diff --git a/Documentation/devicetree/bindings/spi/fsl-imx-cspi.txt b/Documentation/devicetree/bindings/spi/fsl-imx-cspi.txt
index 31b5b21598ff5755d3e12a50aac2abf51d6653f1..5bf13960f7f4a3c826c10b1e15a618df82d82403 100644
--- a/Documentation/devicetree/bindings/spi/fsl-imx-cspi.txt
+++ b/Documentation/devicetree/bindings/spi/fsl-imx-cspi.txt
@@ -9,6 +9,7 @@ Required properties:
   - "fsl,imx31-cspi" for SPI compatible with the one integrated on i.MX31
   - "fsl,imx35-cspi" for SPI compatible with the one integrated on i.MX35
   - "fsl,imx51-ecspi" for SPI compatible with the one integrated on i.MX51
+  - "fsl,imx53-ecspi" for SPI compatible with the one integrated on i.MX53 and later Soc
 - reg : Offset and length of the register set for the device
 - interrupts : Should contain CSPI/eCSPI interrupt
 - cs-gpios : Specifies the gpio pins to be used for chipselects.
diff --git a/Documentation/devicetree/bindings/spi/sh-msiof.txt b/Documentation/devicetree/bindings/spi/sh-msiof.txt
index 64ee489571c42f88a13decf8e0a1679f635f636c..39e5ef7c5e71694eeb6d1ccba4499d0cdd8b125e 100644
--- a/Documentation/devicetree/bindings/spi/sh-msiof.txt
+++ b/Documentation/devicetree/bindings/spi/sh-msiof.txt
@@ -6,6 +6,7 @@ Required properties:
 			 "renesas,msiof-r8a7792" (R-Car V2H)
 			 "renesas,msiof-r8a7793" (R-Car M2-N)
 			 "renesas,msiof-r8a7794" (R-Car E2)
+			 "renesas,msiof-r8a7795" (R-Car H3)
 			 "renesas,msiof-r8a7796" (R-Car M3-W)
 			 "renesas,msiof-sh73a0" (SH-Mobile AG5)
 			 "renesas,sh-mobile-msiof" (generic SH-Mobile compatibile device)
diff --git a/Documentation/devicetree/bindings/spi/spi-rockchip.txt b/Documentation/devicetree/bindings/spi/spi-rockchip.txt
index 83da4931d832688f51e19a56aaa03af8482e721a..6e3ffacbba320a756933255fe8100c0103e01cc6 100644
--- a/Documentation/devicetree/bindings/spi/spi-rockchip.txt
+++ b/Documentation/devicetree/bindings/spi/spi-rockchip.txt
@@ -6,6 +6,7 @@ and display controllers using the SPI communication interface.
 Required Properties:
 
 - compatible: should be one of the following.
+    "rockchip,rv1108-spi" for rv1108 SoCs.
     "rockchip,rk3036-spi" for rk3036 SoCS.
     "rockchip,rk3066-spi" for rk3066 SoCs.
     "rockchip,rk3188-spi" for rk3188 SoCs.
diff --git a/drivers/spi/Kconfig b/drivers/spi/Kconfig
index 9b31351fe429e9776e7f6a44f7485fc9f49c0374..a75f2a2cf7805539262536c11f1d1d46aa51d7aa 100644
--- a/drivers/spi/Kconfig
+++ b/drivers/spi/Kconfig
@@ -55,7 +55,6 @@ comment "SPI Master Controller Drivers"
 
 config SPI_ALTERA
 	tristate "Altera SPI Controller"
-	select SPI_BITBANG
 	help
 	  This is the driver for the Altera SPI Controller.
 
@@ -518,8 +517,8 @@ config SPI_PPC4xx
 
 config SPI_PXA2XX
 	tristate "PXA2xx SSP SPI master"
-	depends on (ARCH_PXA || PCI || ACPI)
-	select PXA_SSP if ARCH_PXA
+	depends on (ARCH_PXA || ARCH_MMP || PCI || ACPI)
+	select PXA_SSP if ARCH_PXA || ARCH_MMP
 	help
 	  This enables using a PXA2xx or Sodaville SSP port as a SPI master
 	  controller. The driver can be configured to use any SSP port and
diff --git a/drivers/spi/spi-altera.c b/drivers/spi/spi-altera.c
index b95010e724526273ef1949a1ceb7d88c06754820..a5adf0d868fc9baf0b836f543d2b313c5fa169ee 100644
--- a/drivers/spi/spi-altera.c
+++ b/drivers/spi/spi-altera.c
@@ -18,7 +18,6 @@
 #include <linux/module.h>
 #include <linux/platform_device.h>
 #include <linux/spi/spi.h>
-#include <linux/spi/spi_bitbang.h>
 #include <linux/io.h>
 #include <linux/of.h>
 
@@ -45,10 +44,6 @@
 #define ALTERA_SPI_CONTROL_SSO_MSK	0x400
 
 struct altera_spi {
-	/* bitbang has to be first */
-	struct spi_bitbang bitbang;
-	struct completion done;
-
 	void __iomem *base;
 	int irq;
 	int len;
@@ -66,59 +61,64 @@ static inline struct altera_spi *altera_spi_to_hw(struct spi_device *sdev)
 	return spi_master_get_devdata(sdev->master);
 }
 
-static void altera_spi_chipsel(struct spi_device *spi, int value)
+static void altera_spi_set_cs(struct spi_device *spi, bool is_high)
 {
 	struct altera_spi *hw = altera_spi_to_hw(spi);
 
-	if (spi->mode & SPI_CS_HIGH) {
-		switch (value) {
-		case BITBANG_CS_INACTIVE:
-			writel(1 << spi->chip_select,
-			       hw->base + ALTERA_SPI_SLAVE_SEL);
-			hw->imr |= ALTERA_SPI_CONTROL_SSO_MSK;
-			writel(hw->imr, hw->base + ALTERA_SPI_CONTROL);
-			break;
-
-		case BITBANG_CS_ACTIVE:
-			hw->imr &= ~ALTERA_SPI_CONTROL_SSO_MSK;
-			writel(hw->imr, hw->base + ALTERA_SPI_CONTROL);
-			writel(0, hw->base + ALTERA_SPI_SLAVE_SEL);
-			break;
-		}
+	if (is_high) {
+		hw->imr &= ~ALTERA_SPI_CONTROL_SSO_MSK;
+		writel(hw->imr, hw->base + ALTERA_SPI_CONTROL);
+		writel(0, hw->base + ALTERA_SPI_SLAVE_SEL);
 	} else {
-		switch (value) {
-		case BITBANG_CS_INACTIVE:
-			hw->imr &= ~ALTERA_SPI_CONTROL_SSO_MSK;
-			writel(hw->imr, hw->base + ALTERA_SPI_CONTROL);
-			break;
+		writel(BIT(spi->chip_select), hw->base + ALTERA_SPI_SLAVE_SEL);
+		hw->imr |= ALTERA_SPI_CONTROL_SSO_MSK;
+		writel(hw->imr, hw->base + ALTERA_SPI_CONTROL);
+	}
+}
+
+static void altera_spi_tx_word(struct altera_spi *hw)
+{
+	unsigned int txd = 0;
 
-		case BITBANG_CS_ACTIVE:
-			writel(1 << spi->chip_select,
-			       hw->base + ALTERA_SPI_SLAVE_SEL);
-			hw->imr |= ALTERA_SPI_CONTROL_SSO_MSK;
-			writel(hw->imr, hw->base + ALTERA_SPI_CONTROL);
+	if (hw->tx) {
+		switch (hw->bytes_per_word) {
+		case 1:
+			txd = hw->tx[hw->count];
+			break;
+		case 2:
+			txd = (hw->tx[hw->count * 2]
+				| (hw->tx[hw->count * 2 + 1] << 8));
 			break;
 		}
 	}
+
+	writel(txd, hw->base + ALTERA_SPI_TXDATA);
 }
 
-static inline unsigned int hw_txbyte(struct altera_spi *hw, int count)
+static void altera_spi_rx_word(struct altera_spi *hw)
 {
-	if (hw->tx) {
+	unsigned int rxd;
+
+	rxd = readl(hw->base + ALTERA_SPI_RXDATA);
+	if (hw->rx) {
 		switch (hw->bytes_per_word) {
 		case 1:
-			return hw->tx[count];
+			hw->rx[hw->count] = rxd;
+			break;
 		case 2:
-			return (hw->tx[count * 2]
-				| (hw->tx[count * 2 + 1] << 8));
+			hw->rx[hw->count * 2] = rxd;
+			hw->rx[hw->count * 2 + 1] = rxd >> 8;
+			break;
 		}
 	}
-	return 0;
+
+	hw->count++;
 }
 
-static int altera_spi_txrx(struct spi_device *spi, struct spi_transfer *t)
+static int altera_spi_txrx(struct spi_master *master,
+	struct spi_device *spi, struct spi_transfer *t)
 {
-	struct altera_spi *hw = altera_spi_to_hw(spi);
+	struct altera_spi *hw = spi_master_get_devdata(master);
 
 	hw->tx = t->tx_buf;
 	hw->rx = t->rx_buf;
@@ -132,67 +132,39 @@ static int altera_spi_txrx(struct spi_device *spi, struct spi_transfer *t)
 		writel(hw->imr, hw->base + ALTERA_SPI_CONTROL);
 
 		/* send the first byte */
-		writel(hw_txbyte(hw, 0), hw->base + ALTERA_SPI_TXDATA);
-
-		wait_for_completion(&hw->done);
-		/* disable receive interrupt */
-		hw->imr &= ~ALTERA_SPI_CONTROL_IRRDY_MSK;
-		writel(hw->imr, hw->base + ALTERA_SPI_CONTROL);
+		altera_spi_tx_word(hw);
 	} else {
 		while (hw->count < hw->len) {
-			unsigned int rxd;
-
-			writel(hw_txbyte(hw, hw->count),
-			       hw->base + ALTERA_SPI_TXDATA);
+			altera_spi_tx_word(hw);
 
 			while (!(readl(hw->base + ALTERA_SPI_STATUS) &
 				 ALTERA_SPI_STATUS_RRDY_MSK))
 				cpu_relax();
 
-			rxd = readl(hw->base + ALTERA_SPI_RXDATA);
-			if (hw->rx) {
-				switch (hw->bytes_per_word) {
-				case 1:
-					hw->rx[hw->count] = rxd;
-					break;
-				case 2:
-					hw->rx[hw->count * 2] = rxd;
-					hw->rx[hw->count * 2 + 1] = rxd >> 8;
-					break;
-				}
-			}
-
-			hw->count++;
+			altera_spi_rx_word(hw);
 		}
+		spi_finalize_current_transfer(master);
 	}
 
-	return hw->count * hw->bytes_per_word;
+	return t->len;
 }
 
 static irqreturn_t altera_spi_irq(int irq, void *dev)
 {
-	struct altera_spi *hw = dev;
-	unsigned int rxd;
+	struct spi_master *master = dev;
+	struct altera_spi *hw = spi_master_get_devdata(master);
 
-	rxd = readl(hw->base + ALTERA_SPI_RXDATA);
-	if (hw->rx) {
-		switch (hw->bytes_per_word) {
-		case 1:
-			hw->rx[hw->count] = rxd;
-			break;
-		case 2:
-			hw->rx[hw->count * 2] = rxd;
-			hw->rx[hw->count * 2 + 1] = rxd >> 8;
-			break;
-		}
-	}
+	altera_spi_rx_word(hw);
 
-	hw->count++;
+	if (hw->count < hw->len) {
+		altera_spi_tx_word(hw);
+	} else {
+		/* disable receive interrupt */
+		hw->imr &= ~ALTERA_SPI_CONTROL_IRRDY_MSK;
+		writel(hw->imr, hw->base + ALTERA_SPI_CONTROL);
 
-	if (hw->count < hw->len)
-		writel(hw_txbyte(hw, hw->count), hw->base + ALTERA_SPI_TXDATA);
-	else
-		complete(&hw->done);
+		spi_finalize_current_transfer(master);
+	}
 
 	return IRQ_HANDLED;
 }
@@ -214,14 +186,10 @@ static int altera_spi_probe(struct platform_device *pdev)
 	master->mode_bits = SPI_CS_HIGH;
 	master->bits_per_word_mask = SPI_BPW_RANGE_MASK(1, 16);
 	master->dev.of_node = pdev->dev.of_node;
+	master->transfer_one = altera_spi_txrx;
+	master->set_cs = altera_spi_set_cs;
 
 	hw = spi_master_get_devdata(master);
-	platform_set_drvdata(pdev, hw);
-
-	/* setup the state for the bitbang driver */
-	hw->bitbang.master = master;
-	hw->bitbang.chipselect = altera_spi_chipsel;
-	hw->bitbang.txrx_bufs = altera_spi_txrx;
 
 	/* find and map our resources */
 	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
@@ -239,15 +207,13 @@ static int altera_spi_probe(struct platform_device *pdev)
 	/* irq is optional */
 	hw->irq = platform_get_irq(pdev, 0);
 	if (hw->irq >= 0) {
-		init_completion(&hw->done);
 		err = devm_request_irq(&pdev->dev, hw->irq, altera_spi_irq, 0,
-				       pdev->name, hw);
+				       pdev->name, master);
 		if (err)
 			goto exit;
 	}
 
-	/* register our spi controller */
-	err = spi_bitbang_start(&hw->bitbang);
+	err = devm_spi_register_master(&pdev->dev, master);
 	if (err)
 		goto exit;
 	dev_info(&pdev->dev, "base %p, irq %d\n", hw->base, hw->irq);
@@ -258,16 +224,6 @@ static int altera_spi_probe(struct platform_device *pdev)
 	return err;
 }
 
-static int altera_spi_remove(struct platform_device *dev)
-{
-	struct altera_spi *hw = platform_get_drvdata(dev);
-	struct spi_master *master = hw->bitbang.master;
-
-	spi_bitbang_stop(&hw->bitbang);
-	spi_master_put(master);
-	return 0;
-}
-
 #ifdef CONFIG_OF
 static const struct of_device_id altera_spi_match[] = {
 	{ .compatible = "ALTR,spi-1.0", },
@@ -279,7 +235,6 @@ MODULE_DEVICE_TABLE(of, altera_spi_match);
 
 static struct platform_driver altera_spi_driver = {
 	.probe = altera_spi_probe,
-	.remove = altera_spi_remove,
 	.driver = {
 		.name = DRV_NAME,
 		.pm = NULL,
diff --git a/drivers/spi/spi-ath79.c b/drivers/spi/spi-ath79.c
index b89cee11f41815be61876bb9206e59cb4d6714b1..0719bd48489175a3fa7825e3269bf9ac42fa3770 100644
--- a/drivers/spi/spi-ath79.c
+++ b/drivers/spi/spi-ath79.c
@@ -39,15 +39,15 @@ struct ath79_spi {
 	u32			reg_ctrl;
 	void __iomem		*base;
 	struct clk		*clk;
-	unsigned		rrw_delay;
+	unsigned int		rrw_delay;
 };
 
-static inline u32 ath79_spi_rr(struct ath79_spi *sp, unsigned reg)
+static inline u32 ath79_spi_rr(struct ath79_spi *sp, unsigned int reg)
 {
 	return ioread32(sp->base + reg);
 }
 
-static inline void ath79_spi_wr(struct ath79_spi *sp, unsigned reg, u32 val)
+static inline void ath79_spi_wr(struct ath79_spi *sp, unsigned int reg, u32 val)
 {
 	iowrite32(val, sp->base + reg);
 }
@@ -57,7 +57,7 @@ static inline struct ath79_spi *ath79_spidev_to_sp(struct spi_device *spi)
 	return spi_master_get_devdata(spi->master);
 }
 
-static inline void ath79_spi_delay(struct ath79_spi *sp, unsigned nsecs)
+static inline void ath79_spi_delay(struct ath79_spi *sp, unsigned int nsecs)
 {
 	if (nsecs > sp->rrw_delay)
 		ndelay(nsecs - sp->rrw_delay);
@@ -148,9 +148,8 @@ static int ath79_spi_setup_cs(struct spi_device *spi)
 
 static void ath79_spi_cleanup_cs(struct spi_device *spi)
 {
-	if (gpio_is_valid(spi->cs_gpio)) {
+	if (gpio_is_valid(spi->cs_gpio))
 		gpio_free(spi->cs_gpio);
-	}
 }
 
 static int ath79_spi_setup(struct spi_device *spi)
@@ -176,7 +175,7 @@ static void ath79_spi_cleanup(struct spi_device *spi)
 	spi_bitbang_cleanup(spi);
 }
 
-static u32 ath79_spi_txrx_mode0(struct spi_device *spi, unsigned nsecs,
+static u32 ath79_spi_txrx_mode0(struct spi_device *spi, unsigned int nsecs,
 			       u32 word, u8 bits)
 {
 	struct ath79_spi *sp = ath79_spidev_to_sp(spi);
diff --git a/drivers/spi/spi-bcm-qspi.c b/drivers/spi/spi-bcm-qspi.c
index b19722ba908c1be3855e42d21b25f2e6fb38eae8..6ef6c44f39f5f7e591710029814ca03128455299 100644
--- a/drivers/spi/spi-bcm-qspi.c
+++ b/drivers/spi/spi-bcm-qspi.c
@@ -25,7 +25,6 @@
 #include <linux/ioport.h>
 #include <linux/kernel.h>
 #include <linux/module.h>
-#include <linux/mtd/spi-nor.h>
 #include <linux/of.h>
 #include <linux/of_irq.h>
 #include <linux/platform_device.h>
@@ -349,76 +348,60 @@ static void bcm_qspi_bspi_set_xfer_params(struct bcm_qspi *qspi, u8 cmd_byte,
 	bcm_qspi_write(qspi, BSPI, BSPI_FLEX_MODE_ENABLE, flex_mode);
 }
 
-static int bcm_qspi_bspi_set_flex_mode(struct bcm_qspi *qspi, int width,
-				       int addrlen, int hp)
+static int bcm_qspi_bspi_set_flex_mode(struct bcm_qspi *qspi,
+				       struct spi_flash_read_message *msg,
+				       int hp)
 {
 	int bpc = 0, bpp = 0;
-	u8 command = SPINOR_OP_READ_FAST;
-	int flex_mode = 1, rv = 0;
-	bool spans_4byte = false;
+	u8 command = msg->read_opcode;
+	int width  = msg->data_nbits ? msg->data_nbits : SPI_NBITS_SINGLE;
+	int addrlen = msg->addr_width;
+	int addr_nbits = msg->addr_nbits ? msg->addr_nbits : SPI_NBITS_SINGLE;
+	int flex_mode = 1;
 
 	dev_dbg(&qspi->pdev->dev, "set flex mode w %x addrlen %x hp %d\n",
 		width, addrlen, hp);
 
-	if (addrlen == BSPI_ADDRLEN_4BYTES) {
+	if (addrlen == BSPI_ADDRLEN_4BYTES)
 		bpp = BSPI_BPP_ADDR_SELECT_MASK;
-		spans_4byte = true;
-	}
 
-	bpp |= 8;
+	bpp |= msg->dummy_bytes * (8/addr_nbits);
 
 	switch (width) {
 	case SPI_NBITS_SINGLE:
 		if (addrlen == BSPI_ADDRLEN_3BYTES)
 			/* default mode, does not need flex_cmd */
 			flex_mode = 0;
-		else
-			command = SPINOR_OP_READ_FAST_4B;
 		break;
 	case SPI_NBITS_DUAL:
 		bpc = 0x00000001;
 		if (hp) {
 			bpc |= 0x00010100; /* address and mode are 2-bit */
 			bpp = BSPI_BPP_MODE_SELECT_MASK;
-			command = OPCODE_DIOR;
-			if (spans_4byte)
-				command = OPCODE_DIOR_4B;
-		} else {
-			command = SPINOR_OP_READ_1_1_2;
-			if (spans_4byte)
-				command = SPINOR_OP_READ_1_1_2_4B;
 		}
 		break;
 	case SPI_NBITS_QUAD:
 		bpc = 0x00000002;
 		if (hp) {
 			bpc |= 0x00020200; /* address and mode are 4-bit */
-			bpp = 4; /* dummy cycles */
-			bpp |= BSPI_BPP_ADDR_SELECT_MASK;
-			command = OPCODE_QIOR;
-			if (spans_4byte)
-				command = OPCODE_QIOR_4B;
-		} else {
-			command = SPINOR_OP_READ_1_1_4;
-			if (spans_4byte)
-				command = SPINOR_OP_READ_1_1_4_4B;
+			bpp |= BSPI_BPP_MODE_SELECT_MASK;
 		}
 		break;
 	default:
-		rv = -EINVAL;
-		break;
+		return -EINVAL;
 	}
 
