Merge tag 'spi-v3.10' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi

Pull spi updates from Mark Brown:
 "A fairly quiet release for SPI, mainly driver work.  A few highlights:

   - Supports bits per word compatibility checking in the core.
   - Allow use of the IP used in Freescale SPI controllers outside
     Freescale SoCs.
   - DMA support for the Atmel SPI driver.
   - New drivers for the BCM2835 and Tegra114"

* tag 'spi-v3.10' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi: (68 commits)
  spi-topcliff-pch: fix to use list_for_each_entry_safe() when delete list items
  spi-topcliff-pch: missing platform_driver_unregister() on error in pch_spi_init()
  ARM: dts: add pinctrl property for spi node for atmel SoC
  ARM: dts: add spi nodes for the atmel boards
  ARM: dts: add spi nodes for atmel SoC
  ARM: at91: add clocks for spi dt entries
  spi/spi-atmel: add dmaengine support
  spi/spi-atmel: add flag to controller data for lock operations
  spi/spi-atmel: add physical base address
  spi/sirf: fix MODULE_DEVICE_TABLE
  MAINTAINERS: Add git repository and update my address
  spi/s3c64xx: Check for errors in dmaengine prepare_transfer()
  spi/s3c64xx: Fix non-dmaengine usage
  spi: omap2-mcspi: fix error return code in omap2_mcspi_probe()
  spi/s3c64xx: let device core setup the default pin configuration
  MAINTAINERS: Update Grant's email address and maintainership
  spi: omap2-mcspi: Fix transfers if DMADEVICES is not set
  spi: s3c64xx: move to generic dmaengine API
  spi-gpio: init CS before spi_bitbang_setup()
  spi: spi-mpc512x-psc: let transmiter/receiver enabled when in xfer loop
  ...

27 files changed, 3372 insertions, 821 deletions

diff --git a/drivers/spi/Kconfig b/drivers/spi/Kconfig
index 2be0de920d67..141d8c10b764 100644
--- a/drivers/spi/Kconfig
+++ b/drivers/spi/Kconfig
@@ -75,6 +75,17 @@ config SPI_ATMEL
 	  This selects a driver for the Atmel SPI Controller, present on
 	  many AT32 (AVR32) and AT91 (ARM) chips.
 
+config SPI_BCM2835
+	tristate "BCM2835 SPI controller"
+	depends on ARCH_BCM2835
+	help
+	  This selects a driver for the Broadcom BCM2835 SPI master.
+
+	  The BCM2835 contains two types of SPI master controller; the
+	  "universal SPI master", and the regular SPI controller. This driver
+	  is for the regular SPI controller. Slave mode operation is not also
+	  not supported.
+
 config SPI_BFIN5XX
 	tristate "SPI controller driver for ADI Blackfin5xx"
 	depends on BLACKFIN
@@ -219,16 +230,23 @@ config SPI_MPC512x_PSC
 
 config SPI_FSL_LIB
 	tristate
+	depends on OF
+
+config SPI_FSL_CPM
+	tristate
 	depends on FSL_SOC
 
 config SPI_FSL_SPI
-	bool "Freescale SPI controller"
-	depends on FSL_SOC
+	bool "Freescale SPI controller and Aeroflex Gaisler GRLIB SPI controller"
+	depends on OF
 	select SPI_FSL_LIB
+	select SPI_FSL_CPM if FSL_SOC
 	help
 	  This enables using the Freescale SPI controllers in master mode.
 	  MPC83xx platform uses the controller in cpu mode or CPM/QE mode.
 	  MPC8569 uses the controller in QE mode, MPC8610 in cpu mode.
+	  This also enables using the Aeroflex Gaisler GRLIB SPI controller in
+	  master mode.
 
 config SPI_FSL_ESPI
 	bool "Freescale eSPI controller"
@@ -398,6 +416,14 @@ config SPI_MXS
 	help
 	  SPI driver for Freescale MXS devices.
 
