Merge remote-tracking branch 'spi/topic/dma' into spi-next

3 files changed, 310 insertions, 108 deletions

diff --git a/drivers/spi/spi-s3c64xx.c b/drivers/spi/spi-s3c64xx.c
index ae907dde1371..25c9bd409a87 100644
--- a/drivers/spi/spi-s3c64xx.c
+++ b/drivers/spi/spi-s3c64xx.c
@@ -381,7 +381,7 @@ static void s3c64xx_spi_dma_stop(struct s3c64xx_spi_driver_data *sdd,
 #else
 
 static void prepare_dma(struct s3c64xx_spi_dma_data *dma,
-					unsigned len, dma_addr_t buf)
+			struct sg_table *sgt)
 {
 	struct s3c64xx_spi_driver_data *sdd;
 	struct dma_slave_config config;
@@ -407,8 +407,8 @@ static void prepare_dma(struct s3c64xx_spi_dma_data *dma,
 		dmaengine_slave_config(dma->ch, &config);
 	}
 
-	desc = dmaengine_prep_slave_single(dma->ch, buf, len,
-					dma->direction, DMA_PREP_INTERRUPT);
+	desc = dmaengine_prep_slave_sg(dma->ch, sgt->sgl, sgt->nents,
+				       dma->direction, DMA_PREP_INTERRUPT);
 
 	desc->callback = s3c64xx_spi_dmacb;
 	desc->callback_param = dma;
@@ -515,7 +515,11 @@ static void enable_datapath(struct s3c64xx_spi_driver_data *sdd,
 		chcfg |= S3C64XX_SPI_CH_TXCH_ON;
 		if (dma_mode) {
 			modecfg |= S3C64XX_SPI_MODE_TXDMA_ON;
+#ifndef CONFIG_S3C_DMA
+			prepare_dma(&sdd->tx_dma, &xfer->tx_sg);
+#else
 			prepare_dma(&sdd->tx_dma, xfer->len, xfer->tx_dma);
+#endif
 		} else {
 			switch (sdd->cur_bpw) {
 			case 32:
@@ -547,7 +551,11 @@ static void enable_datapath(struct s3c64xx_spi_driver_data *sdd,
 			writel(((xfer->len * 8 / sdd->cur_bpw) & 0xffff)
 					| S3C64XX_SPI_PACKET_CNT_EN,
 					regs + S3C64XX_SPI_PACKET_CNT);
+#ifndef CONFIG_S3C_DMA
+			prepare_dma(&sdd->rx_dma, &xfer->rx_sg);
+#else
 			prepare_dma(&sdd->rx_dma, xfer->len, xfer->rx_dma);
+#endif
 		}
 	}
 
@@ -555,23 +563,6 @@ static void enable_datapath(struct s3c64xx_spi_driver_data *sdd,
 	writel(chcfg, regs + S3C64XX_SPI_CH_CFG);
 }
 
-static inline void enable_cs(struct s3c64xx_spi_driver_data *sdd,
-						struct spi_device *spi)
-{
-	if (sdd->tgl_spi != NULL) { /* If last device toggled after mssg */
-		if (sdd->tgl_spi != spi) { /* if last mssg on diff device */
-			/* Deselect the last toggled device */
-			if (spi->cs_gpio >= 0)
-				gpio_set_value(spi->cs_gpio,
-					spi->mode & SPI_CS_HIGH ? 0 : 1);
-		}
-		sdd->tgl_spi = NULL;
-	}
-
-	if (spi->cs_gpio >= 0)
-		gpio_set_value(spi->cs_gpio, spi->mode & SPI_CS_HIGH ? 1 : 0);
-}
-
 static u32 s3c64xx_spi_wait_for_timeout(struct s3c64xx_spi_driver_data *sdd,
 					int timeout_ms)
 {
@@ -593,112 +584,111 @@ static u32 s3c64xx_spi_wait_for_timeout(struct s3c64xx_spi_driver_data *sdd,
 	return RX_FIFO_LVL(status, sdd);
 }
 
