pic32s->rx_irq;
+	cfg.direction = DMA_DEV_TO_MEM;
+	ret = dmaengine_slave_config(master->dma_rx, &cfg);
+	if (ret)
+		dev_err(&master->dev, "rx channel setup failed\n");
+
+	return ret;
+}
+
+static int pic32_spi_set_word_size(struct pic32_spi *pic32s, u8 bits_per_word)
+{
+	enum dma_slave_buswidth dmawidth;
+	u32 buswidth, v;
+
+	switch (bits_per_word) {
+	case 8:
+		pic32s->rx_fifo = pic32_spi_rx_byte;
+		pic32s->tx_fifo = pic32_spi_tx_byte;
+		buswidth = PIC32_BPW_8;
+		dmawidth = DMA_SLAVE_BUSWIDTH_1_BYTE;
+		break;
+	case 16:
+		pic32s->rx_fifo = pic32_spi_rx_word;
+		pic32s->tx_fifo = pic32_spi_tx_word;
+		buswidth = PIC32_BPW_16;
+		dmawidth = DMA_SLAVE_BUSWIDTH_2_BYTES;
+		break;
+	case 32:
+		pic32s->rx_fifo = pic32_spi_rx_dword;
+		pic32s->tx_fifo = pic32_spi_tx_dword;
+		buswidth = PIC32_BPW_32;
+		dmawidth = DMA_SLAVE_BUSWIDTH_4_BYTES;
+		break;
+	default:
+		/* not supported */
+		return -EINVAL;
+	}
+
+	/* calculate maximum number of words fifos can hold */
+	pic32s->fifo_n_elm = DIV_ROUND_UP(pic32s->fifo_n_byte,
+					  bits_per_word / 8);
+	/* set word size */
+	v = readl(&pic32s->regs->ctrl);
+	v &= ~(CTRL_BPW_MASK << CTRL_BPW_SHIFT);
+	v |= buswidth << CTRL_BPW_SHIFT;
+	writel(v, &pic32s->regs->ctrl);
+
+	/* re-configure dma width, if required */
+	if (test_bit(PIC32F_DMA_PREP, &pic32s->flags))
+		pic32_spi_dma_config(pic32s, dmawidth);
+
+	return 0;
+}
+
+static int pic32_spi_prepare_hardware(struct spi_master *master)
+{
+	struct pic32_spi *pic32s = spi_master_get_devdata(master);
+
+	pic32_spi_enable(pic32s);
+
+	return 0;
+}
+
+static int pic32_spi_prepare_message(struct spi_master *master,
+				     struct spi_message *msg)
+{
+	struct pic32_spi *pic32s = spi_master_get_devdata(master);
+	struct spi_device *spi = msg->spi;
+	u32 val;
+
+	/* set device specific bits_per_word */
+	if (pic32s->bits_per_word != spi->bits_per_word) {
+		pic32_spi_set_word_size(pic32s, spi->bits_per_word);
+		pic32s->bits_per_word = spi->bits_per_word;
+	}
+
+	/* device specific speed change */
+	if (pic32s->speed_hz != spi->max_speed_hz) {
+		pic32_spi_set_clk_rate(pic32s, spi->max_speed_hz);
+		pic32s->speed_hz = spi->max_speed_hz;
+	}
+
+	/* device specific mode change */
+	if (pic32s->mode != spi->mode) {
+		val = readl(&pic32s->regs->ctrl);
+		/* active low */
+		if (spi->mode & SPI_CPOL)
+			val |= CTRL_CKP;
+		else
+			val &= ~CTRL_CKP;
+		/* tx on rising edge */
+		if (spi->mode & SPI_CPHA)
+			val &= ~CTRL_CKE;
+		else
+			val |= CTRL_CKE;
+
+		/* rx at end of tx */
+		val |= CTRL_SMP;
+		writel(val, &pic32s->regs->ctrl);
+		pic32s->mode = spi->mode;
+	}
+
+	return 0;
+}
+
+static bool pic32_spi_can_dma(struct spi_master *master,
+			      struct spi_device *spi,
+			      struct spi_transfer *xfer)
+{
+	struct pic32_spi *pic32s = spi_master_get_devdata(master);
+
+	/* skip using DMA on small size transfer to avoid overhead.*/
+	return (xfer->len >= PIC32_DMA_LEN_MIN) &&
+	       test_bit(PIC32F_DMA_PREP, &pic32s->flags);
+}
+
+static int pic32_spi_one_transfer(struct spi_master *master,
+				  struct spi_device *spi,
+				  struct spi_transfer *transfer)
+{
+	struct pic32_spi *pic32s;
+	bool dma_issued = false;
+	int ret;
+
+	pic32s = spi_master_get_devdata(master);
+
+	/* handle transfer specific word size change */
+	if (transfer->bits_per_word &&
+	    (transfer->bits_per_word != pic32s->bits_per_word)) {
