spi: Rework DMA mapped flag

Merge series from Andy Shevchenko <andriy.shevchenko@linux.intel.com>:

The first part of the series (patches 1 to 7) is an introduction
of a new helper followed by the user conversion.

This consolidates the same code and also makes patch 8 (last one)
be localised to the SPI core part.

The last patch is the main rework to get rid of a recently introduced
hack with a dummy SG list and move to the transfer-based DMA mapped
flag.

That said, the patches 1 to 7 may be applied right away since they
have no functional change intended, while the last one needs more
testing and reviewing.

drivers/spi/internals.h

@@ -40,4 +40,12 @@ static inline void spi_unmap_buf(struct spi_controller *ctlr,
 }
 #endif /* CONFIG_HAS_DMA */
 
+static inline bool spi_xfer_is_dma_mapped(struct spi_controller *ctlr,
+					  struct spi_device *spi,
+					  struct spi_transfer *xfer)
+{
+	return ctlr->can_dma && ctlr->can_dma(ctlr, spi, xfer) &&
+	       (xfer->tx_sg_mapped || xfer->rx_sg_mapped);
+}
+
 #endif /* __LINUX_SPI_INTERNALS_H */

drivers/spi/spi-dw-core.c

@@ -19,6 +19,7 @@
 #include <linux/string.h>
 #include <linux/of.h>
 
+#include "internals.h"
 #include "spi-dw.h"
 
 #ifdef CONFIG_DEBUG_FS
@@ -438,8 +439,7 @@ static int dw_spi_transfer_one(struct spi_controller *host,
 	transfer->effective_speed_hz = dws->current_freq;
 
 	/* Check if current transfer is a DMA transaction */
-	if (host->can_dma && host->can_dma(host, spi, transfer))
-		dws->dma_mapped = host->cur_msg_mapped;
+	dws->dma_mapped = spi_xfer_is_dma_mapped(host, spi, transfer);
 
 	/* For poll mode just disable all interrupts */
 	dw_spi_mask_intr(dws, 0xff);

drivers/spi/spi-ingenic.c

@@ -16,6 +16,7 @@
 #include <linux/platform_device.h>
 #include <linux/regmap.h>
 #include <linux/spi/spi.h>
+#include "internals.h"
 
 #define REG_SSIDR	0x0
 #define REG_SSICR0	0x4
@@ -242,11 +243,10 @@ static int spi_ingenic_transfer_one(struct spi_controller *ctlr,
 {
 	struct ingenic_spi *priv = spi_controller_get_devdata(ctlr);
 	unsigned int bits = xfer->bits_per_word ?: spi->bits_per_word;
-	bool can_dma = ctlr->can_dma && ctlr->can_dma(ctlr, spi, xfer);
 
 	spi_ingenic_prepare_transfer(priv, spi, xfer);
 
-	if (ctlr->cur_msg_mapped && can_dma)
+	if (spi_xfer_is_dma_mapped(ctlr, spi, xfer))
 		return spi_ingenic_dma_tx(ctlr, xfer, bits);
 
 	if (bits > 16)

drivers/spi/spi-omap2-mcspi.c

@@ -27,6 +27,8 @@
 
 #include <linux/spi/spi.h>
 
+#include "internals.h"
+
 #include <linux/platform_data/spi-omap2-mcspi.h>
 
 #define OMAP2_MCSPI_MAX_FREQ		48000000
@@ -1208,8 +1210,7 @@ static int omap2_mcspi_transfer_one(struct spi_controller *ctlr,
 		unsigned	count;
 
 		if ((mcspi_dma->dma_rx && mcspi_dma->dma_tx) &&
-		    ctlr->cur_msg_mapped &&
-		    ctlr->can_dma(ctlr, spi, t))
+		    spi_xfer_is_dma_mapped(ctlr, spi, t))
 			omap2_mcspi_set_fifo(spi, t, 1);
 
 		omap2_mcspi_set_enable(spi, 1);
@@ -1220,8 +1221,7 @@ static int omap2_mcspi_transfer_one(struct spi_controller *ctlr,
 					+ OMAP2_MCSPI_TX0);
 
 		if ((mcspi_dma->dma_rx && mcspi_dma->dma_tx) &&
-		    ctlr->cur_msg_mapped &&
-		    ctlr->can_dma(ctlr, spi, t))
+		    spi_xfer_is_dma_mapped(ctlr, spi, t))
 			count = omap2_mcspi_txrx_dma(spi, t);
 		else
 			count = omap2_mcspi_txrx_pio(spi, t);

drivers/spi/spi-pci1xxxx.c

@@ -6,6 +6,7 @@
 
 
 #include <linux/bitfield.h>
+#include <linux/delay.h>
 #include <linux/dma-mapping.h>
 #include <linux/iopoll.h>
 #include <linux/irq.h>
@@ -15,7 +16,7 @@
 #include <linux/pci.h>
 #include <linux/spinlock.h>
 #include <linux/spi/spi.h>
-#include <linux/delay.h>
+#include "internals.h"
 