-static int wait_for_xfer(struct s3c64xx_spi_driver_data *sdd,
-				struct spi_transfer *xfer, int dma_mode)
+static int wait_for_dma(struct s3c64xx_spi_driver_data *sdd,
+			struct spi_transfer *xfer)
 {
 	void __iomem *regs = sdd->regs;
 	unsigned long val;
+	u32 status;
 	int ms;
 
 	/* millisecs to xfer 'len' bytes @ 'cur_speed' */
 	ms = xfer->len * 8 * 1000 / sdd->cur_speed;
 	ms += 10; /* some tolerance */
 
-	if (dma_mode) {
-		val = msecs_to_jiffies(ms) + 10;
-		val = wait_for_completion_timeout(&sdd->xfer_completion, val);
-	} else {
-		u32 status;
-		val = msecs_to_loops(ms);
-		do {
+	val = msecs_to_jiffies(ms) + 10;
+	val = wait_for_completion_timeout(&sdd->xfer_completion, val);
+
+	/*
+	 * If the previous xfer was completed within timeout, then
+	 * proceed further else return -EIO.
+	 * DmaTx returns after simply writing data in the FIFO,
+	 * w/o waiting for real transmission on the bus to finish.
+	 * DmaRx returns only after Dma read data from FIFO which
+	 * needs bus transmission to finish, so we don't worry if
+	 * Xfer involved Rx(with or without Tx).
+	 */
+	if (val && !xfer->rx_buf) {
+		val = msecs_to_loops(10);
+		status = readl(regs + S3C64XX_SPI_STATUS);
+		while ((TX_FIFO_LVL(status, sdd)
+			|| !S3C64XX_SPI_ST_TX_DONE(status, sdd))
+		       && --val) {
+			cpu_relax();
 			status = readl(regs + S3C64XX_SPI_STATUS);
-		} while (RX_FIFO_LVL(status, sdd) < xfer->len && --val);
+		}
+
 	}
 
-	if (dma_mode) {
-		u32 status;
-
-		/*
-		 * If the previous xfer was completed within timeout, then
-		 * proceed further else return -EIO.
-		 * DmaTx returns after simply writing data in the FIFO,
-		 * w/o waiting for real transmission on the bus to finish.
-		 * DmaRx returns only after Dma read data from FIFO which
-		 * needs bus transmission to finish, so we don't worry if
-		 * Xfer involved Rx(with or without Tx).
-		 */
-		if (val && !xfer->rx_buf) {
-			val = msecs_to_loops(10);
-			status = readl(regs + S3C64XX_SPI_STATUS);
-			while ((TX_FIFO_LVL(status, sdd)
-				|| !S3C64XX_SPI_ST_TX_DONE(status, sdd))
-					&& --val) {
-				cpu_relax();
-				status = readl(regs + S3C64XX_SPI_STATUS);
-			}
+	/* If timed out while checking rx/tx status return error */
+	if (!val)
+		return -EIO;
 
-		}
+	return 0;
+}
 
-		/* If timed out while checking rx/tx status return error */
-		if (!val)
-			return -EIO;
-	} else {
-		int loops;
-		u32 cpy_len;
-		u8 *buf;
-
-		/* If it was only Tx */
-		if (!xfer->rx_buf) {
-			sdd->state &= ~TXBUSY;
-			return 0;
-		}
+static int wait_for_pio(struct s3c64xx_spi_driver_data *sdd,
+			struct spi_transfer *xfer)
+{
+	void __iomem *regs = sdd->regs;
+	unsigned long val;
+	u32 status;
+	int loops;
+	u32 cpy_len;
+	u8 *buf;
+	int ms;
 
-		/*
-		 * If the receive length is bigger than the controller fifo
-		 * size, calculate the loops and read the fifo as many times.
-		 * loops = length / max fifo size (calculated by using the
-		 * fifo mask).
-		 * For any size less than the fifo size the below code is
-		 * executed atleast once.
-		 */
-		loops = xfer->len / ((FIFO_LVL_MASK(sdd) >> 1) + 1);
-		buf = xfer->rx_buf;
-		do {
-			/* wait for data to be received in the fifo */
-			cpy_len = s3c64xx_spi_wait_for_timeout(sdd,
-						(loops ? ms : 0));
+	/* millisecs to xfer 'len' bytes @ 'cur_speed' */
+	ms = xfer->len * 8 * 1000 / sdd->cur_speed;
+	ms += 10; /* some tolerance */
 