+		ret = pic32_spi_set_word_size(pic32s, transfer->bits_per_word);
+		if (ret)
+			return ret;
+		pic32s->bits_per_word = transfer->bits_per_word;
+	}
+
+	/* handle transfer specific speed change */
+	if (transfer->speed_hz && (transfer->speed_hz != pic32s->speed_hz)) {
+		pic32_spi_set_clk_rate(pic32s, transfer->speed_hz);
+		pic32s->speed_hz = transfer->speed_hz;
+	}
+
+	reinit_completion(&pic32s->xfer_done);
+
+	/* transact by DMA mode */
+	if (transfer->rx_sg.nents && transfer->tx_sg.nents) {
+		ret = pic32_spi_dma_transfer(pic32s, transfer);
+		if (ret) {
+			dev_err(&spi->dev, "dma submit error\n");
+			return ret;
+		}
+
+		/* DMA issued */
+		dma_issued = true;
+	} else {
+		/* set current transfer information */
+		pic32s->tx = (const void *)transfer->tx_buf;
+		pic32s->rx = (const void *)transfer->rx_buf;
+		pic32s->tx_end = pic32s->tx + transfer->len;
+		pic32s->rx_end = pic32s->rx + transfer->len;
+		pic32s->len = transfer->len;
+
+		/* transact by interrupt driven PIO */
+		enable_irq(pic32s->fault_irq);
+		enable_irq(pic32s->rx_irq);
+		enable_irq(pic32s->tx_irq);
+	}
+
+	/* wait for completion */
+	ret = wait_for_completion_timeout(&pic32s->xfer_done, 2 * HZ);
+	if (ret <= 0) {
+		dev_err(&spi->dev, "wait error/timedout\n");
+		if (dma_issued) {
+			dmaengine_terminate_all(master->dma_rx);
+			dmaengine_terminate_all(master->dma_rx);
+		}
+		ret = -ETIMEDOUT;
+	} else {
+		ret = 0;
+	}
+
+	return ret;
+}
+
+static int pic32_spi_unprepare_message(struct spi_master *master,
+				       struct spi_message *msg)
+{
+	/* nothing to do */
+	return 0;
+}
+
+static int pic32_spi_unprepare_hardware(struct spi_master *master)
+{
+	struct pic32_spi *pic32s = spi_master_get_devdata(master);
+
+	pic32_spi_disable(pic32s);
+
+	return 0;
+}
+
+/* This may be called multiple times by same spi dev */
+static int pic32_spi_setup(struct spi_device *spi)
+{
+	if (!spi->max_speed_hz) {
+		dev_err(&spi->dev, "No max speed HZ parameter\n");
+		return -EINVAL;
+	}
+
+	/* PIC32 spi controller can drive /CS during transfer depending
+	 * on tx fifo fill-level. /CS will stay asserted as long as TX
+	 * fifo is non-empty, else will be deasserted indicating
+	 * completion of the ongoing transfer. This might result into
+	 * unreliable/erroneous SPI transactions.
+	 * To avoid that we will always handle /CS by toggling GPIO.
+	 */
+	if (!gpio_is_valid(spi->cs_gpio))
+		return -EINVAL;
+
+	gpio_direction_output(spi->cs_gpio, !(spi->mode & SPI_CS_HIGH));
+
+	return 0;
+}
+
+static void pic32_spi_cleanup(struct spi_device *spi)
+{
+	/* de-activate cs-gpio */
+	gpio_direction_output(spi->cs_gpio, !(spi->mode & SPI_CS_HIGH));
+}
+
+static void pic32_spi_dma_prep(struct pic32_spi *pic32s, struct device *dev)
+{
+	struct spi_master *master = pic32s->master;
+	dma_cap_mask_t mask;
+
+	dma_cap_zero(mask);
+	dma_cap_set(DMA_SLAVE, mask);
+
+	master->dma_rx = dma_request_slave_channel_compat(mask, NULL, NULL,
+							  dev, "spi-rx");
+	if (!master->dma_rx) {
+		dev_warn(dev, "RX channel not found.\n");
+		goto out_err;
+	}
+
+	master->dma_tx = dma_request_slave_channel_compat(mask, NULL, NULL,
+							  dev, "spi-tx");
+	if (!master->dma_tx) {
+		dev_warn(dev, "TX channel not found.\n");
+		goto out_err;
+	}
+
+	if (pic32_spi_dma_config(pic32s, DMA_SLAVE_BUSWIDTH_1_BYTE))
+		goto out_err;
+
+	/* DMA chnls allocated and prepared */
+	set_bit(PIC32F_DMA_PREP, &pic32s->flags);
+
+	return;
+
+out_err:
