Merge branch 'fix/fsl-dspi' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi into spi-fsl-dspi

drivers/spi/Kconfig

@@ -159,10 +159,9 @@ config SPI_BUTTERFLY
 
 config SPI_CADENCE
 	tristate "Cadence SPI controller"
-	depends on ARM
 	help
 	  This selects the Cadence SPI controller master driver
-	  used by Xilinx Zynq.
+	  used by Xilinx Zynq and ZynqMP.
 
 config SPI_CLPS711X
 	tristate "CLPS711X host SPI controller"
@@ -632,7 +631,7 @@ config SPI_DW_PCI
 
 config SPI_DW_MID_DMA
 	bool "DMA support for DW SPI controller on Intel MID platform"
-	depends on SPI_DW_PCI && INTEL_MID_DMAC
+	depends on SPI_DW_PCI && DW_DMAC_PCI
 
 config SPI_DW_MMIO
 	tristate "Memory-mapped io interface driver for DW SPI core"

drivers/spi/spi-atmel.c

@@ -180,11 +180,17 @@
 	  | SPI_BF(name, value))
 
 /* Register access macros */
+#ifdef CONFIG_AVR32
 #define spi_readl(port, reg) \
 	__raw_readl((port)->regs + SPI_##reg)
 #define spi_writel(port, reg, value) \
 	__raw_writel((value), (port)->regs + SPI_##reg)
-
+#else
+#define spi_readl(port, reg) \
+	readl_relaxed((port)->regs + SPI_##reg)
+#define spi_writel(port, reg, value) \
+	writel_relaxed((value), (port)->regs + SPI_##reg)
+#endif
 /* use PIO for small transfers, avoiding DMA setup/teardown overhead and
  * cache operations; better heuristics consider wordsize and bitrate.
  */
@@ -764,17 +770,17 @@ static void atmel_spi_pdc_next_xfer(struct spi_master *master,
 			(unsigned long long)xfer->rx_dma);
 	}
 
-	/* REVISIT: We're waiting for ENDRX before we start the next
+	/* REVISIT: We're waiting for RXBUFF before we start the next
 	 * transfer because we need to handle some difficult timing
-	 * issues otherwise. If we wait for ENDTX in one transfer and
-	 * then starts waiting for ENDRX in the next, it's difficult
-	 * to tell the difference between the ENDRX interrupt we're
-	 * actually waiting for and the ENDRX interrupt of the
+	 * issues otherwise. If we wait for TXBUFE in one transfer and
+	 * then starts waiting for RXBUFF in the next, it's difficult
+	 * to tell the difference between the RXBUFF interrupt we're
+	 * actually waiting for and the RXBUFF interrupt of the
 	 * previous transfer.
 	 *
 	 * It should be doable, though. Just not now...
 	 */
-	spi_writel(as, IER, SPI_BIT(ENDRX) | SPI_BIT(OVRES));
+	spi_writel(as, IER, SPI_BIT(RXBUFF) | SPI_BIT(OVRES));
 	spi_writel(as, PTCR, SPI_BIT(TXTEN) | SPI_BIT(RXTEN));
 }
 

drivers/spi/spi-bcm53xx.c

@@ -44,7 +44,7 @@ static int bcm53xxspi_wait(struct bcm53xxspi *b53spi, unsigned int timeout_ms)
 	u32 tmp;
 
 	/* SPE bit has to be 0 before we read MSPI STATUS */
-	deadline = jiffies + BCM53XXSPI_SPE_TIMEOUT_MS * HZ / 1000;
+	deadline = jiffies + msecs_to_jiffies(BCM53XXSPI_SPE_TIMEOUT_MS);
 	do {
 		tmp = bcm53xxspi_read(b53spi, B53SPI_MSPI_SPCR2);
 		if (!(tmp & B53SPI_MSPI_SPCR2_SPE))
@@ -56,7 +56,7 @@ static int bcm53xxspi_wait(struct bcm53xxspi *b53spi, unsigned int timeout_ms)
 		goto spi_timeout;
 
 	/* Check status */
-	deadline = jiffies + timeout_ms * HZ / 1000;
+	deadline = jiffies + msecs_to_jiffies(timeout_ms);
 	do {
 		tmp = bcm53xxspi_read(b53spi, B53SPI_MSPI_MSPI_STATUS);
 		if (tmp & B53SPI_MSPI_MSPI_STATUS_SPIF) {

drivers/spi/spi-bfin5xx.c

@@ -559,7 +559,7 @@ static void bfin_spi_pump_transfers(unsigned long data)
 	struct spi_transfer *previous = NULL;
 	struct bfin_spi_slave_data *chip = NULL;
 	unsigned int bits_per_word;
-	u16 cr, cr_width, dma_width, dma_config;
+	u16 cr, cr_width = 0, dma_width, dma_config;
 	u32 tranf_success = 1;
 	u8 full_duplex = 0;
 
@@ -648,7 +648,6 @@ static void bfin_spi_pump_transfers(unsigned long data)
 	} else if (bits_per_word == 8) {
 		drv_data->n_bytes = bits_per_word/8;
 		drv_data->len = transfer->len;
-		cr_width = 0;
 		drv_data->ops = &bfin_bfin_spi_transfer_ops_u8;
 	}
 	cr = bfin_read(&drv_data->regs->ctl) & ~(BIT_CTL_TIMOD | BIT_CTL_WORDSIZE);

drivers/spi/spi-bitbang-txrx.h

@@ -49,12 +49,17 @@ bitbang_txrx_be_cpha0(struct spi_device *spi,
 {
 	/* if (cpol == 0) this is SPI_MODE_0; else this is SPI_MODE_2 */
 
+	bool oldbit = !(word & 1);
 	/* clock starts at inactive polarity */
 	for (word <<= (32 - bits); likely(bits); bits--) {
 
 		/* setup MSB (to slave) on trailing edge */
-		if ((flags & SPI_MASTER_NO_TX) == 0)
-			setmosi(spi, word & (1 << 31));
+		if ((flags & SPI_MASTER_NO_TX) == 0) {
+			if ((word & (1 << 31)) != oldbit) {
+				setmosi(spi, word & (1 << 31));
+				oldbit = word & (1 << 31);
+			}
+		}
 		spidelay(nsecs);	/* T(setup) */
 
 		setsck(spi, !cpol);
@@ -76,13 +81,18 @@ bitbang_txrx_be_cpha1(struct spi_device *spi,
 {
 	/* if (cpol == 0) this is SPI_MODE_1; else this is SPI_MODE_3 */
 
+	bool oldbit = !(word & (1 << 31));
 	/* clock starts at inactive polarity */
 	for (word <<= (32 - bits); likely(bits); bits--) {
 
 		/* setup MSB (to slave) on leading edge */
 		setsck(spi, !cpol);
-		if ((flags & SPI_MASTER_NO_TX) == 0)
-			setmosi(spi, word & (1 << 31));
+		if ((flags & SPI_MASTER_NO_TX) == 0) {
+			if ((word & (1 << 31)) != oldbit) {
+				setmosi(spi, word & (1 << 31));
+				oldbit = word & (1 << 31);
+			}
+		}
 		spidelay(nsecs); /* T(setup) */
 
 		setsck(spi, cpol);

drivers/spi/spi-dw-mid.c

@@ -23,29 +23,31 @@
 #include "spi-dw.h"
 
 #ifdef CONFIG_SPI_DW_MID_DMA
-#include <linux/intel_mid_dma.h>
 #include <linux/pci.h>
+#include <linux/platform_data/dma-dw.h>
 
 #define RX_BUSY		0
 #define TX_BUSY		1
 
-struct mid_dma {
-	struct intel_mid_dma_slave	dmas_tx;
-	struct intel_mid_dma_slave	dmas_rx;
-};
+static struct dw_dma_slave mid_dma_tx = { .dst_id = 1 };
+static struct dw_dma_slave mid_dma_rx = { .src_id = 0 };
 
 static bool mid_spi_dma_chan_filter(struct dma_chan *chan, void *param)
 {
-	struct dw_spi *dws = param;
+	struct dw_dma_slave *s = param;
 
-	return dws->dma_dev == chan->device->dev;
+	if (s->dma_dev != chan->device->dev)
+		return false;
+
+	chan->private = s;
+	return true;
 }
 
 static int mid_spi_dma_init(struct dw_spi *dws)
 {
-	struct mid_dma *dw_dma = dws->dma_priv;
 	struct pci_dev *dma_dev;
-	struct intel_mid_dma_slave *rxs, *txs;
+	struct dw_dma_slave *tx = dws->dma_tx;
+	struct dw_dma_slave *rx = dws->dma_rx;
 	dma_cap_mask_t mask;
 
 	/*
@@ -56,28 +58,22 @@ static int mid_spi_dma_init(struct dw_spi *dws)
 	if (!dma_dev)
 		return -ENODEV;
 
-	dws->dma_dev = &dma_dev->dev;
-
 	dma_cap_zero(mask);
 	dma_cap_set(DMA_SLAVE, mask);
 
 	/* 1. Init rx channel */
-	dws->rxchan = dma_request_channel(mask, mid_spi_dma_chan_filter, dws);
+	rx->dma_dev = &dma_dev->dev;
+	dws->rxchan = dma_request_channel(mask, mid_spi_dma_chan_filter, rx);
 	if (!dws->rxchan)
 		goto err_exit;
-	rxs = &dw_dma->dmas_rx;
-	rxs->hs_mode = LNW_DMA_HW_HS;
-	rxs->cfg_mode = LNW_DMA_PER_TO_MEM;
-	dws->rxchan->private = rxs;
+	dws->master->dma_rx = dws->rxchan;
 
 	/* 2. Init tx channel */
-	dws->txchan = dma_request_channel(mask, mid_spi_dma_chan_filter, dws);
+	tx->dma_dev = &dma_dev->dev;
+	dws->txchan = dma_request_channel(mask, mid_spi_dma_chan_filter, tx);
 	if (!dws->txchan)
 		goto free_rxchan;
-	txs = &dw_dma->dmas_tx;
-	txs->hs_mode = LNW_DMA_HW_HS;
-	txs->cfg_mode = LNW_DMA_MEM_TO_PER;
-	dws->txchan->private = txs;
+	dws->master->dma_tx = dws->txchan;
 
 	dws->dma_inited = 1;
 	return 0;
@@ -100,6 +96,42 @@ static void mid_spi_dma_exit(struct dw_spi *dws)
 	dma_release_channel(dws->rxchan);
 }
 