-			switch (sdd->cur_bpw) {
-			case 32:
-				ioread32_rep(regs + S3C64XX_SPI_RX_DATA,
-					buf, cpy_len / 4);
-				break;
-			case 16:
-				ioread16_rep(regs + S3C64XX_SPI_RX_DATA,
-					buf, cpy_len / 2);
-				break;
-			default:
-				ioread8_rep(regs + S3C64XX_SPI_RX_DATA,
-					buf, cpy_len);
-				break;
-			}
+	val = msecs_to_loops(ms);
+	do {
+		status = readl(regs + S3C64XX_SPI_STATUS);
+	} while (RX_FIFO_LVL(status, sdd) < xfer->len && --val);
 
-			buf = buf + cpy_len;
-		} while (loops--);
-		sdd->state &= ~RXBUSY;
+
+	/* If it was only Tx */
+	if (!xfer->rx_buf) {
+		sdd->state &= ~TXBUSY;
+		return 0;
 	}
 
-	return 0;
-}
+	/*
+	 * If the receive length is bigger than the controller fifo
+	 * size, calculate the loops and read the fifo as many times.
+	 * loops = length / max fifo size (calculated by using the
+	 * fifo mask).
+	 * For any size less than the fifo size the below code is
+	 * executed atleast once.
+	 */
+	loops = xfer->len / ((FIFO_LVL_MASK(sdd) >> 1) + 1);
+	buf = xfer->rx_buf;
+	do {
+		/* wait for data to be received in the fifo */
+		cpy_len = s3c64xx_spi_wait_for_timeout(sdd,
+						       (loops ? ms : 0));
+
+		switch (sdd->cur_bpw) {
+		case 32:
+			ioread32_rep(regs + S3C64XX_SPI_RX_DATA,
+				     buf, cpy_len / 4);
+			break;
+		case 16:
+			ioread16_rep(regs + S3C64XX_SPI_RX_DATA,
+				     buf, cpy_len / 2);
+			break;
+		default:
+			ioread8_rep(regs + S3C64XX_SPI_RX_DATA,
+				    buf, cpy_len);
+			break;
+		}
 
-static inline void disable_cs(struct s3c64xx_spi_driver_data *sdd,
-						struct spi_device *spi)
-{
-	if (sdd->tgl_spi == spi)
-		sdd->tgl_spi = NULL;
+		buf = buf + cpy_len;
+	} while (loops--);
+	sdd->state &= ~RXBUSY;
 
-	if (spi->cs_gpio >= 0)
-		gpio_set_value(spi->cs_gpio, spi->mode & SPI_CS_HIGH ? 0 : 1);
+	return 0;
 }
 
 static void s3c64xx_spi_config(struct s3c64xx_spi_driver_data *sdd)
@@ -929,7 +919,10 @@ static int s3c64xx_spi_transfer_one(struct spi_master *master,
 
 	spin_unlock_irqrestore(&sdd->lock, flags);
 
-	status = wait_for_xfer(sdd, xfer, use_dma);
+	if (use_dma)
+		status = wait_for_dma(sdd, xfer);
+	else
+		status = wait_for_pio(sdd, xfer);
 
 	if (status) {
 		dev_err(&spi->dev, "I/O Error: rx-%d tx-%d res:rx-%c tx-%c len-%d\n",
@@ -1092,14 +1085,12 @@ static int s3c64xx_spi_setup(struct spi_device *spi)
 
 	pm_runtime_put(&sdd->pdev->dev);
 	writel(S3C64XX_SPI_SLAVE_SIG_INACT, sdd->regs + S3C64XX_SPI_SLAVE_SEL);
-	disable_cs(sdd, spi);
 	return 0;
 
 setup_exit:
 	pm_runtime_put(&sdd->pdev->dev);
 	/* setup() returns with device de-selected */
 	writel(S3C64XX_SPI_SLAVE_SIG_INACT, sdd->regs + S3C64XX_SPI_SLAVE_SEL);
-	disable_cs(sdd, spi);
 
