Merge tag 'spi-v4.16' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi

Pull spi updates from Mark Brown:
 "Quite a quiet release for SPI, there are no changes at all to the core
  and not that many changes to drivers. Highlights of those driver
  changes include:

   - SH MSIOF support for GPIO chip selects contributed by Geert
     Uytterhoeven.

   - Full duplex support for a3700 contributed by Maxime Chevallier.

   - Support for DMA transfers on Atmel devices that require a bounce
     buffer contributed by Radu Pirea"

* tag 'spi-v4.16' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi: (31 commits)
  spi: dw: Remove unused members from struct chip_data
  spi: orion: Fix a resource leak if the optional "axi" clk is deferred
  spi: a3700: Remove endianness swapping for full-duplex transfers
  spi: a3700: Remove endianness swapping functions when accessing FIFOs
  spi: a3700: Add full-duplex support
  spi: a3700: Allow to enable or disable FIFO mode
  spi: a3700: Set frequency limits at startup
  spi: a3700: Clear DATA_OUT when performing a read
  spi: orion: Fix clock resource by adding an optional bus clock
  spi: s3c64xx: add SPDX identifier
  spi: imx: do not access registers while clocks disabled
  spi: atmel: Implements transfers with bounce buffer
  spi: sh-msiof: Fix timeout failures for TX-only DMA transfers
  spi: spi-fsl-dspi: account for const type of of_device_id.data
  spi: bcm53xx: simplify reading SPI data
  spi: sirf: account for const type of of_device_id.data
  spi: pxa2xx: Use gpiod_put() not gpiod_free()
  spi: pxa2xx: avoid redundant gpio_to_desc(desc_to_gpio()) round-trip
  spi: sh-msiof: Document hardware limitations related to chip selects
  spi: sh-msiof: Implement cs-gpios configuration
  ...

16 files changed, 348 insertions, 125 deletions

diff --git a/drivers/spi/spi-armada-3700.c b/drivers/spi/spi-armada-3700.c
index d65345312527..7dcb14d303eb 100644
--- a/drivers/spi/spi-armada-3700.c
+++ b/drivers/spi/spi-armada-3700.c
@@ -27,6 +27,8 @@
 
 #define DRIVER_NAME			"armada_3700_spi"
 
+#define A3700_SPI_MAX_SPEED_HZ		100000000
+#define A3700_SPI_MAX_PRESCALE		30
 #define A3700_SPI_TIMEOUT		10
 
 /* SPI Register Offest */
@@ -184,12 +186,15 @@ static int a3700_spi_pin_mode_set(struct a3700_spi *a3700_spi,
 	return 0;
 }
 
-static void a3700_spi_fifo_mode_set(struct a3700_spi *a3700_spi)
+static void a3700_spi_fifo_mode_set(struct a3700_spi *a3700_spi, bool enable)
 {
 	u32 val;
 
 	val = spireg_read(a3700_spi, A3700_SPI_IF_CFG_REG);
-	val |= A3700_SPI_FIFO_MODE;
+	if (enable)
+		val |= A3700_SPI_FIFO_MODE;
+	else
+		val &= ~A3700_SPI_FIFO_MODE;
 	spireg_write(a3700_spi, A3700_SPI_IF_CFG_REG, val);
 }
 
@@ -297,7 +302,7 @@ static int a3700_spi_init(struct a3700_spi *a3700_spi)
 		a3700_spi_deactivate_cs(a3700_spi, i);
 
 	/* Enable FIFO mode */
-	a3700_spi_fifo_mode_set(a3700_spi);
+	a3700_spi_fifo_mode_set(a3700_spi, true);
 
 	/* Set SPI mode */
 	a3700_spi_mode_set(a3700_spi, master->mode_bits);
@@ -416,15 +421,20 @@ static void a3700_spi_transfer_setup(struct spi_device *spi,
 				     struct spi_transfer *xfer)
 {
 	struct a3700_spi *a3700_spi;
-	unsigned int byte_len;
 
 	a3700_spi = spi_master_get_devdata(spi->master);
 
 	a3700_spi_clock_set(a3700_spi, xfer->speed_hz);
 
-	byte_len = xfer->bits_per_word >> 3;
+	/* Use 4 bytes long transfers. Each transfer method has its way to deal
+	 * with the remaining bytes for non 4-bytes aligned transfers.
+	 */
+	a3700_spi_bytelen_set(a3700_spi, 4);
 
-	a3700_spi_fifo_thres_set(a3700_spi, byte_len);
+	/* Initialize the working buffers */
+	a3700_spi->tx_buf  = xfer->tx_buf;
+	a3700_spi->rx_buf  = xfer->rx_buf;
+	a3700_spi->buf_len = xfer->len;
 }
 
 static void a3700_spi_set_cs(struct spi_device *spi, bool enable)
@@ -491,7 +501,7 @@ static int a3700_spi_fifo_write(struct a3700_spi *a3700_spi)
 	u32 val;
 
 	while (!a3700_is_wfifo_full(a3700_spi) && a3700_spi->buf_len) {
-		val = cpu_to_le32(*(u32 *)a3700_spi->tx_buf);
+		val = *(u32 *)a3700_spi->tx_buf;
 		spireg_write(a3700_spi, A3700_SPI_DATA_OUT_REG, val);
 		a3700_spi->buf_len -= 4;
 		a3700_spi->tx_buf += 4;
@@ -514,9 +524,8 @@ static int a3700_spi_fifo_read(struct a3700_spi *a3700_spi)
 	while (!a3700_is_rfifo_empty(a3700_spi) && a3700_spi->buf_len) {
 		val = spireg_read(a3700_spi, A3700_SPI_DATA_IN_REG);
 		if (a3700_spi->buf_len >= 4) {
-			u32 data = le32_to_cpu(val);
 
-			memcpy(a3700_spi->rx_buf, &data, 4);
+			memcpy(a3700_spi->rx_buf, &val, 4);
 
 			a3700_spi->buf_len -= 4;
 			a3700_spi->rx_buf += 4;
@@ -579,27 +588,26 @@ static int a3700_spi_prepare_message(struct spi_master *master,
 	if (ret)
 		return ret;
 