+static irqreturn_t dma_transfer(struct dw_spi *dws)
+{
+	u16 irq_status = dw_readl(dws, DW_SPI_ISR);
+
+	if (!irq_status)
+		return IRQ_NONE;
+
+	dw_readl(dws, DW_SPI_ICR);
+	spi_reset_chip(dws);
+
+	dev_err(&dws->master->dev, "%s: FIFO overrun/underrun\n", __func__);
+	dws->master->cur_msg->status = -EIO;
+	spi_finalize_current_transfer(dws->master);
+	return IRQ_HANDLED;
+}
+
+static bool mid_spi_can_dma(struct spi_master *master, struct spi_device *spi,
+		struct spi_transfer *xfer)
+{
+	struct dw_spi *dws = spi_master_get_devdata(master);
+
+	if (!dws->dma_inited)
+		return false;
+
+	return xfer->len > dws->fifo_len;
+}
+
+static enum dma_slave_buswidth convert_dma_width(u32 dma_width) {
+	if (dma_width == 1)
+		return DMA_SLAVE_BUSWIDTH_1_BYTE;
+	else if (dma_width == 2)
+		return DMA_SLAVE_BUSWIDTH_2_BYTES;
+
+	return DMA_SLAVE_BUSWIDTH_UNDEFINED;
+}
+
 /*
  * dws->dma_chan_busy is set before the dma transfer starts, callback for tx
  * channel will clear a corresponding bit.
@@ -108,37 +140,38 @@ static void dw_spi_dma_tx_done(void *arg)
 {
 	struct dw_spi *dws = arg;
 
-	if (test_and_clear_bit(TX_BUSY, &dws->dma_chan_busy) & BIT(RX_BUSY))
+	clear_bit(TX_BUSY, &dws->dma_chan_busy);
+	if (test_bit(RX_BUSY, &dws->dma_chan_busy))
 		return;
-	dw_spi_xfer_done(dws);
+	spi_finalize_current_transfer(dws->master);
 }
 
-static struct dma_async_tx_descriptor *dw_spi_dma_prepare_tx(struct dw_spi *dws)
+static struct dma_async_tx_descriptor *dw_spi_dma_prepare_tx(struct dw_spi *dws,
+		struct spi_transfer *xfer)
 {
 	struct dma_slave_config txconf;
 	struct dma_async_tx_descriptor *txdesc;
 
-	if (!dws->tx_dma)
+	if (!xfer->tx_buf)
 		return NULL;
 
 	txconf.direction = DMA_MEM_TO_DEV;
 	txconf.dst_addr = dws->dma_addr;
-	txconf.dst_maxburst = LNW_DMA_MSIZE_16;
+	txconf.dst_maxburst = 16;
 	txconf.src_addr_width = DMA_SLAVE_BUSWIDTH_4_BYTES;
-	txconf.dst_addr_width = dws->dma_width;
+	txconf.dst_addr_width = convert_dma_width(dws->dma_width);
 	txconf.device_fc = false;
 
 	dmaengine_slave_config(dws->txchan, &txconf);
 
-	memset(&dws->tx_sgl, 0, sizeof(dws->tx_sgl));
-	dws->tx_sgl.dma_address = dws->tx_dma;
-	dws->tx_sgl.length = dws->len;
-
 	txdesc = dmaengine_prep_slave_sg(dws->txchan,
-				&dws->tx_sgl,
-				1,
+				xfer->tx_sg.sgl,
+				xfer->tx_sg.nents,
 				DMA_MEM_TO_DEV,
 				DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
+	if (!txdesc)
+		return NULL;
+
 	txdesc->callback = dw_spi_dma_tx_done;
 	txdesc->callback_param = dws;
 
@@ -153,74 +186,74 @@ static void dw_spi_dma_rx_done(void *arg)
 {
 	struct dw_spi *dws = arg;
 
-	if (test_and_clear_bit(RX_BUSY, &dws->dma_chan_busy) & BIT(TX_BUSY))
+	clear_bit(RX_BUSY, &dws->dma_chan_busy);
+	if (test_bit(TX_BUSY, &dws->dma_chan_busy))
 		return;
-	dw_spi_xfer_done(dws);
+	spi_finalize_current_transfer(dws->master);
 }
 
-static struct dma_async_tx_descriptor *dw_spi_dma_prepare_rx(struct dw_spi *dws)
+static struct dma_async_tx_descriptor *dw_spi_dma_prepare_rx(struct dw_spi *dws,
+		struct spi_transfer *xfer)
 {
 	struct dma_slave_config rxconf;
 	struct dma_async_tx_descriptor *rxdesc;
 
-	if (!dws->rx_dma)
+	if (!xfer->rx_buf)
 		return NULL;
 
 	rxconf.direction = DMA_DEV_TO_MEM;
 	rxconf.src_addr = dws->dma_addr;
-	rxconf.src_maxburst = LNW_DMA_MSIZE_16;
+	rxconf.src_maxburst = 16;
 	rxconf.dst_addr_width = DMA_SLAVE_BUSWIDTH_4_BYTES;
-	rxconf.src_addr_width = dws->dma_width;
+	rxconf.src_addr_width = convert_dma_width(dws->dma_width);
 	rxconf.device_fc = false;
 
 	dmaengine_slave_config(dws->rxchan, &rxconf);
 
-	memset(&dws->rx_sgl, 0, sizeof(dws->rx_sgl));
-	dws->rx_sgl.dma_address = dws->rx_dma;
-	dws->rx_sgl.length = dws->len;
-
 	rxdesc = dmaengine_prep_slave_sg(dws->rxchan,
-				&dws->rx_sgl,
-				1,
+				xfer->rx_sg.sgl,
+				xfer->rx_sg.nents,
 				DMA_DEV_TO_MEM,
 				DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
+	if (!rxdesc)
+		return NULL;
+
 	rxdesc->callback = dw_spi_dma_rx_done;
 	rxdesc->callback_param = dws;
 
 	return rxdesc;
 }
 
-static void dw_spi_dma_setup(struct dw_spi *dws)
+static int mid_spi_dma_setup(struct dw_spi *dws, struct spi_transfer *xfer)
 {
 	u16 dma_ctrl = 0;
 
-	spi_enable_chip(dws, 0);
+	dw_writel(dws, DW_SPI_DMARDLR, 0xf);
+	dw_writel(dws, DW_SPI_DMATDLR, 0x10);
 
-	dw_writew(dws, DW_SPI_DMARDLR, 0xf);
-	dw_writew(dws, DW_SPI_DMATDLR, 0x10);
-
-	if (dws->tx_dma)
+	if (xfer->tx_buf)
 		dma_ctrl |= SPI_DMA_TDMAE;
-	if (dws->rx_dma)
+	if (xfer->rx_buf)
 		dma_ctrl |= SPI_DMA_RDMAE;
-	dw_writew(dws, DW_SPI_DMACR, dma_ctrl);
+	dw_writel(dws, DW_SPI_DMACR, dma_ctrl);
 
-	spi_enable_chip(dws, 1);
+	/* Set the interrupt mask */
+	spi_umask_intr(dws, SPI_INT_TXOI | SPI_INT_RXUI | SPI_INT_RXOI);
+
+	dws->transfer_handler = dma_transfer;
+
+	return 0;
 }
 
-static int mid_spi_dma_transfer(struct dw_spi *dws, int cs_change)
+static int mid_spi_dma_transfer(struct dw_spi *dws, struct spi_transfer *xfer)
 {
 	struct dma_async_tx_descriptor *txdesc, *rxdesc;
 
-	/* 1. setup DMA related registers */
-	if (cs_change)
-		dw_spi_dma_setup(dws);
+	/* Prepare the TX dma transfer */
+	txdesc = dw_spi_dma_prepare_tx(dws, xfer);
 
-	/* 2. Prepare the TX dma transfer */
-	txdesc = dw_spi_dma_prepare_tx(dws);
-
-	/* 3. Prepare the RX dma transfer */
-	rxdesc = dw_spi_dma_prepare_rx(dws);
+	/* Prepare the RX dma transfer */
+	rxdesc = dw_spi_dma_prepare_rx(dws, xfer);
 
 	/* rx must be started before tx due to spi instinct */
 	if (rxdesc) {
@@ -238,10 +271,25 @@ static int mid_spi_dma_transfer(struct dw_spi *dws, int cs_change)
 	return 0;
 }
 
+static void mid_spi_dma_stop(struct dw_spi *dws)
+{
+	if (test_bit(TX_BUSY, &dws->dma_chan_busy)) {
+		dmaengine_terminate_all(dws->txchan);
+		clear_bit(TX_BUSY, &dws->dma_chan_busy);
+	}
+	if (test_bit(RX_BUSY, &dws->dma_chan_busy)) {
+		dmaengine_terminate_all(dws->rxchan);
+		clear_bit(RX_BUSY, &dws->dma_chan_busy);
+	}
+}
+
 static struct dw_spi_dma_ops mid_dma_ops = {
 	.dma_init	= mid_spi_dma_init,
 	.dma_exit	= mid_spi_dma_exit,
+	.dma_setup	= mid_spi_dma_setup,
+	.can_dma	= mid_spi_can_dma,
 	.dma_transfer	= mid_spi_dma_transfer,
+	.dma_stop	= mid_spi_dma_stop,
 };
 #endif
 
@@ -274,9 +322,8 @@ int dw_spi_mid_init(struct dw_spi *dws)
 	iounmap(clk_reg);
 
 #ifdef CONFIG_SPI_DW_MID_DMA
-	dws->dma_priv = kzalloc(sizeof(struct mid_dma), GFP_KERNEL);
-	if (!dws->dma_priv)
-		return -ENOMEM;
+	dws->dma_tx = &mid_dma_tx;
+	dws->dma_rx = &mid_dma_rx;
 	dws->dma_ops = &mid_dma_ops;
 #endif
 	return 0;

drivers/spi/spi-dw-pci.c

@@ -36,13 +36,13 @@ struct spi_pci_desc {
 
 static struct spi_pci_desc spi_pci_mid_desc_1 = {
 	.setup = dw_spi_mid_init,
-	.num_cs = 32,
+	.num_cs = 5,
 	.bus_num = 0,
 };
 
 static struct spi_pci_desc spi_pci_mid_desc_2 = {
 	.setup = dw_spi_mid_init,
-	.num_cs = 4,
+	.num_cs = 2,
 	.bus_num = 1,
 };
 

drivers/spi/spi-dw.h

@@ -91,12 +91,15 @@ struct dw_spi;
 struct dw_spi_dma_ops {
 	int (*dma_init)(struct dw_spi *dws);
 	void (*dma_exit)(struct dw_spi *dws);
-	int (*dma_transfer)(struct dw_spi *dws, int cs_change);
+	int (*dma_setup)(struct dw_spi *dws, struct spi_transfer *xfer);
+	bool (*can_dma)(struct spi_master *master, struct spi_device *spi,
+			struct spi_transfer *xfer);
+	int (*dma_transfer)(struct dw_spi *dws, struct spi_transfer *xfer);
+	void (*dma_stop)(struct dw_spi *dws);
 };
 
 struct dw_spi {
 	struct spi_master	*master;
-	struct spi_device	*cur_dev;
 	enum dw_ssi_type	type;
 	char			name[16];
 