+	if (master->dma_rx)
+		dma_release_channel(master->dma_rx);
+
+	if (master->dma_tx)
+		dma_release_channel(master->dma_tx);
+}
+
+static void pic32_spi_dma_unprep(struct pic32_spi *pic32s)
+{
+	if (!test_bit(PIC32F_DMA_PREP, &pic32s->flags))
+		return;
+
+	clear_bit(PIC32F_DMA_PREP, &pic32s->flags);
+	if (pic32s->master->dma_rx)
+		dma_release_channel(pic32s->master->dma_rx);
+
+	if (pic32s->master->dma_tx)
+		dma_release_channel(pic32s->master->dma_tx);
+}
+
+static void pic32_spi_hw_init(struct pic32_spi *pic32s)
+{
+	u32 ctrl;
+
+	/* disable hardware */
+	pic32_spi_disable(pic32s);
+
+	ctrl = readl(&pic32s->regs->ctrl);
+	/* enable enhanced fifo of 128bit deep */
+	ctrl |= CTRL_ENHBUF;
+	pic32s->fifo_n_byte = 16;
+
+	/* disable framing mode */
+	ctrl &= ~CTRL_FRMEN;
+
+	/* enable master mode while disabled */
+	ctrl |= CTRL_MSTEN;
+
+	/* set tx fifo threshold interrupt */
+	ctrl &= ~(0x3 << CTRL_TX_INT_SHIFT);
+	ctrl |= (TX_FIFO_HALF_EMPTY << CTRL_TX_INT_SHIFT);
+
+	/* set rx fifo threshold interrupt */
+	ctrl &= ~(0x3 << CTRL_RX_INT_SHIFT);
+	ctrl |= (RX_FIFO_NOT_EMPTY << CTRL_RX_INT_SHIFT);
+
+	/* select clk source */
+	ctrl &= ~CTRL_MCLKSEL;
+
+	/* set manual /CS mode */
+	ctrl &= ~CTRL_MSSEN;
+
+	writel(ctrl, &pic32s->regs->ctrl);
+
+	/* enable error reporting */
+	ctrl = CTRL2_TX_UR_EN | CTRL2_RX_OV_EN | CTRL2_FRM_ERR_EN;
+	writel(ctrl, &pic32s->regs->ctrl2_set);
+}
+
+static int pic32_spi_hw_probe(struct platform_device *pdev,
+			      struct pic32_spi *pic32s)
+{
+	struct resource *mem;
+	int ret;
+
+	mem = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	pic32s->regs = devm_ioremap_resource(&pdev->dev, mem);
+	if (IS_ERR(pic32s->regs))
+		return PTR_ERR(pic32s->regs);
+
+	pic32s->dma_base = mem->start;
+
+	/* get irq resources: err-irq, rx-irq, tx-irq */
+	pic32s->fault_irq = platform_get_irq_byname(pdev, "fault");
+	if (pic32s->fault_irq < 0) {
+		dev_err(&pdev->dev, "fault-irq not found\n");
+		return pic32s->fault_irq;
+	}
+
+	pic32s->rx_irq = platform_get_irq_byname(pdev, "rx");
+	if (pic32s->rx_irq < 0) {
+		dev_err(&pdev->dev, "rx-irq not found\n");
+		return pic32s->rx_irq;
+	}
+
+	pic32s->tx_irq = platform_get_irq_byname(pdev, "tx");
+	if (pic32s->tx_irq < 0) {
+		dev_err(&pdev->dev, "tx-irq not found\n");
+		return pic32s->tx_irq;
+	}
+
+	/* get clock */
+	pic32s->clk = devm_clk_get(&pdev->dev, "mck0");
+	if (IS_ERR(pic32s->clk)) {
+		dev_err(&pdev->dev, "clk not found\n");
+		ret = PTR_ERR(pic32s->clk);
+		goto err_unmap_mem;
+	}
+
+	ret = clk_prepare_enable(pic32s->clk);
+	if (ret)
+		goto err_unmap_mem;
+
+	pic32_spi_hw_init(pic32s);
+
+	return 0;
+
+err_unmap_mem:
+	dev_err(&pdev->dev, "%s failed, err %d\n", __func__, ret);
+	return ret;
+}
+
+static int pic32_spi_probe(struct platform_device *pdev)
+{
+	struct spi_master *master;
+	struct pic32_spi *pic32s;
+	int ret;
+
+	master = spi_alloc_master(&pdev->dev, sizeof(*pic32s));
+	if (!master)
+		return -ENOMEM;
+
+	pic32s = spi_master_get_devdata(master);
+	pic32s->master = master;
+
+	ret = pic32_spi_hw_probe(pdev, pic32s);
+	if (ret)
+		goto err_master;
+
+	master->dev.of_node	= of_node_get(pdev->dev.of_node);
+	master->mode_bits	= SPI_MODE_3 | SPI_MODE_0 | SPI_CS_HIGH;
+	master->num_chipselect	= 1; /* single chip-select */
+	master->max_speed_hz	= clk_get_rate(pic32s->clk);
+	master->setup		= pic32_spi_setup;
+	master->cleanup		= pic32_spi_cleanup;