-	a3700_spi_bytelen_set(a3700_spi, 4);
-
 	a3700_spi_mode_set(a3700_spi, spi->mode);
 
 	return 0;
 }
 
-static int a3700_spi_transfer_one(struct spi_master *master,
+static int a3700_spi_transfer_one_fifo(struct spi_master *master,
 				  struct spi_device *spi,
 				  struct spi_transfer *xfer)
 {
 	struct a3700_spi *a3700_spi = spi_master_get_devdata(master);
 	int ret = 0, timeout = A3700_SPI_TIMEOUT;
-	unsigned int nbits = 0;
+	unsigned int nbits = 0, byte_len;
 	u32 val;
 
-	a3700_spi_transfer_setup(spi, xfer);
+	/* Make sure we use FIFO mode */
+	a3700_spi_fifo_mode_set(a3700_spi, true);
 
-	a3700_spi->tx_buf  = xfer->tx_buf;
-	a3700_spi->rx_buf  = xfer->rx_buf;
-	a3700_spi->buf_len = xfer->len;
+	/* Configure FIFO thresholds */
+	byte_len = xfer->bits_per_word >> 3;
+	a3700_spi_fifo_thres_set(a3700_spi, byte_len);
 
 	if (xfer->tx_buf)
 		nbits = xfer->tx_nbits;
@@ -615,6 +623,11 @@ static int a3700_spi_transfer_one(struct spi_master *master,
 	a3700_spi_header_set(a3700_spi);
 
 	if (xfer->rx_buf) {
+		/* Clear WFIFO, since it's last 2 bytes are shifted out during
+		 * a read operation
+		 */
+		spireg_write(a3700_spi, A3700_SPI_DATA_OUT_REG, 0);
+
 		/* Set read data length */
 		spireg_write(a3700_spi, A3700_SPI_IF_DIN_CNT_REG,
 			     a3700_spi->buf_len);
@@ -729,6 +742,63 @@ out:
 	return ret;
 }
 
+static int a3700_spi_transfer_one_full_duplex(struct spi_master *master,
+				  struct spi_device *spi,
+				  struct spi_transfer *xfer)
+{
+	struct a3700_spi *a3700_spi = spi_master_get_devdata(master);
+	u32 val;
+
+	/* Disable FIFO mode */
+	a3700_spi_fifo_mode_set(a3700_spi, false);
+
+	while (a3700_spi->buf_len) {
+
+		/* When we have less than 4 bytes to transfer, switch to 1 byte
+		 * mode. This is reset after each transfer
+		 */
+		if (a3700_spi->buf_len < 4)
+			a3700_spi_bytelen_set(a3700_spi, 1);
+
+		if (a3700_spi->byte_len == 1)
+			val = *a3700_spi->tx_buf;
+		else
+			val = *(u32 *)a3700_spi->tx_buf;
+
+		spireg_write(a3700_spi, A3700_SPI_DATA_OUT_REG, val);
+
+		/* Wait for all the data to be shifted in / out */
+		while (!(spireg_read(a3700_spi, A3700_SPI_IF_CTRL_REG) &
+				A3700_SPI_XFER_DONE))
+			cpu_relax();
+
+		val = spireg_read(a3700_spi, A3700_SPI_DATA_IN_REG);
+
+		memcpy(a3700_spi->rx_buf, &val, a3700_spi->byte_len);
+
+		a3700_spi->buf_len -= a3700_spi->byte_len;
+		a3700_spi->tx_buf += a3700_spi->byte_len;
+		a3700_spi->rx_buf += a3700_spi->byte_len;
+
+	}
+
+	spi_finalize_current_transfer(master);
+
+	return 0;
+}
+
+static int a3700_spi_transfer_one(struct spi_master *master,
+				  struct spi_device *spi,
+				  struct spi_transfer *xfer)
+{
+	a3700_spi_transfer_setup(spi, xfer);
+
+	if (xfer->tx_buf && xfer->rx_buf)
+		return a3700_spi_transfer_one_full_duplex(master, spi, xfer);
+
+	return a3700_spi_transfer_one_fifo(master, spi, xfer);
+}
+
 static int a3700_spi_unprepare_message(struct spi_master *master,
 				       struct spi_message *message)
 {
@@ -778,7 +848,6 @@ static int a3700_spi_probe(struct platform_device *pdev)
 	master->transfer_one = a3700_spi_transfer_one;
 	master->unprepare_message = a3700_spi_unprepare_message;
 	master->set_cs = a3700_spi_set_cs;
-	master->flags = SPI_MASTER_HALF_DUPLEX;
 	master->mode_bits |= (SPI_RX_DUAL | SPI_TX_DUAL |
 			      SPI_RX_QUAD | SPI_TX_QUAD);
 
@@ -818,6 +887,11 @@ static int a3700_spi_probe(struct platform_device *pdev)
 		goto error;
 	}
 
+	master->max_speed_hz = min_t(unsigned long, A3700_SPI_MAX_SPEED_HZ,
+					clk_get_rate(spi->clk));
+	master->min_speed_hz = DIV_ROUND_UP(clk_get_rate(spi->clk),
+						A3700_SPI_MAX_PRESCALE);
+
 	ret = a3700_spi_init(spi);
 	if (ret)
 		goto error_clk;
diff --git a/drivers/spi/spi-atmel.c b/drivers/spi/spi-atmel.c
index 669470971023..4a11fc0d4136 100644
--- a/drivers/spi/spi-atmel.c
+++ b/drivers/spi/spi-atmel.c
@@ -291,6 +291,10 @@ struct atmel_spi {
 	struct spi_transfer	*current_transfer;
 	int			current_remaining_bytes;
 	int			done_status;
+	dma_addr_t		dma_addr_rx_bbuf;
+	dma_addr_t		dma_addr_tx_bbuf;
+	void			*addr_rx_bbuf;
+	void			*addr_tx_bbuf;
 
 	struct completion	xfer_completion;
 
@@ -436,6 +440,11 @@ static void atmel_spi_unlock(struct atmel_spi *as) __releases(&as->lock)
 	spin_unlock_irqrestore(&as->lock, as->flags);
 }
 
+static inline bool atmel_spi_is_vmalloc_xfer(struct spi_transfer *xfer)
+{
+	return is_vmalloc_addr(xfer->tx_buf) || is_vmalloc_addr(xfer->rx_buf);
+}
+
 static inline bool atmel_spi_use_dma(struct atmel_spi *as,
 				struct spi_transfer *xfer)
 {
@@ -448,7 +457,12 @@ static bool atmel_spi_can_dma(struct spi_master *master,
 {
 	struct atmel_spi *as = spi_master_get_devdata(master);
 