-	if (rv == 0)
-		bcm_qspi_bspi_set_xfer_params(qspi, command, bpp, bpc,
-					      flex_mode);
+	bcm_qspi_bspi_set_xfer_params(qspi, command, bpp, bpc, flex_mode);
 
-	return rv;
+	return 0;
 }
 
-static int bcm_qspi_bspi_set_override(struct bcm_qspi *qspi, int width,
-				      int addrlen, int hp)
+static int bcm_qspi_bspi_set_override(struct bcm_qspi *qspi,
+				      struct spi_flash_read_message *msg,
+				      int hp)
 {
+	int width = msg->data_nbits ? msg->data_nbits : SPI_NBITS_SINGLE;
+	int addrlen = msg->addr_width;
 	u32 data = bcm_qspi_read(qspi, BSPI, BSPI_STRAP_OVERRIDE_CTRL);
 
 	dev_dbg(&qspi->pdev->dev, "set override mode w %x addrlen %x hp %d\n",
@@ -430,7 +413,6 @@ static int bcm_qspi_bspi_set_override(struct bcm_qspi *qspi, int width,
 		data &= ~(BSPI_STRAP_OVERRIDE_CTRL_DATA_QUAD |
 			  BSPI_STRAP_OVERRIDE_CTRL_DATA_DUAL);
 		break;
-
 	case SPI_NBITS_QUAD:
 		/* clear dual mode and set quad mode */
 		data &= ~BSPI_STRAP_OVERRIDE_CTRL_DATA_DUAL;
@@ -455,15 +437,17 @@ static int bcm_qspi_bspi_set_override(struct bcm_qspi *qspi, int width,
 	/* set the override mode */
 	data |=	BSPI_STRAP_OVERRIDE_CTRL_OVERRIDE;
 	bcm_qspi_write(qspi, BSPI, BSPI_STRAP_OVERRIDE_CTRL, data);
-	bcm_qspi_bspi_set_xfer_params(qspi, SPINOR_OP_READ_FAST, 0, 0, 0);
+	bcm_qspi_bspi_set_xfer_params(qspi, msg->read_opcode, 0, 0, 0);
 
 	return 0;
 }
 
 static int bcm_qspi_bspi_set_mode(struct bcm_qspi *qspi,
-				  int width, int addrlen, int hp)
+				  struct spi_flash_read_message *msg, int hp)
 {
 	int error = 0;
+	int width = msg->data_nbits ? msg->data_nbits : SPI_NBITS_SINGLE;
+	int addrlen = msg->addr_width;
 
 	/* default mode */
 	qspi->xfer_mode.flex_mode = true;
@@ -475,23 +459,13 @@ static int bcm_qspi_bspi_set_mode(struct bcm_qspi *qspi,
 		mask = BSPI_STRAP_OVERRIDE_CTRL_OVERRIDE;
 		if (val & mask || qspi->s3_strap_override_ctrl & mask) {
 			qspi->xfer_mode.flex_mode = false;
-			bcm_qspi_write(qspi, BSPI, BSPI_FLEX_MODE_ENABLE,
-				       0);
-
-			if ((val | qspi->s3_strap_override_ctrl) &
-			    BSPI_STRAP_OVERRIDE_CTRL_DATA_DUAL)
-				width = SPI_NBITS_DUAL;
-			else if ((val |  qspi->s3_strap_override_ctrl) &
-				 BSPI_STRAP_OVERRIDE_CTRL_DATA_QUAD)
-				width = SPI_NBITS_QUAD;
-
-			error = bcm_qspi_bspi_set_override(qspi, width, addrlen,
-							   hp);
+			bcm_qspi_write(qspi, BSPI, BSPI_FLEX_MODE_ENABLE, 0);
+			error = bcm_qspi_bspi_set_override(qspi, msg, hp);
 		}
 	}
 
 	if (qspi->xfer_mode.flex_mode)
-		error = bcm_qspi_bspi_set_flex_mode(qspi, width, addrlen, hp);
+		error = bcm_qspi_bspi_set_flex_mode(qspi, msg, hp);
 
 	if (error) {
 		dev_warn(&qspi->pdev->dev,
@@ -981,7 +955,7 @@ static int bcm_qspi_flash_read(struct spi_device *spi,
 	struct bcm_qspi *qspi = spi_master_get_devdata(spi->master);
 	int ret = 0;
 	bool mspi_read = false;
-	u32 io_width, addrlen, addr, len;
+	u32 addr, len;
 	u_char *buf;
 
 	buf = msg->buf;
@@ -1010,9 +984,7 @@ static int bcm_qspi_flash_read(struct spi_device *spi,
 	if (mspi_read)
 		return bcm_qspi_mspi_flash_read(spi, msg);
 
-	io_width = msg->data_nbits ? msg->data_nbits : SPI_NBITS_SINGLE;
-	addrlen = msg->addr_width;
-	ret = bcm_qspi_bspi_set_mode(qspi, io_width, addrlen, -1);
+	ret = bcm_qspi_bspi_set_mode(qspi, msg, -1);
 
 	if (!ret)
 		ret = bcm_qspi_bspi_flash_read(spi, msg);
@@ -1422,6 +1394,11 @@ static int __maybe_unused bcm_qspi_suspend(struct device *dev)
 {
 	struct bcm_qspi *qspi = dev_get_drvdata(dev);
 
+	/* store the override strap value */
+	if (!bcm_qspi_bspi_ver_three(qspi))
+		qspi->s3_strap_override_ctrl =
+			bcm_qspi_read(qspi, BSPI, BSPI_STRAP_OVERRIDE_CTRL);
+
 	spi_master_suspend(qspi->master);
 	clk_disable(qspi->clk);
 	bcm_qspi_hw_uninit(qspi);
diff --git a/drivers/spi/spi-bcm63xx-hsspi.c b/drivers/spi/spi-bcm63xx-hsspi.c
index 4da2d4a524ca48075d7220737303274d559f064b..cbcba614b253375d024cf45f03408fa53337d8f5 100644
--- a/drivers/spi/spi-bcm63xx-hsspi.c
+++ b/drivers/spi/spi-bcm63xx-hsspi.c
@@ -108,7 +108,7 @@ struct bcm63xx_hsspi {
 	u8 cs_polarity;
 };
 
-static void bcm63xx_hsspi_set_cs(struct bcm63xx_hsspi *bs, unsigned cs,
+static void bcm63xx_hsspi_set_cs(struct bcm63xx_hsspi *bs, unsigned int cs,
 				 bool active)
 {
 	u32 reg;
@@ -127,7 +127,7 @@ static void bcm63xx_hsspi_set_cs(struct bcm63xx_hsspi *bs, unsigned cs,
 static void bcm63xx_hsspi_set_clk(struct bcm63xx_hsspi *bs,
 				  struct spi_device *spi, int hz)
 {
-	unsigned profile = spi->chip_select;
+	unsigned int profile = spi->chip_select;
 	u32 reg;
 
 	reg = DIV_ROUND_UP(2048, DIV_ROUND_UP(bs->speed_hz, hz));
@@ -154,7 +154,7 @@ static void bcm63xx_hsspi_set_clk(struct bcm63xx_hsspi *bs,
 static int bcm63xx_hsspi_do_txrx(struct spi_device *spi, struct spi_transfer *t)
 {
 	struct bcm63xx_hsspi *bs = spi_master_get_devdata(spi->master);
-	unsigned chip_select = spi->chip_select;
+	unsigned int chip_select = spi->chip_select;
 	u16 opcode = 0;
 	int pending = t->len;
 	int step_size = HSSPI_BUFFER_LEN;
@@ -338,8 +338,8 @@ static int bcm63xx_hsspi_probe(struct platform_device *pdev)
 
 	irq = platform_get_irq(pdev, 0);
 	if (irq < 0) {
-		dev_err(dev, "no irq\n");
-		return -ENXIO;
+		dev_err(dev, "no irq: %d\n", irq);
+		return irq;
 	}
 
 	res_mem = platform_get_resource(pdev, IORESOURCE_MEM, 0);
diff --git a/drivers/spi/spi-bcm63xx.c b/drivers/spi/spi-bcm63xx.c
index 84c7356ce5b491554c77934e7cccec35653b1df5..bfe5754768f9738c7e96d27ea8795685f17d7be8 100644
--- a/drivers/spi/spi-bcm63xx.c
+++ b/drivers/spi/spi-bcm63xx.c
@@ -530,8 +530,8 @@ static int bcm63xx_spi_probe(struct platform_device *pdev)
 
 	irq = platform_get_irq(pdev, 0);
 	if (irq < 0) {
-		dev_err(dev, "no irq\n");
-		return -ENXIO;
+		dev_err(dev, "no irq: %d\n", irq);
+		return irq;
 	}
 
 	clk = devm_clk_get(dev, "spi");
diff --git a/drivers/spi/spi-cadence.c b/drivers/spi/spi-cadence.c
index f0b5c7b91f37b0921410f599cf485a775152e691..5c9516ae4942e5cf8b2ef381d2ecd496803cbf14 100644
--- a/drivers/spi/spi-cadence.c
+++ b/drivers/spi/spi-cadence.c
@@ -576,10 +576,10 @@ 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);
+	pm_runtime_enable(&pdev->dev);
 
 	ret = of_property_read_u32(pdev->dev.of_node, "num-cs", &num_cs);
 	if (ret < 0)
@@ -704,7 +704,9 @@ 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);
+	struct cdns_spi *xspi = spi_master_get_devdata(master);
 
+	cdns_spi_init_hw(xspi);
 	return spi_master_resume(master);
 }
 
diff --git a/drivers/spi/spi-ep93xx.c b/drivers/spi/spi-ep93xx.c
index b5d766064b7b59390d50e888ad0832605f47699f..e5cc07357746afd20c55c75ae3a407d35c2d14d0 100644
--- a/drivers/spi/spi-ep93xx.c
+++ b/drivers/spi/spi-ep93xx.c
@@ -70,12 +70,9 @@
 
 /**
  * struct ep93xx_spi - EP93xx SPI controller structure
- * @pdev: pointer to platform device
  * @clk: clock for the controller
- * @regs_base: pointer to ioremap()'d registers
+ * @mmio: pointer to ioremap()'d registers
  * @sspdr_phys: physical address of the SSPDR register
- * @wait: wait here until given transfer is completed
- * @current_msg: message that is currently processed (or %NULL if none)
  * @tx: current byte in transfer to transmit
  * @rx: current byte in transfer to receive
  * @fifo_level: how full is FIFO (%0..%SPI_FIFO_SIZE - %1). Receiving one
@@ -90,12 +87,9 @@
  *            the client
  */
 struct ep93xx_spi {
-	const struct platform_device	*pdev;
 	struct clk			*clk;
-	void __iomem			*regs_base;
+	void __iomem			*mmio;
 	unsigned long			sspdr_phys;
-	struct completion		wait;
-	struct spi_message		*current_msg;
 	size_t				tx;
 	size_t				rx;
 	size_t				fifo_level;
@@ -111,91 +105,23 @@ struct ep93xx_spi {
 /* converts bits per word to CR0.DSS value */
 #define bits_per_word_to_dss(bpw)	((bpw) - 1)
 
-static void ep93xx_spi_write_u8(const struct ep93xx_spi *espi,
-				u16 reg, u8 value)
-{
-	writeb(value, espi->regs_base + reg);
-}
-
-static u8 ep93xx_spi_read_u8(const struct ep93xx_spi *spi, u16 reg)
-{
-	return readb(spi->regs_base + reg);
-}
-
-static void ep93xx_spi_write_u16(const struct ep93xx_spi *espi,
-				 u16 reg, u16 value)
-{
-	writew(value, espi->regs_base + reg);
-}
-
-static u16 ep93xx_spi_read_u16(const struct ep93xx_spi *spi, u16 reg)
-{
-	return readw(spi->regs_base + reg);
-}
-
-static int ep93xx_spi_enable(const struct ep93xx_spi *espi)
-{
-	u8 regval;
-	int err;
-
-	err = clk_enable(espi->clk);
-	if (err)
-		return err;
-
-	regval = ep93xx_spi_read_u8(espi, SSPCR1);
-	regval |= SSPCR1_SSE;
-	ep93xx_spi_write_u8(espi, SSPCR1, regval);
-
-	return 0;
-}
-
-static void ep93xx_spi_disable(const struct ep93xx_spi *espi)
-{
-	u8 regval;
-
-	regval = ep93xx_spi_read_u8(espi, SSPCR1);
-	regval &= ~SSPCR1_SSE;
-	ep93xx_spi_write_u8(espi, SSPCR1, regval);
-
-	clk_disable(espi->clk);
-}
-
-static void ep93xx_spi_enable_interrupts(const struct ep93xx_spi *espi)
-{
-	u8 regval;
-
-	regval = ep93xx_spi_read_u8(espi, SSPCR1);
-	regval |= (SSPCR1_RORIE | SSPCR1_TIE | SSPCR1_RIE);
-	ep93xx_spi_write_u8(espi, SSPCR1, regval);
-}
-
-static void ep93xx_spi_disable_interrupts(const struct ep93xx_spi *espi)
-{
-	u8 regval;
-
-	regval = ep93xx_spi_read_u8(espi, SSPCR1);
-	regval &= ~(SSPCR1_RORIE | SSPCR1_TIE | SSPCR1_RIE);
-	ep93xx_spi_write_u8(espi, SSPCR1, regval);
-}
-
 /**
  * ep93xx_spi_calc_divisors() - calculates SPI clock divisors
- * @espi: ep93xx SPI controller struct
+ * @master: SPI master
  * @rate: desired SPI output clock rate
  * @div_cpsr: pointer to return the cpsr (pre-scaler) divider
  * @div_scr: pointer to return the scr divider
  */
-static int ep93xx_spi_calc_divisors(const struct ep93xx_spi *espi,
+static int ep93xx_spi_calc_divisors(struct spi_master *master,
 				    u32 rate, u8 *div_cpsr, u8 *div_scr)
 {
-	struct spi_master *master = platform_get_drvdata(espi->pdev);
+	struct ep93xx_spi *espi = spi_master_get_devdata(master);
 	unsigned long spi_clk_rate = clk_get_rate(espi->clk);
 	int cpsr, scr;
 
 	/*
 	 * Make sure that max value is between values supported by the
-	 * controller. Note that minimum value is already checked in
-	 * ep93xx_spi_transfer_one_message().
+	 * controller.
 	 */
 	rate = clamp(rate, master->min_speed_hz, master->max_speed_hz);
 
@@ -220,26 +146,18 @@ static int ep93xx_spi_calc_divisors(const struct ep93xx_spi *espi,
 	return -EINVAL;
 }
 
-static void ep93xx_spi_cs_control(struct spi_device *spi, bool enable)
-{
-	if (spi->mode & SPI_CS_HIGH)
-		enable = !enable;
-
-	if (gpio_is_valid(spi->cs_gpio))
-		gpio_set_value(spi->cs_gpio, !enable);
-}
-
-static int ep93xx_spi_chip_setup(const struct ep93xx_spi *espi,
+static int ep93xx_spi_chip_setup(struct spi_master *master,
 				 struct spi_device *spi,
 				 struct spi_transfer *xfer)
 {
+	struct ep93xx_spi *espi = spi_master_get_devdata(master);
 	u8 dss = bits_per_word_to_dss(xfer->bits_per_word);
 	u8 div_cpsr = 0;
 	u8 div_scr = 0;
 	u16 cr0;
 	int err;
 
-	err = ep93xx_spi_calc_divisors(espi, xfer->speed_hz,
+	err = ep93xx_spi_calc_divisors(master, xfer->speed_hz,
 				       &div_cpsr, &div_scr);
 	if (err)
 		return err;
@@ -248,51 +166,49 @@ static int ep93xx_spi_chip_setup(const struct ep93xx_spi *espi,
 	cr0 |= (spi->mode & (SPI_CPHA | SPI_CPOL)) << SSPCR0_MODE_SHIFT;
 	cr0 |= dss;
 
-	dev_dbg(&espi->pdev->dev, "setup: mode %d, cpsr %d, scr %d, dss %d\n",
+	dev_dbg(&master->dev, "setup: mode %d, cpsr %d, scr %d, dss %d\n",
 		spi->mode, div_cpsr, div_scr, dss);
-	dev_dbg(&espi->pdev->dev, "setup: cr0 %#x\n", cr0);
+	dev_dbg(&master->dev, "setup: cr0 %#x\n", cr0);
 
-	ep93xx_spi_write_u8(espi, SSPCPSR, div_cpsr);
-	ep93xx_spi_write_u16(espi, SSPCR0, cr0);
+	writel(div_cpsr, espi->mmio + SSPCPSR);
+	writel(cr0, espi->mmio + SSPCR0);
 
 	return 0;
 }
 
-static void ep93xx_do_write(struct ep93xx_spi *espi, struct spi_transfer *t)
+static void ep93xx_do_write(struct spi_master *master)
 {
-	if (t->bits_per_word > 8) {
-		u16 tx_val = 0;
+	struct ep93xx_spi *espi = spi_master_get_devdata(master);
+	struct spi_transfer *xfer = master->cur_msg->state;
+	u32 val = 0;
 
-		if (t->tx_buf)
-			tx_val = ((u16 *)t->tx_buf)[espi->tx];
-		ep93xx_spi_write_u16(espi, SSPDR, tx_val);
-		espi->tx += sizeof(tx_val);
+	if (xfer->bits_per_word > 8) {
+		if (xfer->tx_buf)
+			val = ((u16 *)xfer->tx_buf)[espi->tx];
+		espi->tx += 2;
 	} else {
-		u8 tx_val = 0;
-
-		if (t->tx_buf)
-			tx_val = ((u8 *)t->tx_buf)[espi->tx];
-		ep93xx_spi_write_u8(espi, SSPDR, tx_val);
-		espi->tx += sizeof(tx_val);
+		if (xfer->tx_buf)
+			val = ((u8 *)xfer->tx_buf)[espi->tx];
+		espi->tx += 1;
 	}
+	writel(val, espi->mmio + SSPDR);
 }
 
-static void ep93xx_do_read(struct ep93xx_spi *espi, struct spi_transfer *t)
+static void ep93xx_do_read(struct spi_master *master)
 {
-	if (t->bits_per_word > 8) {
-		u16 rx_val;
-
-		rx_val = ep93xx_spi_read_u16(espi, SSPDR);
-		if (t->rx_buf)
-			((u16 *)t->rx_buf)[espi->rx] = rx_val;
-		espi->rx += sizeof(rx_val);
+	struct ep93xx_spi *espi = spi_master_get_devdata(master);
+	struct spi_transfer *xfer = master->cur_msg->state;
+	u32 val;
+
+	val = readl(espi->mmio + SSPDR);
+	if (xfer->bits_per_word > 8) {
+		if (xfer->rx_buf)
+			((u16 *)xfer->rx_buf)[espi->rx] = val;
+		espi->rx += 2;
 	} else {
-		u8 rx_val;
-
-		rx_val = ep93xx_spi_read_u8(espi, SSPDR);
-		if (t->rx_buf)
-			((u8 *)t->rx_buf)[espi->rx] = rx_val;
-		espi->rx += sizeof(rx_val);
+		if (xfer->rx_buf)
+			((u8 *)xfer->rx_buf)[espi->rx] = val;
+		espi->rx += 1;
 	}
 }
 