+config SPI_TEGRA114
+	tristate "NVIDIA Tegra114 SPI Controller"
+	depends on ARCH_TEGRA && TEGRA20_APB_DMA
+	help
+	  SPI driver for NVIDIA Tegra114 SPI Controller interface. This controller
+	  is different than the older SoCs SPI controller and also register interface
+	  get changed with this controller.
+
 config SPI_TEGRA20_SFLASH
 	tristate "Nvidia Tegra20 Serial flash Controller"
 	depends on ARCH_TEGRA
diff --git a/drivers/spi/Makefile b/drivers/spi/Makefile
index e53c30941340..33f9c09561e7 100644
--- a/drivers/spi/Makefile
+++ b/drivers/spi/Makefile
@@ -14,6 +14,7 @@ obj-$(CONFIG_SPI_ALTERA)		+= spi-altera.o
 obj-$(CONFIG_SPI_ATMEL)			+= spi-atmel.o
 obj-$(CONFIG_SPI_ATH79)			+= spi-ath79.o
 obj-$(CONFIG_SPI_AU1550)		+= spi-au1550.o
+obj-$(CONFIG_SPI_BCM2835)		+= spi-bcm2835.o
 obj-$(CONFIG_SPI_BCM63XX)		+= spi-bcm63xx.o
 obj-$(CONFIG_SPI_BFIN5XX)		+= spi-bfin5xx.o
 obj-$(CONFIG_SPI_BFIN_SPORT)		+= spi-bfin-sport.o
@@ -28,6 +29,7 @@ obj-$(CONFIG_SPI_DW_PCI)		+= spi-dw-midpci.o
 spi-dw-midpci-objs			:= spi-dw-pci.o spi-dw-mid.o
 obj-$(CONFIG_SPI_EP93XX)		+= spi-ep93xx.o
 obj-$(CONFIG_SPI_FALCON)		+= spi-falcon.o
+obj-$(CONFIG_SPI_FSL_CPM)		+= spi-fsl-cpm.o
 obj-$(CONFIG_SPI_FSL_LIB)		+= spi-fsl-lib.o
 obj-$(CONFIG_SPI_FSL_ESPI)		+= spi-fsl-espi.o
 obj-$(CONFIG_SPI_FSL_SPI)		+= spi-fsl-spi.o
@@ -63,6 +65,7 @@ obj-$(CONFIG_SPI_SH_HSPI)		+= spi-sh-hspi.o
 obj-$(CONFIG_SPI_SH_MSIOF)		+= spi-sh-msiof.o
 obj-$(CONFIG_SPI_SH_SCI)		+= spi-sh-sci.o
 obj-$(CONFIG_SPI_SIRF)		+= spi-sirf.o
+obj-$(CONFIG_SPI_TEGRA114)		+= spi-tegra114.o
 obj-$(CONFIG_SPI_TEGRA20_SFLASH)	+= spi-tegra20-sflash.o
 obj-$(CONFIG_SPI_TEGRA20_SLINK)		+= spi-tegra20-slink.o
 obj-$(CONFIG_SPI_TI_SSP)		+= spi-ti-ssp.o
diff --git a/drivers/spi/spi-atmel.c b/drivers/spi/spi-atmel.c
index 656d137db253..787bd2c22bca 100644
--- a/drivers/spi/spi-atmel.c
+++ b/drivers/spi/spi-atmel.c
@@ -15,16 +15,17 @@
 #include <linux/platform_device.h>
 #include <linux/delay.h>
 #include <linux/dma-mapping.h>
+#include <linux/dmaengine.h>
 #include <linux/err.h>
 #include <linux/interrupt.h>
 #include <linux/spi/spi.h>
 #include <linux/slab.h>
 #include <linux/platform_data/atmel.h>
+#include <linux/platform_data/dma-atmel.h>
 #include <linux/of.h>
 
-#include <asm/io.h>
-#include <asm/gpio.h>
-#include <mach/cpu.h>
+#include <linux/io.h>
+#include <linux/gpio.h>
 
 /* SPI register offsets */
 #define SPI_CR					0x0000
@@ -39,6 +40,7 @@
 #define SPI_CSR1				0x0034
 #define SPI_CSR2				0x0038
 #define SPI_CSR3				0x003c
+#define SPI_VERSION				0x00fc
 #define SPI_RPR					0x0100
 #define SPI_RCR					0x0104
 #define SPI_TPR					0x0108
@@ -71,6 +73,8 @@
 #define SPI_FDIV_SIZE				1
 #define SPI_MODFDIS_OFFSET			4
 #define SPI_MODFDIS_SIZE			1
+#define SPI_WDRBT_OFFSET			5
+#define SPI_WDRBT_SIZE				1
 #define SPI_LLB_OFFSET				7
 #define SPI_LLB_SIZE				1
 #define SPI_PCS_OFFSET				16
@@ -180,6 +184,27 @@
 #define spi_writel(port,reg,value) \
 	__raw_writel((value), (port)->regs + SPI_##reg)
 
+/* use PIO for small transfers, avoiding DMA setup/teardown overhead and
+ * cache operations; better heuristics consider wordsize and bitrate.
+ */
+#define DMA_MIN_BYTES	16
+
+struct atmel_spi_dma {
+	struct dma_chan			*chan_rx;
+	struct dma_chan			*chan_tx;
+	struct scatterlist		sgrx;
+	struct scatterlist		sgtx;
+	struct dma_async_tx_descriptor	*data_desc_rx;
+	struct dma_async_tx_descriptor	*data_desc_tx;
+
+	struct at_dma_slave	dma_slave;
+};
+
+struct atmel_spi_caps {
+	bool	is_spi2;
+	bool	has_wdrbt;
+	bool	has_dma_support;
+};
 