+	master->flags		= SPI_MASTER_MUST_TX | SPI_MASTER_MUST_RX;
+	master->bits_per_word_mask	= SPI_BPW_MASK(8) | SPI_BPW_MASK(16) |
+					  SPI_BPW_MASK(32);
+	master->transfer_one		= pic32_spi_one_transfer;
+	master->prepare_message		= pic32_spi_prepare_message;
+	master->unprepare_message	= pic32_spi_unprepare_message;
+	master->prepare_transfer_hardware	= pic32_spi_prepare_hardware;
+	master->unprepare_transfer_hardware	= pic32_spi_unprepare_hardware;
+
+	/* optional DMA support */
+	pic32_spi_dma_prep(pic32s, &pdev->dev);
+	if (test_bit(PIC32F_DMA_PREP, &pic32s->flags))
+		master->can_dma	= pic32_spi_can_dma;
+
+	init_completion(&pic32s->xfer_done);
+	pic32s->mode = -1;
+
+	/* install irq handlers (with irq-disabled) */
+	irq_set_status_flags(pic32s->fault_irq, IRQ_NOAUTOEN);
+	ret = devm_request_irq(&pdev->dev, pic32s->fault_irq,
+			       pic32_spi_fault_irq, IRQF_NO_THREAD,
+			       dev_name(&pdev->dev), pic32s);
+	if (ret < 0) {
+		dev_err(&pdev->dev, "request fault-irq %d\n", pic32s->rx_irq);
+		goto err_bailout;
+	}
+
+	/* receive interrupt handler */
+	irq_set_status_flags(pic32s->rx_irq, IRQ_NOAUTOEN);
+	ret = devm_request_irq(&pdev->dev, pic32s->rx_irq,
+			       pic32_spi_rx_irq, IRQF_NO_THREAD,
+			       dev_name(&pdev->dev), pic32s);
+	if (ret < 0) {
+		dev_err(&pdev->dev, "request rx-irq %d\n", pic32s->rx_irq);
+		goto err_bailout;
+	}
+
+	/* transmit interrupt handler */
+	irq_set_status_flags(pic32s->tx_irq, IRQ_NOAUTOEN);
+	ret = devm_request_irq(&pdev->dev, pic32s->tx_irq,
+			       pic32_spi_tx_irq, IRQF_NO_THREAD,
+			       dev_name(&pdev->dev), pic32s);
+	if (ret < 0) {
+		dev_err(&pdev->dev, "request tx-irq %d\n", pic32s->tx_irq);
+		goto err_bailout;
+	}
+
+	/* register master */
+	ret = devm_spi_register_master(&pdev->dev, master);
+	if (ret) {
+		dev_err(&master->dev, "failed registering spi master\n");
+		goto err_bailout;
+	}
+
+	platform_set_drvdata(pdev, pic32s);
+
+	return 0;
+
+err_bailout:
+	clk_disable_unprepare(pic32s->clk);
+err_master:
+	spi_master_put(master);
+	return ret;
+}
+
+static int pic32_spi_remove(struct platform_device *pdev)
+{
+	struct pic32_spi *pic32s;
+
+	pic32s = platform_get_drvdata(pdev);
+	pic32_spi_disable(pic32s);
+	clk_disable_unprepare(pic32s->clk);
+	pic32_spi_dma_unprep(pic32s);
+
+	return 0;
+}
+
+static const struct of_device_id pic32_spi_of_match[] = {
+	{.compatible = "microchip,pic32mzda-spi",},
+	{},
+};
+MODULE_DEVICE_TABLE(of, pic32_spi_of_match);
+
+static struct platform_driver pic32_spi_driver = {
+	.driver = {
+		.name = "spi-pic32",
+		.of_match_table = of_match_ptr(pic32_spi_of_match),
+	},
+	.probe = pic32_spi_probe,
+	.remove = pic32_spi_remove,
+};
+
+module_platform_driver(pic32_spi_driver);
+
+MODULE_AUTHOR("Purna Chandra Mandal <purna.mandal@microchip.com>");
+MODULE_DESCRIPTION("Microchip SPI driver for PIC32 SPI controller.");
+MODULE_LICENSE("GPL v2");
diff --git a/drivers/spi/spi-pxa2xx-dma.c b/drivers/spi/spi-pxa2xx-dma.c
index 365fc22c3572..a18a03d0afb7 100644
--- a/drivers/spi/spi-pxa2xx-dma.c
+++ b/drivers/spi/spi-pxa2xx-dma.c
@@ -33,12 +33,10 @@ static int pxa2xx_spi_map_dma_buffer(struct driver_data *drv_data,
 		dmadev = drv_data->tx_chan->device->dev;
 		sgt = &drv_data->tx_sgt;
 		buf = drv_data->tx;
-		drv_data->tx_map_len = len;
 	} else {
 		dmadev = drv_data->rx_chan->device->dev;
 		sgt = &drv_data->rx_sgt;
 		buf = drv_data->rx;
-		drv_data->rx_map_len = len;
 	}
 