-	return atmel_spi_use_dma(as, xfer);
+	if (IS_ENABLED(CONFIG_SOC_SAM_V4_V5))
+		return atmel_spi_use_dma(as, xfer) &&
+			!atmel_spi_is_vmalloc_xfer(xfer);
+	else
+		return atmel_spi_use_dma(as, xfer);
+
 }
 
 static int atmel_spi_dma_slave_config(struct atmel_spi *as,
@@ -594,6 +608,11 @@ static void dma_callback(void *data)
 	struct spi_master	*master = data;
 	struct atmel_spi	*as = spi_master_get_devdata(master);
 
+	if (is_vmalloc_addr(as->current_transfer->rx_buf) &&
+	    IS_ENABLED(CONFIG_SOC_SAM_V4_V5)) {
+		memcpy(as->current_transfer->rx_buf, as->addr_rx_bbuf,
+		       as->current_transfer->len);
+	}
 	complete(&as->xfer_completion);
 }
 
@@ -744,17 +763,41 @@ static int atmel_spi_next_xfer_dma_submit(struct spi_master *master,
 		goto err_exit;
 
 	/* Send both scatterlists */
-	rxdesc = dmaengine_prep_slave_sg(rxchan,
-					 xfer->rx_sg.sgl, xfer->rx_sg.nents,
-					 DMA_FROM_DEVICE,
-					 DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
+	if (atmel_spi_is_vmalloc_xfer(xfer) &&
+	    IS_ENABLED(CONFIG_SOC_SAM_V4_V5)) {
+		rxdesc = dmaengine_prep_slave_single(rxchan,
+						     as->dma_addr_rx_bbuf,
+						     xfer->len,
+						     DMA_FROM_DEVICE,
+						     DMA_PREP_INTERRUPT |
+						     DMA_CTRL_ACK);
+	} else {
+		rxdesc = dmaengine_prep_slave_sg(rxchan,
+						 xfer->rx_sg.sgl,
+						 xfer->rx_sg.nents,
+						 DMA_FROM_DEVICE,
+						 DMA_PREP_INTERRUPT |
+						 DMA_CTRL_ACK);
+	}
 	if (!rxdesc)
 		goto err_dma;
 
-	txdesc = dmaengine_prep_slave_sg(txchan,
-					 xfer->tx_sg.sgl, xfer->tx_sg.nents,
-					 DMA_TO_DEVICE,
-					 DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
+	if (atmel_spi_is_vmalloc_xfer(xfer) &&
+	    IS_ENABLED(CONFIG_SOC_SAM_V4_V5)) {
+		memcpy(as->addr_tx_bbuf, xfer->tx_buf, xfer->len);
+		txdesc = dmaengine_prep_slave_single(txchan,
+						     as->dma_addr_tx_bbuf,
+						     xfer->len, DMA_TO_DEVICE,
+						     DMA_PREP_INTERRUPT |
+						     DMA_CTRL_ACK);
+	} else {
+		txdesc = dmaengine_prep_slave_sg(txchan,
+						 xfer->tx_sg.sgl,
+						 xfer->tx_sg.nents,
+						 DMA_TO_DEVICE,
+						 DMA_PREP_INTERRUPT |
+						 DMA_CTRL_ACK);
+	}
 	if (!txdesc)
 		goto err_dma;
 
@@ -1426,27 +1469,7 @@ static void atmel_get_caps(struct atmel_spi *as)
 
 	as->caps.is_spi2 = version > 0x121;
 	as->caps.has_wdrbt = version >= 0x210;
-#ifdef CONFIG_SOC_SAM_V4_V5
-	/*
-	 * Atmel SoCs based on ARM9 (SAM9x) cores should not use spi_map_buf()
-	 * since this later function tries to map buffers with dma_map_sg()
-	 * even if they have not been allocated inside DMA-safe areas.
-	 * On SoCs based on Cortex A5 (SAMA5Dx), it works anyway because for
-	 * those ARM cores, the data cache follows the PIPT model.
-	 * Also the L2 cache controller of SAMA5D2 uses the PIPT model too.
-	 * In case of PIPT caches, there cannot be cache aliases.
-	 * However on ARM9 cores, the data cache follows the VIVT model, hence
-	 * the cache aliases issue can occur when buffers are allocated from
-	 * DMA-unsafe areas, by vmalloc() for instance, where cache coherency is
-	 * not taken into account or at least not handled completely (cache
-	 * lines of aliases are not invalidated).
-	 * This is not a theorical issue: it was reproduced when trying to mount
-	 * a UBI file-system on a at91sam9g35ek board.
-	 */
-	as->caps.has_dma_support = false;
-#else
 	as->caps.has_dma_support = version >= 0x212;
-#endif
 	as->caps.has_pdc_support = version < 0x212;
 }
 
@@ -1592,6 +1615,30 @@ static int atmel_spi_probe(struct platform_device *pdev)
 		as->use_pdc = true;
 	}
 
+	if (IS_ENABLED(CONFIG_SOC_SAM_V4_V5)) {
+		as->addr_rx_bbuf = dma_alloc_coherent(&pdev->dev,
+						      SPI_MAX_DMA_XFER,
+						      &as->dma_addr_rx_bbuf,
+						      GFP_KERNEL | GFP_DMA);
+		if (!as->addr_rx_bbuf) {
+			as->use_dma = false;
+		} else {
+			as->addr_tx_bbuf = dma_alloc_coherent(&pdev->dev,
+					SPI_MAX_DMA_XFER,
+					&as->dma_addr_tx_bbuf,
+					GFP_KERNEL | GFP_DMA);
+			if (!as->addr_tx_bbuf) {
+				as->use_dma = false;
+				dma_free_coherent(&pdev->dev, SPI_MAX_DMA_XFER,
+						  as->addr_rx_bbuf,
+						  as->dma_addr_rx_bbuf);
+			}
+		}
+		if (!as->use_dma)
+			dev_info(master->dev.parent,
+				 "  can not allocate dma coherent memory\n");
+	}
+
 	if (as->caps.has_dma_support && !as->use_dma)
 		dev_info(&pdev->dev, "Atmel SPI Controller using PIO only\n");
 