 	gpio_free(cs->line);
 	spi_set_ctldata(spi, NULL);
diff --git a/drivers/spi/spi.c b/drivers/spi/spi.c
index 0a8f399c907f..32e4603d5fc8 100644
--- a/drivers/spi/spi.c
+++ b/drivers/spi/spi.c
@@ -24,6 +24,8 @@
 #include <linux/device.h>
 #include <linux/init.h>
 #include <linux/cache.h>
+#include <linux/dma-mapping.h>
+#include <linux/dmaengine.h>
 #include <linux/mutex.h>
 #include <linux/of_device.h>
 #include <linux/of_irq.h>
@@ -578,6 +580,169 @@ static void spi_set_cs(struct spi_device *spi, bool enable)
 		spi->master->set_cs(spi, !enable);
 }
 
+static int spi_map_buf(struct spi_master *master, struct device *dev,
+		       struct sg_table *sgt, void *buf, size_t len,
+		       enum dma_data_direction dir)
+{
+	const bool vmalloced_buf = is_vmalloc_addr(buf);
+	const int desc_len = vmalloced_buf ? PAGE_SIZE : master->max_dma_len;
+	const int sgs = DIV_ROUND_UP(len, desc_len);
+	struct page *vm_page;
+	void *sg_buf;
+	size_t min;
+	int i, ret;
+
+	ret = sg_alloc_table(sgt, sgs, GFP_KERNEL);
+	if (ret != 0)
+		return ret;
+
+	for (i = 0; i < sgs; i++) {
+		min = min_t(size_t, len, desc_len);
+
+		if (vmalloced_buf) {
+			vm_page = vmalloc_to_page(buf);
+			if (!vm_page) {
+				sg_free_table(sgt);
+				return -ENOMEM;
+			}
+			sg_buf = page_address(vm_page) +
+				((size_t)buf & ~PAGE_MASK);
+		} else {
+			sg_buf = buf;
+		}
+
+		sg_set_buf(&sgt->sgl[i], sg_buf, min);
+
+		buf += min;
+		len -= min;
+	}
+
+	ret = dma_map_sg(dev, sgt->sgl, sgt->nents, dir);
+	if (ret < 0) {
+		sg_free_table(sgt);
+		return ret;
+	}
+
+	sgt->nents = ret;
+
+	return 0;
+}
+
+static void spi_unmap_buf(struct spi_master *master, struct device *dev,
+			  struct sg_table *sgt, enum dma_data_direction dir)
+{
+	if (sgt->orig_nents) {
+		dma_unmap_sg(dev, sgt->sgl, sgt->orig_nents, dir);
+		sg_free_table(sgt);
+	}
+}
+
+static int spi_map_msg(struct spi_master *master, struct spi_message *msg)
+{
+	struct device *tx_dev, *rx_dev;
+	struct spi_transfer *xfer;
+	void *tmp;
+	unsigned int max_tx, max_rx;
+	int ret;
+
+	if (master->flags & (SPI_MASTER_MUST_RX | SPI_MASTER_MUST_TX)) {
+		max_tx = 0;
+		max_rx = 0;
+
+		list_for_each_entry(xfer, &msg->transfers, transfer_list) {
+			if ((master->flags & SPI_MASTER_MUST_TX) &&
+			    !xfer->tx_buf)
+				max_tx = max(xfer->len, max_tx);
+			if ((master->flags & SPI_MASTER_MUST_RX) &&
+			    !xfer->rx_buf)
+				max_rx = max(xfer->len, max_rx);
+		}
+
+		if (max_tx) {
+			tmp = krealloc(master->dummy_tx, max_tx,