 	nents = DIV_ROUND_UP(len, SZ_2K);
@@ -55,11 +53,7 @@ static int pxa2xx_spi_map_dma_buffer(struct driver_data *drv_data,
 	for_each_sg(sgt->sgl, sg, sgt->nents, i) {
 		size_t bytes = min_t(size_t, len, SZ_2K);
 
-		if (buf)
-			sg_set_buf(sg, pbuf, bytes);
-		else
-			sg_set_buf(sg, drv_data->dummy, bytes);
-
+		sg_set_buf(sg, pbuf, bytes);
 		pbuf += bytes;
 		len -= bytes;
 	}
@@ -133,9 +127,6 @@ static void pxa2xx_spi_dma_transfer_complete(struct driver_data *drv_data,
 		if (!error) {
 			pxa2xx_spi_unmap_dma_buffers(drv_data);
 
-			drv_data->tx += drv_data->tx_map_len;
-			drv_data->rx += drv_data->rx_map_len;
-
 			msg->actual_length += drv_data->len;
 			msg->state = pxa2xx_spi_next_transfer(drv_data);
 		} else {
@@ -267,19 +258,22 @@ irqreturn_t pxa2xx_spi_dma_transfer(struct driver_data *drv_data)
 int pxa2xx_spi_dma_prepare(struct driver_data *drv_data, u32 dma_burst)
 {
 	struct dma_async_tx_descriptor *tx_desc, *rx_desc;
+	int err = 0;
 