@@ -1664,6 +1711,14 @@ static int atmel_spi_remove(struct platform_device *pdev)
 	if (as->use_dma) {
 		atmel_spi_stop_dma(master);
 		atmel_spi_release_dma(master);
+		if (IS_ENABLED(CONFIG_SOC_SAM_V4_V5)) {
+			dma_free_coherent(&pdev->dev, SPI_MAX_DMA_XFER,
+					  as->addr_tx_bbuf,
+					  as->dma_addr_tx_bbuf);
+			dma_free_coherent(&pdev->dev, SPI_MAX_DMA_XFER,
+					  as->addr_rx_bbuf,
+					  as->dma_addr_rx_bbuf);
+		}
 	}
 
 	spin_lock_irq(&as->lock);
diff --git a/drivers/spi/spi-bcm53xx.c b/drivers/spi/spi-bcm53xx.c
index 6e409eabe1c9..d02ceb7a29d1 100644
--- a/drivers/spi/spi-bcm53xx.c
+++ b/drivers/spi/spi-bcm53xx.c
@@ -27,8 +27,6 @@ struct bcm53xxspi {
 	struct bcma_device *core;
 	struct spi_master *master;
 	void __iomem *mmio_base;
-
-	size_t read_offset;
 	bool bspi;				/* Boot SPI mode with memory mapping */
 };
 
@@ -172,8 +170,6 @@ static void bcm53xxspi_buf_write(struct bcm53xxspi *b53spi, u8 *w_buf,
 
 	if (!cont)
 		bcm53xxspi_write(b53spi, B53SPI_MSPI_WRITE_LOCK, 0);
-
-	b53spi->read_offset = len;
 }
 
 static void bcm53xxspi_buf_read(struct bcm53xxspi *b53spi, u8 *r_buf,
@@ -182,10 +178,10 @@ static void bcm53xxspi_buf_read(struct bcm53xxspi *b53spi, u8 *r_buf,
 	u32 tmp;
 	int i;
 
-	for (i = 0; i < b53spi->read_offset + len; i++) {
+	for (i = 0; i < len; i++) {
 		tmp = B53SPI_CDRAM_CONT | B53SPI_CDRAM_PCS_DISABLE_ALL |
 		      B53SPI_CDRAM_PCS_DSCK;
-		if (!cont && i == b53spi->read_offset + len - 1)
+		if (!cont && i == len - 1)
 			tmp &= ~B53SPI_CDRAM_CONT;
 		tmp &= ~0x1;
 		/* Command Register File */
@@ -194,8 +190,7 @@ static void bcm53xxspi_buf_read(struct bcm53xxspi *b53spi, u8 *r_buf,
 
 	/* Set queue pointers */
 	bcm53xxspi_write(b53spi, B53SPI_MSPI_NEWQP, 0);
-	bcm53xxspi_write(b53spi, B53SPI_MSPI_ENDQP,
-			 b53spi->read_offset + len - 1);
+	bcm53xxspi_write(b53spi, B53SPI_MSPI_ENDQP, len - 1);
 
 	if (cont)
 		bcm53xxspi_write(b53spi, B53SPI_MSPI_WRITE_LOCK, 1);
@@ -214,13 +209,11 @@ static void bcm53xxspi_buf_read(struct bcm53xxspi *b53spi, u8 *r_buf,
 		bcm53xxspi_write(b53spi, B53SPI_MSPI_WRITE_LOCK, 0);
 
 	for (i = 0; i < len; ++i) {
-		int offset = b53spi->read_offset + i;
+		u16 reg = B53SPI_MSPI_RXRAM + 4 * (1 + i * 2);
 
 		/* Data stored in the transmit register file LSB */
-		r_buf[i] = (u8)bcm53xxspi_read(b53spi, B53SPI_MSPI_RXRAM + 4 * (1 + offset * 2));
+		r_buf[i] = (u8)bcm53xxspi_read(b53spi, reg);
 	}
-
-	b53spi->read_offset = 0;
 }
 
 static int bcm53xxspi_transfer_one(struct spi_master *master,
@@ -238,7 +231,8 @@ static int bcm53xxspi_transfer_one(struct spi_master *master,
 		left = t->len;
 		while (left) {
 			size_t to_write = min_t(size_t, 16, left);
-			bool cont = left - to_write > 0;
+			bool cont = !spi_transfer_is_last(master, t) ||
+				    left - to_write > 0;
 
 			bcm53xxspi_buf_write(b53spi, buf, to_write, cont);
 			left -= to_write;
@@ -250,9 +244,9 @@ static int bcm53xxspi_transfer_one(struct spi_master *master,
 		buf = (u8 *)t->rx_buf;
 		left = t->len;
 		while (left) {
-			size_t to_read = min_t(size_t, 16 - b53spi->read_offset,
-					       left);
-			bool cont = left - to_read > 0;
+			size_t to_read = min_t(size_t, 16, left);
+			bool cont = !spi_transfer_is_last(master, t) ||
+				    left - to_read > 0;
 
 			bcm53xxspi_buf_read(b53spi, buf, to_read, cont);
 			left -= to_read;
diff --git a/drivers/spi/spi-davinci.c b/drivers/spi/spi-davinci.c
index 6ddb6ef1fda4..60d59b003aa4 100644
--- a/drivers/spi/spi-davinci.c
+++ b/drivers/spi/spi-davinci.c
@@ -945,6 +945,8 @@ static int davinci_spi_probe(struct platform_device *pdev)
 		goto free_master;
 	}
 
+	init_completion(&dspi->done);
+
 	ret = platform_get_irq(pdev, 0);
 	if (ret == 0)
 		ret = -EINVAL;
@@ -1021,8 +1023,6 @@ static int davinci_spi_probe(struct platform_device *pdev)
 	dspi->get_rx = davinci_spi_rx_buf_u8;
 	dspi->get_tx = davinci_spi_tx_buf_u8;
 
-	init_completion(&dspi->done);
-
 	/* Reset In/OUT SPI module */
 	iowrite32(0, dspi->base + SPIGCR0);
 	udelay(100);
diff --git a/drivers/spi/spi-dw.c b/drivers/spi/spi-dw.c
index b217c22ff72f..211cc7d75bf8 100644
--- a/drivers/spi/spi-dw.c
+++ b/drivers/spi/spi-dw.c
@@ -30,13 +30,11 @@
 