@@ -109,41 +112,26 @@ struct dw_spi {
 	u16			bus_num;
 	u16			num_cs;		/* supported slave numbers */
 
-	/* Message Transfer pump */
-	struct tasklet_struct	pump_transfers;
-
 	/* Current message transfer state info */
-	struct spi_message	*cur_msg;
-	struct spi_transfer	*cur_transfer;
-	struct chip_data	*cur_chip;
-	struct chip_data	*prev_chip;
 	size_t			len;
 	void			*tx;
 	void			*tx_end;
 	void			*rx;
 	void			*rx_end;
 	int			dma_mapped;
-	dma_addr_t		rx_dma;
-	dma_addr_t		tx_dma;
-	size_t			rx_map_len;
-	size_t			tx_map_len;
 	u8			n_bytes;	/* current is a 1/2 bytes op */
-	u8			max_bits_per_word;	/* maxim is 16b */
 	u32			dma_width;
 	irqreturn_t		(*transfer_handler)(struct dw_spi *dws);
-	void			(*cs_control)(u32 command);
 
-	/* Dma info */
+	/* DMA info */
 	int			dma_inited;
 	struct dma_chan		*txchan;
-	struct scatterlist	tx_sgl;
 	struct dma_chan		*rxchan;
-	struct scatterlist	rx_sgl;
 	unsigned long		dma_chan_busy;
-	struct device		*dma_dev;
 	dma_addr_t		dma_addr; /* phy address of the Data register */
 	struct dw_spi_dma_ops	*dma_ops;
-	void			*dma_priv; /* platform relate info */
+	void			*dma_tx;
+	void			*dma_rx;
 
 	/* Bus interface info */
 	void			*priv;
@@ -162,16 +150,6 @@ static inline void dw_writel(struct dw_spi *dws, u32 offset, u32 val)
 	__raw_writel(val, dws->regs + offset);
 }
 
-static inline u16 dw_readw(struct dw_spi *dws, u32 offset)
-{
-	return __raw_readw(dws->regs + offset);
-}
-
-static inline void dw_writew(struct dw_spi *dws, u32 offset, u16 val)
-{
-	__raw_writew(val, dws->regs + offset);
-}
-
 static inline void spi_enable_chip(struct dw_spi *dws, int enable)
 {
 	dw_writel(dws, DW_SPI_SSIENR, (enable ? 1 : 0));
@@ -182,22 +160,6 @@ static inline void spi_set_clk(struct dw_spi *dws, u16 div)
 	dw_writel(dws, DW_SPI_BAUDR, div);
 }
 
-static inline void spi_chip_sel(struct dw_spi *dws, struct spi_device *spi,
-		int active)
-{
-	u16 cs = spi->chip_select;
-	int gpio_val = active ? (spi->mode & SPI_CS_HIGH) :
-		!(spi->mode & SPI_CS_HIGH);
-
-	if (dws->cs_control)
-		dws->cs_control(active);
-	if (gpio_is_valid(spi->cs_gpio))
-		gpio_set_value(spi->cs_gpio, gpio_val);
-
-	if (active)
-		dw_writel(dws, DW_SPI_SER, 1 << cs);
-}
-
 /* Disable IRQ bits */
 static inline void spi_mask_intr(struct dw_spi *dws, u32 mask)
 {
@@ -216,16 +178,27 @@ static inline void spi_umask_intr(struct dw_spi *dws, u32 mask)
 	dw_writel(dws, DW_SPI_IMR, new_mask);
 }
 
+/*
+ * This does disable the SPI controller, interrupts, and re-enable the
+ * controller back. Transmit and receive FIFO buffers are cleared when the
+ * device is disabled.
+ */
+static inline void spi_reset_chip(struct dw_spi *dws)
+{
+	spi_enable_chip(dws, 0);
+	spi_mask_intr(dws, 0xff);
+	spi_enable_chip(dws, 1);
+}
+
 /*
  * Each SPI slave device to work with dw_api controller should
- * has such a structure claiming its working mode (PIO/DMA etc),
+ * has such a structure claiming its working mode (poll or PIO/DMA),
  * which can be save in the "controller_data" member of the
  * struct spi_device.
  */
 struct dw_spi_chip {
 	u8 poll_mode;	/* 1 for controller polling mode */
 	u8 type;	/* SPI/SSP/MicroWire */
-	u8 enable_dma;
 	void (*cs_control)(u32 command);
 };
 
@@ -233,7 +206,6 @@ extern int dw_spi_add_host(struct device *dev, struct dw_spi *dws);
 extern void dw_spi_remove_host(struct dw_spi *dws);
 extern int dw_spi_suspend_host(struct dw_spi *dws);
 extern int dw_spi_resume_host(struct dw_spi *dws);
-extern void dw_spi_xfer_done(struct dw_spi *dws);
 
 /* platform related setup */
 extern int dw_spi_mid_init(struct dw_spi *dws); /* Intel MID platforms */

drivers/spi/spi-fsl-dspi.c

@@ -20,6 +20,7 @@
 #include <linux/interrupt.h>
 #include <linux/io.h>
 #include <linux/kernel.h>
+#include <linux/math64.h>
 #include <linux/module.h>
 #include <linux/of.h>
 #include <linux/of_device.h>
@@ -29,6 +30,7 @@
 #include <linux/sched.h>
 #include <linux/spi/spi.h>
 #include <linux/spi/spi_bitbang.h>
+#include <linux/time.h>
 
 #define DRIVER_NAME "fsl-dspi"
 
@@ -51,7 +53,7 @@
 #define SPI_CTAR_CPOL(x)	((x) << 26)
 #define SPI_CTAR_CPHA(x)	((x) << 25)
 #define SPI_CTAR_LSBFE(x)	((x) << 24)
-#define SPI_CTAR_PCSSCR(x)	(((x) & 0x00000003) << 22)
+#define SPI_CTAR_PCSSCK(x)	(((x) & 0x00000003) << 22)
 #define SPI_CTAR_PASC(x)	(((x) & 0x00000003) << 20)
 #define SPI_CTAR_PDT(x)	(((x) & 0x00000003) << 18)
 #define SPI_CTAR_PBR(x)	(((x) & 0x00000003) << 16)
@@ -59,6 +61,7 @@
 #define SPI_CTAR_ASC(x)	(((x) & 0x0000000f) << 8)
 #define SPI_CTAR_DT(x)		(((x) & 0x0000000f) << 4)
 #define SPI_CTAR_BR(x)		((x) & 0x0000000f)
+#define SPI_CTAR_SCALE_BITS	0xf
 
 #define SPI_CTAR0_SLAVE	0x0c
 
@@ -148,23 +151,66 @@ static void hz_to_spi_baud(char *pbr, char *br, int speed_hz,
 		16,	32,	64,	128,
 		256,	512,	1024,	2048,
 		4096,	8192,	16384,	32768 };
-	int temp, i = 0, j = 0;
+	int scale_needed, scale, minscale = INT_MAX;
+	int i, j;
 
-	temp = clkrate / 2 / speed_hz;
+	scale_needed = clkrate / speed_hz;
+	if (clkrate % speed_hz)
+		scale_needed++;
 
-	for (i = 0; i < ARRAY_SIZE(pbr_tbl); i++)
-		for (j = 0; j < ARRAY_SIZE(brs); j++) {
-			if (pbr_tbl[i] * brs[j] >= temp) {
-				*pbr = i;
-				*br = j;
-				return;
+	for (i = 0; i < ARRAY_SIZE(brs); i++)
+		for (j = 0; j < ARRAY_SIZE(pbr_tbl); j++) {
+			scale = brs[i] * pbr_tbl[j];
+			if (scale >= scale_needed) {
+				if (scale < minscale) {
+					minscale = scale;
+					*br = i;
+					*pbr = j;
+				}
+				break;
 			}
 		}
 
-	pr_warn("Can not find valid baud rate,speed_hz is %d,clkrate is %ld\
-		,we use the max prescaler value.\n", speed_hz, clkrate);
-	*pbr = ARRAY_SIZE(pbr_tbl) - 1;
-	*br =  ARRAY_SIZE(brs) - 1;
+	if (minscale == INT_MAX) {
+		pr_warn("Can not find valid baud rate,speed_hz is %d,clkrate is %ld, we use the max prescaler value.\n",
+			speed_hz, clkrate);
+		*pbr = ARRAY_SIZE(pbr_tbl) - 1;
+		*br =  ARRAY_SIZE(brs) - 1;
+	}
+}
+
+static void ns_delay_scale(char *psc, char *sc, int delay_ns,
+		unsigned long clkrate)
+{
+	int pscale_tbl[4] = {1, 3, 5, 7};
+	int scale_needed, scale, minscale = INT_MAX;
+	int i, j;
+	u32 remainder;
+
+	scale_needed = div_u64_rem((u64)delay_ns * clkrate, NSEC_PER_SEC,
+			&remainder);
+	if (remainder)
+		scale_needed++;
+
+	for (i = 0; i < ARRAY_SIZE(pscale_tbl); i++)
+		for (j = 0; j <= SPI_CTAR_SCALE_BITS; j++) {
+			scale = pscale_tbl[i] * (2 << j);
+			if (scale >= scale_needed) {
+				if (scale < minscale) {
+					minscale = scale;
+					*psc = i;
+					*sc = j;
+				}
+				break;
+			}
+		}
+
+	if (minscale == INT_MAX) {
+		pr_warn("Cannot find correct scale values for %dns delay at clkrate %ld, using max prescaler value",
+			delay_ns, clkrate);
+		*psc = ARRAY_SIZE(pscale_tbl) - 1;
+		*sc = SPI_CTAR_SCALE_BITS;
+	}
 }
 
 static int dspi_transfer_write(struct fsl_dspi *dspi)
@@ -295,10 +341,10 @@ static int dspi_transfer_one_message(struct spi_master *master,
 		dspi->cur_msg = message;
 		dspi->cur_chip = spi_get_ctldata(spi);
 		dspi->cs = spi->chip_select;
+		dspi->cs_change = 0;
 		if (dspi->cur_transfer->transfer_list.next
 				== &dspi->cur_msg->transfers)
-			transfer->cs_change = 1;
-		dspi->cs_change = transfer->cs_change;
+			dspi->cs_change = 1;
 		dspi->void_write_data = dspi->cur_chip->void_write_data;
 
 		dspi->dataflags = 0;
@@ -345,7 +391,10 @@ static int dspi_setup(struct spi_device *spi)
 {
 	struct chip_data *chip;
 	struct fsl_dspi *dspi = spi_master_get_devdata(spi->master);
-	unsigned char br = 0, pbr = 0, fmsz = 0;
+	u32 cs_sck_delay = 0, sck_cs_delay = 0;
+	unsigned char br = 0, pbr = 0, pcssck = 0, cssck = 0;
+	unsigned char pasc = 0, asc = 0, fmsz = 0;
+	unsigned long clkrate;
 
 	if ((spi->bits_per_word >= 4) && (spi->bits_per_word <= 16)) {
 		fmsz = spi->bits_per_word - 1;
@@ -362,18 +411,34 @@ static int dspi_setup(struct spi_device *spi)
 			return -ENOMEM;
 	}
 