 #define DRV_NAME "spi-pci1xxxx"
 
@@ -567,7 +568,7 @@ static int pci1xxxx_spi_transfer_with_dma(struct spi_controller *spi_ctlr,
 static int pci1xxxx_spi_transfer_one(struct spi_controller *spi_ctlr,
 				     struct spi_device *spi, struct spi_transfer *xfer)
 {
-	if (spi_ctlr->can_dma(spi_ctlr, spi, xfer) && spi_ctlr->cur_msg_mapped)
+	if (spi_xfer_is_dma_mapped(spi_ctlr, spi, xfer))
 		return pci1xxxx_spi_transfer_with_dma(spi_ctlr, spi, xfer);
 	else
 		return pci1xxxx_spi_transfer_with_io(spi_ctlr, spi, xfer);

drivers/spi/spi-pxa2xx.c

@@ -26,6 +26,7 @@
 
 #include <linux/spi/spi.h>
 
+#include "internals.h"
 #include "spi-pxa2xx.h"
 
 #define TIMOUT_DFLT		1000
@@ -993,11 +994,8 @@ static int pxa2xx_spi_transfer_one(struct spi_controller *controller,
 	}
 
 	dma_thresh = SSCR1_RxTresh(RX_THRESH_DFLT) | SSCR1_TxTresh(TX_THRESH_DFLT);
-	dma_mapped = controller->can_dma &&
-		     controller->can_dma(controller, spi, transfer) &&
-		     controller->cur_msg_mapped;
+	dma_mapped = spi_xfer_is_dma_mapped(controller, spi, transfer);
 	if (dma_mapped) {
-
 		/* Ensure we have the correct interrupt handler */
 		drv_data->transfer_handler = pxa2xx_spi_dma_transfer;
 

drivers/spi/spi-qup.c

@@ -5,6 +5,8 @@
 
 #include <linux/clk.h>
 #include <linux/delay.h>
+#include <linux/dma-mapping.h>
+#include <linux/dmaengine.h>
 #include <linux/err.h>
 #include <linux/interconnect.h>
 #include <linux/interrupt.h>
@@ -16,8 +18,7 @@
 #include <linux/pm_opp.h>
 #include <linux/pm_runtime.h>
 #include <linux/spi/spi.h>
-#include <linux/dmaengine.h>
-#include <linux/dma-mapping.h>
+#include "internals.h"
 
 #define QUP_CONFIG			0x0000
 #define QUP_STATE			0x0004
@@ -709,9 +710,7 @@ static int spi_qup_io_prep(struct spi_device *spi, struct spi_transfer *xfer)
 
 	if (controller->n_words <= (controller->in_fifo_sz / sizeof(u32)))
 		controller->mode = QUP_IO_M_MODE_FIFO;
-	else if (spi->controller->can_dma &&
-		 spi->controller->can_dma(spi->controller, spi, xfer) &&
-		 spi->controller->cur_msg_mapped)
+	else if (spi_xfer_is_dma_mapped(spi->controller, spi, xfer))
 		controller->mode = QUP_IO_M_MODE_BAM;
 	else
 		controller->mode = QUP_IO_M_MODE_BLOCK;

drivers/spi/spi.c

@@ -1220,11 +1220,6 @@ void spi_unmap_buf(struct spi_controller *ctlr, struct device *dev,
 	spi_unmap_buf_attrs(ctlr, dev, sgt, dir, 0);
 }
 