 /* Slave spi_dev related */
 struct chip_data {
-	u8 cs;			/* chip select pin */
 	u8 tmode;		/* TR/TO/RO/EEPROM */
 	u8 type;		/* SPI/SSP/MicroWire */
 
 	u8 poll_mode;		/* 1 means use poll mode */
 
-	u8 enable_dma;
 	u16 clk_div;		/* baud rate divider */
 	u32 speed_hz;		/* baud rate */
 	void (*cs_control)(u32 command);
diff --git a/drivers/spi/spi-fsl-dspi.c b/drivers/spi/spi-fsl-dspi.c
index f652f70cb8db..0630962ce442 100644
--- a/drivers/spi/spi-fsl-dspi.c
+++ b/drivers/spi/spi-fsl-dspi.c
@@ -903,10 +903,9 @@ static irqreturn_t dspi_interrupt(int irq, void *dev_id)
 }
 
 static const struct of_device_id fsl_dspi_dt_ids[] = {
-	{ .compatible = "fsl,vf610-dspi", .data = (void *)&vf610_data, },
-	{ .compatible = "fsl,ls1021a-v1.0-dspi",
-		.data = (void *)&ls1021a_v1_data, },
-	{ .compatible = "fsl,ls2085a-dspi", .data = (void *)&ls2085a_data, },
+	{ .compatible = "fsl,vf610-dspi", .data = &vf610_data, },
+	{ .compatible = "fsl,ls1021a-v1.0-dspi", .data = &ls1021a_v1_data, },
+	{ .compatible = "fsl,ls2085a-dspi", .data = &ls2085a_data, },
 	{ /* sentinel */ }
 };
 MODULE_DEVICE_TABLE(of, fsl_dspi_dt_ids);
@@ -980,7 +979,7 @@ static int dspi_probe(struct platform_device *pdev)
 	master->dev.of_node = pdev->dev.of_node;
 
 	master->cleanup = dspi_cleanup;
-	master->mode_bits = SPI_CPOL | SPI_CPHA;
+	master->mode_bits = SPI_CPOL | SPI_CPHA | SPI_LSB_FIRST;
 	master->bits_per_word_mask = SPI_BPW_MASK(4) | SPI_BPW_MASK(8) |
 					SPI_BPW_MASK(16);
 
diff --git a/drivers/spi/spi-imx.c b/drivers/spi/spi-imx.c
index 79ddefe4180d..6f57592a7f95 100644
--- a/drivers/spi/spi-imx.c
+++ b/drivers/spi/spi-imx.c
@@ -1622,6 +1622,11 @@ static int spi_imx_probe(struct platform_device *pdev)
 	spi_imx->devtype_data->intctrl(spi_imx, 0);
 
 	master->dev.of_node = pdev->dev.of_node;
+	ret = spi_bitbang_start(&spi_imx->bitbang);
+	if (ret) {
+		dev_err(&pdev->dev, "bitbang start failed with %d\n", ret);
+		goto out_clk_put;
+	}
 
 	/* Request GPIO CS lines, if any */
 	if (!spi_imx->slave_mode && master->cs_gpios) {
@@ -1640,12 +1645,6 @@ static int spi_imx_probe(struct platform_device *pdev)
 		}
 	}
 
-	ret = spi_bitbang_start(&spi_imx->bitbang);
-	if (ret) {
-		dev_err(&pdev->dev, "bitbang start failed with %d\n", ret);
-		goto out_clk_put;
-	}
-
 	dev_info(&pdev->dev, "probed\n");
 
 	clk_disable(spi_imx->clk_ipg);
@@ -1668,12 +1667,23 @@ static int spi_imx_remove(struct platform_device *pdev)
 {
 	struct spi_master *master = platform_get_drvdata(pdev);
 	struct spi_imx_data *spi_imx = spi_master_get_devdata(master);
+	int ret;
 
 	spi_bitbang_stop(&spi_imx->bitbang);
 
+	ret = clk_enable(spi_imx->clk_per);
+	if (ret)
+		return ret;
+
+	ret = clk_enable(spi_imx->clk_ipg);
+	if (ret) {
+		clk_disable(spi_imx->clk_per);
+		return ret;
+	}
+
 	writel(0, spi_imx->base + MXC_CSPICTRL);
-	clk_unprepare(spi_imx->clk_ipg);
-	clk_unprepare(spi_imx->clk_per);
+	clk_disable_unprepare(spi_imx->clk_ipg);
+	clk_disable_unprepare(spi_imx->clk_per);
 	spi_imx_sdma_exit(spi_imx);
 	spi_master_put(master);
 
diff --git a/drivers/spi/spi-jcore.c b/drivers/spi/spi-jcore.c
index cebfea5faa4b..dafed6280df3 100644
--- a/drivers/spi/spi-jcore.c
+++ b/drivers/spi/spi-jcore.c
@@ -198,8 +198,10 @@ static int jcore_spi_probe(struct platform_device *pdev)
 
 	/* Register our spi controller */
 	err = devm_spi_register_master(&pdev->dev, master);
-	if (err)
+	if (err) {
+		clk_disable(clk);
 		goto exit;
+	}
 
 	return 0;
 
diff --git a/drivers/spi/spi-meson-spicc.c b/drivers/spi/spi-meson-spicc.c
index 7f8429635502..5c82910e3480 100644
--- a/drivers/spi/spi-meson-spicc.c
+++ b/drivers/spi/spi-meson-spicc.c
@@ -599,6 +599,7 @@ static int meson_spicc_remove(struct platform_device *pdev)
 
 static const struct of_device_id meson_spicc_of_match[] = {
 	{ .compatible = "amlogic,meson-gx-spicc", },
+	{ .compatible = "amlogic,meson-axg-spicc", },
 	{ /* sentinel */ }
 };
 MODULE_DEVICE_TABLE(of, meson_spicc_of_match);
diff --git a/drivers/spi/spi-orion.c b/drivers/spi/spi-orion.c
index 8974bb340b3a..deca63e82ff6 100644
--- a/drivers/spi/spi-orion.c
+++ b/drivers/spi/spi-orion.c
@@ -94,6 +94,7 @@ struct orion_spi {
 	struct spi_master	*master;
 	void __iomem		*base;
 	struct clk              *clk;
+	struct clk              *axi_clk;
 	const struct orion_spi_dev *devdata;
 
 	struct orion_direct_acc	direct_access[ORION_NUM_CHIPSELECTS];
@@ -634,6 +635,16 @@ static int orion_spi_probe(struct platform_device *pdev)
 	if (status)
 		goto out;
 