+	of_property_read_u32(spi->dev.of_node, "fsl,spi-cs-sck-delay",
+			&cs_sck_delay);
+
+	of_property_read_u32(spi->dev.of_node, "fsl,spi-sck-cs-delay",
+			&sck_cs_delay);
+
 	chip->mcr_val = SPI_MCR_MASTER | SPI_MCR_PCSIS |
 		SPI_MCR_CLR_TXF | SPI_MCR_CLR_RXF;
 
 	chip->void_write_data = 0;
 
-	hz_to_spi_baud(&pbr, &br,
-			spi->max_speed_hz, clk_get_rate(dspi->clk));
+	clkrate = clk_get_rate(dspi->clk);
+	hz_to_spi_baud(&pbr, &br, spi->max_speed_hz, clkrate);
+
+	/* Set PCS to SCK delay scale values */
+	ns_delay_scale(&pcssck, &cssck, cs_sck_delay, clkrate);
+
+	/* Set After SCK delay scale values */
+	ns_delay_scale(&pasc, &asc, sck_cs_delay, clkrate);
 
 	chip->ctar_val =  SPI_CTAR_FMSZ(fmsz)
 		| SPI_CTAR_CPOL(spi->mode & SPI_CPOL ? 1 : 0)
 		| SPI_CTAR_CPHA(spi->mode & SPI_CPHA ? 1 : 0)
 		| SPI_CTAR_LSBFE(spi->mode & SPI_LSB_FIRST ? 1 : 0)
+		| SPI_CTAR_PCSSCK(pcssck)
+		| SPI_CTAR_CSSCK(cssck)
+		| SPI_CTAR_PASC(pasc)
+		| SPI_CTAR_ASC(asc)
 		| SPI_CTAR_PBR(pbr)
 		| SPI_CTAR_BR(br);
 

drivers/spi/spi-img-spfi.c

@@ -12,6 +12,7 @@
 #include <linux/clk.h>
 #include <linux/delay.h>
 #include <linux/dmaengine.h>
+#include <linux/gpio.h>
 #include <linux/interrupt.h>
 #include <linux/io.h>
 #include <linux/irq.h>
@@ -122,36 +123,31 @@ static inline void spfi_start(struct img_spfi *spfi)
 	spfi_writel(spfi, val, SPFI_CONTROL);
 }
 
-static inline void spfi_stop(struct img_spfi *spfi)
-{
-	u32 val;
-
-	val = spfi_readl(spfi, SPFI_CONTROL);
-	val &= ~SPFI_CONTROL_SPFI_EN;
-	spfi_writel(spfi, val, SPFI_CONTROL);
-}
-
 static inline void spfi_reset(struct img_spfi *spfi)
 {
 	spfi_writel(spfi, SPFI_CONTROL_SOFT_RESET, SPFI_CONTROL);
-	udelay(1);
 	spfi_writel(spfi, 0, SPFI_CONTROL);
 }
 
-static void spfi_flush_tx_fifo(struct img_spfi *spfi)
+static int spfi_wait_all_done(struct img_spfi *spfi)
 {
-	unsigned long timeout = jiffies + msecs_to_jiffies(10);
+	unsigned long timeout = jiffies + msecs_to_jiffies(50);
 
-	spfi_writel(spfi, SPFI_INTERRUPT_SDE, SPFI_INTERRUPT_CLEAR);
 	while (time_before(jiffies, timeout)) {
-		if (spfi_readl(spfi, SPFI_INTERRUPT_STATUS) &
-		    SPFI_INTERRUPT_SDE)
-			return;
+		u32 status = spfi_readl(spfi, SPFI_INTERRUPT_STATUS);
+
+		if (status & SPFI_INTERRUPT_ALLDONETRIG) {
+			spfi_writel(spfi, SPFI_INTERRUPT_ALLDONETRIG,
+				    SPFI_INTERRUPT_CLEAR);
+			return 0;
+		}
 		cpu_relax();
 	}
 
-	dev_err(spfi->dev, "Timed out waiting for FIFO to drain\n");
+	dev_err(spfi->dev, "Timed out waiting for transaction to complete\n");
 	spfi_reset(spfi);
+
+	return -ETIMEDOUT;
 }
 
 static unsigned int spfi_pio_write32(struct img_spfi *spfi, const u32 *buf,
@@ -237,6 +233,7 @@ static int img_spfi_start_pio(struct spi_master *master,
 	const void *tx_buf = xfer->tx_buf;
 	void *rx_buf = xfer->rx_buf;
 	unsigned long timeout;
+	int ret;
 
 	if (tx_buf)
 		tx_bytes = xfer->len;
@@ -269,16 +266,15 @@ static int img_spfi_start_pio(struct spi_master *master,
 		cpu_relax();
 	}
 
+	ret = spfi_wait_all_done(spfi);
+	if (ret < 0)
+		return ret;
+
 	if (rx_bytes > 0 || tx_bytes > 0) {
 		dev_err(spfi->dev, "PIO transfer timed out\n");
-		spfi_reset(spfi);
 		return -ETIMEDOUT;
 	}
 
-	if (tx_buf)
-		spfi_flush_tx_fifo(spfi);
-	spfi_stop(spfi);
-
 	return 0;
 }
 
@@ -287,14 +283,12 @@ static void img_spfi_dma_rx_cb(void *data)
 	struct img_spfi *spfi = data;
 	unsigned long flags;
 
+	spfi_wait_all_done(spfi);
+
 	spin_lock_irqsave(&spfi->lock, flags);
-
 	spfi->rx_dma_busy = false;
-	if (!spfi->tx_dma_busy) {
-		spfi_stop(spfi);
+	if (!spfi->tx_dma_busy)
 		spi_finalize_current_transfer(spfi->master);
-	}
-
 	spin_unlock_irqrestore(&spfi->lock, flags);
 }
 
@@ -303,16 +297,12 @@ static void img_spfi_dma_tx_cb(void *data)
 	struct img_spfi *spfi = data;
 	unsigned long flags;
 
-	spfi_flush_tx_fifo(spfi);
+	spfi_wait_all_done(spfi);
 
 	spin_lock_irqsave(&spfi->lock, flags);
-
 	spfi->tx_dma_busy = false;
-	if (!spfi->rx_dma_busy) {
-		spfi_stop(spfi);
+	if (!spfi->rx_dma_busy)
 		spi_finalize_current_transfer(spfi->master);
-	}
-
 	spin_unlock_irqrestore(&spfi->lock, flags);
 }
 
@@ -397,6 +387,75 @@ stop_dma:
 	return -EIO;
 }
 
+static void img_spfi_handle_err(struct spi_master *master,
+				struct spi_message *msg)
+{
+	struct img_spfi *spfi = spi_master_get_devdata(master);
+	unsigned long flags;
+
+	/*
+	 * Stop all DMA and reset the controller if the previous transaction
+	 * timed-out and never completed it's DMA.
+	 */
+	spin_lock_irqsave(&spfi->lock, flags);
+	if (spfi->tx_dma_busy || spfi->rx_dma_busy) {
+		spfi->tx_dma_busy = false;
+		spfi->rx_dma_busy = false;
+
+		dmaengine_terminate_all(spfi->tx_ch);
+		dmaengine_terminate_all(spfi->rx_ch);
+	}
+	spin_unlock_irqrestore(&spfi->lock, flags);
+}
+
+static int img_spfi_prepare(struct spi_master *master, struct spi_message *msg)
+{
+	struct img_spfi *spfi = spi_master_get_devdata(master);
+	u32 val;
+
+	val = spfi_readl(spfi, SPFI_PORT_STATE);
+	if (msg->spi->mode & SPI_CPHA)
+		val |= SPFI_PORT_STATE_CK_PHASE(msg->spi->chip_select);
+	else
+		val &= ~SPFI_PORT_STATE_CK_PHASE(msg->spi->chip_select);
+	if (msg->spi->mode & SPI_CPOL)
+		val |= SPFI_PORT_STATE_CK_POL(msg->spi->chip_select);
+	else
+		val &= ~SPFI_PORT_STATE_CK_POL(msg->spi->chip_select);
+	spfi_writel(spfi, val, SPFI_PORT_STATE);
+
+	return 0;
+}
+
+static int img_spfi_unprepare(struct spi_master *master,
+			      struct spi_message *msg)
+{
+	struct img_spfi *spfi = spi_master_get_devdata(master);
+
+	spfi_reset(spfi);
+
+	return 0;
+}
+
+static int img_spfi_setup(struct spi_device *spi)
+{
+	int ret;
+
+	ret = gpio_request_one(spi->cs_gpio, (spi->mode & SPI_CS_HIGH) ?
+			       GPIOF_OUT_INIT_LOW : GPIOF_OUT_INIT_HIGH,
+			       dev_name(&spi->dev));
+	if (ret)
+		dev_err(&spi->dev, "can't request chipselect gpio %d\n",
+				spi->cs_gpio);
+
+	return ret;
+}
+
+static void img_spfi_cleanup(struct spi_device *spi)
+{
+	gpio_free(spi->cs_gpio);
+}
+
 static void img_spfi_config(struct spi_master *master, struct spi_device *spi,
 			    struct spi_transfer *xfer)
 {
@@ -405,10 +464,10 @@ static void img_spfi_config(struct spi_master *master, struct spi_device *spi,
 