 /*
  * The core SPI transfer engine just talks to a register bank to set up
@@ -188,7 +213,9 @@
  */
 struct atmel_spi {
 	spinlock_t		lock;
+	unsigned long		flags;
 
+	phys_addr_t		phybase;
 	void __iomem		*regs;
 	int			irq;
 	struct clk		*clk;
@@ -197,13 +224,23 @@ struct atmel_spi {
 
 	u8			stopping;
 	struct list_head	queue;
+	struct tasklet_struct	tasklet;
 	struct spi_transfer	*current_transfer;
 	unsigned long		current_remaining_bytes;
 	struct spi_transfer	*next_transfer;
 	unsigned long		next_remaining_bytes;
+	int			done_status;
 
+	/* scratch buffer */
 	void			*buffer;
 	dma_addr_t		buffer_dma;
+
+	struct atmel_spi_caps	caps;
+
+	bool			use_dma;
+	bool			use_pdc;
+	/* dmaengine data */
+	struct atmel_spi_dma	dma;
 };
 
 /* Controller-specific per-slave state */
@@ -222,14 +259,10 @@ struct atmel_spi_device {
  *  - SPI_SR.TXEMPTY, SPI_SR.NSSR (and corresponding irqs)
  *  - SPI_CSRx.CSAAT
  *  - SPI_CSRx.SBCR allows faster clocking
- *
- * We can determine the controller version by reading the VERSION
- * register, but I haven't checked that it exists on all chips, and
- * this is cheaper anyway.
  */
-static bool atmel_spi_is_v2(void)
+static bool atmel_spi_is_v2(struct atmel_spi *as)
 {
-	return !cpu_is_at91rm9200();
+	return as->caps.is_spi2;
 }
 
 /*
@@ -250,11 +283,6 @@ static bool atmel_spi_is_v2(void)
  * Master on Chip Select 0.")  No workaround exists for that ... so for
  * nCS0 on that chip, we (a) don't use the GPIO, (b) can't support CS_HIGH,
  * and (c) will trigger that first erratum in some cases.
- *
- * TODO: Test if the atmel_spi_is_v2() branch below works on
- * AT91RM9200 if we use some other register than CSR0. However, don't
- * do this unconditionally since AP7000 has an errata where the BITS
- * field in CSR0 overrides all other CSRs.
  */
 
 static void cs_activate(struct atmel_spi *as, struct spi_device *spi)
@@ -263,15 +291,25 @@ static void cs_activate(struct atmel_spi *as, struct spi_device *spi)
 	unsigned active = spi->mode & SPI_CS_HIGH;
 	u32 mr;
 
-	if (atmel_spi_is_v2()) {
-		/*
-		 * Always use CSR0. This ensures that the clock
-		 * switches to the correct idle polarity before we
-		 * toggle the CS.
+	if (atmel_spi_is_v2(as)) {
+		spi_writel(as, CSR0 + 4 * spi->chip_select, asd->csr);
+		/* For the low SPI version, there is a issue that PDC transfer
+		 * on CS1,2,3 needs SPI_CSR0.BITS config as SPI_CSR1,2,3.BITS
 		 */
 		spi_writel(as, CSR0, asd->csr);
-		spi_writel(as, MR, SPI_BF(PCS, 0x0e) | SPI_BIT(MODFDIS)
-				| SPI_BIT(MSTR));
+		if (as->caps.has_wdrbt) {
+			spi_writel(as, MR,
+					SPI_BF(PCS, ~(0x01 << spi->chip_select))
+					| SPI_BIT(WDRBT)
+					| SPI_BIT(MODFDIS)
+					| SPI_BIT(MSTR));
+		} else {
+			spi_writel(as, MR,
+					SPI_BF(PCS, ~(0x01 << spi->chip_select))
+					| SPI_BIT(MODFDIS)
+					| SPI_BIT(MSTR));
+		}
+
 		mr = spi_readl(as, MR);
 		gpio_set_value(asd->npcs_pin, active);
 	} else {
@@ -318,10 +356,26 @@ static void cs_deactivate(struct atmel_spi *as, struct spi_device *spi)
 			asd->npcs_pin, active ? " (low)" : "",
 			mr);
 
-	if (atmel_spi_is_v2() || spi->chip_select != 0)
+	if (atmel_spi_is_v2(as) || spi->chip_select != 0)
 		gpio_set_value(asd->npcs_pin, !active);
 }
 