+	/* The following clock is only used by some SoCs */
+	spi->axi_clk = devm_clk_get(&pdev->dev, "axi");
+	if (IS_ERR(spi->axi_clk) &&
+	    PTR_ERR(spi->axi_clk) == -EPROBE_DEFER) {
+		status = -EPROBE_DEFER;
+		goto out_rel_clk;
+	}
+	if (!IS_ERR(spi->axi_clk))
+		clk_prepare_enable(spi->axi_clk);
+
 	tclk_hz = clk_get_rate(spi->clk);
 
 	/*
@@ -658,7 +669,7 @@ static int orion_spi_probe(struct platform_device *pdev)
 	spi->base = devm_ioremap_resource(&pdev->dev, r);
 	if (IS_ERR(spi->base)) {
 		status = PTR_ERR(spi->base);
-		goto out_rel_clk;
+		goto out_rel_axi_clk;
 	}
 
 	/* Scan all SPI devices of this controller for direct mapped devices */
@@ -696,7 +707,7 @@ static int orion_spi_probe(struct platform_device *pdev)
 							    PAGE_SIZE);
 		if (!spi->direct_access[cs].vaddr) {
 			status = -ENOMEM;
-			goto out_rel_clk;
+			goto out_rel_axi_clk;
 		}
 		spi->direct_access[cs].size = PAGE_SIZE;
 
@@ -724,6 +735,8 @@ static int orion_spi_probe(struct platform_device *pdev)
 
 out_rel_pm:
 	pm_runtime_disable(&pdev->dev);
+out_rel_axi_clk:
+	clk_disable_unprepare(spi->axi_clk);
 out_rel_clk:
 	clk_disable_unprepare(spi->clk);
 out:
@@ -738,6 +751,7 @@ static int orion_spi_remove(struct platform_device *pdev)
 	struct orion_spi *spi = spi_master_get_devdata(master);
 
 	pm_runtime_get_sync(&pdev->dev);
+	clk_disable_unprepare(spi->axi_clk);
 	clk_disable_unprepare(spi->clk);
 
 	spi_unregister_master(master);
@@ -754,6 +768,7 @@ static int orion_spi_runtime_suspend(struct device *dev)
 	struct spi_master *master = dev_get_drvdata(dev);
 	struct orion_spi *spi = spi_master_get_devdata(master);
 
+	clk_disable_unprepare(spi->axi_clk);
 	clk_disable_unprepare(spi->clk);
 	return 0;
 }
@@ -763,6 +778,8 @@ static int orion_spi_runtime_resume(struct device *dev)
 	struct spi_master *master = dev_get_drvdata(dev);
 	struct orion_spi *spi = spi_master_get_devdata(master);
 
+	if (!IS_ERR(spi->axi_clk))
+		clk_prepare_enable(spi->axi_clk);
 	return clk_prepare_enable(spi->clk);
 }
 #endif
diff --git a/drivers/spi/spi-pxa2xx.c b/drivers/spi/spi-pxa2xx.c
index 4cb515a3104c..b0822d1dba29 100644
--- a/drivers/spi/spi-pxa2xx.c
+++ b/drivers/spi/spi-pxa2xx.c
@@ -1237,7 +1237,7 @@ static int setup_cs(struct spi_device *spi, struct chip_data *chip,
 	 * different chip_info, release previously requested GPIO
 	 */
 	if (chip->gpiod_cs) {
-		gpio_free(desc_to_gpio(chip->gpiod_cs));
+		gpiod_put(chip->gpiod_cs);
 		chip->gpiod_cs = NULL;
 	}
 
@@ -1417,7 +1417,7 @@ static void cleanup(struct spi_device *spi)
 
 	if (drv_data->ssp_type != CE4100_SSP && !drv_data->cs_gpiods &&
 	    chip->gpiod_cs)
-		gpio_free(desc_to_gpio(chip->gpiod_cs));
+		gpiod_put(chip->gpiod_cs);
 
 	kfree(chip);
 }
diff --git a/drivers/spi/spi-s3c64xx.c b/drivers/spi/spi-s3c64xx.c
index de7df20f8712..baa3a9fa2638 100644
--- a/drivers/spi/spi-s3c64xx.c
+++ b/drivers/spi/spi-s3c64xx.c
@@ -1,17 +1,7 @@
-/*
- * Copyright (C) 2009 Samsung Electronics Ltd.
- *	Jaswinder Singh <jassi.brar@samsung.com>
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- */
+// SPDX-License-Identifier: GPL-2.0+
+//
+// Copyright (c) 2009 Samsung Electronics Co., Ltd.
+//      Jaswinder Singh <jassi.brar@samsung.com>
 
 #include <linux/init.h>
 #include <linux/module.h>
diff --git a/drivers/spi/spi-sh-msiof.c b/drivers/spi/spi-sh-msiof.c
index fcd261f98b9f..c5dcfb434a49 100644
--- a/drivers/spi/spi-sh-msiof.c
+++ b/drivers/spi/spi-sh-msiof.c
@@ -19,6 +19,7 @@
 #include <linux/dmaengine.h>
 #include <linux/err.h>
 #include <linux/gpio.h>
+#include <linux/gpio/consumer.h>
 #include <linux/interrupt.h>
 #include <linux/io.h>
 #include <linux/kernel.h>
@@ -55,9 +56,14 @@ struct sh_msiof_spi_priv {
 	void *rx_dma_page;
 	dma_addr_t tx_dma_addr;
 	dma_addr_t rx_dma_addr;
+	unsigned short unused_ss;
+	bool native_cs_inited;
+	bool native_cs_high;
 	bool slave_aborted;
 };
 