+				       GFP_KERNEL | GFP_DMA);
+			if (!tmp)
+				return -ENOMEM;
+			master->dummy_tx = tmp;
+			memset(tmp, 0, max_tx);
+		}
+
+		if (max_rx) {
+			tmp = krealloc(master->dummy_rx, max_rx,
+				       GFP_KERNEL | GFP_DMA);
+			if (!tmp)
+				return -ENOMEM;
+			master->dummy_rx = tmp;
+		}
+
+		if (max_tx || max_rx) {
+			list_for_each_entry(xfer, &msg->transfers,
+					    transfer_list) {
+				if (!xfer->tx_buf)
+					xfer->tx_buf = master->dummy_tx;
+				if (!xfer->rx_buf)
+					xfer->rx_buf = master->dummy_rx;
+			}
+		}
+	}
+
+	if (!master->can_dma)
+		return 0;
+
+	tx_dev = &master->dma_tx->dev->device;
+	rx_dev = &master->dma_rx->dev->device;
+
+	list_for_each_entry(xfer, &msg->transfers, transfer_list) {
+		if (!master->can_dma(master, msg->spi, xfer))
+			continue;
+
+		if (xfer->tx_buf != NULL) {
+			ret = spi_map_buf(master, tx_dev, &xfer->tx_sg,
+					  (void *)xfer->tx_buf, xfer->len,
+					  DMA_TO_DEVICE);
+			if (ret != 0)
+				return ret;
+		}
+
+		if (xfer->rx_buf != NULL) {
+			ret = spi_map_buf(master, rx_dev, &xfer->rx_sg,
+					  xfer->rx_buf, xfer->len,
+					  DMA_FROM_DEVICE);
+			if (ret != 0) {
+				spi_unmap_buf(master, tx_dev, &xfer->tx_sg,
+					      DMA_TO_DEVICE);
+				return ret;
+			}
+		}
+	}
+
+	master->cur_msg_mapped = true;
+
+	return 0;
+}
+
+static int spi_unmap_msg(struct spi_master *master, struct spi_message *msg)
+{
+	struct spi_transfer *xfer;
+	struct device *tx_dev, *rx_dev;
+
+	if (!master->cur_msg_mapped || !master->can_dma)
+		return 0;
+
+	tx_dev = &master->dma_tx->dev->device;
+	rx_dev = &master->dma_rx->dev->device;
+
+	list_for_each_entry(xfer, &msg->transfers, transfer_list) {
+		if (!master->can_dma(master, msg->spi, xfer))
+			continue;
+
+		spi_unmap_buf(master, rx_dev, &xfer->rx_sg, DMA_FROM_DEVICE);
+		spi_unmap_buf(master, tx_dev, &xfer->tx_sg, DMA_TO_DEVICE);
+	}
+
+	return 0;
+}
+
 /*
  * spi_transfer_one_message - Default implementation of transfer_one_message()
  *
@@ -684,6 +849,10 @@ static void spi_pump_messages(struct kthread_work *work)
 		}
 		master->busy = false;
 		spin_unlock_irqrestore(&master->queue_lock, flags);
+		kfree(master->dummy_rx);
+		master->dummy_rx = NULL;
+		kfree(master->dummy_tx);
+		master->dummy_tx = NULL;
 		if (master->unprepare_transfer_hardware &&
 		    master->unprepare_transfer_hardware(master))
 			dev_err(&master->dev,
@@ -750,6 +919,13 @@ static void spi_pump_messages(struct kthread_work *work)
 		master->cur_msg_prepared = true;
 	}
 