-/* Dummy SG for unidirect transfers */
-static struct scatterlist dummy_sg = {
-	.page_link = SG_END,
-};
-
 static int __spi_map_msg(struct spi_controller *ctlr, struct spi_message *msg)
 {
 	struct device *tx_dev, *rx_dev;
@@ -1263,8 +1258,8 @@ static int __spi_map_msg(struct spi_controller *ctlr, struct spi_message *msg)
 						attrs);
 			if (ret != 0)
 				return ret;
-		} else {
-			xfer->tx_sg.sgl = &dummy_sg;
+
+			xfer->tx_sg_mapped = true;
 		}
 
 		if (xfer->rx_buf != NULL) {
@@ -1278,8 +1273,8 @@ static int __spi_map_msg(struct spi_controller *ctlr, struct spi_message *msg)
 
 				return ret;
 			}
-		} else {
-			xfer->rx_sg.sgl = &dummy_sg;
+
+			xfer->rx_sg_mapped = true;
 		}
 	}
 	/* No transfer has been mapped, bail out with success */
@@ -1288,7 +1283,6 @@ static int __spi_map_msg(struct spi_controller *ctlr, struct spi_message *msg)
 
 	ctlr->cur_rx_dma_dev = rx_dev;
 	ctlr->cur_tx_dma_dev = tx_dev;
-	ctlr->cur_msg_mapped = true;
 
 	return 0;
 }
@@ -1299,57 +1293,46 @@ static int __spi_unmap_msg(struct spi_controller *ctlr, struct spi_message *msg)
 	struct device *tx_dev = ctlr->cur_tx_dma_dev;
 	struct spi_transfer *xfer;
 
-	if (!ctlr->cur_msg_mapped || !ctlr->can_dma)
-		return 0;
-
 	list_for_each_entry(xfer, &msg->transfers, transfer_list) {
 		/* The sync has already been done after each transfer. */
 		unsigned long attrs = DMA_ATTR_SKIP_CPU_SYNC;
 
-		if (!ctlr->can_dma(ctlr, msg->spi, xfer))
-			continue;
+		if (xfer->rx_sg_mapped)
+			spi_unmap_buf_attrs(ctlr, rx_dev, &xfer->rx_sg,
+					    DMA_FROM_DEVICE, attrs);
+		xfer->rx_sg_mapped = false;
 
-		spi_unmap_buf_attrs(ctlr, rx_dev, &xfer->rx_sg,
-				    DMA_FROM_DEVICE, attrs);
-		spi_unmap_buf_attrs(ctlr, tx_dev, &xfer->tx_sg,
-				    DMA_TO_DEVICE, attrs);
+		if (xfer->tx_sg_mapped)
+			spi_unmap_buf_attrs(ctlr, tx_dev, &xfer->tx_sg,
+					    DMA_TO_DEVICE, attrs);
+		xfer->tx_sg_mapped = false;
 	}
 
-	ctlr->cur_msg_mapped = false;
-
 	return 0;
 }
 
-static void spi_dma_sync_for_device(struct spi_controller *ctlr, struct spi_message *msg,
+static void spi_dma_sync_for_device(struct spi_controller *ctlr,
 				    struct spi_transfer *xfer)
 {
 	struct device *rx_dev = ctlr->cur_rx_dma_dev;
 	struct device *tx_dev = ctlr->cur_tx_dma_dev;
 
-	if (!ctlr->cur_msg_mapped)
-		return;
-
-	if (!ctlr->can_dma(ctlr, msg->spi, xfer))
-		return;
-
-	dma_sync_sgtable_for_device(tx_dev, &xfer->tx_sg, DMA_TO_DEVICE);
-	dma_sync_sgtable_for_device(rx_dev, &xfer->rx_sg, DMA_FROM_DEVICE);
+	if (xfer->tx_sg_mapped)
+		dma_sync_sgtable_for_device(tx_dev, &xfer->tx_sg, DMA_TO_DEVICE);
+	if (xfer->rx_sg_mapped)
+		dma_sync_sgtable_for_device(rx_dev, &xfer->rx_sg, DMA_FROM_DEVICE);
 }
 
-static void spi_dma_sync_for_cpu(struct spi_controller *ctlr, struct spi_message *msg,
+static void spi_dma_sync_for_cpu(struct spi_controller *ctlr,
 				 struct spi_transfer *xfer)
 {
 	struct device *rx_dev = ctlr->cur_rx_dma_dev;
 	struct device *tx_dev = ctlr->cur_tx_dma_dev;
 