+#define MAX_SS	3	/* Maximum number of native chip selects */
+
 #define TMDR1	0x00	/* Transmit Mode Register 1 */
 #define TMDR2	0x04	/* Transmit Mode Register 2 */
 #define TMDR3	0x08	/* Transmit Mode Register 3 */
@@ -91,6 +97,8 @@ struct sh_msiof_spi_priv {
 #define MDR1_XXSTP	 0x00000001 /* Transmission/Reception Stop on FIFO */
 /* TMDR1 */
 #define TMDR1_PCON	 0x40000000 /* Transfer Signal Connection */
+#define TMDR1_SYNCCH_MASK 0xc000000 /* Synchronization Signal Channel Select */
+#define TMDR1_SYNCCH_SHIFT	 26 /* 0=MSIOF_SYNC, 1=MSIOF_SS1, 2=MSIOF_SS2 */
 
 /* TMDR2 and RMDR2 */
 #define MDR2_BITLEN1(i)	(((i) - 1) << 24) /* Data Size (8-32 bits) */
@@ -324,7 +332,7 @@ static u32 sh_msiof_spi_get_dtdl_and_syncdl(struct sh_msiof_spi_priv *p)
 	return val;
 }
 
-static void sh_msiof_spi_set_pin_regs(struct sh_msiof_spi_priv *p,
+static void sh_msiof_spi_set_pin_regs(struct sh_msiof_spi_priv *p, u32 ss,
 				      u32 cpol, u32 cpha,
 				      u32 tx_hi_z, u32 lsb_first, u32 cs_high)
 {
@@ -342,10 +350,13 @@ static void sh_msiof_spi_set_pin_regs(struct sh_msiof_spi_priv *p,
 	tmp |= !cs_high << MDR1_SYNCAC_SHIFT;
 	tmp |= lsb_first << MDR1_BITLSB_SHIFT;
 	tmp |= sh_msiof_spi_get_dtdl_and_syncdl(p);
-	if (spi_controller_is_slave(p->master))
+	if (spi_controller_is_slave(p->master)) {
 		sh_msiof_write(p, TMDR1, tmp | TMDR1_PCON);
-	else
-		sh_msiof_write(p, TMDR1, tmp | MDR1_TRMD | TMDR1_PCON);
+	} else {
+		sh_msiof_write(p, TMDR1,
+			       tmp | MDR1_TRMD | TMDR1_PCON |
+			       (ss < MAX_SS ? ss : 0) << TMDR1_SYNCCH_SHIFT);
+	}
 	if (p->master->flags & SPI_MASTER_MUST_TX) {
 		/* These bits are reserved if RX needs TX */
 		tmp &= ~0x0000ffff;
@@ -528,8 +539,7 @@ static int sh_msiof_spi_setup(struct spi_device *spi)
 {
 	struct device_node	*np = spi->master->dev.of_node;
 	struct sh_msiof_spi_priv *p = spi_master_get_devdata(spi->master);
-
-	pm_runtime_get_sync(&p->pdev->dev);
+	u32 clr, set, tmp;
 
 	if (!np) {
 		/*
@@ -539,19 +549,31 @@ static int sh_msiof_spi_setup(struct spi_device *spi)
 		spi->cs_gpio = (uintptr_t)spi->controller_data;
 	}
 
-	/* Configure pins before deasserting CS */
-	sh_msiof_spi_set_pin_regs(p, !!(spi->mode & SPI_CPOL),
-				  !!(spi->mode & SPI_CPHA),
-				  !!(spi->mode & SPI_3WIRE),
-				  !!(spi->mode & SPI_LSB_FIRST),
-				  !!(spi->mode & SPI_CS_HIGH));
+	if (gpio_is_valid(spi->cs_gpio)) {
+		gpio_direction_output(spi->cs_gpio, !(spi->mode & SPI_CS_HIGH));
+		return 0;
+	}
 
-	if (spi->cs_gpio >= 0)
-		gpio_set_value(spi->cs_gpio, !(spi->mode & SPI_CS_HIGH));
+	if (spi_controller_is_slave(p->master))
+		return 0;
 
+	if (p->native_cs_inited &&
+	    (p->native_cs_high == !!(spi->mode & SPI_CS_HIGH)))
+		return 0;
 
+	/* Configure native chip select mode/polarity early */
+	clr = MDR1_SYNCMD_MASK;
+	set = MDR1_TRMD | TMDR1_PCON | MDR1_SYNCMD_SPI;
+	if (spi->mode & SPI_CS_HIGH)
+		clr |= BIT(MDR1_SYNCAC_SHIFT);
+	else
+		set |= BIT(MDR1_SYNCAC_SHIFT);
+	pm_runtime_get_sync(&p->pdev->dev);
+	tmp = sh_msiof_read(p, TMDR1) & ~clr;
+	sh_msiof_write(p, TMDR1, tmp | set);
 	pm_runtime_put(&p->pdev->dev);
-
+	p->native_cs_high = spi->mode & SPI_CS_HIGH;
+	p->native_cs_inited = true;
 	return 0;
 }
 
@@ -560,13 +582,20 @@ static int sh_msiof_prepare_message(struct spi_master *master,
 {
 	struct sh_msiof_spi_priv *p = spi_master_get_devdata(master);
 	const struct spi_device *spi = msg->spi;
+	u32 ss, cs_high;
 
 	/* Configure pins before asserting CS */
-	sh_msiof_spi_set_pin_regs(p, !!(spi->mode & SPI_CPOL),
+	if (gpio_is_valid(spi->cs_gpio)) {
+		ss = p->unused_ss;
+		cs_high = p->native_cs_high;
+	} else {
+		ss = spi->chip_select;
+		cs_high = !!(spi->mode & SPI_CS_HIGH);
+	}
+	sh_msiof_spi_set_pin_regs(p, ss, !!(spi->mode & SPI_CPOL),
 				  !!(spi->mode & SPI_CPHA),
 				  !!(spi->mode & SPI_3WIRE),
-				  !!(spi->mode & SPI_LSB_FIRST),
-				  !!(spi->mode & SPI_CS_HIGH));
+				  !!(spi->mode & SPI_LSB_FIRST), cs_high);
 	return 0;
 }
 
@@ -784,11 +813,21 @@ static int sh_msiof_dma_once(struct sh_msiof_spi_priv *p, const void *tx,
 		goto stop_dma;
 	}
 
-	/* wait for tx fifo to be emptied / rx fifo to be filled */
+	/* wait for tx/rx DMA completion */
 	ret = sh_msiof_wait_for_completion(p);
 	if (ret)
 		goto stop_reset;
 