+	ret = spi_map_msg(master, master->cur_msg);
+	if (ret) {
+		master->cur_msg->status = ret;
+		spi_finalize_current_message(master);
+		return;
+	}
+
 	ret = master->transfer_one_message(master, master->cur_msg);
 	if (ret) {
 		dev_err(&master->dev,
@@ -837,6 +1013,8 @@ void spi_finalize_current_message(struct spi_master *master)
 	queue_kthread_work(&master->kworker, &master->pump_messages);
 	spin_unlock_irqrestore(&master->queue_lock, flags);
 
+	spi_unmap_msg(master, mesg);
+
 	if (master->cur_msg_prepared && master->unprepare_message) {
 		ret = master->unprepare_message(master, mesg);
 		if (ret) {
@@ -1370,6 +1548,8 @@ int spi_register_master(struct spi_master *master)
 	mutex_init(&master->bus_lock_mutex);
 	master->bus_lock_flag = 0;
 	init_completion(&master->xfer_completion);
+	if (!master->max_dma_len)
+		master->max_dma_len = INT_MAX;
 
 	/* register the device, then userspace will see it.
 	 * registration fails if the bus ID is in use.
diff --git a/include/linux/spi/spi.h b/include/linux/spi/spi.h
index 4203c66d8803..36c86ef51ff3 100644
--- a/include/linux/spi/spi.h
+++ b/include/linux/spi/spi.h
@@ -24,6 +24,9 @@
 #include <linux/slab.h>
 #include <linux/kthread.h>
 #include <linux/completion.h>
+#include <linux/scatterlist.h>
+
+struct dma_chan;
 
 /*
  * INTERFACES between SPI master-side drivers and SPI infrastructure.
@@ -266,6 +269,7 @@ static inline void spi_unregister_driver(struct spi_driver *sdrv)
  * @auto_runtime_pm: the core should ensure a runtime PM reference is held
  *                   while the hardware is prepared, using the parent
  *                   device for the spidev
+ * @max_dma_len: Maximum length of a DMA transfer for the device.
  * @prepare_transfer_hardware: a message will soon arrive from the queue
  *	so the subsystem requests the driver to prepare the transfer hardware
  *	by issuing this call
@@ -348,6 +352,8 @@ struct spi_master {
 #define SPI_MASTER_HALF_DUPLEX	BIT(0)		/* can't do full duplex */
 #define SPI_MASTER_NO_RX	BIT(1)		/* can't do buffer read */
 #define SPI_MASTER_NO_TX	BIT(2)		/* can't do buffer write */
+#define SPI_MASTER_MUST_RX      BIT(3)		/* requires rx */
+#define SPI_MASTER_MUST_TX      BIT(4)		/* requires tx */
 
 	/* lock and mutex for SPI bus locking */
 	spinlock_t		bus_lock_spinlock;
@@ -390,6 +396,17 @@ struct spi_master {
 	void			(*cleanup)(struct spi_device *spi);
 
 	/*
+	 * Used to enable core support for DMA handling, if can_dma()
+	 * exists and returns true then the transfer will be mapped
+	 * prior to transfer_one() being called.  The driver should
+	 * not modify or store xfer and dma_tx and dma_rx must be set
+	 * while the device is prepared.
+	 */
+	bool			(*can_dma)(struct spi_master *master,
+					   struct spi_device *spi,
+					   struct spi_transfer *xfer);
+
+	/*
 	 * These hooks are for drivers that want to use the generic
 	 * master transfer queueing mechanism. If these are used, the
 	 * transfer() function above must NOT be specified by the driver.
@@ -407,7 +424,9 @@ struct spi_master {
 	bool				rt;
 	bool				auto_runtime_pm;
 	bool                            cur_msg_prepared;
+	bool				cur_msg_mapped;
 	struct completion               xfer_completion;
+	size_t				max_dma_len;
 
 	int (*prepare_transfer_hardware)(struct spi_master *master);
 	int (*transfer_one_message)(struct spi_master *master,
@@ -428,6 +447,14 @@ struct spi_master {
 
 	/* gpio chip select */
 	int			*cs_gpios;
+
+	/* DMA channels for use with core dmaengine helpers */
+	struct dma_chan		*dma_tx;
+	struct dma_chan		*dma_rx;
+
+	/* dummy data for full duplex devices */
+	void			*dummy_rx;
+	void			*dummy_tx;
 };
 
 static inline void *spi_master_get_devdata(struct spi_master *master)
@@ -512,6 +539,8 @@ extern struct spi_master *spi_busnum_to_master(u16 busnum);
  *	(optionally) changing the chipselect status, then starting
  *	the next transfer or completing this @spi_message.
  * @transfer_list: transfers are sequenced through @spi_message.transfers
+ * @tx_sg: Scatterlist for transmit, currently not for client use
+ * @rx_sg: Scatterlist for receive, currently not for client use
  *
  * SPI transfers always write the same number of bytes as they read.
  * Protocol drivers should always provide @rx_buf and/or @tx_buf.
@@ -579,6 +608,8 @@ struct spi_transfer {
 
 	dma_addr_t	tx_dma;
 	dma_addr_t	rx_dma;
+	struct sg_table tx_sg;
+	struct sg_table rx_sg;
 
 	unsigned	cs_change:1;
 	unsigned	tx_nbits:3;