@@ -307,44 +223,32 @@ static void ep93xx_do_read(struct ep93xx_spi *espi, struct spi_transfer *t)
  * When this function is finished, RX FIFO should be empty and TX FIFO should be
  * full.
  */
-static int ep93xx_spi_read_write(struct ep93xx_spi *espi)
+static int ep93xx_spi_read_write(struct spi_master *master)
 {
-	struct spi_message *msg = espi->current_msg;
-	struct spi_transfer *t = msg->state;
+	struct ep93xx_spi *espi = spi_master_get_devdata(master);
+	struct spi_transfer *xfer = master->cur_msg->state;
 
 	/* read as long as RX FIFO has frames in it */
-	while ((ep93xx_spi_read_u8(espi, SSPSR) & SSPSR_RNE)) {
-		ep93xx_do_read(espi, t);
+	while ((readl(espi->mmio + SSPSR) & SSPSR_RNE)) {
+		ep93xx_do_read(master);
 		espi->fifo_level--;
 	}
 
 	/* write as long as TX FIFO has room */
-	while (espi->fifo_level < SPI_FIFO_SIZE && espi->tx < t->len) {
-		ep93xx_do_write(espi, t);
+	while (espi->fifo_level < SPI_FIFO_SIZE && espi->tx < xfer->len) {
+		ep93xx_do_write(master);
 		espi->fifo_level++;
 	}
 
-	if (espi->rx == t->len)
+	if (espi->rx == xfer->len)
 		return 0;
 
 	return -EINPROGRESS;
 }
 
-static void ep93xx_spi_pio_transfer(struct ep93xx_spi *espi)
-{
-	/*
-	 * Now everything is set up for the current transfer. We prime the TX
-	 * FIFO, enable interrupts, and wait for the transfer to complete.
-	 */
-	if (ep93xx_spi_read_write(espi)) {
-		ep93xx_spi_enable_interrupts(espi);
-		wait_for_completion(&espi->wait);
-	}
-}
-
 /**
  * ep93xx_spi_dma_prepare() - prepares a DMA transfer
- * @espi: ep93xx SPI controller struct
+ * @master: SPI master
  * @dir: DMA transfer direction
  *
  * Function configures the DMA, maps the buffer and prepares the DMA
@@ -352,9 +256,11 @@ static void ep93xx_spi_pio_transfer(struct ep93xx_spi *espi)
  * in case of failure.
  */
 static struct dma_async_tx_descriptor *
-ep93xx_spi_dma_prepare(struct ep93xx_spi *espi, enum dma_transfer_direction dir)
+ep93xx_spi_dma_prepare(struct spi_master *master,
+		       enum dma_transfer_direction dir)
 {
-	struct spi_transfer *t = espi->current_msg->state;
+	struct ep93xx_spi *espi = spi_master_get_devdata(master);
+	struct spi_transfer *xfer = master->cur_msg->state;
 	struct dma_async_tx_descriptor *txd;
 	enum dma_slave_buswidth buswidth;
 	struct dma_slave_config conf;
@@ -362,10 +268,10 @@ ep93xx_spi_dma_prepare(struct ep93xx_spi *espi, enum dma_transfer_direction dir)
 	struct sg_table *sgt;
 	struct dma_chan *chan;
 	const void *buf, *pbuf;
-	size_t len = t->len;
+	size_t len = xfer->len;
 	int i, ret, nents;
 
-	if (t->bits_per_word > 8)
+	if (xfer->bits_per_word > 8)
 		buswidth = DMA_SLAVE_BUSWIDTH_2_BYTES;
 	else
 		buswidth = DMA_SLAVE_BUSWIDTH_1_BYTE;
@@ -375,14 +281,14 @@ ep93xx_spi_dma_prepare(struct ep93xx_spi *espi, enum dma_transfer_direction dir)
 
 	if (dir == DMA_DEV_TO_MEM) {
 		chan = espi->dma_rx;
-		buf = t->rx_buf;
+		buf = xfer->rx_buf;
 		sgt = &espi->rx_sgt;
 
 		conf.src_addr = espi->sspdr_phys;
 		conf.src_addr_width = buswidth;
 	} else {
 		chan = espi->dma_tx;
-		buf = t->tx_buf;
+		buf = xfer->tx_buf;
 		sgt = &espi->tx_sgt;
 
 		conf.dst_addr = espi->sspdr_phys;
@@ -429,7 +335,7 @@ ep93xx_spi_dma_prepare(struct ep93xx_spi *espi, enum dma_transfer_direction dir)
 	}
 
 	if (WARN_ON(len)) {
-		dev_warn(&espi->pdev->dev, "len = %zu expected 0!\n", len);
+		dev_warn(&master->dev, "len = %zu expected 0!\n", len);
 		return ERR_PTR(-EINVAL);
 	}
 
@@ -447,15 +353,16 @@ ep93xx_spi_dma_prepare(struct ep93xx_spi *espi, enum dma_transfer_direction dir)
 
 /**
  * ep93xx_spi_dma_finish() - finishes with a DMA transfer
- * @espi: ep93xx SPI controller struct
+ * @master: SPI master
  * @dir: DMA transfer direction
  *
  * Function finishes with the DMA transfer. After this, the DMA buffer is
  * unmapped.
  */
-static void ep93xx_spi_dma_finish(struct ep93xx_spi *espi,
+static void ep93xx_spi_dma_finish(struct spi_master *master,
 				  enum dma_transfer_direction dir)
 {
+	struct ep93xx_spi *espi = spi_master_get_devdata(master);
 	struct dma_chan *chan;
 	struct sg_table *sgt;
 
@@ -472,72 +379,107 @@ static void ep93xx_spi_dma_finish(struct ep93xx_spi *espi,
 
 static void ep93xx_spi_dma_callback(void *callback_param)
 {
-	complete(callback_param);
+	struct spi_master *master = callback_param;
+
+	ep93xx_spi_dma_finish(master, DMA_MEM_TO_DEV);
+	ep93xx_spi_dma_finish(master, DMA_DEV_TO_MEM);
+
+	spi_finalize_current_transfer(master);
 }
 
-static void ep93xx_spi_dma_transfer(struct ep93xx_spi *espi)
+static int ep93xx_spi_dma_transfer(struct spi_master *master)
 {
-	struct spi_message *msg = espi->current_msg;
+	struct ep93xx_spi *espi = spi_master_get_devdata(master);
 	struct dma_async_tx_descriptor *rxd, *txd;
 
-	rxd = ep93xx_spi_dma_prepare(espi, DMA_DEV_TO_MEM);
+	rxd = ep93xx_spi_dma_prepare(master, DMA_DEV_TO_MEM);
 	if (IS_ERR(rxd)) {
-		dev_err(&espi->pdev->dev, "DMA RX failed: %ld\n", PTR_ERR(rxd));
-		msg->status = PTR_ERR(rxd);
-		return;
+		dev_err(&master->dev, "DMA RX failed: %ld\n", PTR_ERR(rxd));
+		return PTR_ERR(rxd);
 	}
 
-	txd = ep93xx_spi_dma_prepare(espi, DMA_MEM_TO_DEV);
+	txd = ep93xx_spi_dma_prepare(master, DMA_MEM_TO_DEV);
 	if (IS_ERR(txd)) {
-		ep93xx_spi_dma_finish(espi, DMA_DEV_TO_MEM);
-		dev_err(&espi->pdev->dev, "DMA TX failed: %ld\n", PTR_ERR(txd));
-		msg->status = PTR_ERR(txd);
-		return;
+		ep93xx_spi_dma_finish(master, DMA_DEV_TO_MEM);
+		dev_err(&master->dev, "DMA TX failed: %ld\n", PTR_ERR(txd));
+		return PTR_ERR(txd);
 	}
 
 	/* We are ready when RX is done */
 	rxd->callback = ep93xx_spi_dma_callback;
-	rxd->callback_param = &espi->wait;
+	rxd->callback_param = master;
 
-	/* Now submit both descriptors and wait while they finish */
+	/* Now submit both descriptors and start DMA */
 	dmaengine_submit(rxd);
 	dmaengine_submit(txd);
 
 	dma_async_issue_pending(espi->dma_rx);
 	dma_async_issue_pending(espi->dma_tx);
 
-	wait_for_completion(&espi->wait);
-
-	ep93xx_spi_dma_finish(espi, DMA_MEM_TO_DEV);
-	ep93xx_spi_dma_finish(espi, DMA_DEV_TO_MEM);
+	/* signal that we need to wait for completion */
+	return 1;
 }
 
-/**
- * ep93xx_spi_process_transfer() - processes one SPI transfer
- * @espi: ep93xx SPI controller struct
- * @msg: current message
- * @t: transfer to process
- *
- * This function processes one SPI transfer given in @t. Function waits until
- * transfer is complete (may sleep) and updates @msg->status based on whether
- * transfer was successfully processed or not.
- */
-static void ep93xx_spi_process_transfer(struct ep93xx_spi *espi,
-					struct spi_message *msg,
-					struct spi_transfer *t)
+static irqreturn_t ep93xx_spi_interrupt(int irq, void *dev_id)
 {
-	int err;
+	struct spi_master *master = dev_id;
+	struct ep93xx_spi *espi = spi_master_get_devdata(master);
+	u32 val;
+
+	/*
+	 * If we got ROR (receive overrun) interrupt we know that something is
+	 * wrong. Just abort the message.
+	 */
+	if (readl(espi->mmio + SSPIIR) & SSPIIR_RORIS) {
+		/* clear the overrun interrupt */
+		writel(0, espi->mmio + SSPICR);
+		dev_warn(&master->dev,
+			 "receive overrun, aborting the message\n");
+		master->cur_msg->status = -EIO;
+	} else {
+		/*
+		 * Interrupt is either RX (RIS) or TX (TIS). For both cases we
+		 * simply execute next data transfer.
+		 */
+		if (ep93xx_spi_read_write(master)) {
+			/*
+			 * In normal case, there still is some processing left
+			 * for current transfer. Let's wait for the next
+			 * interrupt then.
+			 */
+			return IRQ_HANDLED;
+		}
+	}
 
-	msg->state = t;
+	/*
+	 * Current transfer is finished, either with error or with success. In
+	 * any case we disable interrupts and notify the worker to handle
+	 * any post-processing of the message.
+	 */
+	val = readl(espi->mmio + SSPCR1);
+	val &= ~(SSPCR1_RORIE | SSPCR1_TIE | SSPCR1_RIE);
+	writel(val, espi->mmio + SSPCR1);
+
+	spi_finalize_current_transfer(master);
+
+	return IRQ_HANDLED;
+}
 
-	err = ep93xx_spi_chip_setup(espi, msg->spi, t);
-	if (err) {
-		dev_err(&espi->pdev->dev,
-			"failed to setup chip for transfer\n");
-		msg->status = err;
-		return;
+static int ep93xx_spi_transfer_one(struct spi_master *master,
+				   struct spi_device *spi,
+				   struct spi_transfer *xfer)
+{
+	struct ep93xx_spi *espi = spi_master_get_devdata(master);
+	u32 val;
+	int ret;
+
+	ret = ep93xx_spi_chip_setup(master, spi, xfer);
+	if (ret) {
+		dev_err(&master->dev, "failed to setup chip for transfer\n");
+		return ret;
 	}
 
+	master->cur_msg->state = xfer;
 	espi->rx = 0;
 	espi->tx = 0;
 
@@ -546,83 +488,37 @@ static void ep93xx_spi_process_transfer(struct ep93xx_spi *espi,
 	 * fit into the FIFO and can be transferred with a single interrupt.
 	 * So in these cases we will be using PIO and don't bother for DMA.
 	 */
-	if (espi->dma_rx && t->len > SPI_FIFO_SIZE)
-		ep93xx_spi_dma_transfer(espi);
-	else
-		ep93xx_spi_pio_transfer(espi);
+	if (espi->dma_rx && xfer->len > SPI_FIFO_SIZE)
+		return ep93xx_spi_dma_transfer(master);
 
-	/*
-	 * In case of error during transmit, we bail out from processing
-	 * the message.
-	 */
-	if (msg->status)
-		return;
+	/* Using PIO so prime the TX FIFO and enable interrupts */
+	ep93xx_spi_read_write(master);
 
-	msg->actual_length += t->len;
+	val = readl(espi->mmio + SSPCR1);
+	val |= (SSPCR1_RORIE | SSPCR1_TIE | SSPCR1_RIE);
+	writel(val, espi->mmio + SSPCR1);
 
-	/*
-	 * After this transfer is finished, perform any possible
-	 * post-transfer actions requested by the protocol driver.
-	 */
-	if (t->delay_usecs) {
-		set_current_state(TASK_UNINTERRUPTIBLE);
-		schedule_timeout(usecs_to_jiffies(t->delay_usecs));
-	}
-	if (t->cs_change) {
-		if (!list_is_last(&t->transfer_list, &msg->transfers)) {
-			/*
-			 * In case protocol driver is asking us to drop the
-			 * chipselect briefly, we let the scheduler to handle
-			 * any "delay" here.
-			 */
-			ep93xx_spi_cs_control(msg->spi, false);
-			cond_resched();
-			ep93xx_spi_cs_control(msg->spi, true);
-		}
-	}
+	/* signal that we need to wait for completion */
+	return 1;
 }
 
-/*
- * ep93xx_spi_process_message() - process one SPI message
- * @espi: ep93xx SPI controller struct
- * @msg: message to process
- *
- * This function processes a single SPI message. We go through all transfers in
- * the message and pass them to ep93xx_spi_process_transfer(). Chipselect is
- * asserted during the whole message (unless per transfer cs_change is set).
- *
- * @msg->status contains %0 in case of success or negative error code in case of
- * failure.
- */
-static void ep93xx_spi_process_message(struct ep93xx_spi *espi,
-				       struct spi_message *msg)
+static int ep93xx_spi_prepare_message(struct spi_master *master,
+				      struct spi_message *msg)
 {
+	struct ep93xx_spi *espi = spi_master_get_devdata(master);
 	unsigned long timeout;
-	struct spi_transfer *t;
-	int err;
-
-	/*
-	 * Enable the SPI controller and its clock.
-	 */
-	err = ep93xx_spi_enable(espi);
-	if (err) {
-		dev_err(&espi->pdev->dev, "failed to enable SPI controller\n");
-		msg->status = err;
-		return;
-	}
 
 	/*
 	 * Just to be sure: flush any data from RX FIFO.
 	 */
 	timeout = jiffies + msecs_to_jiffies(SPI_TIMEOUT);
-	while (ep93xx_spi_read_u16(espi, SSPSR) & SSPSR_RNE) {
+	while (readl(espi->mmio + SSPSR) & SSPSR_RNE) {
 		if (time_after(jiffies, timeout)) {
-			dev_warn(&espi->pdev->dev,
+			dev_warn(&master->dev,
 				 "timeout while flushing RX FIFO\n");
-			msg->status = -ETIMEDOUT;
-			return;
+			return -ETIMEDOUT;
 		}
-		ep93xx_spi_read_u16(espi, SSPDR);
+		readl(espi->mmio + SSPDR);
 	}
 
 	/*
@@ -631,81 +527,38 @@ static void ep93xx_spi_process_message(struct ep93xx_spi *espi,
 	 */
 	espi->fifo_level = 0;
 
-	/*
-	 * Assert the chipselect.
-	 */
-	ep93xx_spi_cs_control(msg->spi, true);
-
-	list_for_each_entry(t, &msg->transfers, transfer_list) {
-		ep93xx_spi_process_transfer(espi, msg, t);
-		if (msg->status)
-			break;
-	}
-
-	/*
-	 * Now the whole message is transferred (or failed for some reason). We
-	 * deselect the device and disable the SPI controller.
-	 */
-	ep93xx_spi_cs_control(msg->spi, false);
-	ep93xx_spi_disable(espi);
+	return 0;
 }
 
-static int ep93xx_spi_transfer_one_message(struct spi_master *master,
-					   struct spi_message *msg)
+static int ep93xx_spi_prepare_hardware(struct spi_master *master)
 {
 	struct ep93xx_spi *espi = spi_master_get_devdata(master);
+	u32 val;
+	int ret;
 
-	msg->state = NULL;
-	msg->status = 0;
-	msg->actual_length = 0;
-
-	espi->current_msg = msg;
-	ep93xx_spi_process_message(espi, msg);
-	espi->current_msg = NULL;
+	ret = clk_enable(espi->clk);
+	if (ret)
+		return ret;
 
-	spi_finalize_current_message(master);
+	val = readl(espi->mmio + SSPCR1);
+	val |= SSPCR1_SSE;
+	writel(val, espi->mmio + SSPCR1);
 
 	return 0;
 }
 
-static irqreturn_t ep93xx_spi_interrupt(int irq, void *dev_id)
+static int ep93xx_spi_unprepare_hardware(struct spi_master *master)
 {
-	struct ep93xx_spi *espi = dev_id;
-	u8 irq_status = ep93xx_spi_read_u8(espi, SSPIIR);
+	struct ep93xx_spi *espi = spi_master_get_devdata(master);
+	u32 val;
 
-	/*
-	 * If we got ROR (receive overrun) interrupt we know that something is
-	 * wrong. Just abort the message.
-	 */
-	if (unlikely(irq_status & SSPIIR_RORIS)) {
-		/* clear the overrun interrupt */
-		ep93xx_spi_write_u8(espi, SSPICR, 0);
-		dev_warn(&espi->pdev->dev,
-			 "receive overrun, aborting the message\n");
-		espi->current_msg->status = -EIO;
-	} else {
-		/*
-		 * Interrupt is either RX (RIS) or TX (TIS). For both cases we
-		 * simply execute next data transfer.
-		 */
-		if (ep93xx_spi_read_write(espi)) {
-			/*
-			 * In normal case, there still is some processing left
-			 * for current transfer. Let's wait for the next
-			 * interrupt then.
-			 */
-			return IRQ_HANDLED;
-		}
-	}
+	val = readl(espi->mmio + SSPCR1);
+	val &= ~SSPCR1_SSE;
+	writel(val, espi->mmio + SSPCR1);
 
-	/*
-	 * Current transfer is finished, either with error or with success. In
-	 * any case we disable interrupts and notify the worker to handle
-	 * any post-processing of the message.
-	 */
-	ep93xx_spi_disable_interrupts(espi);
-	complete(&espi->wait);
-	return IRQ_HANDLED;
+	clk_disable(espi->clk);
+
+	return 0;
 }
 
 static bool ep93xx_spi_dma_filter(struct dma_chan *chan, void *filter_param)
@@ -809,7 +662,10 @@ static int ep93xx_spi_probe(struct platform_device *pdev)
 	if (!master)
 		return -ENOMEM;
 
-	master->transfer_one_message = ep93xx_spi_transfer_one_message;
+	master->prepare_transfer_hardware = ep93xx_spi_prepare_hardware;
+	master->unprepare_transfer_hardware = ep93xx_spi_unprepare_hardware;
+	master->prepare_message = ep93xx_spi_prepare_message;
+	master->transfer_one = ep93xx_spi_transfer_one;
 	master->bus_num = pdev->id;
 	master->mode_bits = SPI_CPOL | SPI_CPHA | SPI_CS_HIGH;
 	master->bits_per_word_mask = SPI_BPW_RANGE_MASK(4, 16);
@@ -850,26 +706,23 @@ static int ep93xx_spi_probe(struct platform_device *pdev)
 		goto fail_release_master;
 	}
 
-	init_completion(&espi->wait);
-
 	/*
 	 * Calculate maximum and minimum supported clock rates
 	 * for the controller.
 	 */
 	master->max_speed_hz = clk_get_rate(espi->clk) / 2;
 	master->min_speed_hz = clk_get_rate(espi->clk) / (254 * 256);
-	espi->pdev = pdev;
 
 	espi->sspdr_phys = res->start + SSPDR;
 
-	espi->regs_base = devm_ioremap_resource(&pdev->dev, res);
-	if (IS_ERR(espi->regs_base)) {
-		error = PTR_ERR(espi->regs_base);
+	espi->mmio = devm_ioremap_resource(&pdev->dev, res);
+	if (IS_ERR(espi->mmio)) {
+		error = PTR_ERR(espi->mmio);
 		goto fail_release_master;
 	}
 
 	error = devm_request_irq(&pdev->dev, irq, ep93xx_spi_interrupt,
-				0, "ep93xx-spi", espi);
+				0, "ep93xx-spi", master);
 	if (error) {
 		dev_err(&pdev->dev, "failed to request irq\n");
 		goto fail_release_master;
@@ -879,7 +732,7 @@ static int ep93xx_spi_probe(struct platform_device *pdev)
 		dev_warn(&pdev->dev, "DMA setup failed. Falling back to PIO\n");
 