+	if (!rx) {
+		reinit_completion(&p->done);
+		sh_msiof_write(p, IER, IER_TEOFE);
+
+		/* wait for tx fifo to be emptied */
+		ret = sh_msiof_wait_for_completion(p);
+		if (ret)
+			goto stop_reset;
+	}
+
 	/* clear status bits */
 	sh_msiof_reset_str(p);
 
@@ -912,9 +951,8 @@ static int sh_msiof_transfer_one(struct spi_master *master,
 
 		ret = sh_msiof_dma_once(p, tx_buf, rx_buf, l);
 		if (ret == -EAGAIN) {
-			pr_warn_once("%s %s: DMA not available, falling back to PIO\n",
-				     dev_driver_string(&p->pdev->dev),
-				     dev_name(&p->pdev->dev));
+			dev_warn_once(&p->pdev->dev,
+				"DMA not available, falling back to PIO\n");
 			break;
 		}
 		if (ret)
@@ -1071,6 +1109,45 @@ static struct sh_msiof_spi_info *sh_msiof_spi_parse_dt(struct device *dev)
 }
 #endif
 
+static int sh_msiof_get_cs_gpios(struct sh_msiof_spi_priv *p)
+{
+	struct device *dev = &p->pdev->dev;
+	unsigned int used_ss_mask = 0;
+	unsigned int cs_gpios = 0;
+	unsigned int num_cs, i;
+	int ret;
+
+	ret = gpiod_count(dev, "cs");
+	if (ret <= 0)
+		return 0;
+
+	num_cs = max_t(unsigned int, ret, p->master->num_chipselect);
+	for (i = 0; i < num_cs; i++) {
+		struct gpio_desc *gpiod;
+
+		gpiod = devm_gpiod_get_index(dev, "cs", i, GPIOD_ASIS);
+		if (!IS_ERR(gpiod)) {
+			cs_gpios++;
+			continue;
+		}
+
+		if (PTR_ERR(gpiod) != -ENOENT)
+			return PTR_ERR(gpiod);
+
+		if (i >= MAX_SS) {
+			dev_err(dev, "Invalid native chip select %d\n", i);
+			return -EINVAL;
+		}
+		used_ss_mask |= BIT(i);
+	}
+	p->unused_ss = ffz(used_ss_mask);
+	if (cs_gpios && p->unused_ss >= MAX_SS) {
+		dev_err(dev, "No unused native chip select available\n");
+		return -EINVAL;
+	}
+	return 0;
+}
+
 static struct dma_chan *sh_msiof_request_dma_chan(struct device *dev,
 	enum dma_transfer_direction dir, unsigned int id, dma_addr_t port_addr)
 {
@@ -1284,13 +1361,18 @@ static int sh_msiof_spi_probe(struct platform_device *pdev)
 	if (p->info->rx_fifo_override)
 		p->rx_fifo_size = p->info->rx_fifo_override;
 
+	/* Setup GPIO chip selects */
+	master->num_chipselect = p->info->num_chipselect;
+	ret = sh_msiof_get_cs_gpios(p);
+	if (ret)
+		goto err1;
+
 	/* init master code */
 	master->mode_bits = SPI_CPOL | SPI_CPHA | SPI_CS_HIGH;
 	master->mode_bits |= SPI_LSB_FIRST | SPI_3WIRE;
 	master->flags = chipdata->master_flags;
 	master->bus_num = pdev->id;
 	master->dev.of_node = pdev->dev.of_node;
-	master->num_chipselect = p->info->num_chipselect;
 	master->setup = sh_msiof_spi_setup;
 	master->prepare_message = sh_msiof_prepare_message;
 	master->slave_abort = sh_msiof_slave_abort;
diff --git a/drivers/spi/spi-sirf.c b/drivers/spi/spi-sirf.c
index bbb1a275f718..f009d76f96b1 100644
--- a/drivers/spi/spi-sirf.c
+++ b/drivers/spi/spi-sirf.c
@@ -1072,7 +1072,7 @@ static int spi_sirfsoc_probe(struct platform_device *pdev)
 	struct sirfsoc_spi *sspi;
 	struct spi_master *master;
 	struct resource *mem_res;
-	struct sirf_spi_comp_data *spi_comp_data;
+	const struct sirf_spi_comp_data *spi_comp_data;
 	int irq;
 	int ret;
 	const struct of_device_id *match;
@@ -1092,7 +1092,7 @@ static int spi_sirfsoc_probe(struct platform_device *pdev)
 	platform_set_drvdata(pdev, master);
 	sspi = spi_master_get_devdata(master);
 	sspi->fifo_full_offset = ilog2(sspi->fifo_size);
-	spi_comp_data = (struct sirf_spi_comp_data *)match->data;
+	spi_comp_data = match->data;
 	sspi->regs = spi_comp_data->regs;
 	sspi->type = spi_comp_data->type;
 	sspi->fifo_level_chk_mask = (sspi->fifo_size / 4) - 1;
diff --git a/drivers/spi/spi-sun6i.c b/drivers/spi/spi-sun6i.c
index fb38234249a8..8533f4edd00a 100644
--- a/drivers/spi/spi-sun6i.c
+++ b/drivers/spi/spi-sun6i.c
@@ -541,7 +541,7 @@ err_free_master:
 
 static int sun6i_spi_remove(struct platform_device *pdev)
 {
-	pm_runtime_disable(&pdev->dev);
+	pm_runtime_force_suspend(&pdev->dev);
 
 	return 0;
 }
diff --git a/drivers/spi/spi-xilinx.c b/drivers/spi/spi-xilinx.c
index e0b9fe1d0e37..63fedc49ae9c 100644
--- a/drivers/spi/spi-xilinx.c
+++ b/drivers/spi/spi-xilinx.c
@@ -381,6 +381,7 @@ static int xilinx_spi_find_buffer_size(struct xilinx_spi *xspi)
 }
 
 static const struct of_device_id xilinx_spi_of_match[] = {
+	{ .compatible = "xlnx,axi-quad-spi-1.00.a", },
 	{ .compatible = "xlnx,xps-spi-2.00.a", },
 	{ .compatible = "xlnx,xps-spi-2.00.b", },
 	{}