 	tx_desc = pxa2xx_spi_dma_prepare_one(drv_data, DMA_MEM_TO_DEV);
 	if (!tx_desc) {
 		dev_err(&drv_data->pdev->dev,
 			"failed to get DMA TX descriptor\n");
-		return -EBUSY;
+		err = -EBUSY;
+		goto err_tx;
 	}
 
 	rx_desc = pxa2xx_spi_dma_prepare_one(drv_data, DMA_DEV_TO_MEM);
 	if (!rx_desc) {
 		dev_err(&drv_data->pdev->dev,
 			"failed to get DMA RX descriptor\n");
-		return -EBUSY;
+		err = -EBUSY;
+		goto err_rx;
 	}
 
 	/* We are ready when RX completes */
@@ -289,6 +283,12 @@ int pxa2xx_spi_dma_prepare(struct driver_data *drv_data, u32 dma_burst)
 	dmaengine_submit(rx_desc);
 	dmaengine_submit(tx_desc);
 	return 0;
+
+err_rx:
+	dmaengine_terminate_async(drv_data->tx_chan);
+err_tx:
+	pxa2xx_spi_unmap_dma_buffers(drv_data);
+	return err;
 }
 
 void pxa2xx_spi_dma_start(struct driver_data *drv_data)
@@ -308,10 +308,6 @@ int pxa2xx_spi_dma_setup(struct driver_data *drv_data)
 	dma_cap_zero(mask);
 	dma_cap_set(DMA_SLAVE, mask);
 
-	drv_data->dummy = devm_kzalloc(dev, SZ_2K, GFP_KERNEL);
-	if (!drv_data->dummy)
-		return -ENOMEM;
-
 	drv_data->tx_chan = dma_request_slave_channel_compat(mask,
 				pdata->dma_filter, pdata->tx_param, dev, "tx");
 	if (!drv_data->tx_chan)
diff --git a/drivers/spi/spi-pxa2xx-pci.c b/drivers/spi/spi-pxa2xx-pci.c
index 4fd7f9802f1b..5202de94f792 100644
--- a/drivers/spi/spi-pxa2xx-pci.c
+++ b/drivers/spi/spi-pxa2xx-pci.c
@@ -173,8 +173,8 @@ static int pxa2xx_spi_pci_probe(struct pci_dev *dev,
 	ssp->type = c->type;
 
 	snprintf(buf, sizeof(buf), "pxa2xx-spi.%d", ssp->port_id);
-	ssp->clk = clk_register_fixed_rate(&dev->dev, buf , NULL,
-					CLK_IS_ROOT, c->max_clk_rate);
+	ssp->clk = clk_register_fixed_rate(&dev->dev, buf , NULL, 0,
+					   c->max_clk_rate);
 	 if (IS_ERR(ssp->clk))
 		return PTR_ERR(ssp->clk);
 
diff --git a/drivers/spi/spi-pxa2xx.c b/drivers/spi/spi-pxa2xx.c
index 86138e4101b0..fe07c0592b44 100644
--- a/drivers/spi/spi-pxa2xx.c
+++ b/drivers/spi/spi-pxa2xx.c
@@ -570,9 +570,8 @@ static void giveback(struct driver_data *drv_data)
 		/* see if the next and current messages point
 		 * to the same chip
 		 */
-		if (next_msg && next_msg->spi != msg->spi)
-			next_msg = NULL;
-		if (!next_msg || msg->state == ERROR_STATE)
+		if ((next_msg && next_msg->spi != msg->spi) ||
+		    msg->state == ERROR_STATE)
 			cs_deassert(drv_data);
 	}
 
@@ -928,6 +927,7 @@ static void pump_transfers(unsigned long data)
 	u32 dma_thresh = drv_data->cur_chip->dma_threshold;
 	u32 dma_burst = drv_data->cur_chip->dma_burst_size;
 	u32 change_mask = pxa2xx_spi_get_ssrc1_change_mask(drv_data);
+	int err;
 
 	/* Get current state information */
 	message = drv_data->cur_msg;
@@ -1047,7 +1047,12 @@ static void pump_transfers(unsigned long data)
 		/* Ensure we have the correct interrupt handler */
 		drv_data->transfer_handler = pxa2xx_spi_dma_transfer;
 
-		pxa2xx_spi_dma_prepare(drv_data, dma_burst);
+		err = pxa2xx_spi_dma_prepare(drv_data, dma_burst);
+		if (err) {
+			message->status = err;
+			giveback(drv_data);
+			return;
+		}
 
 		/* Clear status and start DMA engine */
 		cr1 = chip->cr1 | dma_thresh | drv_data->dma_cr1;
@@ -1543,7 +1548,6 @@ static int pxa2xx_spi_probe(struct platform_device *pdev)
 	drv_data->pdev = pdev;
 	drv_data->ssp = ssp;
 