 	/* make sure that the hardware is disabled */
-	ep93xx_spi_write_u8(espi, SSPCR1, 0);
+	writel(0, espi->mmio + SSPCR1);
 
 	error = devm_spi_register_master(&pdev->dev, master);
 	if (error) {
diff --git a/drivers/spi/spi-falcon.c b/drivers/spi/spi-falcon.c
index 286b2c81fc6bb0d7500568df47753b47094da8a2..f8638e82e5db95e6d152473cdbaabb7f40f2cd59 100644
--- a/drivers/spi/spi-falcon.c
+++ b/drivers/spi/spi-falcon.c
@@ -395,11 +395,6 @@ static int falcon_sflash_probe(struct platform_device *pdev)
 	struct spi_master *master;
 	int ret;
 
-	if (ltq_boot_select() != BS_SPI) {
-		dev_err(&pdev->dev, "invalid bootstrap options\n");
-		return -ENODEV;
-	}
-
 	master = spi_alloc_master(&pdev->dev, sizeof(*priv));
 	if (!master)
 		return -ENOMEM;
diff --git a/drivers/spi/spi-imx.c b/drivers/spi/spi-imx.c
index f9698b7aeb3b50bbcd79ee01a950faf48908e33c..babb15f079951021ca2669e68898adcc44fb646c 100644
--- a/drivers/spi/spi-imx.c
+++ b/drivers/spi/spi-imx.c
@@ -56,6 +56,7 @@
 
 /* The maximum  bytes that a sdma BD can transfer.*/
 #define MAX_SDMA_BD_BYTES  (1 << 15)
+#define MX51_ECSPI_CTRL_MAX_BURST	512
 
 enum spi_imx_devtype {
 	IMX1_CSPI,
@@ -63,7 +64,8 @@ enum spi_imx_devtype {
 	IMX27_CSPI,
 	IMX31_CSPI,
 	IMX35_CSPI,	/* CSPI on all i.mx except above */
-	IMX51_ECSPI,	/* ECSPI on i.mx51 and later */
+	IMX51_ECSPI,	/* ECSPI on i.mx51 */
+	IMX53_ECSPI,	/* ECSPI on i.mx53 and later */
 };
 
 struct spi_imx_data;
@@ -74,6 +76,9 @@ struct spi_imx_devtype_data {
 	void (*trigger)(struct spi_imx_data *);
 	int (*rx_available)(struct spi_imx_data *);
 	void (*reset)(struct spi_imx_data *);
+	bool has_dmamode;
+	unsigned int fifo_size;
+	bool dynamic_burst;
 	enum spi_imx_devtype devtype;
 };
 
@@ -94,12 +99,14 @@ struct spi_imx_data {
 	unsigned int bits_per_word;
 	unsigned int spi_drctl;
 
-	unsigned int count;
+	unsigned int count, remainder;
 	void (*tx)(struct spi_imx_data *);
 	void (*rx)(struct spi_imx_data *);
 	void *rx_buf;
 	const void *tx_buf;
 	unsigned int txfifo; /* number of words pushed in tx FIFO */
+	unsigned int dynamic_burst, read_u32;
+	unsigned int word_mask;
 
 	/* DMA */
 	bool usedma;
@@ -125,9 +132,9 @@ static inline int is_imx51_ecspi(struct spi_imx_data *d)
 	return d->devtype_data->devtype == IMX51_ECSPI;
 }
 
-static inline unsigned spi_imx_get_fifosize(struct spi_imx_data *d)
+static inline int is_imx53_ecspi(struct spi_imx_data *d)
 {
-	return is_imx51_ecspi(d) ? 64 : 8;
+	return d->devtype_data->devtype == IMX53_ECSPI;
 }
 
 #define MXC_SPI_BUF_RX(type)						\
@@ -219,7 +226,7 @@ static bool spi_imx_can_dma(struct spi_master *master, struct spi_device *spi,
 	if (bytes_per_word != 1 && bytes_per_word != 2 && bytes_per_word != 4)
 		return false;
 
-	for (i = spi_imx_get_fifosize(spi_imx) / 2; i > 0; i--) {
+	for (i = spi_imx->devtype_data->fifo_size / 2; i > 0; i--) {
 		if (!(transfer->len % (i * bytes_per_word)))
 			break;
 	}
@@ -228,6 +235,7 @@ static bool spi_imx_can_dma(struct spi_master *master, struct spi_device *spi,
 		return false;
 
 	spi_imx->wml = i;
+	spi_imx->dynamic_burst = 0;
 
 	return true;
 }
@@ -242,6 +250,7 @@ static bool spi_imx_can_dma(struct spi_master *master, struct spi_device *spi,
 #define MX51_ECSPI_CTRL_PREDIV_OFFSET	12
 #define MX51_ECSPI_CTRL_CS(cs)		((cs) << 18)
 #define MX51_ECSPI_CTRL_BL_OFFSET	20
+#define MX51_ECSPI_CTRL_BL_MASK		(0xfff << 20)
 
 #define MX51_ECSPI_CONFIG	0x0c
 #define MX51_ECSPI_CONFIG_SCLKPHA(cs)	(1 << ((cs) +  0))
@@ -269,6 +278,106 @@ static bool spi_imx_can_dma(struct spi_master *master, struct spi_device *spi,
 #define MX51_ECSPI_TESTREG	0x20
 #define MX51_ECSPI_TESTREG_LBC	BIT(31)
 
+static void spi_imx_buf_rx_swap_u32(struct spi_imx_data *spi_imx)
+{
+	unsigned int val = readl(spi_imx->base + MXC_CSPIRXDATA);
+#ifdef __LITTLE_ENDIAN
+	unsigned int bytes_per_word;
+#endif
+
+	if (spi_imx->rx_buf) {
+#ifdef __LITTLE_ENDIAN
+		bytes_per_word = spi_imx_bytes_per_word(spi_imx->bits_per_word);
+		if (bytes_per_word == 1)
+			val = cpu_to_be32(val);
+		else if (bytes_per_word == 2)
+			val = (val << 16) | (val >> 16);
+#endif
+		val &= spi_imx->word_mask;
+		*(u32 *)spi_imx->rx_buf = val;
+		spi_imx->rx_buf += sizeof(u32);
+	}
+}
+
+static void spi_imx_buf_rx_swap(struct spi_imx_data *spi_imx)
+{
+	unsigned int bytes_per_word;
+
+	bytes_per_word = spi_imx_bytes_per_word(spi_imx->bits_per_word);
+	if (spi_imx->read_u32) {
+		spi_imx_buf_rx_swap_u32(spi_imx);
+		return;
+	}
+
+	if (bytes_per_word == 1)
+		spi_imx_buf_rx_u8(spi_imx);
+	else if (bytes_per_word == 2)
+		spi_imx_buf_rx_u16(spi_imx);
+}
+
+static void spi_imx_buf_tx_swap_u32(struct spi_imx_data *spi_imx)
+{
+	u32 val = 0;
+#ifdef __LITTLE_ENDIAN
+	unsigned int bytes_per_word;
+#endif
+
+	if (spi_imx->tx_buf) {
+		val = *(u32 *)spi_imx->tx_buf;
+		val &= spi_imx->word_mask;
+		spi_imx->tx_buf += sizeof(u32);
+	}
+
+	spi_imx->count -= sizeof(u32);
+#ifdef __LITTLE_ENDIAN
+	bytes_per_word = spi_imx_bytes_per_word(spi_imx->bits_per_word);
+
+	if (bytes_per_word == 1)
+		val = cpu_to_be32(val);
+	else if (bytes_per_word == 2)
+		val = (val << 16) | (val >> 16);
+#endif
+	writel(val, spi_imx->base + MXC_CSPITXDATA);
+}
+
+static void spi_imx_buf_tx_swap(struct spi_imx_data *spi_imx)
+{
+	u32 ctrl, val;
+	unsigned int bytes_per_word;
+
+	if (spi_imx->count == spi_imx->remainder) {
+		ctrl = readl(spi_imx->base + MX51_ECSPI_CTRL);
+		ctrl &= ~MX51_ECSPI_CTRL_BL_MASK;
+		if (spi_imx->count > MX51_ECSPI_CTRL_MAX_BURST) {
+			spi_imx->remainder = spi_imx->count %
+					     MX51_ECSPI_CTRL_MAX_BURST;
+			val = MX51_ECSPI_CTRL_MAX_BURST * 8 - 1;
+		} else if (spi_imx->count >= sizeof(u32)) {
+			spi_imx->remainder = spi_imx->count % sizeof(u32);
+			val = (spi_imx->count - spi_imx->remainder) * 8 - 1;
+		} else {
+			spi_imx->remainder = 0;
+			val = spi_imx->bits_per_word - 1;
+			spi_imx->read_u32 = 0;
+		}
+
+		ctrl |= (val << MX51_ECSPI_CTRL_BL_OFFSET);
+		writel(ctrl, spi_imx->base + MX51_ECSPI_CTRL);
+	}
+
+	if (spi_imx->count >= sizeof(u32)) {
+		spi_imx_buf_tx_swap_u32(spi_imx);
+		return;
+	}
+
+	bytes_per_word = spi_imx_bytes_per_word(spi_imx->bits_per_word);
+
+	if (bytes_per_word == 1)
+		spi_imx_buf_tx_u8(spi_imx);
+	else if (bytes_per_word == 2)
+		spi_imx_buf_tx_u16(spi_imx);
+}
+
 /* MX51 eCSPI */
 static unsigned int mx51_ecspi_clkdiv(struct spi_imx_data *spi_imx,
 				      unsigned int fspi, unsigned int *fres)
@@ -513,8 +622,8 @@ static int mx31_config(struct spi_device *spi)
 		reg |= MX31_CSPICTRL_POL;
 	if (spi->mode & SPI_CS_HIGH)
 		reg |= MX31_CSPICTRL_SSPOL;
-	if (spi->cs_gpio < 0)
-		reg |= (spi->cs_gpio + 32) <<
+	if (!gpio_is_valid(spi->cs_gpio))
+		reg |= (spi->chip_select) <<
 			(is_imx35_cspi(spi_imx) ? MX35_CSPICTRL_CS_SHIFT :
 						  MX31_CSPICTRL_CS_SHIFT);
 
@@ -605,8 +714,8 @@ static int mx21_config(struct spi_device *spi)
 		reg |= MX21_CSPICTRL_POL;
 	if (spi->mode & SPI_CS_HIGH)
 		reg |= MX21_CSPICTRL_SSPOL;
-	if (spi->cs_gpio < 0)
-		reg |= (spi->cs_gpio + 32) << MX21_CSPICTRL_CS_SHIFT;
+	if (!gpio_is_valid(spi->cs_gpio))
+		reg |= spi->chip_select << MX21_CSPICTRL_CS_SHIFT;
 
 	writel(reg, spi_imx->base + MXC_CSPICTRL);
 
@@ -693,6 +802,9 @@ static struct spi_imx_devtype_data imx1_cspi_devtype_data = {
 	.trigger = mx1_trigger,
 	.rx_available = mx1_rx_available,
 	.reset = mx1_reset,
+	.fifo_size = 8,
+	.has_dmamode = false,
+	.dynamic_burst = false,
 	.devtype = IMX1_CSPI,
 };
 
@@ -702,6 +814,9 @@ static struct spi_imx_devtype_data imx21_cspi_devtype_data = {
 	.trigger = mx21_trigger,
 	.rx_available = mx21_rx_available,
 	.reset = mx21_reset,
+	.fifo_size = 8,
+	.has_dmamode = false,
+	.dynamic_burst = false,
 	.devtype = IMX21_CSPI,
 };
 
@@ -712,6 +827,9 @@ static struct spi_imx_devtype_data imx27_cspi_devtype_data = {
 	.trigger = mx21_trigger,
 	.rx_available = mx21_rx_available,
 	.reset = mx21_reset,
+	.fifo_size = 8,
+	.has_dmamode = false,
+	.dynamic_burst = false,
 	.devtype = IMX27_CSPI,
 };
 
@@ -721,6 +839,9 @@ static struct spi_imx_devtype_data imx31_cspi_devtype_data = {
 	.trigger = mx31_trigger,
 	.rx_available = mx31_rx_available,
 	.reset = mx31_reset,
+	.fifo_size = 8,
+	.has_dmamode = false,
+	.dynamic_burst = false,
 	.devtype = IMX31_CSPI,
 };
 
@@ -731,6 +852,9 @@ static struct spi_imx_devtype_data imx35_cspi_devtype_data = {
 	.trigger = mx31_trigger,
 	.rx_available = mx31_rx_available,
 	.reset = mx31_reset,
+	.fifo_size = 8,
+	.has_dmamode = true,
+	.dynamic_burst = false,
 	.devtype = IMX35_CSPI,
 };
 
@@ -740,9 +864,23 @@ static struct spi_imx_devtype_data imx51_ecspi_devtype_data = {
 	.trigger = mx51_ecspi_trigger,
 	.rx_available = mx51_ecspi_rx_available,
 	.reset = mx51_ecspi_reset,
+	.fifo_size = 64,
+	.has_dmamode = true,
+	.dynamic_burst = true,
 	.devtype = IMX51_ECSPI,
 };
 
+static struct spi_imx_devtype_data imx53_ecspi_devtype_data = {
+	.intctrl = mx51_ecspi_intctrl,
+	.config = mx51_ecspi_config,
+	.trigger = mx51_ecspi_trigger,
+	.rx_available = mx51_ecspi_rx_available,
+	.reset = mx51_ecspi_reset,
+	.fifo_size = 64,
+	.has_dmamode = true,
+	.devtype = IMX53_ECSPI,
+};
+
 static const struct platform_device_id spi_imx_devtype[] = {
 	{
 		.name = "imx1-cspi",
@@ -762,6 +900,9 @@ static const struct platform_device_id spi_imx_devtype[] = {
 	}, {
 		.name = "imx51-ecspi",
 		.driver_data = (kernel_ulong_t) &imx51_ecspi_devtype_data,
+	}, {
+		.name = "imx53-ecspi",
+		.driver_data = (kernel_ulong_t) &imx53_ecspi_devtype_data,
 	}, {
 		/* sentinel */
 	}
@@ -774,6 +915,7 @@ static const struct of_device_id spi_imx_dt_ids[] = {
 	{ .compatible = "fsl,imx31-cspi", .data = &imx31_cspi_devtype_data, },
 	{ .compatible = "fsl,imx35-cspi", .data = &imx35_cspi_devtype_data, },
 	{ .compatible = "fsl,imx51-ecspi", .data = &imx51_ecspi_devtype_data, },
+	{ .compatible = "fsl,imx53-ecspi", .data = &imx53_ecspi_devtype_data, },
 	{ /* sentinel */ }
 };
 MODULE_DEVICE_TABLE(of, spi_imx_dt_ids);
@@ -783,6 +925,9 @@ static void spi_imx_chipselect(struct spi_device *spi, int is_active)
 	int active = is_active != BITBANG_CS_INACTIVE;
 	int dev_is_lowactive = !(spi->mode & SPI_CS_HIGH);
 
+	if (spi->mode & SPI_NO_CS)
+		return;
+
 	if (!gpio_is_valid(spi->cs_gpio))
 		return;
 
@@ -791,9 +936,11 @@ static void spi_imx_chipselect(struct spi_device *spi, int is_active)
 
 static void spi_imx_push(struct spi_imx_data *spi_imx)
 {
-	while (spi_imx->txfifo < spi_imx_get_fifosize(spi_imx)) {
+	while (spi_imx->txfifo < spi_imx->devtype_data->fifo_size) {
 		if (!spi_imx->count)
 			break;
+		if (spi_imx->txfifo && (spi_imx->count == spi_imx->remainder))
+			break;
 		spi_imx->tx(spi_imx);
 		spi_imx->txfifo++;
 	}
@@ -887,15 +1034,37 @@ static int spi_imx_setupxfer(struct spi_device *spi,
 	spi_imx->speed_hz  = t->speed_hz;
 
 	/* Initialize the functions for transfer */
-	if (spi_imx->bits_per_word <= 8) {
-		spi_imx->rx = spi_imx_buf_rx_u8;
-		spi_imx->tx = spi_imx_buf_tx_u8;
-	} else if (spi_imx->bits_per_word <= 16) {
-		spi_imx->rx = spi_imx_buf_rx_u16;
-		spi_imx->tx = spi_imx_buf_tx_u16;
+	if (spi_imx->devtype_data->dynamic_burst) {
+		u32 mask;
+
+		spi_imx->dynamic_burst = 0;
+		spi_imx->remainder = 0;
+		spi_imx->read_u32  = 1;
+
+		mask = (1 << spi_imx->bits_per_word) - 1;
+		spi_imx->rx = spi_imx_buf_rx_swap;
+		spi_imx->tx = spi_imx_buf_tx_swap;
+		spi_imx->dynamic_burst = 1;
+		spi_imx->remainder = t->len;
+
+		if (spi_imx->bits_per_word <= 8)
+			spi_imx->word_mask = mask << 24 | mask << 16
+					     | mask << 8 | mask;
+		else if (spi_imx->bits_per_word <= 16)
+			spi_imx->word_mask = mask << 16 | mask;
+		else
+			spi_imx->word_mask = mask;
 	} else {
-		spi_imx->rx = spi_imx_buf_rx_u32;
-		spi_imx->tx = spi_imx_buf_tx_u32;
+		if (spi_imx->bits_per_word <= 8) {
+			spi_imx->rx = spi_imx_buf_rx_u8;
+			spi_imx->tx = spi_imx_buf_tx_u8;
+		} else if (spi_imx->bits_per_word <= 16) {
+			spi_imx->rx = spi_imx_buf_rx_u16;
+			spi_imx->tx = spi_imx_buf_tx_u16;
+		} else {
+			spi_imx->rx = spi_imx_buf_rx_u32;
+			spi_imx->tx = spi_imx_buf_tx_u32;
+		}
 	}
 
 	if (spi_imx_can_dma(spi_imx->bitbang.master, spi, t))
@@ -938,7 +1107,7 @@ static int spi_imx_sdma_init(struct device *dev, struct spi_imx_data *spi_imx,
 	if (of_machine_is_compatible("fsl,imx6dl"))
 		return 0;
 
-	spi_imx->wml = spi_imx_get_fifosize(spi_imx) / 2;
+	spi_imx->wml = spi_imx->devtype_data->fifo_size / 2;
 
 	/* Prepare for TX DMA: */
 	master->dma_tx = dma_request_slave_channel_reason(dev, "tx");
@@ -1109,6 +1278,9 @@ static int spi_imx_setup(struct spi_device *spi)
 	dev_dbg(&spi->dev, "%s: mode %d, %u bpw, %d hz\n", __func__,
 		 spi->mode, spi->bits_per_word, spi->max_speed_hz);
 