+static void atmel_spi_lock(struct atmel_spi *as)
+{
+	spin_lock_irqsave(&as->lock, as->flags);
+}
+
+static void atmel_spi_unlock(struct atmel_spi *as)
+{
+	spin_unlock_irqrestore(&as->lock, as->flags);
+}
+
+static inline bool atmel_spi_use_dma(struct atmel_spi *as,
+				struct spi_transfer *xfer)
+{
+	return as->use_dma && xfer->len >= DMA_MIN_BYTES;
+}
+
 static inline int atmel_spi_xfer_is_last(struct spi_message *msg,
 					struct spi_transfer *xfer)
 {
@@ -333,6 +387,265 @@ static inline int atmel_spi_xfer_can_be_chained(struct spi_transfer *xfer)
 	return xfer->delay_usecs == 0 && !xfer->cs_change;
 }
 
+static int atmel_spi_dma_slave_config(struct atmel_spi *as,
+				struct dma_slave_config *slave_config,
+				u8 bits_per_word)
+{
+	int err = 0;
+
+	if (bits_per_word > 8) {
+		slave_config->dst_addr_width = DMA_SLAVE_BUSWIDTH_2_BYTES;
+		slave_config->src_addr_width = DMA_SLAVE_BUSWIDTH_2_BYTES;
+	} else {
+		slave_config->dst_addr_width = DMA_SLAVE_BUSWIDTH_1_BYTE;
+		slave_config->src_addr_width = DMA_SLAVE_BUSWIDTH_1_BYTE;
+	}
+
+	slave_config->dst_addr = (dma_addr_t)as->phybase + SPI_TDR;
+	slave_config->src_addr = (dma_addr_t)as->phybase + SPI_RDR;
+	slave_config->src_maxburst = 1;
+	slave_config->dst_maxburst = 1;
+	slave_config->device_fc = false;
+
+	slave_config->direction = DMA_MEM_TO_DEV;
+	if (dmaengine_slave_config(as->dma.chan_tx, slave_config)) {
+		dev_err(&as->pdev->dev,
+			"failed to configure tx dma channel\n");
+		err = -EINVAL;
+	}
+
+	slave_config->direction = DMA_DEV_TO_MEM;
+	if (dmaengine_slave_config(as->dma.chan_rx, slave_config)) {
+		dev_err(&as->pdev->dev,
+			"failed to configure rx dma channel\n");
+		err = -EINVAL;
+	}
+
+	return err;
+}
+
+static bool filter(struct dma_chan *chan, void *slave)
+{
+	struct	at_dma_slave *sl = slave;
+
+	if (sl->dma_dev == chan->device->dev) {
+		chan->private = sl;
+		return true;
+	} else {
+		return false;
+	}
+}
+
+static int atmel_spi_configure_dma(struct atmel_spi *as)
+{
+	struct at_dma_slave *sdata = &as->dma.dma_slave;
+	struct dma_slave_config	slave_config;
+	int err;
+
+	if (sdata && sdata->dma_dev) {
+		dma_cap_mask_t mask;
+
+		/* Try to grab two DMA channels */
+		dma_cap_zero(mask);
+		dma_cap_set(DMA_SLAVE, mask);
+		as->dma.chan_tx = dma_request_channel(mask, filter, sdata);
+		if (as->dma.chan_tx)
+			as->dma.chan_rx =
+				dma_request_channel(mask, filter, sdata);
+	}
+	if (!as->dma.chan_rx || !as->dma.chan_tx) {
+		dev_err(&as->pdev->dev,
+			"DMA channel not available, SPI unable to use DMA\n");
+		err = -EBUSY;
+		goto error;
+	}
+
+	err = atmel_spi_dma_slave_config(as, &slave_config, 8);
+	if (err)
+		goto error;
+
+	dev_info(&as->pdev->dev,
+			"Using %s (tx) and %s (rx) for DMA transfers\n",
+			dma_chan_name(as->dma.chan_tx),
+			dma_chan_name(as->dma.chan_rx));
+	return 0;
+error:
+	if (as->dma.chan_rx)
+		dma_release_channel(as->dma.chan_rx);
+	if (as->dma.chan_tx)
+		dma_release_channel(as->dma.chan_tx);
+	return err;
+}
+
+static void atmel_spi_stop_dma(struct atmel_spi *as)
+{
+	if (as->dma.chan_rx)
+		as->dma.chan_rx->device->device_control(as->dma.chan_rx,
+							DMA_TERMINATE_ALL, 0);
+	if (as->dma.chan_tx)
+		as->dma.chan_tx->device->device_control(as->dma.chan_tx,
+							DMA_TERMINATE_ALL, 0);
+}
+
+static void atmel_spi_release_dma(struct atmel_spi *as)
+{
+	if (as->dma.chan_rx)
+		dma_release_channel(as->dma.chan_rx);
+	if (as->dma.chan_tx)
+		dma_release_channel(as->dma.chan_tx);
+}
+
+/* This function is called by the DMA driver from tasklet context */
+static void dma_callback(void *data)
+{
+	struct spi_master	*master = data;
+	struct atmel_spi	*as = spi_master_get_devdata(master);
+
+	/* trigger SPI tasklet */
+	tasklet_schedule(&as->tasklet);
+}
+
+/*
+ * Next transfer using PIO.
+ * lock is held, spi tasklet is blocked
+ */
+static void atmel_spi_next_xfer_pio(struct spi_master *master,
+				struct spi_transfer *xfer)
+{