-	master->dev.parent = &pdev->dev;
 	master->dev.of_node = pdev->dev.of_node;
 	/* the spi->mode bits understood by this driver: */
 	master->mode_bits = SPI_CPOL | SPI_CPHA | SPI_CS_HIGH | SPI_LOOP;
@@ -1556,6 +1560,7 @@ static int pxa2xx_spi_probe(struct platform_device *pdev)
 	master->unprepare_transfer_hardware = pxa2xx_spi_unprepare_transfer;
 	master->fw_translate_cs = pxa2xx_spi_fw_translate_cs;
 	master->auto_runtime_pm = true;
+	master->flags = SPI_MASTER_MUST_RX | SPI_MASTER_MUST_TX;
 
 	drv_data->ssp_type = ssp->type;
 
diff --git a/drivers/spi/spi-pxa2xx.h b/drivers/spi/spi-pxa2xx.h
index a1ef88948144..e6b09000ff14 100644
--- a/drivers/spi/spi-pxa2xx.h
+++ b/drivers/spi/spi-pxa2xx.h
@@ -56,7 +56,6 @@ struct driver_data {
 	struct sg_table tx_sgt;
 	int rx_nents;
 	int tx_nents;
-	void *dummy;
 	atomic_t dma_running;
 
 	/* Current message transfer state info */
@@ -69,8 +68,6 @@ struct driver_data {
 	void *rx;
 	void *rx_end;
 	int dma_mapped;
-	size_t rx_map_len;
-	size_t tx_map_len;
 	u8 n_bytes;
 	int (*write)(struct driver_data *drv_data);
 	int (*read)(struct driver_data *drv_data);
diff --git a/drivers/spi/spi-qup.c b/drivers/spi/spi-qup.c
index 810a7fae3479..c338ef1136f6 100644
--- a/drivers/spi/spi-qup.c
+++ b/drivers/spi/spi-qup.c
@@ -937,6 +937,10 @@ static int spi_qup_pm_suspend_runtime(struct device *device)
 	config = readl(controller->base + QUP_CONFIG);
 	config |= QUP_CONFIG_CLOCK_AUTO_GATE;
 	writel_relaxed(config, controller->base + QUP_CONFIG);
+
+	clk_disable_unprepare(controller->cclk);
+	clk_disable_unprepare(controller->iclk);
+
 	return 0;
 }
 
@@ -945,6 +949,15 @@ static int spi_qup_pm_resume_runtime(struct device *device)
 	struct spi_master *master = dev_get_drvdata(device);
 	struct spi_qup *controller = spi_master_get_devdata(master);
 	u32 config;
+	int ret;
+
+	ret = clk_prepare_enable(controller->iclk);
+	if (ret)
+		return ret;
+
+	ret = clk_prepare_enable(controller->cclk);
+	if (ret)
+		return ret;
 
 	/* Disable clocks auto gaiting */
 	config = readl_relaxed(controller->base + QUP_CONFIG);
@@ -1017,6 +1030,8 @@ static int spi_qup_remove(struct platform_device *pdev)
 
 	pm_runtime_put_noidle(&pdev->dev);
 	pm_runtime_disable(&pdev->dev);
+	spi_master_put(master);
+
 	return 0;
 }
 
diff --git a/drivers/spi/spi-rockchip.c b/drivers/spi/spi-rockchip.c
index 6c6c0013ec7a..cd89682065b9 100644
--- a/drivers/spi/spi-rockchip.c
+++ b/drivers/spi/spi-rockchip.c
@@ -744,10 +744,8 @@ static int rockchip_spi_probe(struct platform_device *pdev)
 	rs->dma_rx.ch = dma_request_chan(rs->dev, "rx");
 	if (IS_ERR(rs->dma_rx.ch)) {
 		if (PTR_ERR(rs->dma_rx.ch) == -EPROBE_DEFER) {
-			dma_release_channel(rs->dma_tx.ch);
-			rs->dma_tx.ch = NULL;
 			ret = -EPROBE_DEFER;
-			goto err_get_fifo_len;
+			goto err_free_dma_tx;
 		}
 		dev_warn(rs->dev, "Failed to request RX DMA channel\n");
 		rs->dma_rx.ch = NULL;
@@ -775,10 +773,11 @@ static int rockchip_spi_probe(struct platform_device *pdev)
 