+	if (spi->mode & SPI_NO_CS)
+		return 0;
+
 	if (gpio_is_valid(spi->cs_gpio))
 		gpio_direction_output(spi->cs_gpio,
 				      spi->mode & SPI_CS_HIGH ? 0 : 1);
@@ -1208,8 +1380,10 @@ static int spi_imx_probe(struct platform_device *pdev)
 	spi_imx->bitbang.master->cleanup = spi_imx_cleanup;
 	spi_imx->bitbang.master->prepare_message = spi_imx_prepare_message;
 	spi_imx->bitbang.master->unprepare_message = spi_imx_unprepare_message;
-	spi_imx->bitbang.master->mode_bits = SPI_CPOL | SPI_CPHA | SPI_CS_HIGH;
-	if (is_imx35_cspi(spi_imx) || is_imx51_ecspi(spi_imx))
+	spi_imx->bitbang.master->mode_bits = SPI_CPOL | SPI_CPHA | SPI_CS_HIGH \
+					     | SPI_NO_CS;
+	if (is_imx35_cspi(spi_imx) || is_imx51_ecspi(spi_imx) ||
+	    is_imx53_ecspi(spi_imx))
 		spi_imx->bitbang.master->mode_bits |= SPI_LOOP | SPI_READY;
 
 	spi_imx->spi_drctl = spi_drctl;
@@ -1262,7 +1436,7 @@ static int spi_imx_probe(struct platform_device *pdev)
 	 * Only validated on i.mx35 and i.mx6 now, can remove the constraint
 	 * if validated on other chips.
 	 */
-	if (is_imx35_cspi(spi_imx) || is_imx51_ecspi(spi_imx)) {
+	if (spi_imx->devtype_data->has_dmamode) {
 		ret = spi_imx_sdma_init(&pdev->dev, spi_imx, master);
 		if (ret == -EPROBE_DEFER)
 			goto out_clk_put;
diff --git a/drivers/spi/spi-loopback-test.c b/drivers/spi/spi-loopback-test.c
index 3459965004f86f387156ce8aa9a513cdd9c3b5f6..bed7403bb6b3a8b1e172dd5ddb286ad058ad8e83 100644
--- a/drivers/spi/spi-loopback-test.c
+++ b/drivers/spi/spi-loopback-test.c
@@ -32,39 +32,50 @@
 #include "spi-test.h"
 
 /* flag to only simulate transfers */
-int simulate_only;
+static int simulate_only;
 module_param(simulate_only, int, 0);
 MODULE_PARM_DESC(simulate_only, "if not 0 do not execute the spi message");
 
 /* dump spi messages */
-int dump_messages;
+static int dump_messages;
 module_param(dump_messages, int, 0);
 MODULE_PARM_DESC(dump_messages,
 		 "=1 dump the basic spi_message_structure, " \
 		 "=2 dump the spi_message_structure including data, " \
 		 "=3 dump the spi_message structure before and after execution");
 /* the device is jumpered for loopback - enabling some rx_buf tests */
-int loopback;
+static int loopback;
 module_param(loopback, int, 0);
 MODULE_PARM_DESC(loopback,
 		 "if set enable loopback mode, where the rx_buf "	\
 		 "is checked to match tx_buf after the spi_message "	\
 		 "is executed");
 
+static int loop_req;
+module_param(loop_req, int, 0);
+MODULE_PARM_DESC(loop_req,
+		 "if set controller will be asked to enable test loop mode. " \
+		 "If controller supported it, MISO and MOSI will be connected");
+
+static int no_cs;
+module_param(no_cs, int, 0);
+MODULE_PARM_DESC(no_cs,
+		 "if set Chip Select (CS) will not be used");
+
 /* run only a specific test */
-int run_only_test = -1;
+static int run_only_test = -1;
 module_param(run_only_test, int, 0);
 MODULE_PARM_DESC(run_only_test,
 		 "only run the test with this number (0-based !)");
 
 /* use vmalloc'ed buffers */
-int use_vmalloc;
+static int use_vmalloc;
 module_param(use_vmalloc, int, 0644);
 MODULE_PARM_DESC(use_vmalloc,
 		 "use vmalloc'ed buffers instead of kmalloc'ed");
 
 /* check rx ranges */
-int check_ranges = 1;
+static int check_ranges = 1;
 module_param(check_ranges, int, 0644);
 MODULE_PARM_DESC(check_ranges,
 		 "checks rx_buffer pattern are valid");
@@ -313,6 +324,17 @@ static int spi_loopback_test_probe(struct spi_device *spi)
 {
 	int ret;
 
+	if (loop_req || no_cs) {
+		spi->mode |= loop_req ? SPI_LOOP : 0;
+		spi->mode |= no_cs ? SPI_NO_CS : 0;
+		ret = spi_setup(spi);
+		if (ret) {
+			dev_err(&spi->dev, "SPI setup with SPI_LOOP or SPI_NO_CS failed (%d)\n",
+				ret);
+			return ret;
+		}
+	}
+
 	dev_info(&spi->dev, "Executing spi-loopback-tests\n");
 
 	ret = spi_test_run_tests(spi, spi_tests);
diff --git a/drivers/spi/spi-omap2-mcspi.c b/drivers/spi/spi-omap2-mcspi.c
index e048268d8ba25e58d8e23e0011f7470e7d99cbfe..9bf64e6eca9ba706b64e46a34fa9d0763680433d 100644
--- a/drivers/spi/spi-omap2-mcspi.c
+++ b/drivers/spi/spi-omap2-mcspi.c
@@ -1338,7 +1338,6 @@ static int omap2_mcspi_probe(struct platform_device *pdev)
 	struct resource		*r;
 	int			status = 0, i;
 	u32			regs_offset = 0;
-	static int		bus_num = 1;
 	struct device_node	*node = pdev->dev.of_node;
 	const struct of_device_id *match;
 
@@ -1374,14 +1373,11 @@ static int omap2_mcspi_probe(struct platform_device *pdev)
 
 		of_property_read_u32(node, "ti,spi-num-cs", &num_cs);
 		master->num_chipselect = num_cs;
-		master->bus_num = bus_num++;
 		if (of_get_property(node, "ti,pindir-d0-out-d1-in", NULL))
 			mcspi->pin_dir = MCSPI_PINDIR_D0_OUT_D1_IN;
 	} else {
 		pdata = dev_get_platdata(&pdev->dev);
 		master->num_chipselect = pdata->num_cs;
-		if (pdev->id != -1)
-			master->bus_num = pdev->id;
 		mcspi->pin_dir = pdata->pin_dir;
 	}
 	regs_offset = pdata->regs_offset;
diff --git a/drivers/spi/spi-orion.c b/drivers/spi/spi-orion.c
index 28fc9f161b9db488767cf01c4b4e318cc06c5ab8..4b6dd73b80da0f034976e5781dd34089377e408c 100644
--- a/drivers/spi/spi-orion.c
+++ b/drivers/spi/spi-orion.c
@@ -669,8 +669,8 @@ static int orion_spi_probe(struct platform_device *pdev)
 		status = of_property_read_u32(np, "reg", &cs);
 		if (status) {
 			dev_err(&pdev->dev,
-				"%s has no valid 'reg' property (%d)\n",
-				np->full_name, status);
+				"%pOF has no valid 'reg' property (%d)\n",
+				np, status);
 			status = 0;
 			continue;
 		}
diff --git a/drivers/spi/spi-pic32.c b/drivers/spi/spi-pic32.c
index fefb688a34328f863953b106052d1e9589e526f6..f8a45af1fa9f2434689b7946435d26859f895306 100644
--- a/drivers/spi/spi-pic32.c
+++ b/drivers/spi/spi-pic32.c
@@ -52,14 +52,14 @@ struct pic32_spi_regs {
 
 /* Bit fields of SPI Control Register */
 #define CTRL_RX_INT_SHIFT	0  /* Rx interrupt generation */
-#define  RX_FIFO_EMTPY		0
+#define  RX_FIFO_EMPTY		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_EMPTY		1 /* empty */
 #define  TX_FIFO_HALF_EMPTY	2 /* empty by half or more */
 #define  TX_FIFO_NOT_FULL	3 /* atleast one empty */
 
diff --git a/drivers/spi/spi-pl022.c b/drivers/spi/spi-pl022.c
index 2f76e022cc590093518d3c6785830e218d92b9cc..4797c57f426303b87d351d9b28a1dcb41fa12df1 100644
--- a/drivers/spi/spi-pl022.c
+++ b/drivers/spi/spi-pl022.c
@@ -2429,7 +2429,7 @@ static struct vendor_data vendor_lsi = {
 	.internal_cs_ctrl = true,
 };
 
-static struct amba_id pl022_ids[] = {
+static const struct amba_id pl022_ids[] = {
 	{
 		/*
 		 * ARM PL022 variant, this has a 16bit wide
diff --git a/drivers/spi/spi-pxa2xx.c b/drivers/spi/spi-pxa2xx.c
index 38d0536828927c0a05192e1765eb2915164685f0..4cb515a3104c1759b451c1f2d313953c009fb3ce 100644
--- a/drivers/spi/spi-pxa2xx.c
+++ b/drivers/spi/spi-pxa2xx.c
@@ -402,8 +402,8 @@ static void cs_assert(struct driver_data *drv_data)
 		return;
 	}
 
-	if (gpio_is_valid(chip->gpio_cs)) {
-		gpio_set_value(chip->gpio_cs, chip->gpio_cs_inverted);
+	if (chip->gpiod_cs) {
+		gpiod_set_value(chip->gpiod_cs, chip->gpio_cs_inverted);
 		return;
 	}
 
@@ -424,8 +424,8 @@ static void cs_deassert(struct driver_data *drv_data)
 		return;
 	}
 
-	if (gpio_is_valid(chip->gpio_cs)) {
-		gpio_set_value(chip->gpio_cs, !chip->gpio_cs_inverted);
+	if (chip->gpiod_cs) {
+		gpiod_set_value(chip->gpiod_cs, !chip->gpio_cs_inverted);
 		return;
 	}
 
@@ -1213,17 +1213,16 @@ static int setup_cs(struct spi_device *spi, struct chip_data *chip,
 		    struct pxa2xx_spi_chip *chip_info)
 {
 	struct driver_data *drv_data = spi_master_get_devdata(spi->master);
+	struct gpio_desc *gpiod;
 	int err = 0;
 
 	if (chip == NULL)
 		return 0;
 
 	if (drv_data->cs_gpiods) {
-		struct gpio_desc *gpiod;
-
 		gpiod = drv_data->cs_gpiods[spi->chip_select];
 		if (gpiod) {
-			chip->gpio_cs = desc_to_gpio(gpiod);
+			chip->gpiod_cs = gpiod;
 			chip->gpio_cs_inverted = spi->mode & SPI_CS_HIGH;
 			gpiod_set_value(gpiod, chip->gpio_cs_inverted);
 		}
@@ -1237,8 +1236,10 @@ static int setup_cs(struct spi_device *spi, struct chip_data *chip,
 	/* NOTE: setup() can be called multiple times, possibly with
 	 * different chip_info, release previously requested GPIO
 	 */
-	if (gpio_is_valid(chip->gpio_cs))
-		gpio_free(chip->gpio_cs);
+	if (chip->gpiod_cs) {
+		gpio_free(desc_to_gpio(chip->gpiod_cs));
+		chip->gpiod_cs = NULL;
+	}
 
 	/* If (*cs_control) is provided, ignore GPIO chip select */
 	if (chip_info->cs_control) {
@@ -1254,11 +1255,11 @@ static int setup_cs(struct spi_device *spi, struct chip_data *chip,
 			return err;
 		}
 
-		chip->gpio_cs = chip_info->gpio_cs;
+		gpiod = gpio_to_desc(chip_info->gpio_cs);
+		chip->gpiod_cs = gpiod;
 		chip->gpio_cs_inverted = spi->mode & SPI_CS_HIGH;
 
-		err = gpio_direction_output(chip->gpio_cs,
-					!chip->gpio_cs_inverted);
+		err = gpiod_direction_output(gpiod, !chip->gpio_cs_inverted);
 	}
 
 	return err;
@@ -1317,8 +1318,7 @@ static int setup(struct spi_device *spi)
 			}
 
 			chip->frm = spi->chip_select;
-		} else
-			chip->gpio_cs = -1;
+		}
 		chip->enable_dma = drv_data->master_info->enable_dma;
 		chip->timeout = TIMOUT_DFLT;
 	}
@@ -1416,8 +1416,8 @@ static void cleanup(struct spi_device *spi)
 		return;
 
 	if (drv_data->ssp_type != CE4100_SSP && !drv_data->cs_gpiods &&
-	    gpio_is_valid(chip->gpio_cs))
-		gpio_free(chip->gpio_cs);
+	    chip->gpiod_cs)
+		gpio_free(desc_to_gpio(chip->gpiod_cs));
 
 	kfree(chip);
 }
@@ -1769,8 +1769,7 @@ static int pxa2xx_spi_probe(struct platform_device *pdev)
 		for (i = 0; i < master->num_chipselect; i++) {
 			struct gpio_desc *gpiod;
 
-			gpiod = devm_gpiod_get_index(dev, "cs", i,
-						     GPIOD_OUT_HIGH);
+			gpiod = devm_gpiod_get_index(dev, "cs", i, GPIOD_ASIS);
 			if (IS_ERR(gpiod)) {
 				/* Means use native chip select */
 				if (PTR_ERR(gpiod) == -ENOENT)
diff --git a/drivers/spi/spi-pxa2xx.h b/drivers/spi/spi-pxa2xx.h
index 2823a00a940591a631b73a945c0c1bff88c1617b..94f7b0713281929c5706555af3a9ed8418045670 100644
--- a/drivers/spi/spi-pxa2xx.h
+++ b/drivers/spi/spi-pxa2xx.h
@@ -83,7 +83,7 @@ struct chip_data {
 	u16 lpss_tx_threshold;
 	u8 enable_dma;
 	union {
-		int gpio_cs;
+		struct gpio_desc *gpiod_cs;
 		unsigned int frm;
 	};
 	int gpio_cs_inverted;
diff --git a/drivers/spi/spi-qup.c b/drivers/spi/spi-qup.c
index 1bfa889b8427a2801f4d4eb2ab3be067de28fb27..974a8ce58b68bc31e38d717ff16564aec398232e 100644
--- a/drivers/spi/spi-qup.c
+++ b/drivers/spi/spi-qup.c
@@ -19,6 +19,7 @@
 #include <linux/list.h>
 #include <linux/module.h>
 #include <linux/of.h>
+#include <linux/of_device.h>
 #include <linux/platform_device.h>
 #include <linux/pm_runtime.h>
 #include <linux/spi/spi.h>
@@ -82,6 +83,8 @@
 #define QUP_IO_M_MODE_BAM		3
 
 /* QUP_OPERATIONAL fields */
+#define QUP_OP_IN_BLOCK_READ_REQ	BIT(13)
+#define QUP_OP_OUT_BLOCK_WRITE_REQ	BIT(12)
 #define QUP_OP_MAX_INPUT_DONE_FLAG	BIT(11)
 #define QUP_OP_MAX_OUTPUT_DONE_FLAG	BIT(10)
 #define QUP_OP_IN_SERVICE_FLAG		BIT(9)
@@ -118,7 +121,7 @@
 
 #define SPI_NUM_CHIPSELECTS		4
 
-#define SPI_MAX_DMA_XFER		(SZ_64K - 64)
+#define SPI_MAX_XFER			(SZ_64K - 64)
 
 /* high speed mode is when bus rate is greater then 26MHz */
 #define SPI_HS_MIN_RATE			26000000
@@ -147,13 +150,37 @@ struct spi_qup {
 	int			n_words;
 	int			tx_bytes;
 	int			rx_bytes;
+	const u8		*tx_buf;
+	u8			*rx_buf;
 	int			qup_v1;
 
-	int			use_dma;
+	int			mode;
 	struct dma_slave_config	rx_conf;
 	struct dma_slave_config	tx_conf;
 };
 
+static int spi_qup_io_config(struct spi_device *spi, struct spi_transfer *xfer);
+
+static inline bool spi_qup_is_flag_set(struct spi_qup *controller, u32 flag)
+{
+	u32 opflag = readl_relaxed(controller->base + QUP_OPERATIONAL);
+
+	return (opflag & flag) != 0;
+}
+
+static inline bool spi_qup_is_dma_xfer(int mode)
+{
+	if (mode == QUP_IO_M_MODE_DMOV || mode == QUP_IO_M_MODE_BAM)
+		return true;
+
+	return false;
+}
+
+/* get's the transaction size length */
+static inline unsigned int spi_qup_len(struct spi_qup *controller)
+{
+	return controller->n_words * controller->w_size;
+}
 
 static inline bool spi_qup_is_valid_state(struct spi_qup *controller)
 {
@@ -207,29 +234,26 @@ static int spi_qup_set_state(struct spi_qup *controller, u32 state)
 	return 0;
 }
 
-static void spi_qup_fifo_read(struct spi_qup *controller,
-			    struct spi_transfer *xfer)
+static void spi_qup_read_from_fifo(struct spi_qup *controller, u32 num_words)
 {
-	u8 *rx_buf = xfer->rx_buf;
-	u32 word, state;
-	int idx, shift, w_size;
-
-	w_size = controller->w_size;
+	u8 *rx_buf = controller->rx_buf;
+	int i, shift, num_bytes;
+	u32 word;
 
-	while (controller->rx_bytes < xfer->len) {
-
-		state = readl_relaxed(controller->base + QUP_OPERATIONAL);
-		if (0 == (state & QUP_OP_IN_FIFO_NOT_EMPTY))
-			break;
+	for (; num_words; num_words--) {
 
 		word = readl_relaxed(controller->base + QUP_INPUT_FIFO);
 
+		num_bytes = min_t(int, spi_qup_len(controller) -
+				       controller->rx_bytes,
+				       controller->w_size);
+
 		if (!rx_buf) {
-			controller->rx_bytes += w_size;
+			controller->rx_bytes += num_bytes;
 			continue;
 		}
 
-		for (idx = 0; idx < w_size; idx++, controller->rx_bytes++) {
+		for (i = 0; i < num_bytes; i++, controller->rx_bytes++) {
 			/*
 			 * The data format depends on bytes per SPI word:
 			 *  4 bytes: 0x12345678
@@ -237,38 +261,81 @@ static void spi_qup_fifo_read(struct spi_qup *controller,
 			 *  1 byte : 0x00000012
 			 */
 			shift = BITS_PER_BYTE;
-			shift *= (w_size - idx - 1);
+			shift *= (controller->w_size - i - 1);
 			rx_buf[controller->rx_bytes] = word >> shift;
 		}
 	}
 }
 
-static void spi_qup_fifo_write(struct spi_qup *controller,
-			    struct spi_transfer *xfer)
+static void spi_qup_read(struct spi_qup *controller, u32 *opflags)
 {
-	const u8 *tx_buf = xfer->tx_buf;
-	u32 word, state, data;
-	int idx, w_size;
+	u32 remainder, words_per_block, num_words;
+	bool is_block_mode = controller->mode == QUP_IO_M_MODE_BLOCK;
+
+	remainder = DIV_ROUND_UP(spi_qup_len(controller) - controller->rx_bytes,
+				 controller->w_size);
+	words_per_block = controller->in_blk_sz >> 2;
+
+	do {
+		/* ACK by clearing service flag */
+		writel_relaxed(QUP_OP_IN_SERVICE_FLAG,
+			       controller->base + QUP_OPERATIONAL);
+
+		if (is_block_mode) {
+			num_words = (remainder > words_per_block) ?
+					words_per_block : remainder;
+		} else {
+			if (!spi_qup_is_flag_set(controller,
+						 QUP_OP_IN_FIFO_NOT_EMPTY))
+				break;
 
-	w_size = controller->w_size;
+			num_words = 1;
+		}
 
-	while (controller->tx_bytes < xfer->len) {
+		/* read up to the maximum transfer size available */
+		spi_qup_read_from_fifo(controller, num_words);
 
-		state = readl_relaxed(controller->base + QUP_OPERATIONAL);
-		if (state & QUP_OP_OUT_FIFO_FULL)
+		remainder -= num_words;
+
+		/* if block mode, check to see if next block is available */
+		if (is_block_mode && !spi_qup_is_flag_set(controller,
+					QUP_OP_IN_BLOCK_READ_REQ))
 			break;
 
+	} while (remainder);
+
+	/*
+	 * Due to extra stickiness of the QUP_OP_IN_SERVICE_FLAG during block
+	 * reads, it has to be cleared again at the very end.  However, be sure
+	 * to refresh opflags value because MAX_INPUT_DONE_FLAG may now be
+	 * present and this is used to determine if transaction is complete
+	 */
+	*opflags = readl_relaxed(controller->base + QUP_OPERATIONAL);
+	if (is_block_mode && *opflags & QUP_OP_MAX_INPUT_DONE_FLAG)
+		writel_relaxed(QUP_OP_IN_SERVICE_FLAG,
+			       controller->base + QUP_OPERATIONAL);
+
+}
+
+static void spi_qup_write_to_fifo(struct spi_qup *controller, u32 num_words)
+{
+	const u8 *tx_buf = controller->tx_buf;
+	int i, num_bytes;
+	u32 word, data;
+
+	for (; num_words; num_words--) {
 		word = 0;
-		for (idx = 0; idx < w_size; idx++, controller->tx_bytes++) {
 