+	struct atmel_spi	*as = spi_master_get_devdata(master);
+
+	dev_vdbg(master->dev.parent, "atmel_spi_next_xfer_pio\n");
+
+	as->current_remaining_bytes = xfer->len;
+
+	/* Make sure data is not remaining in RDR */
+	spi_readl(as, RDR);
+	while (spi_readl(as, SR) & SPI_BIT(RDRF)) {
+		spi_readl(as, RDR);
+		cpu_relax();
+	}
+
+	if (xfer->tx_buf)
+		spi_writel(as, TDR, *(u8 *)(xfer->tx_buf));
+	else
+		spi_writel(as, TDR, 0);
+
+	dev_dbg(master->dev.parent,
+		"  start pio xfer %p: len %u tx %p rx %p\n",
+		xfer, xfer->len, xfer->tx_buf, xfer->rx_buf);
+
+	/* Enable relevant interrupts */
+	spi_writel(as, IER, SPI_BIT(RDRF) | SPI_BIT(OVRES));
+}
+
+/*
+ * Submit next transfer for DMA.
+ * lock is held, spi tasklet is blocked
+ */
+static int atmel_spi_next_xfer_dma_submit(struct spi_master *master,
+				struct spi_transfer *xfer,
+				u32 *plen)
+{
+	struct atmel_spi	*as = spi_master_get_devdata(master);
+	struct dma_chan		*rxchan = as->dma.chan_rx;
+	struct dma_chan		*txchan = as->dma.chan_tx;
+	struct dma_async_tx_descriptor *rxdesc;
+	struct dma_async_tx_descriptor *txdesc;
+	struct dma_slave_config	slave_config;
+	dma_cookie_t		cookie;
+	u32	len = *plen;
+
+	dev_vdbg(master->dev.parent, "atmel_spi_next_xfer_dma_submit\n");
+
+	/* Check that the channels are available */
+	if (!rxchan || !txchan)
+		return -ENODEV;
+
+	/* release lock for DMA operations */
+	atmel_spi_unlock(as);
+
+	/* prepare the RX dma transfer */
+	sg_init_table(&as->dma.sgrx, 1);
+	if (xfer->rx_buf) {
+		as->dma.sgrx.dma_address = xfer->rx_dma + xfer->len - *plen;
+	} else {
+		as->dma.sgrx.dma_address = as->buffer_dma;
+		if (len > BUFFER_SIZE)
+			len = BUFFER_SIZE;
+	}
+
+	/* prepare the TX dma transfer */
+	sg_init_table(&as->dma.sgtx, 1);
+	if (xfer->tx_buf) {
+		as->dma.sgtx.dma_address = xfer->tx_dma + xfer->len - *plen;
+	} else {
+		as->dma.sgtx.dma_address = as->buffer_dma;
+		if (len > BUFFER_SIZE)
+			len = BUFFER_SIZE;
+		memset(as->buffer, 0, len);
+	}
+
+	sg_dma_len(&as->dma.sgtx) = len;
+	sg_dma_len(&as->dma.sgrx) = len;
+
+	*plen = len;
+
+	if (atmel_spi_dma_slave_config(as, &slave_config, 8))
+		goto err_exit;
+
+	/* Send both scatterlists */
+	rxdesc = rxchan->device->device_prep_slave_sg(rxchan,
+					&as->dma.sgrx,
+					1,
+					DMA_FROM_DEVICE,
+					DMA_PREP_INTERRUPT | DMA_CTRL_ACK,
+					NULL);
+	if (!rxdesc)
+		goto err_dma;
+
+	txdesc = txchan->device->device_prep_slave_sg(txchan,
+					&as->dma.sgtx,
+					1,
+					DMA_TO_DEVICE,
+					DMA_PREP_INTERRUPT | DMA_CTRL_ACK,
+					NULL);
+	if (!txdesc)
+		goto err_dma;
+
+	dev_dbg(master->dev.parent,
+		"  start dma xfer %p: len %u tx %p/%08x rx %p/%08x\n",
+		xfer, xfer->len, xfer->tx_buf, xfer->tx_dma,
+		xfer->rx_buf, xfer->rx_dma);
+
+	/* Enable relevant interrupts */
+	spi_writel(as, IER, SPI_BIT(OVRES));
+
+	/* Put the callback on the RX transfer only, that should finish last */
+	rxdesc->callback = dma_callback;
+	rxdesc->callback_param = master;
+
+	/* Submit and fire RX and TX with TX last so we're ready to read! */
+	cookie = rxdesc->tx_submit(rxdesc);
+	if (dma_submit_error(cookie))
+		goto err_dma;
+	cookie = txdesc->tx_submit(txdesc);
+	if (dma_submit_error(cookie))
+		goto err_dma;
+	rxchan->device->device_issue_pending(rxchan);
+	txchan->device->device_issue_pending(txchan);
+
+	/* take back lock */
+	atmel_spi_lock(as);
+	return 0;
+
+err_dma:
+	spi_writel(as, IDR, SPI_BIT(OVRES));
+	atmel_spi_stop_dma(as);
+err_exit:
+	atmel_spi_lock(as);
+	return -ENOMEM;
+}
+
 static void atmel_spi_next_xfer_data(struct spi_master *master,
 				struct spi_transfer *xfer,
 				dma_addr_t *tx_dma,
@@ -350,6 +663,7 @@ static void atmel_spi_next_xfer_data(struct spi_master *master,
 		if (len > BUFFER_SIZE)
 			len = BUFFER_SIZE;
 	}
+
 	if (xfer->tx_buf)
 		*tx_dma = xfer->tx_dma + xfer->len - *plen;
 	else {
@@ -365,10 +679,10 @@ static void atmel_spi_next_xfer_data(struct spi_master *master,
 }
 