 	/*
 	 * output = spfi_clk * (BITCLK / 512), where BITCLK must be a
-	 * power of 2 up to 256 (where 255 == 256 since BITCLK is 8 bits)
+	 * power of 2 up to 128
 	 */
-	div = DIV_ROUND_UP(master->max_speed_hz, xfer->speed_hz);
-	div = clamp(512 / (1 << get_count_order(div)), 1, 255);
+	div = DIV_ROUND_UP(clk_get_rate(spfi->spfi_clk), xfer->speed_hz);
+	div = clamp(512 / (1 << get_count_order(div)), 1, 128);
 
 	val = spfi_readl(spfi, SPFI_DEVICE_PARAMETER(spi->chip_select));
 	val &= ~(SPFI_DEVICE_PARAMETER_BITCLK_MASK <<
@@ -416,6 +475,9 @@ static void img_spfi_config(struct spi_master *master, struct spi_device *spi,
 	val |= div << SPFI_DEVICE_PARAMETER_BITCLK_SHIFT;
 	spfi_writel(spfi, val, SPFI_DEVICE_PARAMETER(spi->chip_select));
 
+	spfi_writel(spfi, xfer->len << SPFI_TRANSACTION_TSIZE_SHIFT,
+		    SPFI_TRANSACTION);
+
 	val = spfi_readl(spfi, SPFI_CONTROL);
 	val &= ~(SPFI_CONTROL_SEND_DMA | SPFI_CONTROL_GET_DMA);
 	if (xfer->tx_buf)
@@ -429,25 +491,7 @@ static void img_spfi_config(struct spi_master *master, struct spi_device *spi,
 	else if (xfer->tx_nbits == SPI_NBITS_QUAD &&
 		 xfer->rx_nbits == SPI_NBITS_QUAD)
 		val |= SPFI_CONTROL_TMODE_QUAD << SPFI_CONTROL_TMODE_SHIFT;
-	val &= ~SPFI_CONTROL_CONTINUE;
-	if (!xfer->cs_change && !list_is_last(&xfer->transfer_list,
-					      &master->cur_msg->transfers))
-		val |= SPFI_CONTROL_CONTINUE;
 	spfi_writel(spfi, val, SPFI_CONTROL);
-
-	val = spfi_readl(spfi, SPFI_PORT_STATE);
-	if (spi->mode & SPI_CPHA)
-		val |= SPFI_PORT_STATE_CK_PHASE(spi->chip_select);
-	else
-		val &= ~SPFI_PORT_STATE_CK_PHASE(spi->chip_select);
-	if (spi->mode & SPI_CPOL)
-		val |= SPFI_PORT_STATE_CK_POL(spi->chip_select);
-	else
-		val &= ~SPFI_PORT_STATE_CK_POL(spi->chip_select);
-	spfi_writel(spfi, val, SPFI_PORT_STATE);
-
-	spfi_writel(spfi, xfer->len << SPFI_TRANSACTION_TSIZE_SHIFT,
-		    SPFI_TRANSACTION);
 }
 
 static int img_spfi_transfer_one(struct spi_master *master,
@@ -455,25 +499,13 @@ static int img_spfi_transfer_one(struct spi_master *master,
 				 struct spi_transfer *xfer)
 {
 	struct img_spfi *spfi = spi_master_get_devdata(spi->master);
-	bool dma_reset = false;
-	unsigned long flags;
 	int ret;
 
-	/*
-	 * Stop all DMA and reset the controller if the previous transaction
-	 * timed-out and never completed it's DMA.
-	 */
-	spin_lock_irqsave(&spfi->lock, flags);
-	if (spfi->tx_dma_busy || spfi->rx_dma_busy) {
-		dev_err(spfi->dev, "SPI DMA still busy\n");
-		dma_reset = true;
-	}
-	spin_unlock_irqrestore(&spfi->lock, flags);
-
-	if (dma_reset) {
-		dmaengine_terminate_all(spfi->tx_ch);
-		dmaengine_terminate_all(spfi->rx_ch);
-		spfi_reset(spfi);
+	if (xfer->len > SPFI_TRANSACTION_TSIZE_MASK) {
+		dev_err(spfi->dev,
+			"Transfer length (%d) is greater than the max supported (%d)",
+			xfer->len, SPFI_TRANSACTION_TSIZE_MASK);
+		return -EINVAL;
 	}
 
 	img_spfi_config(master, spi, xfer);
@@ -485,17 +517,6 @@ static int img_spfi_transfer_one(struct spi_master *master,
 	return ret;
 }
 
-static void img_spfi_set_cs(struct spi_device *spi, bool enable)
-{
-	struct img_spfi *spfi = spi_master_get_devdata(spi->master);
-	u32 val;
-
-	val = spfi_readl(spfi, SPFI_PORT_STATE);
-	val &= ~(SPFI_PORT_STATE_DEV_SEL_MASK << SPFI_PORT_STATE_DEV_SEL_SHIFT);
-	val |= spi->chip_select << SPFI_PORT_STATE_DEV_SEL_SHIFT;
-	spfi_writel(spfi, val, SPFI_PORT_STATE);
-}
-
 static bool img_spfi_can_dma(struct spi_master *master, struct spi_device *spi,
 			     struct spi_transfer *xfer)
 {
@@ -584,14 +605,17 @@ static int img_spfi_probe(struct platform_device *pdev)
 	master->mode_bits = SPI_CPOL | SPI_CPHA | SPI_TX_DUAL | SPI_RX_DUAL;
 	if (of_property_read_bool(spfi->dev->of_node, "img,supports-quad-mode"))
 		master->mode_bits |= SPI_TX_QUAD | SPI_RX_QUAD;
-	master->num_chipselect = 5;
 	master->dev.of_node = pdev->dev.of_node;
 	master->bits_per_word_mask = SPI_BPW_MASK(32) | SPI_BPW_MASK(8);
-	master->max_speed_hz = clk_get_rate(spfi->spfi_clk);
-	master->min_speed_hz = master->max_speed_hz / 512;
+	master->max_speed_hz = clk_get_rate(spfi->spfi_clk) / 4;
+	master->min_speed_hz = clk_get_rate(spfi->spfi_clk) / 512;
 
-	master->set_cs = img_spfi_set_cs;
+	master->setup = img_spfi_setup;
+	master->cleanup = img_spfi_cleanup;
 	master->transfer_one = img_spfi_transfer_one;
+	master->prepare_message = img_spfi_prepare;
+	master->unprepare_message = img_spfi_unprepare;
+	master->handle_err = img_spfi_handle_err;
 
 	spfi->tx_ch = dma_request_slave_channel(spfi->dev, "tx");
 	spfi->rx_ch = dma_request_slave_channel(spfi->dev, "rx");

drivers/spi/spi-imx.c

@@ -370,8 +370,6 @@ static int __maybe_unused mx51_ecspi_config(struct spi_imx_data *spi_imx,
 	if (spi_imx->dma_is_inited) {
 		dma = readl(spi_imx->base + MX51_ECSPI_DMA);
 
-		spi_imx->tx_wml = spi_imx_get_fifosize(spi_imx) / 2;
-		spi_imx->rx_wml = spi_imx_get_fifosize(spi_imx) / 2;
 		spi_imx->rxt_wml = spi_imx_get_fifosize(spi_imx) / 2;
 		rx_wml_cfg = spi_imx->rx_wml << MX51_ECSPI_DMA_RX_WML_OFFSET;
 		tx_wml_cfg = spi_imx->tx_wml << MX51_ECSPI_DMA_TX_WML_OFFSET;
@@ -868,6 +866,8 @@ static int spi_imx_sdma_init(struct device *dev, struct spi_imx_data *spi_imx,
 	master->max_dma_len = MAX_SDMA_BD_BYTES;
 	spi_imx->bitbang.master->flags = SPI_MASTER_MUST_RX |
 					 SPI_MASTER_MUST_TX;
+	spi_imx->tx_wml = spi_imx_get_fifosize(spi_imx) / 2;
+	spi_imx->rx_wml = spi_imx_get_fifosize(spi_imx) / 2;
 	spi_imx->dma_is_inited = 1;
 
 	return 0;
@@ -903,7 +903,7 @@ static int spi_imx_dma_transfer(struct spi_imx_data *spi_imx,
 
 	if (tx) {
 		desc_tx = dmaengine_prep_slave_sg(master->dma_tx,
-					tx->sgl, tx->nents, DMA_TO_DEVICE,
+					tx->sgl, tx->nents, DMA_MEM_TO_DEV,
 					DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
 		if (!desc_tx)
 			goto no_dma;
@@ -915,7 +915,7 @@ static int spi_imx_dma_transfer(struct spi_imx_data *spi_imx,
 
 	if (rx) {
 		desc_rx = dmaengine_prep_slave_sg(master->dma_rx,
-					rx->sgl, rx->nents, DMA_FROM_DEVICE,
+					rx->sgl, rx->nents, DMA_DEV_TO_MEM,
 					DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
 		if (!desc_rx)
 			goto no_dma;

drivers/spi/spi-mpc512x-psc.c

@@ -588,7 +588,7 @@ static int mpc512x_psc_spi_of_remove(struct platform_device *op)
 	return mpc512x_psc_spi_do_remove(&op->dev);
 }
 
-static struct of_device_id mpc512x_psc_spi_of_match[] = {
+static const struct of_device_id mpc512x_psc_spi_of_match[] = {
 	{ .compatible = "fsl,mpc5121-psc-spi", },
 	{},
 };

drivers/spi/spi-octeon.c

@@ -238,7 +238,7 @@ static int octeon_spi_remove(struct platform_device *pdev)
 	return 0;
 }
 
-static struct of_device_id octeon_spi_match[] = {
+static const struct of_device_id octeon_spi_match[] = {
 	{ .compatible = "cavium,octeon-3010-spi", },
 	{},
 };

drivers/spi/spi-omap-100k.c

@@ -24,6 +24,7 @@
 #include <linux/device.h>
 #include <linux/delay.h>
 #include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
 #include <linux/err.h>
 #include <linux/clk.h>
 #include <linux/io.h>
@@ -294,16 +295,6 @@ static int omap1_spi100k_setup(struct spi_device *spi)
 	return ret;
 }
 
-static int omap1_spi100k_prepare_hardware(struct spi_master *master)
-{
-	struct omap1_spi100k *spi100k = spi_master_get_devdata(master);
-
-	clk_prepare_enable(spi100k->ick);
-	clk_prepare_enable(spi100k->fck);
-
-	return 0;
-}
-
 static int omap1_spi100k_transfer_one_message(struct spi_master *master,
 					      struct spi_message *m)
 {
@@ -372,16 +363,6 @@ static int omap1_spi100k_transfer_one_message(struct spi_master *master,
 	return status;
 }
 