-			if (!tx_buf) {
-				controller->tx_bytes += w_size;
-				break;
+		num_bytes = min_t(int, spi_qup_len(controller) -
+				       controller->tx_bytes,
+				       controller->w_size);
+		if (tx_buf)
+			for (i = 0; i < num_bytes; i++) {
+				data = tx_buf[controller->tx_bytes + i];
+				word |= data << (BITS_PER_BYTE * (3 - i));
 			}
 
-			data = tx_buf[controller->tx_bytes];
-			word |= data << (BITS_PER_BYTE * (3 - idx));
-		}
+		controller->tx_bytes += num_bytes;
 
 		writel_relaxed(word, controller->base + QUP_OUTPUT_FIFO);
 	}
@@ -281,31 +348,61 @@ static void spi_qup_dma_done(void *data)
 	complete(&qup->done);
 }
 
-static int spi_qup_prep_sg(struct spi_master *master, struct spi_transfer *xfer,
-			   enum dma_transfer_direction dir,
+static void spi_qup_write(struct spi_qup *controller)
+{
+	bool is_block_mode = controller->mode == QUP_IO_M_MODE_BLOCK;
+	u32 remainder, words_per_block, num_words;
+
+	remainder = DIV_ROUND_UP(spi_qup_len(controller) - controller->tx_bytes,
+				 controller->w_size);
+	words_per_block = controller->out_blk_sz >> 2;
+
+	do {
+		/* ACK by clearing service flag */
+		writel_relaxed(QUP_OP_OUT_SERVICE_FLAG,
+			       controller->base + QUP_OPERATIONAL);
+
+		if (is_block_mode) {
+			num_words = (remainder > words_per_block) ?
+				words_per_block : remainder;
+		} else {
+			if (spi_qup_is_flag_set(controller,
+						QUP_OP_OUT_FIFO_FULL))
+				break;
+
+			num_words = 1;
+		}
+
+		spi_qup_write_to_fifo(controller, num_words);
+
+		remainder -= num_words;
+
+		/* if block mode, check to see if next block is available */
+		if (is_block_mode && !spi_qup_is_flag_set(controller,
+					QUP_OP_OUT_BLOCK_WRITE_REQ))
+			break;
+
+	} while (remainder);
+}
+
+static int spi_qup_prep_sg(struct spi_master *master, struct scatterlist *sgl,
+			   unsigned int nents, enum dma_transfer_direction dir,
 			   dma_async_tx_callback callback)
 {
 	struct spi_qup *qup = spi_master_get_devdata(master);
 	unsigned long flags = DMA_PREP_INTERRUPT | DMA_PREP_FENCE;
 	struct dma_async_tx_descriptor *desc;
-	struct scatterlist *sgl;
 	struct dma_chan *chan;
 	dma_cookie_t cookie;
-	unsigned int nents;
 
-	if (dir == DMA_MEM_TO_DEV) {
+	if (dir == DMA_MEM_TO_DEV)
 		chan = master->dma_tx;
-		nents = xfer->tx_sg.nents;
-		sgl = xfer->tx_sg.sgl;
-	} else {
+	else
 		chan = master->dma_rx;
-		nents = xfer->rx_sg.nents;
-		sgl = xfer->rx_sg.sgl;
-	}
 
 	desc = dmaengine_prep_slave_sg(chan, sgl, nents, dir, flags);
-	if (!desc)
-		return -EINVAL;
+	if (IS_ERR_OR_NULL(desc))
+		return desc ? PTR_ERR(desc) : -EINVAL;
 
 	desc->callback = callback;
 	desc->callback_param = qup;
@@ -324,9 +421,33 @@ static void spi_qup_dma_terminate(struct spi_master *master,
 		dmaengine_terminate_all(master->dma_rx);
 }
 
-static int spi_qup_do_dma(struct spi_master *master, struct spi_transfer *xfer)
+static u32 spi_qup_sgl_get_nents_len(struct scatterlist *sgl, u32 max,
+				     u32 *nents)
+{
+	struct scatterlist *sg;
+	u32 total = 0;
+
+	for (sg = sgl; sg; sg = sg_next(sg)) {
+		unsigned int len = sg_dma_len(sg);
+
+		/* check for overflow as well as limit */
+		if (((total + len) < total) || ((total + len) > max))
+			break;
+
+		total += len;
+		(*nents)++;
+	}
+
+	return total;
+}
+
+static int spi_qup_do_dma(struct spi_device *spi, struct spi_transfer *xfer,
+			  unsigned long timeout)
 {
 	dma_async_tx_callback rx_done = NULL, tx_done = NULL;
+	struct spi_master *master = spi->master;
+	struct spi_qup *qup = spi_master_get_devdata(master);
+	struct scatterlist *tx_sgl, *rx_sgl;
 	int ret;
 
 	if (xfer->rx_buf)
@@ -334,43 +455,122 @@ static int spi_qup_do_dma(struct spi_master *master, struct spi_transfer *xfer)
 	else if (xfer->tx_buf)
 		tx_done = spi_qup_dma_done;
 
-	if (xfer->rx_buf) {
-		ret = spi_qup_prep_sg(master, xfer, DMA_DEV_TO_MEM, rx_done);
-		if (ret)
-			return ret;
+	rx_sgl = xfer->rx_sg.sgl;
+	tx_sgl = xfer->tx_sg.sgl;
 
-		dma_async_issue_pending(master->dma_rx);
-	}
+	do {
+		u32 rx_nents = 0, tx_nents = 0;
 
-	if (xfer->tx_buf) {
-		ret = spi_qup_prep_sg(master, xfer, DMA_MEM_TO_DEV, tx_done);
+		if (rx_sgl)
+			qup->n_words = spi_qup_sgl_get_nents_len(rx_sgl,
+					SPI_MAX_XFER, &rx_nents) / qup->w_size;
+		if (tx_sgl)
+			qup->n_words = spi_qup_sgl_get_nents_len(tx_sgl,
+					SPI_MAX_XFER, &tx_nents) / qup->w_size;
+		if (!qup->n_words)
+			return -EIO;
+
+		ret = spi_qup_io_config(spi, xfer);
 		if (ret)
 			return ret;
 
-		dma_async_issue_pending(master->dma_tx);
-	}
+		/* before issuing the descriptors, set the QUP to run */
+		ret = spi_qup_set_state(qup, QUP_STATE_RUN);
+		if (ret) {
+			dev_warn(qup->dev, "cannot set RUN state\n");
+			return ret;
+		}
+		if (rx_sgl) {
+			ret = spi_qup_prep_sg(master, rx_sgl, rx_nents,
+					      DMA_DEV_TO_MEM, rx_done);
+			if (ret)
+				return ret;
+			dma_async_issue_pending(master->dma_rx);
+		}
+
+		if (tx_sgl) {
+			ret = spi_qup_prep_sg(master, tx_sgl, tx_nents,
+					      DMA_MEM_TO_DEV, tx_done);
+			if (ret)
+				return ret;
+
+			dma_async_issue_pending(master->dma_tx);
+		}
+
+		if (!wait_for_completion_timeout(&qup->done, timeout))
+			return -ETIMEDOUT;
+
+		for (; rx_sgl && rx_nents--; rx_sgl = sg_next(rx_sgl))
+			;
+		for (; tx_sgl && tx_nents--; tx_sgl = sg_next(tx_sgl))
+			;
+
+	} while (rx_sgl || tx_sgl);
 
 	return 0;
 }
 
-static int spi_qup_do_pio(struct spi_master *master, struct spi_transfer *xfer)
+static int spi_qup_do_pio(struct spi_device *spi, struct spi_transfer *xfer,
+			  unsigned long timeout)
 {
+	struct spi_master *master = spi->master;
 	struct spi_qup *qup = spi_master_get_devdata(master);
-	int ret;
+	int ret, n_words, iterations, offset = 0;
 
-	ret = spi_qup_set_state(qup, QUP_STATE_RUN);
-	if (ret) {
-		dev_warn(qup->dev, "cannot set RUN state\n");
-		return ret;
-	}
+	n_words = qup->n_words;
+	iterations = n_words / SPI_MAX_XFER; /* round down */
+	qup->rx_buf = xfer->rx_buf;
+	qup->tx_buf = xfer->tx_buf;
 
-	ret = spi_qup_set_state(qup, QUP_STATE_PAUSE);
-	if (ret) {
-		dev_warn(qup->dev, "cannot set PAUSE state\n");
-		return ret;
-	}
+	do {
+		if (iterations)
+			qup->n_words = SPI_MAX_XFER;
+		else
+			qup->n_words = n_words % SPI_MAX_XFER;
+
+		if (qup->tx_buf && offset)
+			qup->tx_buf = xfer->tx_buf + offset * SPI_MAX_XFER;
+
+		if (qup->rx_buf && offset)
+			qup->rx_buf = xfer->rx_buf + offset * SPI_MAX_XFER;
+
+		/*
+		 * if the transaction is small enough, we need
+		 * to fallback to FIFO mode
+		 */
+		if (qup->n_words <= (qup->in_fifo_sz / sizeof(u32)))
+			qup->mode = QUP_IO_M_MODE_FIFO;
 
-	spi_qup_fifo_write(qup, xfer);
+		ret = spi_qup_io_config(spi, xfer);
+		if (ret)
+			return ret;
+
+		ret = spi_qup_set_state(qup, QUP_STATE_RUN);
+		if (ret) {
+			dev_warn(qup->dev, "cannot set RUN state\n");
+			return ret;
+		}
+
+		ret = spi_qup_set_state(qup, QUP_STATE_PAUSE);
+		if (ret) {
+			dev_warn(qup->dev, "cannot set PAUSE state\n");
+			return ret;
+		}
+
+		if (qup->mode == QUP_IO_M_MODE_FIFO)
+			spi_qup_write(qup);
+
+		ret = spi_qup_set_state(qup, QUP_STATE_RUN);
+		if (ret) {
+			dev_warn(qup->dev, "cannot set RUN state\n");
+			return ret;
+		}
+
+		if (!wait_for_completion_timeout(&qup->done, timeout))
+			return -ETIMEDOUT;
+
+		offset++;
+	} while (iterations--);
 
 	return 0;
 }
@@ -378,29 +578,15 @@ static int spi_qup_do_pio(struct spi_master *master, struct spi_transfer *xfer)
 static irqreturn_t spi_qup_qup_irq(int irq, void *dev_id)
 {
 	struct spi_qup *controller = dev_id;
-	struct spi_transfer *xfer;
 	u32 opflags, qup_err, spi_err;
-	unsigned long flags;
 	int error = 0;
 
-	spin_lock_irqsave(&controller->lock, flags);
-	xfer = controller->xfer;
-	controller->xfer = NULL;
-	spin_unlock_irqrestore(&controller->lock, flags);
-
 	qup_err = readl_relaxed(controller->base + QUP_ERROR_FLAGS);
 	spi_err = readl_relaxed(controller->base + SPI_ERROR_FLAGS);
 	opflags = readl_relaxed(controller->base + QUP_OPERATIONAL);
 
 	writel_relaxed(qup_err, controller->base + QUP_ERROR_FLAGS);
 	writel_relaxed(spi_err, controller->base + SPI_ERROR_FLAGS);
-	writel_relaxed(opflags, controller->base + QUP_OPERATIONAL);
-
-	if (!xfer) {
-		dev_err_ratelimited(controller->dev, "unexpected irq %08x %08x %08x\n",
-				    qup_err, spi_err, opflags);
-		return IRQ_HANDLED;
-	}
 
 	if (qup_err) {
 		if (qup_err & QUP_ERROR_OUTPUT_OVER_RUN)
@@ -424,54 +610,27 @@ static irqreturn_t spi_qup_qup_irq(int irq, void *dev_id)
 		error = -EIO;
 	}
 
-	if (!controller->use_dma) {
+	if (spi_qup_is_dma_xfer(controller->mode)) {
+		writel_relaxed(opflags, controller->base + QUP_OPERATIONAL);
+	} else {
 		if (opflags & QUP_OP_IN_SERVICE_FLAG)
-			spi_qup_fifo_read(controller, xfer);
+			spi_qup_read(controller, &opflags);
 
 		if (opflags & QUP_OP_OUT_SERVICE_FLAG)
-			spi_qup_fifo_write(controller, xfer);
+			spi_qup_write(controller);
 	}
 
-	spin_lock_irqsave(&controller->lock, flags);
-	controller->error = error;
-	controller->xfer = xfer;
-	spin_unlock_irqrestore(&controller->lock, flags);
-
-	if (controller->rx_bytes == xfer->len || error)
+	if ((opflags & QUP_OP_MAX_INPUT_DONE_FLAG) || error)
 		complete(&controller->done);
 
 	return IRQ_HANDLED;
 }
 
-static u32
-spi_qup_get_mode(struct spi_master *master, struct spi_transfer *xfer)
-{
-	struct spi_qup *qup = spi_master_get_devdata(master);
-	u32 mode;
-
-	qup->w_size = 4;
-
-	if (xfer->bits_per_word <= 8)
-		qup->w_size = 1;
-	else if (xfer->bits_per_word <= 16)
-		qup->w_size = 2;
-
-	qup->n_words = xfer->len / qup->w_size;
-
-	if (qup->n_words <= (qup->in_fifo_sz / sizeof(u32)))
-		mode = QUP_IO_M_MODE_FIFO;
-	else
-		mode = QUP_IO_M_MODE_BLOCK;
-
-	return mode;
-}
-
-/* set clock freq ... bits per word */
-static int spi_qup_io_config(struct spi_device *spi, struct spi_transfer *xfer)
+/* set clock freq ... bits per word, determine mode */
+static int spi_qup_io_prep(struct spi_device *spi, struct spi_transfer *xfer)
 {
 	struct spi_qup *controller = spi_master_get_devdata(spi->master);
-	u32 config, iomode, mode, control;
-	int ret, n_words;
+	int ret;
 
 	if (spi->mode & SPI_LOOP && xfer->len > controller->in_fifo_sz) {
 		dev_err(controller->dev, "too big size for loopback %d > %d\n",
@@ -486,30 +645,59 @@ static int spi_qup_io_config(struct spi_device *spi, struct spi_transfer *xfer)
 		return -EIO;
 	}
 
+	controller->w_size = DIV_ROUND_UP(xfer->bits_per_word, 8);
+	controller->n_words = xfer->len / controller->w_size;
+
+	if (controller->n_words <= (controller->in_fifo_sz / sizeof(u32)))
+		controller->mode = QUP_IO_M_MODE_FIFO;
+	else if (spi->master->can_dma &&
+		 spi->master->can_dma(spi->master, spi, xfer) &&
+		 spi->master->cur_msg_mapped)
+		controller->mode = QUP_IO_M_MODE_BAM;
+	else
+		controller->mode = QUP_IO_M_MODE_BLOCK;
+
+	return 0;
+}
+
+/* prep qup for another spi transaction of specific type */
+static int spi_qup_io_config(struct spi_device *spi, struct spi_transfer *xfer)
+{
+	struct spi_qup *controller = spi_master_get_devdata(spi->master);
+	u32 config, iomode, control;
+	unsigned long flags;
+
+	spin_lock_irqsave(&controller->lock, flags);
+	controller->xfer     = xfer;
+	controller->error    = 0;
+	controller->rx_bytes = 0;
+	controller->tx_bytes = 0;
+	spin_unlock_irqrestore(&controller->lock, flags);
+
+
 	if (spi_qup_set_state(controller, QUP_STATE_RESET)) {
 		dev_err(controller->dev, "cannot set RESET state\n");
 		return -EIO;
 	}
 
-	mode = spi_qup_get_mode(spi->master, xfer);
-	n_words = controller->n_words;
-
-	if (mode == QUP_IO_M_MODE_FIFO) {
-		writel_relaxed(n_words, controller->base + QUP_MX_READ_CNT);
-		writel_relaxed(n_words, controller->base + QUP_MX_WRITE_CNT);
+	switch (controller->mode) {
+	case QUP_IO_M_MODE_FIFO:
+		writel_relaxed(controller->n_words,
+			       controller->base + QUP_MX_READ_CNT);
+		writel_relaxed(controller->n_words,
+			       controller->base + QUP_MX_WRITE_CNT);
 		/* must be zero for FIFO */
 		writel_relaxed(0, controller->base + QUP_MX_INPUT_CNT);
 		writel_relaxed(0, controller->base + QUP_MX_OUTPUT_CNT);
-	} else if (!controller->use_dma) {
-		writel_relaxed(n_words, controller->base + QUP_MX_INPUT_CNT);
-		writel_relaxed(n_words, controller->base + QUP_MX_OUTPUT_CNT);
+		break;
+	case QUP_IO_M_MODE_BAM:
+		writel_relaxed(controller->n_words,
+			       controller->base + QUP_MX_INPUT_CNT);
+		writel_relaxed(controller->n_words,
+			       controller->base + QUP_MX_OUTPUT_CNT);
 		/* must be zero for BLOCK and BAM */
 		writel_relaxed(0, controller->base + QUP_MX_READ_CNT);
 		writel_relaxed(0, controller->base + QUP_MX_WRITE_CNT);
-	} else {
-		mode = QUP_IO_M_MODE_BAM;
-		writel_relaxed(0, controller->base + QUP_MX_READ_CNT);
-		writel_relaxed(0, controller->base + QUP_MX_WRITE_CNT);
 
 		if (!controller->qup_v1) {
 			void __iomem *input_cnt;
@@ -524,23 +712,38 @@ static int spi_qup_io_config(struct spi_device *spi, struct spi_transfer *xfer)
 			if (xfer->tx_buf)
 				writel_relaxed(0, input_cnt);
 			else
-				writel_relaxed(n_words, input_cnt);
+				writel_relaxed(controller->n_words, input_cnt);
 
 			writel_relaxed(0, controller->base + QUP_MX_OUTPUT_CNT);
 		}
+		break;
+	case QUP_IO_M_MODE_BLOCK:
+		reinit_completion(&controller->done);
+		writel_relaxed(controller->n_words,
+			       controller->base + QUP_MX_INPUT_CNT);
+		writel_relaxed(controller->n_words,
+			       controller->base + QUP_MX_OUTPUT_CNT);
+		/* must be zero for BLOCK and BAM */
+		writel_relaxed(0, controller->base + QUP_MX_READ_CNT);
+		writel_relaxed(0, controller->base + QUP_MX_WRITE_CNT);
+		break;
+	default:
+		dev_err(controller->dev, "unknown mode = %d\n",
+				controller->mode);
+		return -EIO;
 	}
 
 	iomode = readl_relaxed(controller->base + QUP_IO_M_MODES);
 	/* Set input and output transfer mode */
 	iomode &= ~(QUP_IO_M_INPUT_MODE_MASK | QUP_IO_M_OUTPUT_MODE_MASK);
 
-	if (!controller->use_dma)
+	if (!spi_qup_is_dma_xfer(controller->mode))
 		iomode &= ~(QUP_IO_M_PACK_EN | QUP_IO_M_UNPACK_EN);
 	else
 		iomode |= QUP_IO_M_PACK_EN | QUP_IO_M_UNPACK_EN;
 
-	iomode |= (mode << QUP_IO_M_OUTPUT_MODE_MASK_SHIFT);
-	iomode |= (mode << QUP_IO_M_INPUT_MODE_MASK_SHIFT);
+	iomode |= (controller->mode << QUP_IO_M_OUTPUT_MODE_MASK_SHIFT);
+	iomode |= (controller->mode << QUP_IO_M_INPUT_MODE_MASK_SHIFT);
 
 	writel_relaxed(iomode, controller->base + QUP_IO_M_MODES);
 
@@ -581,7 +784,7 @@ static int spi_qup_io_config(struct spi_device *spi, struct spi_transfer *xfer)
 	config |= xfer->bits_per_word - 1;
 	config |= QUP_CONFIG_SPI_MODE;
 