 /*
- * Submit next transfer for DMA.
+ * Submit next transfer for PDC.
  * lock is held, spi irq is blocked
  */
-static void atmel_spi_next_xfer(struct spi_master *master,
+static void atmel_spi_pdc_next_xfer(struct spi_master *master,
 				struct spi_message *msg)
 {
 	struct atmel_spi	*as = spi_master_get_devdata(master);
@@ -465,6 +779,48 @@ static void atmel_spi_next_xfer(struct spi_master *master,
 	spi_writel(as, PTCR, SPI_BIT(TXTEN) | SPI_BIT(RXTEN));
 }
 
+/*
+ * Choose way to submit next transfer and start it.
+ * lock is held, spi tasklet is blocked
+ */
+static void atmel_spi_dma_next_xfer(struct spi_master *master,
+				struct spi_message *msg)
+{
+	struct atmel_spi	*as = spi_master_get_devdata(master);
+	struct spi_transfer	*xfer;
+	u32	remaining, len;
+
+	remaining = as->current_remaining_bytes;
+	if (remaining) {
+		xfer = as->current_transfer;
+		len = remaining;
+	} else {
+		if (!as->current_transfer)
+			xfer = list_entry(msg->transfers.next,
+				struct spi_transfer, transfer_list);
+		else
+			xfer = list_entry(
+				as->current_transfer->transfer_list.next,
+					struct spi_transfer, transfer_list);
+
+		as->current_transfer = xfer;
+		len = xfer->len;
+	}
+
+	if (atmel_spi_use_dma(as, xfer)) {
+		u32 total = len;
+		if (!atmel_spi_next_xfer_dma_submit(master, xfer, &len)) {
+			as->current_remaining_bytes = total - len;
+			return;
+		} else {
+			dev_err(&msg->spi->dev, "unable to use DMA, fallback to PIO\n");
+		}
+	}
+
+	/* use PIO if error appened using DMA */
+	atmel_spi_next_xfer_pio(master, xfer);
+}
+
 static void atmel_spi_next_message(struct spi_master *master)
 {
 	struct atmel_spi	*as = spi_master_get_devdata(master);
@@ -489,7 +845,10 @@ static void atmel_spi_next_message(struct spi_master *master)
 	} else
 		cs_activate(as, spi);
 
-	atmel_spi_next_xfer(master, msg);
+	if (as->use_pdc)
+		atmel_spi_pdc_next_xfer(master, msg);
+	else
+		atmel_spi_dma_next_xfer(master, msg);
 }
 
 /*
@@ -542,38 +901,213 @@ static void atmel_spi_dma_unmap_xfer(struct spi_master *master,
 				 xfer->len, DMA_FROM_DEVICE);
 }
 