-static int omap1_spi100k_unprepare_hardware(struct spi_master *master)
-{
-	struct omap1_spi100k *spi100k = spi_master_get_devdata(master);
-
-	clk_disable_unprepare(spi100k->ick);
-	clk_disable_unprepare(spi100k->fck);
-
-	return 0;
-}
-
 static int omap1_spi100k_probe(struct platform_device *pdev)
 {
 	struct spi_master       *master;
@@ -402,14 +383,12 @@ static int omap1_spi100k_probe(struct platform_device *pdev)
 
 	master->setup = omap1_spi100k_setup;
 	master->transfer_one_message = omap1_spi100k_transfer_one_message;
-	master->prepare_transfer_hardware = omap1_spi100k_prepare_hardware;
-	master->unprepare_transfer_hardware = omap1_spi100k_unprepare_hardware;
-	master->cleanup = NULL;
 	master->num_chipselect = 2;
 	master->mode_bits = MODEBITS;
 	master->bits_per_word_mask = SPI_BPW_RANGE_MASK(4, 32);
 	master->min_speed_hz = OMAP1_SPI100K_MAX_FREQ/(1<<16);
 	master->max_speed_hz = OMAP1_SPI100K_MAX_FREQ;
+	master->auto_runtime_pm = true;
 
 	spi100k = spi_master_get_devdata(master);
 
@@ -434,22 +413,96 @@ static int omap1_spi100k_probe(struct platform_device *pdev)
 		goto err;
 	}
 
+	status = clk_prepare_enable(spi100k->ick);
+	if (status != 0) {
+		dev_err(&pdev->dev, "failed to enable ick: %d\n", status);
+		goto err;
+	}
+
+	status = clk_prepare_enable(spi100k->fck);
+	if (status != 0) {
+		dev_err(&pdev->dev, "failed to enable fck: %d\n", status);
+		goto err_ick;
+	}
+
+	pm_runtime_enable(&pdev->dev);
+	pm_runtime_set_active(&pdev->dev);
+
 	status = devm_spi_register_master(&pdev->dev, master);
 	if (status < 0)
-		goto err;
+		goto err_fck;
 
 	return status;
 
+err_fck:
+	clk_disable_unprepare(spi100k->fck);
+err_ick:
+	clk_disable_unprepare(spi100k->ick);
 err:
 	spi_master_put(master);
 	return status;
 }
 
+static int omap1_spi100k_remove(struct platform_device *pdev)
+{
+	struct spi_master *master = spi_master_get(platform_get_drvdata(pdev));
+	struct omap1_spi100k *spi100k = spi_master_get_devdata(master);
+
+	pm_runtime_disable(&pdev->dev);
+
+	clk_disable_unprepare(spi100k->fck);
+	clk_disable_unprepare(spi100k->ick);
+
+	return 0;
+}
+
+#ifdef CONFIG_PM
+static int omap1_spi100k_runtime_suspend(struct device *dev)
+{
+	struct spi_master *master = spi_master_get(dev_get_drvdata(dev));
+	struct omap1_spi100k *spi100k = spi_master_get_devdata(master);
+
+	clk_disable_unprepare(spi100k->ick);
+	clk_disable_unprepare(spi100k->fck);
+
+	return 0;
+}
+
+static int omap1_spi100k_runtime_resume(struct device *dev)
+{
+	struct spi_master *master = spi_master_get(dev_get_drvdata(dev));
+	struct omap1_spi100k *spi100k = spi_master_get_devdata(master);
+	int ret;
+
+	ret = clk_prepare_enable(spi100k->ick);
+	if (ret != 0) {
+		dev_err(dev, "Failed to enable ick: %d\n", ret);
+		return ret;
+	}
+
+	ret = clk_prepare_enable(spi100k->fck);
+	if (ret != 0) {
+		dev_err(dev, "Failed to enable fck: %d\n", ret);
+		clk_disable_unprepare(spi100k->ick);
+		return ret;
+	}
+
+	return 0;
+}
+#endif
+
+static const struct dev_pm_ops omap1_spi100k_pm = {
+	SET_RUNTIME_PM_OPS(omap1_spi100k_runtime_suspend,
+			   omap1_spi100k_runtime_resume, NULL)
+};
+
 static struct platform_driver omap1_spi100k_driver = {
 	.driver = {
 		.name		= "omap1_spi100k",
+		.pm		= &omap1_spi100k_pm,
 	},
 	.probe		= omap1_spi100k_probe,
+	.remove		= omap1_spi100k_remove,
 };
 
 module_platform_driver(omap1_spi100k_driver);

drivers/spi/spi-omap-uwire.c

@@ -44,7 +44,6 @@
 #include <linux/module.h>
 #include <linux/io.h>
 
-#include <asm/irq.h>
 #include <mach/hardware.h>
 #include <asm/mach-types.h>
 

drivers/spi/spi-pl022.c

@@ -285,7 +285,12 @@
  */
 #define DEFAULT_SSP_REG_IMSC  0x0UL
 #define DISABLE_ALL_INTERRUPTS DEFAULT_SSP_REG_IMSC
-#define ENABLE_ALL_INTERRUPTS (~DEFAULT_SSP_REG_IMSC)
+#define ENABLE_ALL_INTERRUPTS ( \
+	SSP_IMSC_MASK_RORIM | \
+	SSP_IMSC_MASK_RTIM | \
+	SSP_IMSC_MASK_RXIM | \
+	SSP_IMSC_MASK_TXIM \
+)
 
 #define CLEAR_ALL_INTERRUPTS  0x3
 
@@ -534,12 +539,12 @@ static void giveback(struct pl022 *pl022)
 	pl022->cur_msg = NULL;
 	pl022->cur_transfer = NULL;
 	pl022->cur_chip = NULL;
-	spi_finalize_current_message(pl022->master);
 
 	/* disable the SPI/SSP operation */
 	writew((readw(SSP_CR1(pl022->virtbase)) &
 		(~SSP_CR1_MASK_SSE)), SSP_CR1(pl022->virtbase));
 
+	spi_finalize_current_message(pl022->master);
 }
 
 /**
@@ -1251,7 +1256,6 @@ static irqreturn_t pl022_interrupt_handler(int irq, void *dev_id)
 	struct pl022 *pl022 = dev_id;
 	struct spi_message *msg = pl022->cur_msg;
 	u16 irq_status = 0;
-	u16 flag = 0;
 
 	if (unlikely(!msg)) {
 		dev_err(&pl022->adev->dev,
@@ -1280,9 +1284,6 @@ static irqreturn_t pl022_interrupt_handler(int irq, void *dev_id)
 		if (readw(SSP_SR(pl022->virtbase)) & SSP_SR_MASK_RFF)
 			dev_err(&pl022->adev->dev,
 				"RXFIFO is full\n");
-		if (readw(SSP_SR(pl022->virtbase)) & SSP_SR_MASK_TNF)
-			dev_err(&pl022->adev->dev,
-				"TXFIFO is full\n");
 
 		/*
 		 * Disable and clear interrupts, disable SSP,
@@ -1303,8 +1304,7 @@ static irqreturn_t pl022_interrupt_handler(int irq, void *dev_id)
 
 	readwriter(pl022);
 
-	if ((pl022->tx == pl022->tx_end) && (flag == 0)) {
-		flag = 1;
+	if (pl022->tx == pl022->tx_end) {
 		/* Disable Transmit interrupt, enable receive interrupt */
 		writew((readw(SSP_IMSC(pl022->virtbase)) &
 		       ~SSP_IMSC_MASK_TXIM) | SSP_IMSC_MASK_RXIM,

drivers/spi/spi-pxa2xx.c

@@ -20,6 +20,7 @@
 #include <linux/errno.h>
 #include <linux/err.h>
 #include <linux/interrupt.h>
+#include <linux/kernel.h>
 #include <linux/platform_device.h>
 #include <linux/spi/pxa2xx_spi.h>
 #include <linux/spi/spi.h>
@@ -30,10 +31,6 @@
 #include <linux/pm_runtime.h>
 #include <linux/acpi.h>
 
-#include <asm/io.h>
-#include <asm/irq.h>
-#include <asm/delay.h>
-
 #include "spi-pxa2xx.h"
 
 MODULE_AUTHOR("Stephen Street");
@@ -67,54 +64,6 @@ MODULE_ALIAS("platform:pxa2xx-spi");
 #define LPSS_TX_LOTHRESH_DFLT	160
 #define LPSS_TX_HITHRESH_DFLT	224
 
-struct quark_spi_rate {
-	u32 bitrate;
-	u32 dds_clk_rate;
-	u32 clk_div;
-};
-
-/*
- * 'rate', 'dds', 'clk_div' lookup table, which is defined in
- * the Quark SPI datasheet.
- */
-static const struct quark_spi_rate quark_spi_rate_table[] = {
-/*	bitrate,	dds_clk_rate,	clk_div */
-	{50000000,	0x800000,	0},
-	{40000000,	0x666666,	0},
-	{25000000,	0x400000,	0},
-	{20000000,	0x666666,	1},
-	{16667000,	0x800000,	2},
-	{13333000,	0x666666,	2},
-	{12500000,	0x200000,	0},
-	{10000000,	0x800000,	4},
-	{8000000,	0x666666,	4},
-	{6250000,	0x400000,	3},
-	{5000000,	0x400000,	4},
-	{4000000,	0x666666,	9},
-	{3125000,	0x80000,	0},
-	{2500000,	0x400000,	9},
-	{2000000,	0x666666,	19},
-	{1563000,	0x40000,	0},
-	{1250000,	0x200000,	9},
-	{1000000,	0x400000,	24},
-	{800000,	0x666666,	49},
-	{781250,	0x20000,	0},
-	{625000,	0x200000,	19},
-	{500000,	0x400000,	49},
-	{400000,	0x666666,	99},
-	{390625,	0x10000,	0},
-	{250000,	0x400000,	99},
-	{200000,	0x666666,	199},
-	{195313,	0x8000,		0},
-	{125000,	0x100000,	49},
-	{100000,	0x200000,	124},
-	{50000,		0x100000,	124},
-	{25000,		0x80000,	124},
-	{10016,		0x20000,	77},
-	{5040,		0x20000,	154},
-	{1002,		0x8000,		194},
-};
-
 /* Offset from drv_data->lpss_base */
 #define GENERAL_REG		0x08
 #define GENERAL_REG_RXTO_HOLDOFF_DISABLE BIT(24)
@@ -701,25 +650,124 @@ static irqreturn_t ssp_int(int irq, void *dev_id)
 }
 