-	if (controller->use_dma) {
+	if (spi_qup_is_dma_xfer(controller->mode)) {
 		if (!xfer->tx_buf)
 			config |= QUP_CONFIG_NO_OUTPUT;
 		if (!xfer->rx_buf)
@@ -599,7 +802,7 @@ static int spi_qup_io_config(struct spi_device *spi, struct spi_transfer *xfer)
 		 * status change in BAM mode
 		 */
 
-		if (mode == QUP_IO_M_MODE_BAM)
+		if (spi_qup_is_dma_xfer(controller->mode))
 			mask = QUP_OP_IN_SERVICE_FLAG | QUP_OP_OUT_SERVICE_FLAG;
 
 		writel_relaxed(mask, controller->base + QUP_OPERATIONAL_MASK);
@@ -616,12 +819,13 @@ static int spi_qup_transfer_one(struct spi_master *master,
 	unsigned long timeout, flags;
 	int ret = -EIO;
 
-	ret = spi_qup_io_config(spi, xfer);
+	ret = spi_qup_io_prep(spi, xfer);
 	if (ret)
 		return ret;
 
 	timeout = DIV_ROUND_UP(xfer->speed_hz, MSEC_PER_SEC);
-	timeout = DIV_ROUND_UP(xfer->len * 8, timeout);
+	timeout = DIV_ROUND_UP(min_t(unsigned long, SPI_MAX_XFER,
+				     xfer->len) * 8, timeout);
 	timeout = 100 * msecs_to_jiffies(timeout);
 
 	reinit_completion(&controller->done);
@@ -633,31 +837,22 @@ static int spi_qup_transfer_one(struct spi_master *master,
 	controller->tx_bytes = 0;
 	spin_unlock_irqrestore(&controller->lock, flags);
 
-	if (controller->use_dma)
-		ret = spi_qup_do_dma(master, xfer);
+	if (spi_qup_is_dma_xfer(controller->mode))
+		ret = spi_qup_do_dma(spi, xfer, timeout);
 	else
-		ret = spi_qup_do_pio(master, xfer);
+		ret = spi_qup_do_pio(spi, xfer, timeout);
 
 	if (ret)
 		goto exit;
 
-	if (spi_qup_set_state(controller, QUP_STATE_RUN)) {
-		dev_warn(controller->dev, "cannot set EXECUTE state\n");
-		goto exit;
-	}
-
-	if (!wait_for_completion_timeout(&controller->done, timeout))
-		ret = -ETIMEDOUT;
-
 exit:
 	spi_qup_set_state(controller, QUP_STATE_RESET);
 	spin_lock_irqsave(&controller->lock, flags);
-	controller->xfer = NULL;
 	if (!ret)
 		ret = controller->error;
 	spin_unlock_irqrestore(&controller->lock, flags);
 
-	if (ret && controller->use_dma)
+	if (ret && spi_qup_is_dma_xfer(controller->mode))
 		spi_qup_dma_terminate(master, xfer);
 
 	return ret;
@@ -668,26 +863,28 @@ static bool spi_qup_can_dma(struct spi_master *master, struct spi_device *spi,
 {
 	struct spi_qup *qup = spi_master_get_devdata(master);
 	size_t dma_align = dma_get_cache_alignment();
-	u32 mode;
-
-	qup->use_dma = 0;
+	int n_words;
 
-	if (xfer->rx_buf && (xfer->len % qup->in_blk_sz ||
-	    IS_ERR_OR_NULL(master->dma_rx) ||
-	    !IS_ALIGNED((size_t)xfer->rx_buf, dma_align)))
-		return false;
+	if (xfer->rx_buf) {
+		if (!IS_ALIGNED((size_t)xfer->rx_buf, dma_align) ||
+		    IS_ERR_OR_NULL(master->dma_rx))
+			return false;
+		if (qup->qup_v1 && (xfer->len % qup->in_blk_sz))
+			return false;
+	}
 
-	if (xfer->tx_buf && (xfer->len % qup->out_blk_sz ||
-	    IS_ERR_OR_NULL(master->dma_tx) ||
-	    !IS_ALIGNED((size_t)xfer->tx_buf, dma_align)))
-		return false;
+	if (xfer->tx_buf) {
+		if (!IS_ALIGNED((size_t)xfer->tx_buf, dma_align) ||
+		    IS_ERR_OR_NULL(master->dma_tx))
+			return false;
+		if (qup->qup_v1 && (xfer->len % qup->out_blk_sz))
+			return false;
+	}
 
-	mode = spi_qup_get_mode(master, xfer);
-	if (mode == QUP_IO_M_MODE_FIFO)
+	n_words = xfer->len / DIV_ROUND_UP(xfer->bits_per_word, 8);
+	if (n_words <= (qup->in_fifo_sz / sizeof(u32)))
 		return false;
 
-	qup->use_dma = 1;
-
 	return true;
 }
 
@@ -750,6 +947,24 @@ static int spi_qup_init_dma(struct spi_master *master, resource_size_t base)
 	return ret;
 }
 
+static void spi_qup_set_cs(struct spi_device *spi, bool val)
+{
+	struct spi_qup *controller;
+	u32 spi_ioc;
+	u32 spi_ioc_orig;
+
+	controller = spi_master_get_devdata(spi->master);
+	spi_ioc = readl_relaxed(controller->base + SPI_IO_CONTROL);
+	spi_ioc_orig = spi_ioc;
+	if (!val)
+		spi_ioc |= SPI_IO_C_FORCE_CS;
+	else
+		spi_ioc &= ~SPI_IO_C_FORCE_CS;
+
+	if (spi_ioc != spi_ioc_orig)
+		writel_relaxed(spi_ioc, controller->base + SPI_IO_CONTROL);
+}
+
 static int spi_qup_probe(struct platform_device *pdev)
 {
 	struct spi_master *master;
@@ -824,7 +1039,7 @@ static int spi_qup_probe(struct platform_device *pdev)
 	master->dev.of_node = pdev->dev.of_node;
 	master->auto_runtime_pm = true;
 	master->dma_alignment = dma_get_cache_alignment();
-	master->max_dma_len = SPI_MAX_DMA_XFER;
+	master->max_dma_len = SPI_MAX_XFER;
 
 	platform_set_drvdata(pdev, master);
 
@@ -842,9 +1057,10 @@ static int spi_qup_probe(struct platform_device *pdev)
 	else if (!ret)
 		master->can_dma = spi_qup_can_dma;
 
-	/* set v1 flag if device is version 1 */
-	if (of_device_is_compatible(dev->of_node, "qcom,spi-qup-v1.1.1"))
-		controller->qup_v1 = 1;
+	controller->qup_v1 = (uintptr_t)of_device_get_match_data(dev);
+
+	if (!controller->qup_v1)
+		master->set_cs = spi_qup_set_cs;
 
 	spin_lock_init(&controller->lock);
 	init_completion(&controller->done);
@@ -1037,7 +1253,7 @@ static int spi_qup_remove(struct platform_device *pdev)
 }
 
 static const struct of_device_id spi_qup_dt_match[] = {
-	{ .compatible = "qcom,spi-qup-v1.1.1", },
+	{ .compatible = "qcom,spi-qup-v1.1.1", .data = (void *)1, },
 	{ .compatible = "qcom,spi-qup-v2.1.1", },
 	{ .compatible = "qcom,spi-qup-v2.2.1", },
 	{ }
diff --git a/drivers/spi/spi-rockchip.c b/drivers/spi/spi-rockchip.c
index 0b4a52b3e1dc601234366d5998b7b5dac66506d1..fdcf3076681b5eeffb1885b428c8d00ff2682d2f 100644
--- a/drivers/spi/spi-rockchip.c
+++ b/drivers/spi/spi-rockchip.c
@@ -568,7 +568,13 @@ static void rockchip_spi_config(struct rockchip_spi *rs)
 
 	writel_relaxed(cr0, rs->regs + ROCKCHIP_SPI_CTRLR0);
 
-	writel_relaxed(rs->len - 1, rs->regs + ROCKCHIP_SPI_CTRLR1);
+	if (rs->n_bytes == 1)
+		writel_relaxed(rs->len - 1, rs->regs + ROCKCHIP_SPI_CTRLR1);
+	else if (rs->n_bytes == 2)
+		writel_relaxed((rs->len / 2) - 1, rs->regs + ROCKCHIP_SPI_CTRLR1);
+	else
+		writel_relaxed((rs->len * 2) - 1, rs->regs + ROCKCHIP_SPI_CTRLR1);
+
 	writel_relaxed(rs->fifo_len / 2 - 1, rs->regs + ROCKCHIP_SPI_TXFTLR);
 	writel_relaxed(rs->fifo_len / 2 - 1, rs->regs + ROCKCHIP_SPI_RXFTLR);
 
@@ -666,7 +672,7 @@ static bool rockchip_spi_can_dma(struct spi_master *master,
 
 static int rockchip_spi_probe(struct platform_device *pdev)
 {
-	int ret = 0;
+	int ret;
 	struct rockchip_spi *rs;
 	struct spi_master *master;
 	struct resource *mem;
@@ -703,13 +709,13 @@ static int rockchip_spi_probe(struct platform_device *pdev)
 	}
 
 	ret = clk_prepare_enable(rs->apb_pclk);
-	if (ret) {
+	if (ret < 0) {
 		dev_err(&pdev->dev, "Failed to enable apb_pclk\n");
 		goto err_put_master;
 	}
 
 	ret = clk_prepare_enable(rs->spiclk);
-	if (ret) {
+	if (ret < 0) {
 		dev_err(&pdev->dev, "Failed to enable spi_clk\n");
 		goto err_disable_apbclk;
 	}
@@ -786,7 +792,7 @@ static int rockchip_spi_probe(struct platform_device *pdev)
 	}
 
 	ret = devm_spi_register_master(&pdev->dev, master);
-	if (ret) {
+	if (ret < 0) {
 		dev_err(&pdev->dev, "Failed to register master\n");
 		goto err_free_dma_rx;
 	}
@@ -816,11 +822,15 @@ static int rockchip_spi_remove(struct platform_device *pdev)
 	struct spi_master *master = spi_master_get(platform_get_drvdata(pdev));
 	struct rockchip_spi *rs = spi_master_get_devdata(master);
 
-	pm_runtime_disable(&pdev->dev);
+	pm_runtime_get_sync(&pdev->dev);
 
 	clk_disable_unprepare(rs->spiclk);
 	clk_disable_unprepare(rs->apb_pclk);
 
+	pm_runtime_put_noidle(&pdev->dev);
+	pm_runtime_disable(&pdev->dev);
+	pm_runtime_set_suspended(&pdev->dev);
+
 	if (rs->dma_tx.ch)
 		dma_release_channel(rs->dma_tx.ch);
 	if (rs->dma_rx.ch)
@@ -834,43 +844,34 @@ static int rockchip_spi_remove(struct platform_device *pdev)
 #ifdef CONFIG_PM_SLEEP
 static int rockchip_spi_suspend(struct device *dev)
 {
-	int ret = 0;
+	int ret;
 	struct spi_master *master = dev_get_drvdata(dev);
 	struct rockchip_spi *rs = spi_master_get_devdata(master);
 
 	ret = spi_master_suspend(rs->master);
-	if (ret)
+	if (ret < 0)
 		return ret;
 
-	if (!pm_runtime_suspended(dev)) {
-		clk_disable_unprepare(rs->spiclk);
-		clk_disable_unprepare(rs->apb_pclk);
-	}
+	ret = pm_runtime_force_suspend(dev);
+	if (ret < 0)
+		return ret;
 
 	pinctrl_pm_select_sleep_state(dev);
 
-	return ret;
+	return 0;
 }
 
 static int rockchip_spi_resume(struct device *dev)
 {
-	int ret = 0;
+	int ret;
 	struct spi_master *master = dev_get_drvdata(dev);
 	struct rockchip_spi *rs = spi_master_get_devdata(master);
 
 	pinctrl_pm_select_default_state(dev);
 
-	if (!pm_runtime_suspended(dev)) {
-		ret = clk_prepare_enable(rs->apb_pclk);
-		if (ret < 0)
-			return ret;
-
-		ret = clk_prepare_enable(rs->spiclk);
-		if (ret < 0) {
-			clk_disable_unprepare(rs->apb_pclk);
-			return ret;
-		}
-	}
+	ret = pm_runtime_force_resume(dev);
+	if (ret < 0)
+		return ret;
 
 	ret = spi_master_resume(rs->master);
 	if (ret < 0) {
@@ -878,7 +879,7 @@ static int rockchip_spi_resume(struct device *dev)
 		clk_disable_unprepare(rs->apb_pclk);
 	}
 
-	return ret;
+	return 0;
 }
 #endif /* CONFIG_PM_SLEEP */
 
@@ -901,14 +902,14 @@ static int rockchip_spi_runtime_resume(struct device *dev)
 	struct rockchip_spi *rs = spi_master_get_devdata(master);
 
 	ret = clk_prepare_enable(rs->apb_pclk);
-	if (ret)
+	if (ret < 0)
 		return ret;
 
 	ret = clk_prepare_enable(rs->spiclk);
-	if (ret)
+	if (ret < 0)
 		clk_disable_unprepare(rs->apb_pclk);
 
-	return ret;
+	return 0;
 }
 #endif /* CONFIG_PM */
 
@@ -919,6 +920,7 @@ static const struct dev_pm_ops rockchip_spi_pm = {
 };
 
 static const struct of_device_id rockchip_spi_dt_match[] = {
+	{ .compatible = "rockchip,rv1108-spi", },
 	{ .compatible = "rockchip,rk3036-spi", },
 	{ .compatible = "rockchip,rk3066-spi", },
 	{ .compatible = "rockchip,rk3188-spi", },
diff --git a/drivers/spi/spi-sh-msiof.c b/drivers/spi/spi-sh-msiof.c
index c304c7167866d2dbc9fdfcff902729eefb2db78e..0eb1e95834854fb658806b1aaefc19fb7bec69b3 100644
--- a/drivers/spi/spi-sh-msiof.c
+++ b/drivers/spi/spi-sh-msiof.c
@@ -38,6 +38,7 @@ struct sh_msiof_chipdata {
 	u16 tx_fifo_size;
 	u16 rx_fifo_size;
 	u16 master_flags;
+	u16 min_div;
 };
 
 struct sh_msiof_spi_priv {
@@ -49,6 +50,7 @@ struct sh_msiof_spi_priv {
 	struct completion done;
 	unsigned int tx_fifo_size;
 	unsigned int rx_fifo_size;
+	unsigned int min_div;
 	void *tx_dma_page;
 	void *rx_dma_page;
 	dma_addr_t tx_dma_addr;
@@ -261,6 +263,8 @@ static void sh_msiof_spi_set_clk_regs(struct sh_msiof_spi_priv *p,
 	if (!WARN_ON(!spi_hz || !parent_rate))
 		div = DIV_ROUND_UP(parent_rate, spi_hz);
 
+	div = max_t(unsigned long, div, p->min_div);
+
 	for (k = 0; k < ARRAY_SIZE(sh_msiof_spi_div_table); k++) {
 		brps = DIV_ROUND_UP(div, sh_msiof_spi_div_table[k].div);
 		/* SCR_BRDV_DIV_1 is valid only if BRPS is x 1/1 or x 1/2 */
@@ -998,24 +1002,33 @@ static const struct sh_msiof_chipdata sh_data = {
 	.tx_fifo_size = 64,
 	.rx_fifo_size = 64,
 	.master_flags = 0,
+	.min_div = 1,
+};
+
+static const struct sh_msiof_chipdata rcar_gen2_data = {
+	.tx_fifo_size = 64,
+	.rx_fifo_size = 64,
+	.master_flags = SPI_MASTER_MUST_TX,
+	.min_div = 1,
 };
 
-static const struct sh_msiof_chipdata r8a779x_data = {
+static const struct sh_msiof_chipdata rcar_gen3_data = {
 	.tx_fifo_size = 64,
 	.rx_fifo_size = 64,
 	.master_flags = SPI_MASTER_MUST_TX,
+	.min_div = 2,
 };
 
 static const struct of_device_id sh_msiof_match[] = {
 	{ .compatible = "renesas,sh-mobile-msiof", .data = &sh_data },
-	{ .compatible = "renesas,msiof-r8a7790",   .data = &r8a779x_data },
-	{ .compatible = "renesas,msiof-r8a7791",   .data = &r8a779x_data },
-	{ .compatible = "renesas,msiof-r8a7792",   .data = &r8a779x_data },
-	{ .compatible = "renesas,msiof-r8a7793",   .data = &r8a779x_data },
-	{ .compatible = "renesas,msiof-r8a7794",   .data = &r8a779x_data },
-	{ .compatible = "renesas,rcar-gen2-msiof", .data = &r8a779x_data },
-	{ .compatible = "renesas,msiof-r8a7796",   .data = &r8a779x_data },
-	{ .compatible = "renesas,rcar-gen3-msiof", .data = &r8a779x_data },
+	{ .compatible = "renesas,msiof-r8a7790",   .data = &rcar_gen2_data },
+	{ .compatible = "renesas,msiof-r8a7791",   .data = &rcar_gen2_data },
+	{ .compatible = "renesas,msiof-r8a7792",   .data = &rcar_gen2_data },
+	{ .compatible = "renesas,msiof-r8a7793",   .data = &rcar_gen2_data },
+	{ .compatible = "renesas,msiof-r8a7794",   .data = &rcar_gen2_data },
+	{ .compatible = "renesas,rcar-gen2-msiof", .data = &rcar_gen2_data },
+	{ .compatible = "renesas,msiof-r8a7796",   .data = &rcar_gen3_data },
+	{ .compatible = "renesas,rcar-gen3-msiof", .data = &rcar_gen3_data },
 	{ .compatible = "renesas,sh-msiof",        .data = &sh_data }, /* Deprecated */
 	{},
 };
@@ -1230,6 +1243,7 @@ static int sh_msiof_spi_probe(struct platform_device *pdev)
 	platform_set_drvdata(pdev, p);
 	p->master = master;
 	p->info = info;
+	p->min_div = chipdata->min_div;
 
 	init_completion(&p->done);
 
diff --git a/drivers/spi/spi-sh.c b/drivers/spi/spi-sh.c
index 2bf53f0e27d9e0fe7bb179c2d3a143c012010a55..50e0ea9acf8b8771b568064d16e76c30ddb144bc 100644
--- a/drivers/spi/spi-sh.c
+++ b/drivers/spi/spi-sh.c
@@ -446,8 +446,8 @@ static int spi_sh_probe(struct platform_device *pdev)
 
 	irq = platform_get_irq(pdev, 0);
 	if (irq < 0) {
-		dev_err(&pdev->dev, "platform_get_irq error\n");
-		return -ENODEV;
+		dev_err(&pdev->dev, "platform_get_irq error: %d\n", irq);
+		return irq;
 	}
 
 	master = spi_alloc_master(&pdev->dev, sizeof(struct spi_sh_data));
diff --git a/drivers/spi/spi-stm32.c b/drivers/spi/spi-stm32.c
index 75644bcd938b6d5635505e4da2938d7cb689f90f..680cdf549506143d6041736f43552f868e3bf6db 100644
--- a/drivers/spi/spi-stm32.c
+++ b/drivers/spi/spi-stm32.c
@@ -1132,7 +1132,7 @@ static int stm32_spi_probe(struct platform_device *pdev)
 		goto err_master_put;
 	}
 
-	spi->rst = devm_reset_control_get(&pdev->dev, NULL);
+	spi->rst = devm_reset_control_get_exclusive(&pdev->dev, NULL);
 	if (!IS_ERR(spi->rst)) {
 		reset_control_assert(spi->rst);
 		udelay(2);
diff --git a/drivers/spi/spi-sun6i.c b/drivers/spi/spi-sun6i.c
index 03a773a9531a9d08944c4e5142f1668a3d3f2b6c..fb38234249a80055dc17212829ea2adafa3b43fe 100644
--- a/drivers/spi/spi-sun6i.c
+++ b/drivers/spi/spi-sun6i.c
@@ -502,7 +502,7 @@ static int sun6i_spi_probe(struct platform_device *pdev)
 
 	init_completion(&sspi->done);
 