+static void atmel_spi_disable_pdc_transfer(struct atmel_spi *as)
+{
+	spi_writel(as, PTCR, SPI_BIT(RXTDIS) | SPI_BIT(TXTDIS));
+}
+
 static void
 atmel_spi_msg_done(struct spi_master *master, struct atmel_spi *as,
-		struct spi_message *msg, int status, int stay)
+		struct spi_message *msg, int stay)
 {
-	if (!stay || status < 0)
+	if (!stay || as->done_status < 0)
 		cs_deactivate(as, msg->spi);
 	else
 		as->stay = msg->spi;
 
 	list_del(&msg->queue);
-	msg->status = status;
+	msg->status = as->done_status;
 
 	dev_dbg(master->dev.parent,
 		"xfer complete: %u bytes transferred\n",
 		msg->actual_length);
 
-	spin_unlock(&as->lock);
+	atmel_spi_unlock(as);
 	msg->complete(msg->context);
-	spin_lock(&as->lock);
+	atmel_spi_lock(as);
 
 	as->current_transfer = NULL;
 	as->next_transfer = NULL;
+	as->done_status = 0;
 
 	/* continue if needed */
-	if (list_empty(&as->queue) || as->stopping)
-		spi_writel(as, PTCR, SPI_BIT(RXTDIS) | SPI_BIT(TXTDIS));
-	else
+	if (list_empty(&as->queue) || as->stopping) {
+		if (as->use_pdc)
+			atmel_spi_disable_pdc_transfer(as);
+	} else {
 		atmel_spi_next_message(master);
+	}
+}
+
+/* Called from IRQ
+ * lock is held
+ *
+ * Must update "current_remaining_bytes" to keep track of data
+ * to transfer.
+ */
+static void
+atmel_spi_pump_pio_data(struct atmel_spi *as, struct spi_transfer *xfer)
+{
+	u8		*txp;
+	u8		*rxp;
+	unsigned long	xfer_pos = xfer->len - as->current_remaining_bytes;
+
+	if (xfer->rx_buf) {
+		rxp = ((u8 *)xfer->rx_buf) + xfer_pos;
+		*rxp = spi_readl(as, RDR);
+	} else {
+		spi_readl(as, RDR);
+	}
+
+	as->current_remaining_bytes--;
+
+	if (as->current_remaining_bytes) {
+		if (xfer->tx_buf) {
+			txp = ((u8 *)xfer->tx_buf) + xfer_pos + 1;
+			spi_writel(as, TDR, *txp);
+		} else {
+			spi_writel(as, TDR, 0);
+		}
+	}
+}
+
+/* Tasklet
+ * Called from DMA callback + pio transfer and overrun IRQ.
+ */
+static void atmel_spi_tasklet_func(unsigned long data)
+{
+	struct spi_master	*master = (struct spi_master *)data;
+	struct atmel_spi	*as = spi_master_get_devdata(master);
+	struct spi_message	*msg;
+	struct spi_transfer	*xfer;
+
+	dev_vdbg(master->dev.parent, "atmel_spi_tasklet_func\n");
+
+	atmel_spi_lock(as);
+
+	xfer = as->current_transfer;
+
+	if (xfer == NULL)
+		/* already been there */
+		goto tasklet_out;
+
+	msg = list_entry(as->queue.next, struct spi_message, queue);
+
+	if (as->current_remaining_bytes == 0) {
+		if (as->done_status < 0) {
+			/* error happened (overrun) */
+			if (atmel_spi_use_dma(as, xfer))
+				atmel_spi_stop_dma(as);
+		} else {
+			/* only update length if no error */
+			msg->actual_length += xfer->len;
+		}
+
+		if (atmel_spi_use_dma(as, xfer))
+			if (!msg->is_dma_mapped)
+				atmel_spi_dma_unmap_xfer(master, xfer);
+
+		if (xfer->delay_usecs)
+			udelay(xfer->delay_usecs);
+
+		if (atmel_spi_xfer_is_last(msg, xfer) || as->done_status < 0) {
+			/* report completed (or erroneous) message */
+			atmel_spi_msg_done(master, as, msg, xfer->cs_change);
+		} else {
+			if (xfer->cs_change) {
+				cs_deactivate(as, msg->spi);
+				udelay(1);
+				cs_activate(as, msg->spi);
+			}
+
+			/*
+			 * Not done yet. Submit the next transfer.
+			 *
+			 * FIXME handle protocol options for xfer
+			 */
+			atmel_spi_dma_next_xfer(master, msg);
+		}
+	} else {
+		/*
+		 * Keep going, we still have data to send in
+		 * the current transfer.
+		 */
+		atmel_spi_dma_next_xfer(master, msg);
+	}
+
+tasklet_out:
+	atmel_spi_unlock(as);
 }
 