-	if (!ctlr->cur_msg_mapped)
-		return;
-
-	if (!ctlr->can_dma(ctlr, msg->spi, xfer))
-		return;
-
-	dma_sync_sgtable_for_cpu(rx_dev, &xfer->rx_sg, DMA_FROM_DEVICE);
-	dma_sync_sgtable_for_cpu(tx_dev, &xfer->tx_sg, DMA_TO_DEVICE);
+	if (xfer->rx_sg_mapped)
+		dma_sync_sgtable_for_cpu(rx_dev, &xfer->rx_sg, DMA_FROM_DEVICE);
+	if (xfer->tx_sg_mapped)
+		dma_sync_sgtable_for_cpu(tx_dev, &xfer->tx_sg, DMA_TO_DEVICE);
 }
 #else /* !CONFIG_HAS_DMA */
 static inline int __spi_map_msg(struct spi_controller *ctlr,
@@ -1365,13 +1348,11 @@ static inline int __spi_unmap_msg(struct spi_controller *ctlr,
 }
 
 static void spi_dma_sync_for_device(struct spi_controller *ctrl,
-				    struct spi_message *msg,
 				    struct spi_transfer *xfer)
 {
 }
 
 static void spi_dma_sync_for_cpu(struct spi_controller *ctrl,
-				 struct spi_message *msg,
 				 struct spi_transfer *xfer)
 {
 }
@@ -1643,13 +1624,13 @@ static int spi_transfer_one_message(struct spi_controller *ctlr,
 			reinit_completion(&ctlr->xfer_completion);
 
 fallback_pio:
-			spi_dma_sync_for_device(ctlr, msg, xfer);
+			spi_dma_sync_for_device(ctlr, xfer);
 			ret = ctlr->transfer_one(ctlr, msg->spi, xfer);
 			if (ret < 0) {
-				spi_dma_sync_for_cpu(ctlr, msg, xfer);
+				spi_dma_sync_for_cpu(ctlr, xfer);
 
-				if (ctlr->cur_msg_mapped &&
-				   (xfer->error & SPI_TRANS_FAIL_NO_START)) {
+				if ((xfer->tx_sg_mapped || xfer->rx_sg_mapped) &&
+				    (xfer->error & SPI_TRANS_FAIL_NO_START)) {
 					__spi_unmap_msg(ctlr, msg);
 					ctlr->fallback = true;
 					xfer->error &= ~SPI_TRANS_FAIL_NO_START;
@@ -1671,7 +1652,7 @@ static int spi_transfer_one_message(struct spi_controller *ctlr,
 					msg->status = ret;
 			}
 
-			spi_dma_sync_for_cpu(ctlr, msg, xfer);
+			spi_dma_sync_for_cpu(ctlr, xfer);
 		} else {
 			if (xfer->len)
 				dev_err(&msg->spi->dev,

include/linux/spi/spi.h

@@ -447,7 +447,6 @@ extern struct spi_device *spi_new_ancillary_device(struct spi_device *spi, u8 ch
  * @cur_msg_need_completion: Flag used internally to opportunistically skip
  *	the @cur_msg_completion. This flag is used to signal the context that
  *	is running spi_finalize_current_message() that it needs to complete()
- * @cur_msg_mapped: message has been mapped for DMA
  * @fallback: fallback to PIO if DMA transfer return failure with
  *	SPI_TRANS_FAIL_NO_START.
  * @last_cs_mode_high: was (mode & SPI_CS_HIGH) true on the last call to set_cs.
@@ -708,7 +707,6 @@ struct spi_controller {
 	bool				running;
 	bool				rt;
 	bool				auto_runtime_pm;
-	bool				cur_msg_mapped;
 	bool                            fallback;
 	bool				last_cs_mode_high;
 	s8				last_cs[SPI_CS_CNT_MAX];
@@ -981,6 +979,8 @@ struct spi_res {
  *      transfer this transfer. Set to 0 if the SPI bus driver does
  *      not support it.
  * @transfer_list: transfers are sequenced through @spi_message.transfers
+ * @tx_sg_mapped: If true, the @tx_sg is mapped for DMA
+ * @rx_sg_mapped: If true, the @rx_sg is mapped for DMA
  * @tx_sg: Scatterlist for transmit, currently not for client use
  * @rx_sg: Scatterlist for receive, currently not for client use
  * @ptp_sts_word_pre: The word (subject to bits_per_word semantics) offset
@@ -1077,10 +1077,13 @@ struct spi_transfer {
 #define SPI_TRANS_FAIL_IO	BIT(1)
 	u16		error;
 
-	dma_addr_t	tx_dma;
-	dma_addr_t	rx_dma;
+	bool		tx_sg_mapped;
+	bool		rx_sg_mapped;
+
 	struct sg_table tx_sg;
 	struct sg_table rx_sg;
+	dma_addr_t	tx_dma;
+	dma_addr_t	rx_dma;
 
 	unsigned	dummy_data:1;
 	unsigned	cs_off:1;