-	sspi->rstc = devm_reset_control_get(&pdev->dev, NULL);
+	sspi->rstc = devm_reset_control_get_exclusive(&pdev->dev, NULL);
 	if (IS_ERR(sspi->rstc)) {
 		dev_err(&pdev->dev, "Couldn't get reset controller\n");
 		ret = PTR_ERR(sspi->rstc);
diff --git a/drivers/spi/spi-tegra114.c b/drivers/spi/spi-tegra114.c
index 08012ae5aa66eb3c03396523e5d25a3dc2101c64..44550182a4a364e53ba1b62da634ca6362d2df90 100644
--- a/drivers/spi/spi-tegra114.c
+++ b/drivers/spi/spi-tegra114.c
@@ -1083,7 +1083,7 @@ static int tegra_spi_probe(struct platform_device *pdev)
 		goto exit_free_irq;
 	}
 
-	tspi->rst = devm_reset_control_get(&pdev->dev, "spi");
+	tspi->rst = devm_reset_control_get_exclusive(&pdev->dev, "spi");
 	if (IS_ERR(tspi->rst)) {
 		dev_err(&pdev->dev, "can not get reset\n");
 		ret = PTR_ERR(tspi->rst);
diff --git a/drivers/spi/spi-tegra20-sflash.c b/drivers/spi/spi-tegra20-sflash.c
index 2c797ee2664dee454b308e23f4b46169e7b54cc0..22893a7e0aa0ea896dad668cfcfb24112bf1c6fb 100644
--- a/drivers/spi/spi-tegra20-sflash.c
+++ b/drivers/spi/spi-tegra20-sflash.c
@@ -485,7 +485,7 @@ static int tegra_sflash_probe(struct platform_device *pdev)
 		goto exit_free_irq;
 	}
 
-	tsd->rst = devm_reset_control_get(&pdev->dev, "spi");
+	tsd->rst = devm_reset_control_get_exclusive(&pdev->dev, "spi");
 	if (IS_ERR(tsd->rst)) {
 		dev_err(&pdev->dev, "can not get reset\n");
 		ret = PTR_ERR(tsd->rst);
diff --git a/drivers/spi/spi-tegra20-slink.c b/drivers/spi/spi-tegra20-slink.c
index 0c06ce424210a692f80305f2eadef1b2e6a179af..3e12d5f87ee4412b086649c8f2312fb3cc9e30a8 100644
--- a/drivers/spi/spi-tegra20-slink.c
+++ b/drivers/spi/spi-tegra20-slink.c
@@ -1081,7 +1081,7 @@ static int tegra_slink_probe(struct platform_device *pdev)
 		goto exit_free_irq;
 	}
 
-	tspi->rst = devm_reset_control_get(&pdev->dev, "spi");
+	tspi->rst = devm_reset_control_get_exclusive(&pdev->dev, "spi");
 	if (IS_ERR(tspi->rst)) {
 		dev_err(&pdev->dev, "can not get reset\n");
 		ret = PTR_ERR(tspi->rst);
diff --git a/drivers/spi/spi-xlp.c b/drivers/spi/spi-xlp.c
index 80cb4d6af8924032ed0a6dc5a25085f79325b7dc..74a01b09a8a5819c065f1066c84776cdf75dd7e0 100644
--- a/drivers/spi/spi-xlp.c
+++ b/drivers/spi/spi-xlp.c
@@ -393,8 +393,8 @@ static int xlp_spi_probe(struct platform_device *pdev)
 
 	irq = platform_get_irq(pdev, 0);
 	if (irq < 0) {
-		dev_err(&pdev->dev, "no IRQ resource found\n");
-		return -EINVAL;
+		dev_err(&pdev->dev, "no IRQ resource found: %d\n", irq);
+		return irq;
 	}
 	err = devm_request_irq(&pdev->dev, irq, xlp_spi_interrupt, 0,
 			pdev->name, xspi);
diff --git a/drivers/spi/spi.c b/drivers/spi/spi.c
index 4fcbb0aa71d361e4e5e50600a2d14a7722e22034..cf8ee5e48f73659cae5471157b288f9e88669f64 100644
--- a/drivers/spi/spi.c
+++ b/drivers/spi/spi.c
@@ -40,9 +40,13 @@
 #include <linux/ioport.h>
 #include <linux/acpi.h>
 #include <linux/highmem.h>
+#include <linux/idr.h>
 
 #define CREATE_TRACE_POINTS
 #include <trace/events/spi.h>
+#define SPI_DYN_FIRST_BUS_NUM 0
+
+static DEFINE_IDR(spi_master_idr);
 
 static void spidev_release(struct device *dev)
 {
@@ -321,8 +325,7 @@ static int spi_uevent(struct device *dev, struct kobj_uevent_env *env)
 	if (rc != -ENODEV)
 		return rc;
 
-	add_uevent_var(env, "MODALIAS=%s%s", SPI_MODULE_PREFIX, spi->modalias);
-	return 0;
+	return add_uevent_var(env, "MODALIAS=%s%s", SPI_MODULE_PREFIX, spi->modalias);
 }
 
 struct bus_type spi_bus_type = {
@@ -421,6 +424,7 @@ static LIST_HEAD(spi_controller_list);
 /*
  * Used to protect add/del opertion for board_info list and
  * spi_controller list, and their matching process
+ * also used to protect object of type struct idr
  */
 static DEFINE_MUTEX(board_lock);
 
@@ -1533,15 +1537,15 @@ static int of_spi_parse_dt(struct spi_controller *ctlr, struct spi_device *spi,
 	int rc;
 
 	/* Mode (clock phase/polarity/etc.) */
-	if (of_find_property(nc, "spi-cpha", NULL))
+	if (of_property_read_bool(nc, "spi-cpha"))
 		spi->mode |= SPI_CPHA;
-	if (of_find_property(nc, "spi-cpol", NULL))
+	if (of_property_read_bool(nc, "spi-cpol"))
 		spi->mode |= SPI_CPOL;
-	if (of_find_property(nc, "spi-cs-high", NULL))
+	if (of_property_read_bool(nc, "spi-cs-high"))
 		spi->mode |= SPI_CS_HIGH;
-	if (of_find_property(nc, "spi-3wire", NULL))
+	if (of_property_read_bool(nc, "spi-3wire"))
 		spi->mode |= SPI_3WIRE;
-	if (of_find_property(nc, "spi-lsb-first", NULL))
+	if (of_property_read_bool(nc, "spi-lsb-first"))
 		spi->mode |= SPI_LSB_FIRST;
 
 	/* Device DUAL/QUAD mode */
@@ -1583,8 +1587,8 @@ static int of_spi_parse_dt(struct spi_controller *ctlr, struct spi_device *spi,
 
 	if (spi_controller_is_slave(ctlr)) {
 		if (strcmp(nc->name, "slave")) {
-			dev_err(&ctlr->dev, "%s is not called 'slave'\n",
-				nc->full_name);
+			dev_err(&ctlr->dev, "%pOF is not called 'slave'\n",
+				nc);
 			return -EINVAL;
 		}
 		return 0;
@@ -1593,8 +1597,8 @@ static int of_spi_parse_dt(struct spi_controller *ctlr, struct spi_device *spi,
 	/* Device address */
 	rc = of_property_read_u32(nc, "reg", &value);
 	if (rc) {
-		dev_err(&ctlr->dev, "%s has no valid 'reg' property (%d)\n",
-			nc->full_name, rc);
+		dev_err(&ctlr->dev, "%pOF has no valid 'reg' property (%d)\n",
+			nc, rc);
 		return rc;
 	}
 	spi->chip_select = value;
@@ -1603,8 +1607,7 @@ static int of_spi_parse_dt(struct spi_controller *ctlr, struct spi_device *spi,
 	rc = of_property_read_u32(nc, "spi-max-frequency", &value);
 	if (rc) {
 		dev_err(&ctlr->dev,
-			"%s has no valid 'spi-max-frequency' property (%d)\n",
-			nc->full_name, rc);
+			"%pOF has no valid 'spi-max-frequency' property (%d)\n", nc, rc);
 		return rc;
 	}
 	spi->max_speed_hz = value;
@@ -1621,8 +1624,7 @@ of_register_spi_device(struct spi_controller *ctlr, struct device_node *nc)
 	/* Alloc an spi_device */
 	spi = spi_alloc_device(ctlr);
 	if (!spi) {
-		dev_err(&ctlr->dev, "spi_device alloc error for %s\n",
-			nc->full_name);
+		dev_err(&ctlr->dev, "spi_device alloc error for %pOF\n", nc);
 		rc = -ENOMEM;
 		goto err_out;
 	}
@@ -1631,8 +1633,7 @@ of_register_spi_device(struct spi_controller *ctlr, struct device_node *nc)
 	rc = of_modalias_node(nc, spi->modalias,
 				sizeof(spi->modalias));
 	if (rc < 0) {
-		dev_err(&ctlr->dev, "cannot find modalias for %s\n",
-			nc->full_name);
+		dev_err(&ctlr->dev, "cannot find modalias for %pOF\n", nc);
 		goto err_out;
 	}
 
@@ -1647,8 +1648,7 @@ of_register_spi_device(struct spi_controller *ctlr, struct device_node *nc)
 	/* Register the new device */
 	rc = spi_add_device(spi);
 	if (rc) {
-		dev_err(&ctlr->dev, "spi_device register error %s\n",
-			nc->full_name);
+		dev_err(&ctlr->dev, "spi_device register error %pOF\n", nc);
 		goto err_of_node_put;
 	}
 
@@ -1682,8 +1682,7 @@ static void of_register_spi_devices(struct spi_controller *ctlr)
 		spi = of_register_spi_device(ctlr, nc);
 		if (IS_ERR(spi)) {
 			dev_warn(&ctlr->dev,
-				 "Failed to create SPI device for %s\n",
-				 nc->full_name);
+				 "Failed to create SPI device for %pOF\n", nc);
 			of_node_clear_flag(nc, OF_POPULATED);
 		}
 	}
@@ -2052,11 +2051,10 @@ static int of_spi_register_master(struct spi_controller *ctlr)
  */
 int spi_register_controller(struct spi_controller *ctlr)
 {
-	static atomic_t		dyn_bus_id = ATOMIC_INIT((1<<15) - 1);
 	struct device		*dev = ctlr->dev.parent;
 	struct boardinfo	*bi;
 	int			status = -ENODEV;
-	int			dynamic = 0;
+	int			id;
 
 	if (!dev)
 		return -ENODEV;
@@ -2072,19 +2070,28 @@ int spi_register_controller(struct spi_controller *ctlr)
 	 */
 	if (ctlr->num_chipselect == 0)
 		return -EINVAL;
-
-	if ((ctlr->bus_num < 0) && ctlr->dev.of_node)
-		ctlr->bus_num = of_alias_get_id(ctlr->dev.of_node, "spi");
-
-	/* convention:  dynamically assigned bus IDs count down from the max */
+	/* allocate dynamic bus number using Linux idr */
+	if ((ctlr->bus_num < 0) && ctlr->dev.of_node) {
+		id = of_alias_get_id(ctlr->dev.of_node, "spi");
+		if (id >= 0) {
+			ctlr->bus_num = id;
+			mutex_lock(&board_lock);
+			id = idr_alloc(&spi_master_idr, ctlr, ctlr->bus_num,
+				       ctlr->bus_num + 1, GFP_KERNEL);
+			mutex_unlock(&board_lock);
+			if (WARN(id < 0, "couldn't get idr"))
+				return id == -ENOSPC ? -EBUSY : id;
+		}
+	}
 	if (ctlr->bus_num < 0) {
-		/* FIXME switch to an IDR based scheme, something like
-		 * I2C now uses, so we can't run out of "dynamic" IDs
-		 */
-		ctlr->bus_num = atomic_dec_return(&dyn_bus_id);
-		dynamic = 1;
+		mutex_lock(&board_lock);
+		id = idr_alloc(&spi_master_idr, ctlr, SPI_DYN_FIRST_BUS_NUM, 0,
+			       GFP_KERNEL);
+		mutex_unlock(&board_lock);
+		if (WARN(id < 0, "couldn't get idr"))
+			return id;
+		ctlr->bus_num = id;
 	}
-
 	INIT_LIST_HEAD(&ctlr->queue);
 	spin_lock_init(&ctlr->queue_lock);
 	spin_lock_init(&ctlr->bus_lock_spinlock);
@@ -2100,11 +2107,16 @@ int spi_register_controller(struct spi_controller *ctlr)
 	 */
 	dev_set_name(&ctlr->dev, "spi%u", ctlr->bus_num);
 	status = device_add(&ctlr->dev);
-	if (status < 0)
+	if (status < 0) {
+		/* free bus id */
+		mutex_lock(&board_lock);
+		idr_remove(&spi_master_idr, ctlr->bus_num);
+		mutex_unlock(&board_lock);
 		goto done;
-	dev_dbg(dev, "registered %s %s%s\n",
+	}
+	dev_dbg(dev, "registered %s %s\n",
 			spi_controller_is_slave(ctlr) ? "slave" : "master",
-			dev_name(&ctlr->dev), dynamic ? " (dynamic)" : "");
+			dev_name(&ctlr->dev));
 
 	/* If we're using a queued driver, start the queue */
 	if (ctlr->transfer)
@@ -2113,6 +2125,10 @@ int spi_register_controller(struct spi_controller *ctlr)
 		status = spi_controller_initialize_queue(ctlr);
 		if (status) {
 			device_del(&ctlr->dev);
+			/* free bus id */
+			mutex_lock(&board_lock);
+			idr_remove(&spi_master_idr, ctlr->bus_num);
+			mutex_unlock(&board_lock);
 			goto done;
 		}
 	}
@@ -2191,19 +2207,33 @@ static int __unregister(struct device *dev, void *null)
  */
 void spi_unregister_controller(struct spi_controller *ctlr)
 {
+	struct spi_controller *found;
 	int dummy;
 
+	/* First make sure that this controller was ever added */
+	mutex_lock(&board_lock);
+	found = idr_find(&spi_master_idr, ctlr->bus_num);
+	mutex_unlock(&board_lock);
+	if (found != ctlr) {
+		dev_dbg(&ctlr->dev,
+			"attempting to delete unregistered controller [%s]\n",
+			dev_name(&ctlr->dev));
+		return;
+	}
 	if (ctlr->queued) {
 		if (spi_destroy_queue(ctlr))
 			dev_err(&ctlr->dev, "queue remove failed\n");
 	}
-
 	mutex_lock(&board_lock);
 	list_del(&ctlr->list);
 	mutex_unlock(&board_lock);
 
 	dummy = device_for_each_child(&ctlr->dev, NULL, __unregister);
 	device_unregister(&ctlr->dev);
+	/* free bus id */
+	mutex_lock(&board_lock);
+	idr_remove(&spi_master_idr, ctlr->bus_num);
+	mutex_unlock(&board_lock);
 }
 EXPORT_SYMBOL_GPL(spi_unregister_controller);
 
@@ -3311,8 +3341,8 @@ static int of_spi_notify(struct notifier_block *nb, unsigned long action,
 		put_device(&ctlr->dev);
 
 		if (IS_ERR(spi)) {
-			pr_err("%s: failed to create for '%s'\n",
-					__func__, rd->dn->full_name);
+			pr_err("%s: failed to create for '%pOF'\n",
+					__func__, rd->dn);
 			of_node_clear_flag(rd->dn, OF_POPULATED);
 			return notifier_from_errno(PTR_ERR(spi));
 		}
diff --git a/tools/Makefile b/tools/Makefile
index 658f8adcf89668c87a76a69d00739ed7c80f67c2..9dfede37c8ff8a4729ba05c60c963d3561d4cfcd 100644
--- a/tools/Makefile
+++ b/tools/Makefile
@@ -90,7 +90,7 @@ kvm_stat: FORCE
 	$(call descend,kvm/$@)
 
 all: acpi cgroup cpupower gpio hv firewire liblockdep \
-		perf selftests turbostat usb \
+		perf selftests spi turbostat usb \
 		virtio vm net x86_energy_perf_policy \
 		tmon freefall iio objtool kvm_stat
 
@@ -100,7 +100,7 @@ acpi_install:
 cpupower_install:
 	$(call descend,power/$(@:_install=),install)
 
-cgroup_install firewire_install gpio_install hv_install iio_install perf_install usb_install virtio_install vm_install net_install objtool_install:
+cgroup_install firewire_install gpio_install hv_install iio_install perf_install spi_install usb_install virtio_install vm_install net_install objtool_install:
 	$(call descend,$(@:_install=),install)
 
 liblockdep_install:
diff --git a/tools/spi/Build b/tools/spi/Build
new file mode 100644
index 0000000000000000000000000000000000000000..8e8466033d99b4967fd6bbc0aa8dde7f8f7c0f41
--- /dev/null
+++ b/tools/spi/Build
@@ -0,0 +1,2 @@
+spidev_test-y += spidev_test.o
+spidev_fdx-y += spidev_fdx.o
diff --git a/tools/spi/Makefile b/tools/spi/Makefile
index 3815b18ba0703631de16203bdf3087a40e1a4757..90615e10c79af1c766f0d9a629bdac862f7cb507 100644
--- a/tools/spi/Makefile
+++ b/tools/spi/Makefile
@@ -1,6 +1,66 @@
+include ../scripts/Makefile.include
+
+bindir ?= /usr/bin
+
+ifeq ($(srctree),)
+srctree := $(patsubst %/,%,$(dir $(CURDIR)))
+srctree := $(patsubst %/,%,$(dir $(srctree)))
+endif
+
+# Do not use make's built-in rules
+# (this improves performance and avoids hard-to-debug behaviour);
+MAKEFLAGS += -r
+
 CC = $(CROSS_COMPILE)gcc
+LD = $(CROSS_COMPILE)ld
+CFLAGS += -O2 -Wall -g -D_GNU_SOURCE -I$(OUTPUT)include
+
+ALL_TARGETS := spidev_test spidev_fdx
+ALL_PROGRAMS := $(patsubst %,$(OUTPUT)%,$(ALL_TARGETS))
+
+all: $(ALL_PROGRAMS)
+
+export srctree OUTPUT CC LD CFLAGS
+include $(srctree)/tools/build/Makefile.include
 
-all: spidev_test spidev_fdx
+#
+# We need the following to be outside of kernel tree
+#
+$(OUTPUT)include/linux/spi/spidev.h: ../../include/uapi/linux/spi/spidev.h
+	mkdir -p $(OUTPUT)include/linux/spi 2>&1 || true
+	ln -sf $(CURDIR)/../../include/uapi/linux/spi/spidev.h $@
+
+prepare: $(OUTPUT)include/linux/spi/spidev.h
+
+#
+# spidev_test
+#
+SPIDEV_TEST_IN := $(OUTPUT)spidev_test-in.o
+$(SPIDEV_TEST_IN): prepare FORCE
+	$(Q)$(MAKE) $(build)=spidev_test
+$(OUTPUT)spidev_test: $(SPIDEV_TEST_IN)
+	$(QUIET_LINK)$(CC) $(CFLAGS) $(LDFLAGS) $< -o $@
+
+#
+# spidev_fdx
+#
+SPIDEV_FDX_IN := $(OUTPUT)spidev_fdx-in.o
+$(SPIDEV_FDX_IN): prepare FORCE
+	$(Q)$(MAKE) $(build)=spidev_fdx
+$(OUTPUT)spidev_fdx: $(SPIDEV_FDX_IN)
+	$(QUIET_LINK)$(CC) $(CFLAGS) $(LDFLAGS) $< -o $@
 
 clean:
-	$(RM) spidev_test spidev_fdx
+	rm -f $(ALL_PROGRAMS)
+	rm -f $(OUTPUT)include/linux/spi/spidev.h
+	find $(if $(OUTPUT),$(OUTPUT),.) -name '*.o' -delete -o -name '\.*.d' -delete
+
+install: $(ALL_PROGRAMS)
+	install -d -m 755 $(DESTDIR)$(bindir);		\
+	for program in $(ALL_PROGRAMS); do		\
+		install $$program $(DESTDIR)$(bindir);	\
+	done
+
+FORCE:
+
+.PHONY: all install clean FORCE prepare