+/* Interrupt
+ *
+ * No need for locking in this Interrupt handler: done_status is the
+ * only information modified. What we need is the update of this field
+ * before tasklet runs. This is ensured by using barrier.
+ */
 static irqreturn_t
-atmel_spi_interrupt(int irq, void *dev_id)
+atmel_spi_pio_interrupt(int irq, void *dev_id)
+{
+	struct spi_master	*master = dev_id;
+	struct atmel_spi	*as = spi_master_get_devdata(master);
+	u32			status, pending, imr;
+	struct spi_transfer	*xfer;
+	int			ret = IRQ_NONE;
+
+	imr = spi_readl(as, IMR);
+	status = spi_readl(as, SR);
+	pending = status & imr;
+
+	if (pending & SPI_BIT(OVRES)) {
+		ret = IRQ_HANDLED;
+		spi_writel(as, IDR, SPI_BIT(OVRES));
+		dev_warn(master->dev.parent, "overrun\n");
+
+		/*
+		 * When we get an overrun, we disregard the current
+		 * transfer. Data will not be copied back from any
+		 * bounce buffer and msg->actual_len will not be
+		 * updated with the last xfer.
+		 *
+		 * We will also not process any remaning transfers in
+		 * the message.
+		 *
+		 * All actions are done in tasklet with done_status indication
+		 */
+		as->done_status = -EIO;
+		smp_wmb();
+
+		/* Clear any overrun happening while cleaning up */
+		spi_readl(as, SR);
+
+		tasklet_schedule(&as->tasklet);
+
+	} else if (pending & SPI_BIT(RDRF)) {
+		atmel_spi_lock(as);
+
+		if (as->current_remaining_bytes) {
+			ret = IRQ_HANDLED;
+			xfer = as->current_transfer;
+			atmel_spi_pump_pio_data(as, xfer);
+			if (!as->current_remaining_bytes) {
+				/* no more data to xfer, kick tasklet */
+				spi_writel(as, IDR, pending);
+				tasklet_schedule(&as->tasklet);
+			}
+		}
+
+		atmel_spi_unlock(as);
+	} else {
+		WARN_ONCE(pending, "IRQ not handled, pending = %x\n", pending);
+		ret = IRQ_HANDLED;
+		spi_writel(as, IDR, pending);
+	}
+
+	return ret;
+}
+
+static irqreturn_t
+atmel_spi_pdc_interrupt(int irq, void *dev_id)
 {
 	struct spi_master	*master = dev_id;
 	struct atmel_spi	*as = spi_master_get_devdata(master);
@@ -582,7 +1116,7 @@ atmel_spi_interrupt(int irq, void *dev_id)
 	u32			status, pending, imr;
 	int			ret = IRQ_NONE;
 
-	spin_lock(&as->lock);
+	atmel_spi_lock(as);
 
 	xfer = as->current_transfer;
 	msg = list_entry(as->queue.next, struct spi_message, queue);
@@ -641,7 +1175,8 @@ atmel_spi_interrupt(int irq, void *dev_id)
 		/* Clear any overrun happening while cleaning up */
 		spi_readl(as, SR);
 
-		atmel_spi_msg_done(master, as, msg, -EIO, 0);
+		as->done_status = -EIO;
+		atmel_spi_msg_done(master, as, msg, 0);
 	} else if (pending & (SPI_BIT(RXBUFF) | SPI_BIT(ENDRX))) {
 		ret = IRQ_HANDLED;
 
@@ -659,7 +1194,7 @@ atmel_spi_interrupt(int irq, void *dev_id)
 
 			if (atmel_spi_xfer_is_last(msg, xfer)) {
 				/* report completed message */
-				atmel_spi_msg_done(master, as, msg, 0,
+				atmel_spi_msg_done(master, as, msg,
 						xfer->cs_change);
 			} else {
 				if (xfer->cs_change) {
@@ -673,18 +1208,18 @@ atmel_spi_interrupt(int irq, void *dev_id)
 				 *
 				 * FIXME handle protocol options for xfer
 				 */
-				atmel_spi_next_xfer(master, msg);
+				atmel_spi_pdc_next_xfer(master, msg);
 			}
 		} else {
 			/*
 			 * Keep going, we still have data to send in
 			 * the current transfer.
 			 */
-			atmel_spi_next_xfer(master, msg);
+			atmel_spi_pdc_next_xfer(master, msg);
 		}
 	}
 
-	spin_unlock(&as->lock);
+	atmel_spi_unlock(as);
 
 	return ret;
 }
@@ -719,7 +1254,7 @@ static int atmel_spi_setup(struct spi_device *spi)
 	}
 
 	/* see notes above re chipselect */
-	if (!atmel_spi_is_v2()
+	if (!atmel_spi_is_v2(as)
 			&am