 err_register_master:
 	pm_runtime_disable(&pdev->dev);
-	if (rs->dma_tx.ch)
-		dma_release_channel(rs->dma_tx.ch);
 	if (rs->dma_rx.ch)
 		dma_release_channel(rs->dma_rx.ch);
+err_free_dma_tx:
+	if (rs->dma_tx.ch)
+		dma_release_channel(rs->dma_tx.ch);
 err_get_fifo_len:
 	clk_disable_unprepare(rs->spiclk);
 err_spiclk_enable:
diff --git a/drivers/spi/spi-st-ssc4.c b/drivers/spi/spi-st-ssc4.c
index f17c0abe299f..d5adf9f31602 100644
--- a/drivers/spi/spi-st-ssc4.c
+++ b/drivers/spi/spi-st-ssc4.c
@@ -345,12 +345,13 @@ static int spi_st_probe(struct platform_device *pdev)
 	spi_st->clk = devm_clk_get(&pdev->dev, "ssc");
 	if (IS_ERR(spi_st->clk)) {
 		dev_err(&pdev->dev, "Unable to request clock\n");
-		return PTR_ERR(spi_st->clk);
+		ret = PTR_ERR(spi_st->clk);
+		goto put_master;
 	}
 
 	ret = spi_st_clk_enable(spi_st);
 	if (ret)
-		return ret;
+		goto put_master;
 
 	init_completion(&spi_st->done);
 
@@ -408,7 +409,8 @@ static int spi_st_probe(struct platform_device *pdev)
 
 clk_disable:
 	spi_st_clk_disable(spi_st);
-
+put_master:
+	spi_master_put(master);
 	return ret;
 }
 
diff --git a/drivers/spi/spi-zynqmp-gqspi.c b/drivers/spi/spi-zynqmp-gqspi.c
index aab9b492c627..18aeaceee286 100644
--- a/drivers/spi/spi-zynqmp-gqspi.c
+++ b/drivers/spi/spi-zynqmp-gqspi.c
@@ -360,7 +360,7 @@ static int zynqmp_prepare_transfer_hardware(struct spi_master *master)
 
 	ret = clk_enable(xqspi->refclk);
 	if (ret)
-		goto clk_err;
+		return ret;
 
 	ret = clk_enable(xqspi->pclk);
 	if (ret)
@@ -369,6 +369,7 @@ static int zynqmp_prepare_transfer_hardware(struct spi_master *master)
 	zynqmp_gqspi_write(xqspi, GQSPI_EN_OFST, GQSPI_EN_MASK);
 	return 0;
 clk_err:
+	clk_disable(xqspi->refclk);
 	return ret;
 }
 
diff --git a/drivers/spi/spi.c b/drivers/spi/spi.c
index 0239b45eed92..77e6e45951f4 100644
--- a/drivers/spi/spi.c
+++ b/drivers/spi/spi.c
@@ -717,9 +717,11 @@ static int spi_map_buf(struct spi_master *master, struct device *dev,
 	if (vmalloced_buf) {
 		desc_len = min_t(int, max_seg_size, PAGE_SIZE);
 		sgs = DIV_ROUND_UP(len + offset_in_page(buf), desc_len);
-	} else {
+	} else if (virt_addr_valid(buf)) {
 		desc_len = min_t(int, max_seg_size, master->max_dma_len);
 		sgs = DIV_ROUND_UP(len, desc_len);
+	} else {
+		return -EINVAL;
 	}
 
 	ret = sg_alloc_table(sgt, sgs, GFP_KERNEL);
@@ -933,7 +935,7 @@ static int spi_map_msg(struct spi_master *master, struct spi_message *msg)
  * spi_transfer_one_message - Default implementation of transfer_one_message()
  *
  * This is a standard implementation of transfer_one_message() for
- * drivers which impelment a transfer_one() operation.  It provides
+ * drivers which implement a transfer_one() operation.  It provides
  * standard handling of delays and chip select management.
  */
 static int spi_transfer_one_message(struct spi_master *master,
@@ -1764,6 +1766,7 @@ struct spi_master *spi_alloc_master(struct device *dev, unsigned size)
 	master->num_chipselect = 1;
 	master->dev.class = &spi_master_class;
 	master->dev.parent = dev;
+	pm_suspend_ignore_children(&master->dev, true);
 	spi_master_set_devdata(master, &master[1]);
 
 	return master;