 /*
- * The Quark SPI data sheet gives a table, and for the given 'rate',
- * the 'dds' and 'clk_div' can be found in the table.
+ * The Quark SPI has an additional 24 bit register (DDS_CLK_RATE) to multiply
+ * input frequency by fractions of 2^24. It also has a divider by 5.
+ *
+ * There are formulas to get baud rate value for given input frequency and
+ * divider parameters, such as DDS_CLK_RATE and SCR:
+ *
+ * Fsys = 200MHz
+ *
+ * Fssp = Fsys * DDS_CLK_RATE / 2^24			(1)
+ * Baud rate = Fsclk = Fssp / (2 * (SCR + 1))		(2)
+ *
+ * DDS_CLK_RATE either 2^n or 2^n / 5.
+ * SCR is in range 0 .. 255
+ *
+ * Divisor = 5^i * 2^j * 2 * k
+ *       i = [0, 1]      i = 1 iff j = 0 or j > 3
+ *       j = [0, 23]     j = 0 iff i = 1
+ *       k = [1, 256]
+ * Special case: j = 0, i = 1: Divisor = 2 / 5
+ *
+ * Accordingly to the specification the recommended values for DDS_CLK_RATE
+ * are:
+ *	Case 1:		2^n, n = [0, 23]
+ *	Case 2:		2^24 * 2 / 5 (0x666666)
+ *	Case 3:		less than or equal to 2^24 / 5 / 16 (0x33333)
+ *
+ * In all cases the lowest possible value is better.
+ *
+ * The function calculates parameters for all cases and chooses the one closest
+ * to the asked baud rate.
  */
-static u32 quark_x1000_set_clk_regvals(u32 rate, u32 *dds, u32 *clk_div)
+static unsigned int quark_x1000_get_clk_div(int rate, u32 *dds)
 {
-	unsigned int i;
+	unsigned long xtal = 200000000;
+	unsigned long fref = xtal / 2;		/* mandatory division by 2,
+						   see (2) */
+						/* case 3 */
+	unsigned long fref1 = fref / 2;		/* case 1 */
+	unsigned long fref2 = fref * 2 / 5;	/* case 2 */
+	unsigned long scale;
+	unsigned long q, q1, q2;
+	long r, r1, r2;
+	u32 mul;
 
-	for (i = 0; i < ARRAY_SIZE(quark_spi_rate_table); i++) {
-		if (rate >= quark_spi_rate_table[i].bitrate) {
-			*dds = quark_spi_rate_table[i].dds_clk_rate;
-			*clk_div = quark_spi_rate_table[i].clk_div;
-			return quark_spi_rate_table[i].bitrate;
+	/* Case 1 */
+
+	/* Set initial value for DDS_CLK_RATE */
+	mul = (1 << 24) >> 1;
+
+	/* Calculate initial quot */
+	q1 = DIV_ROUND_CLOSEST(fref1, rate);
+
+	/* Scale q1 if it's too big */
+	if (q1 > 256) {
+		/* Scale q1 to range [1, 512] */
+		scale = fls_long(q1 - 1);
+		if (scale > 9) {
+			q1 >>= scale - 9;
+			mul >>= scale - 9;
+		}
+
+		/* Round the result if we have a remainder */
+		q1 += q1 & 1;
+	}
+
+	/* Decrease DDS_CLK_RATE as much as we can without loss in precision */
+	scale = __ffs(q1);
+	q1 >>= scale;
+	mul >>= scale;
+
+	/* Get the remainder */
+	r1 = abs(fref1 / (1 << (24 - fls_long(mul))) / q1 - rate);
+
+	/* Case 2 */
+
+	q2 = DIV_ROUND_CLOSEST(fref2, rate);
+	r2 = abs(fref2 / q2 - rate);
+
+	/*
+	 * Choose the best between two: less remainder we have the better. We
+	 * can't go case 2 if q2 is greater than 256 since SCR register can
+	 * hold only values 0 .. 255.
+	 */
+	if (r2 >= r1 || q2 > 256) {
+		/* case 1 is better */
+		r = r1;
+		q = q1;
+	} else {
+		/* case 2 is better */
+		r = r2;
+		q = q2;
+		mul = (1 << 24) * 2 / 5;
+	}
+
+	/* Check case 3 only If the divisor is big enough */
+	if (fref / rate >= 80) {
+		u64 fssp;
+		u32 m;
+
+		/* Calculate initial quot */
+		q1 = DIV_ROUND_CLOSEST(fref, rate);
+		m = (1 << 24) / q1;
+
+		/* Get the remainder */
+		fssp = (u64)fref * m;
+		do_div(fssp, 1 << 24);
+		r1 = abs(fssp - rate);
+
+		/* Choose this one if it suits better */
+		if (r1 < r) {
+			/* case 3 is better */
+			q = 1;
+			mul = m;
 		}
 	}
 
-	*dds = quark_spi_rate_table[i-1].dds_clk_rate;
-	*clk_div = quark_spi_rate_table[i-1].clk_div;
-
-	return quark_spi_rate_table[i-1].bitrate;
+	*dds = mul;
+	return q - 1;
 }
 
 static unsigned int ssp_get_clk_div(struct driver_data *drv_data, int rate)
@@ -730,23 +778,25 @@ static unsigned int ssp_get_clk_div(struct driver_data *drv_data, int rate)
 	rate = min_t(int, ssp_clk, rate);
 
 	if (ssp->type == PXA25x_SSP || ssp->type == CE4100_SSP)
-		return ((ssp_clk / (2 * rate) - 1) & 0xff) << 8;
+		return (ssp_clk / (2 * rate) - 1) & 0xff;
 	else
-		return ((ssp_clk / rate - 1) & 0xfff) << 8;
+		return (ssp_clk / rate - 1) & 0xfff;
 }
 
 static unsigned int pxa2xx_ssp_get_clk_div(struct driver_data *drv_data,
 					   struct chip_data *chip, int rate)
 {
-	u32 clk_div;
+	unsigned int clk_div;
 
 	switch (drv_data->ssp_type) {
 	case QUARK_X1000_SSP:
-		quark_x1000_set_clk_regvals(rate, &chip->dds_rate, &clk_div);
-		return clk_div << 8;
+		clk_div = quark_x1000_get_clk_div(rate, &chip->dds_rate);
+		break;
 	default:
-		return ssp_get_clk_div(drv_data, rate);
+		clk_div = ssp_get_clk_div(drv_data, rate);
+		break;
 	}
+	return clk_div << 8;
 }
 
 static void pump_transfers(unsigned long data)

drivers/spi/spi-qup.c

@@ -22,6 +22,8 @@
 #include <linux/platform_device.h>
 #include <linux/pm_runtime.h>
 #include <linux/spi/spi.h>
+#include <linux/dmaengine.h>
+#include <linux/dma-mapping.h>
 
 #define QUP_CONFIG			0x0000
 #define QUP_STATE			0x0004
@@ -116,6 +118,8 @@
 
 #define SPI_NUM_CHIPSELECTS		4
 
+#define SPI_MAX_DMA_XFER		(SZ_64K - 64)
+
 /* high speed mode is when bus rate is greater then 26MHz */
 #define SPI_HS_MIN_RATE			26000000
 #define SPI_MAX_RATE			50000000
@@ -140,9 +144,14 @@ struct spi_qup {
 	struct completion	done;
 	int			error;
 	int			w_size;	/* bytes per SPI word */
+	int			n_words;
 	int			tx_bytes;
 	int			rx_bytes;
 	int			qup_v1;
+
+	int			use_dma;
+	struct dma_slave_config	rx_conf;
+	struct dma_slave_config	tx_conf;
 };
 
 
@@ -198,7 +207,6 @@ 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)
 {
@@ -266,6 +274,107 @@ static void spi_qup_fifo_write(struct spi_qup *controller,
 	}
 }
 
+static void spi_qup_dma_done(void *data)
+{
+	struct spi_qup *qup = data;
+
+	complete(&qup->done);
+}
+
+static int spi_qup_prep_sg(struct spi_master *master, struct spi_transfer *xfer,
+			   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) {
+		chan = master->dma_tx;
+		nents = xfer->tx_sg.nents;
+		sgl = xfer->tx_sg.sgl;
+	} 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;
+
+	desc->callback = callback;
+	desc->callback_param = qup;
+
+	cookie = dmaengine_submit(desc);
+
+	return dma_submit_error(cookie);
+}
+
+static void spi_qup_dma_terminate(struct spi_master *master,
+				  struct spi_transfer *xfer)
+{
+	if (xfer->tx_buf)
+		dmaengine_terminate_all(master->dma_tx);
+	if (xfer->rx_buf)
+		dmaengine_terminate_all(master->dma_rx);
+}
+
+static int spi_qup_do_dma(struct spi_master *master, struct spi_transfer *xfer)
+{
+	dma_async_tx_callback rx_done = NULL, tx_done = NULL;
+	int ret;
+
+	if (xfer->rx_buf)
+		rx_done = spi_qup_dma_done;
+	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;
+
+		dma_async_issue_pending(master->dma_rx);
+	}
+
+	if (xfer->tx_buf) {
+		ret = spi_qup_prep_sg(master, xfer, DMA_MEM_TO_DEV, tx_done);
+		if (ret)
+			return ret;
+
+		dma_async_issue_pending(master->dma_tx);
+	}
+
+	return 0;
+}
+
+static int spi_qup_do_pio(struct spi_master *master, struct spi_transfer *xfer)
+{
+	struct spi_qup *qup = spi_master_get_devdata(master);
+	int 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;
+	}
+
+	spi_qup_fifo_write(qup, xfer);
+
+	return 0;
+}
+
 static irqreturn_t spi_qup_qup_irq(int irq, void *dev_id)
 {
 	struct spi_qup *controller = dev_id;
@@ -315,11 +424,13 @@ static irqreturn_t spi_qup_qup_irq(int irq, void *dev_id)
 		error = -EIO;
 	}
 
-	if (opflags & QUP_OP_IN_SERVICE_FLAG)
-		spi_qup_fifo_read(controller, xfer);
+	if (!controller->use_dma) {
+		if (opflags & QUP_OP_IN_SERVICE_FLAG)
+			spi_qup_fifo_read(controller, xfer);
 
-	if (opflags & QUP_OP_OUT_SERVICE_FLAG)
-		spi_qup_fifo_write(controller, xfer);
+		if (opflags & QUP_OP_OUT_SERVICE_FLAG)
+			spi_qup_fifo_write(controller, xfer);
+	}
 
 	spin_lock_irqsave(&controller->lock, flags);
 	controller->error = error;
@@ -332,13 +443,35 @@ static irqreturn_t spi_qup_qup_irq(int irq, void *dev_id)
 	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)
 {
 	struct spi_qup *controller = spi_master_get_devdata(spi->master);
 	u32 config, iomode, mode, control;
-	int ret, n_words, w_size;
+	int ret, n_words;
 
 	if (spi->mode & SPI_LOOP && xfer->len > controller->in_fifo_sz) {
 		dev_err(controller->dev, "too big size for loopback %d > %d\n",
@@ -358,35 +491,54 @@ static int spi_qup_io_config(struct spi_device *spi, struct spi_transfer *xfer)
 		return -EIO;
 	}
 
-	w_size = 4;
-	if (xfer->bits_per_word <= 8)
-		w_size = 1;
-	else if (xfer->bits_per_word <= 16)
-		w_size = 2;
+	mode = spi_qup_get_mode(spi->master, xfer);
+	n_words = controller->n_words;
 
-	n_words = xfer->len / w_size;
-	controller->w_size = w_size;
-
-	if (n_words <= (controller->in_fifo_sz / sizeof(u32))) {
-		mode = QUP_IO_M_MODE_FIFO;
+	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);
 		/* must be zero for FIFO */
 		writel_relaxed(0, controller->base + QUP_MX_INPUT_CNT);
 		writel_relaxed(0, controller->base + QUP_MX_OUTPUT_CNT);
-	} else {
-		mode = QUP_IO_M_MODE_BLOCK;
+	} 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);
 		/* 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;
+
+			input_cnt = controller->base + QUP_MX_INPUT_CNT;
+			/*
+			 * for DMA transfers, both QUP_MX_INPUT_CNT and
+			 * QUP_MX_OUTPUT_CNT must be zero to all cases but one.
+			 * That case is a non-balanced transfer when there is
+			 * only a rx_buf.
+			 */
+			if (xfer->tx_buf)
+				writel_relaxed(0, input_cnt);
+			else
+				writel_relaxed(n_words, input_cnt);
+
+			writel_relaxed(0, controller->base + QUP_MX_OUTPUT_CNT);
+		}
 	}
 
 	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);
-	iomode &= ~(QUP_IO_M_PACK_EN | QUP_IO_M_UNPACK_EN);
+
+	if (!controller->use_dma)
+		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);
 
@@ -428,11 +580,31 @@ static int spi_qup_io_config(struct spi_device *spi, struct spi_transfer *xfer)
 	config &= ~(QUP_CONFIG_NO_INPUT | QUP_CONFIG_NO_OUTPUT | QUP_CONFIG_N);
 	config |= xfer->bits_per_word - 1;
 	config |= QUP_CONFIG_SPI_MODE;
+
+	if (controller->use_dma) {
+		if (!xfer->tx_buf)
+			config |= QUP_CONFIG_NO_OUTPUT;
+		if (!xfer->rx_buf)
+			config |= QUP_CONFIG_NO_INPUT;
+	}
+
 	writel_relaxed(config, controller->base + QUP_CONFIG);
 
 	/* only write to OPERATIONAL_MASK when register is present */
-	if (!controller->qup_v1)
-		writel_relaxed(0, controller->base + QUP_OPERATIONAL_MASK);
+	if (!controller->qup_v1) {
+		u32 mask = 0;
+
+		/*
+		 * mask INPUT and OUTPUT service flags to prevent IRQs on FIFO
+		 * status change in BAM mode
+		 */
+
+		if (mode == QUP_IO_M_MODE_BAM)
+			mask = QUP_OP_IN_SERVICE_FLAG | QUP_OP_OUT_SERVICE_FLAG;
+
+		writel_relaxed(mask, controller->base + QUP_OPERATIONAL_MASK);
+	}
+
 	return 0;
 }
 
@@ -461,17 +633,13 @@ static int spi_qup_transfer_one(struct spi_master *master,
 	controller->tx_bytes = 0;
 	spin_unlock_irqrestore(&controller->lock, flags);
 
-	if (spi_qup_set_state(controller, QUP_STATE_RUN)) {
-		dev_warn(controller->dev, "cannot set RUN state\n");
-		goto exit;
-	}
+	if (controller->use_dma)
+		ret = spi_qup_do_dma(master, xfer);
+	else
+		ret = spi_qup_do_pio(master, xfer);
 
-	if (spi_qup_set_state(controller, QUP_STATE_PAUSE)) {
-		dev_warn(controller->dev, "cannot set PAUSE state\n");
+	if (ret)
 		goto exit;
-	}
-
-	spi_qup_fifo_write(controller, xfer);
 
 	if (spi_qup_set_state(controller, QUP_STATE_RUN)) {
 		dev_warn(controller->dev, "cannot set EXECUTE state\n");
@@ -480,6 +648,7 @@ static int spi_qup_transfer_one(struct spi_master *master,
 
 	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);
@@ -487,6 +656,97 @@ exit:
 	if (!ret)
 		ret = controller->error;
 	spin_unlock_irqrestore(&controller->lock, flags);
+
+	if (ret && controller->use_dma)
+		spi_qup_dma_terminate(master, xfer);
+
+	return ret;
+}
+
+static bool spi_qup_can_dma(struct spi_master *master, struct spi_device *spi,
+			    struct spi_transfer *xfer)
+{
+	struct spi_qup *qup = spi_master_get_devdata(master);
+	size_t dma_align = dma_get_cache_alignment();
+	u32 mode;
+
+	qup->use_dma = 0;
+
+	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->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;
+
+	mode = spi_qup_get_mode(master, xfer);
+	if (mode == QUP_IO_M_MODE_FIFO)
+		return false;
+
+	qup->use_dma = 1;
+
+	return true;
+}
+
+static void spi_qup_release_dma(struct spi_master *master)
+{
+	if (!IS_ERR_OR_NULL(master->dma_rx))
+		dma_release_channel(master->dma_rx);
+	if (!IS_ERR_OR_NULL(master->dma_tx))
+		dma_release_channel(master->dma_tx);
+}
+
+static int spi_qup_init_dma(struct spi_master *master, resource_size_t base)
+{
+	struct spi_qup *spi = spi_master_get_devdata(master);
+	struct dma_slave_config *rx_conf = &spi->rx_conf,
+				*tx_conf = &spi->tx_conf;
+	struct device *dev = spi->dev;
+	int ret;
+
+	/* allocate dma resources, if available */
+	master->dma_rx = dma_request_slave_channel_reason(dev, "rx");
+	if (IS_ERR(master->dma_rx))
+		return PTR_ERR(master->dma_rx);
+
+	master->dma_tx = dma_request_slave_channel_reason(dev, "tx");
+	if (IS_ERR(master->dma_tx)) {
+		ret = PTR_ERR(master->dma_tx);
+		goto err_tx;
+	}
+
+	/* set DMA parameters */
+	rx_conf->direction = DMA_DEV_TO_MEM;
+	rx_conf->device_fc = 1;
+	rx_conf->src_addr = base + QUP_INPUT_FIFO;
+	rx_conf->src_maxburst = spi->in_blk_sz;
+
+	tx_conf->direction = DMA_MEM_TO_DEV;
+	tx_conf->device_fc = 1;
+	tx_conf->dst_addr = base + QUP_OUTPUT_FIFO;
+	tx_conf->dst_maxburst = spi->out_blk_sz;
+
+	ret = dmaengine_slave_config(master->dma_rx, rx_conf);
+	if (ret) {
+		dev_err(dev, "failed to configure RX channel\n");
+		goto err;
+	}
+
+	ret = dmaengine_slave_config(master->dma_tx, tx_conf);
+	if (ret) {
+		dev_err(dev, "failed to configure TX channel\n");
+		goto err;
+	}
+
+	return 0;
+
+err:
+	dma_release_channel(master->dma_tx);
+err_tx:
+	dma_release_channel(master->dma_rx);
 	return ret;
 }
 
@@ -498,7 +758,7 @@ static int spi_qup_probe(struct platform_device *pdev)
 	struct resource *res;
 	struct device *dev;
 	void __iomem *base;
-	u32 max_freq, iomode;
+	u32 max_freq, iomode, num_cs;
 	int ret, irq, size;
 
 	dev = &pdev->dev;
@@ -550,10 +810,11 @@ static int spi_qup_probe(struct platform_device *pdev)
 	}
 
 	/* use num-cs unless not present or out of range */
-	if (of_property_read_u16(dev->of_node, "num-cs",
-			&master->num_chipselect) ||
-			(master->num_chipselect > SPI_NUM_CHIPSELECTS))
+	if (of_property_read_u32(dev->of_node, "num-cs", &num_cs) ||
+	    num_cs > SPI_NUM_CHIPSELECTS)
 		master->num_chipselect = SPI_NUM_CHIPSELECTS;
+	else
+		master->num_chipselect = num_cs;
 
 	master->bus_num = pdev->id;
 	master->mode_bits = SPI_CPOL | SPI_CPHA | SPI_CS_HIGH | SPI_LOOP;
@@ -562,6 +823,8 @@ static int spi_qup_probe(struct platform_device *pdev)
 	master->transfer_one = spi_qup_transfer_one;
 	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;
 
 	platform_set_drvdata(pdev, master);
 
@@ -573,6 +836,12 @@ static int spi_qup_probe(struct platform_device *pdev)
 	controller->cclk = cclk;
 	controller->irq = irq;
 
+	ret = spi_qup_init_dma(master, res->start);
+	if (ret == -EPROBE_DEFER)
+		goto error;
+	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;
@@ -609,7 +878,7 @@ static int spi_qup_probe(struct platform_device *pdev)
 	ret = spi_qup_set_state(controller, QUP_STATE_RESET);
 	if (ret) {
 		dev_err(dev, "cannot set RESET state\n");
-		goto error;
+		goto error_dma;
 	}
 
 	writel_relaxed(0, base + QUP_OPERATIONAL);
@@ -633,7 +902,7 @@ static int spi_qup_probe(struct platform_device *pdev)
 	ret = devm_request_irq(dev, irq, spi_qup_qup_irq,
 			       IRQF_TRIGGER_HIGH, pdev->name, controller);
 	if (ret)
-		goto error;
+		goto error_dma;
 
 	pm_runtime_set_autosuspend_delay(dev, MSEC_PER_SEC);
 	pm_runtime_use_autosuspend(dev);
@@ -648,6 +917,8 @@ static int spi_qup_probe(struct platform_device *pdev)
 
 disable_pm:
 	pm_runtime_disable(&pdev->dev);
+error_dma:
+	spi_qup_release_dma(master);
 error:
 	clk_disable_unprepare(cclk);
 	clk_disable_unprepare(iclk);
@@ -739,6 +1010,8 @@ static int spi_qup_remove(struct platform_device *pdev)
 	if (ret)
 		return ret;
 
+	spi_qup_release_dma(master);
+
 	clk_disable_unprepare(controller->cclk);
 	clk_disable_unprepare(controller->iclk);