The DMA transfers happen only if the transfered data are larger than 512 bytes. Otherwise PIO is used. This is a small speed optimization.

The DMA transfer doesn't work if unaligned transfer is requested due to the limitation of the DMA controller. This has to be fixed by introducing generic bounce buffer. Therefore the DMA feature is now disabled by default.

Signed-off-by: Marek Vasut marex@denx.de Cc: Fabio Estevam festevam@freescale.com Cc: Otavio Salvador otavio@ossystems.com.br Cc: Stefano Babic sbabic@denx.de Cc: Wolfgang Denk wd@denx.de --- drivers/spi/mxs_spi.c | 105 ++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 104 insertions(+), 1 deletion(-)

diff --git a/drivers/spi/mxs_spi.c b/drivers/spi/mxs_spi.c index 0459b3b..aec7514 100644 --- a/drivers/spi/mxs_spi.c +++ b/drivers/spi/mxs_spi.c @@ -31,17 +31,32 @@ #include <asm/arch/clock.h> #include <asm/arch/imx-regs.h> #include <asm/arch/sys_proto.h> +#include <asm/arch/dma.h>

#define MXS_SPI_MAX_TIMEOUT 1000000 #define MXS_SPI_PORT_OFFSET 0x2000 #define MXS_SSP_CHIPSELECT_MASK 0x00300000 #define MXS_SSP_CHIPSELECT_SHIFT 20

+#define MXS_SSP_SMALL_TRANSFER 512 + +/* + * CONFIG_MXS_SPI_DMA_ENABLE: Experimental mixed PIO/DMA support for MXS SPI + * host. Use with utmost caution! + * + * Enabling this is not yet recommended since this + * still doesn't support transfers to/from unaligned + * addresses. Therefore this driver will not work + * for example with saving environment. This is + * caused by DMA alignment constraints on MXS. + */ + struct mxs_spi_slave { struct spi_slave slave; uint32_t max_khz; uint32_t mode; struct mx28_ssp_regs *regs; + struct mxs_dma_desc *desc; };

static inline struct mxs_spi_slave *to_mxs_slave(struct spi_slave *slave) @@ -69,6 +84,7 @@ struct spi_slave *spi_setup_slave(unsigned int bus, unsigned int cs, uint32_t addr; struct mx28_ssp_regs *ssp_regs; int reg; + struct mxs_dma_desc *desc;

if (!spi_cs_is_valid(bus, cs)) { printf("mxs_spi: invalid bus %d / chip select %d\n", bus, cs); @@ -79,6 +95,13 @@ struct spi_slave *spi_setup_slave(unsigned int bus, unsigned int cs, if (!mxs_slave) return NULL;

+ desc = mxs_dma_desc_alloc(); + if (!desc) + goto err_desc; + + if (mxs_dma_init_channel(bus)) + goto err_init; + addr = MXS_SSP0_BASE + (bus * MXS_SPI_PORT_OFFSET);

mxs_slave->slave.bus = bus; @@ -86,6 +109,7 @@ struct spi_slave *spi_setup_slave(unsigned int bus, unsigned int cs, mxs_slave->max_khz = max_hz / 1000; mxs_slave->mode = mode; mxs_slave->regs = (struct mx28_ssp_regs *)addr; + mxs_slave->desc = desc; ssp_regs = mxs_slave->regs;

reg = readl(&ssp_regs->hw_ssp_ctrl0); @@ -94,11 +118,18 @@ struct spi_slave *spi_setup_slave(unsigned int bus, unsigned int cs,

writel(reg, &ssp_regs->hw_ssp_ctrl0); return &mxs_slave->slave; + +err_init: + mxs_dma_desc_free(desc); +err_desc: + free(mxs_slave); + return NULL; }

void spi_free_slave(struct spi_slave *slave) { struct mxs_spi_slave *mxs_slave = to_mxs_slave(slave); + mxs_dma_desc_free(mxs_slave->desc); free(mxs_slave); }

@@ -195,15 +226,81 @@ static int mxs_spi_xfer_pio(struct mxs_spi_slave *slave,

}

+static int mxs_spi_xfer_dma(struct mxs_spi_slave *slave, + char *data, int length, int write, unsigned long flags) +{ + struct mxs_dma_desc *desc = slave->desc; + struct mx28_ssp_regs *ssp_regs = slave->regs; + uint32_t ctrl0 = SSP_CTRL0_DATA_XFER; + uint32_t cache_data_count; + int dmach; + + memset(desc, 0, sizeof(struct mxs_dma_desc)); + desc->address = (dma_addr_t)desc; + + if (flags & SPI_XFER_BEGIN) + ctrl0 |= SSP_CTRL0_LOCK_CS; + if (flags & SPI_XFER_END) + ctrl0 |= SSP_CTRL0_IGNORE_CRC; + if (!write) + ctrl0 |= SSP_CTRL0_READ; + + writel(length, &ssp_regs->hw_ssp_xfer_size); + + if (length % ARCH_DMA_MINALIGN) + cache_data_count = roundup(length, ARCH_DMA_MINALIGN); + else + cache_data_count = length; + + if (!write) { + slave->desc->cmd.data = MXS_DMA_DESC_COMMAND_DMA_WRITE; + slave->desc->cmd.address = (dma_addr_t)data; + } else { + slave->desc->cmd.data = MXS_DMA_DESC_COMMAND_DMA_READ; + slave->desc->cmd.address = (dma_addr_t)data; + + /* Flush data to DRAM so DMA can pick them up */ + flush_dcache_range((uint32_t)data, + (uint32_t)(data + cache_data_count)); + } + + slave->desc->cmd.data |= MXS_DMA_DESC_IRQ | MXS_DMA_DESC_DEC_SEM | + (length << MXS_DMA_DESC_BYTES_OFFSET) | + (1 << MXS_DMA_DESC_PIO_WORDS_OFFSET) | + MXS_DMA_DESC_WAIT4END; + + slave->desc->cmd.pio_words[0] = ctrl0; + + dmach = MXS_DMA_CHANNEL_AHB_APBH_SSP0 + slave->slave.bus; + mxs_dma_desc_append(dmach, slave->desc); + if (mxs_dma_go(dmach)) + return -EINVAL; + + /* The data arrived into DRAM, invalidate cache over them */ + if (!write) { + invalidate_dcache_range((uint32_t)data, + (uint32_t)(data + cache_data_count)); + } + + return 0; +} + int spi_xfer(struct spi_slave *slave, unsigned int bitlen, const void *dout, void *din, unsigned long flags) { struct mxs_spi_slave *mxs_slave = to_mxs_slave(slave); + struct mx28_ssp_regs *ssp_regs = mxs_slave->regs; int len = bitlen / 8; char dummy; int write = 0; char *data = NULL;

+#ifdef CONFIG_MXS_SPI_DMA_ENABLE + const int dma = 1; +#else + const int dma = 0; +#endif + if (bitlen == 0) { if (flags & SPI_XFER_END) { din = (void *)&dummy; @@ -226,5 +323,11 @@ int spi_xfer(struct spi_slave *slave, unsigned int bitlen, write = 0; }

- return mxs_spi_xfer_pio(mxs_slave, data, len, write, flags); + if (!dma || (len < MXS_SSP_SMALL_TRANSFER)) { + writel(SSP_CTRL1_DMA_ENABLE, &ssp_regs->hw_ssp_ctrl1_clr); + return mxs_spi_xfer_pio(mxs_slave, data, len, write, flags); + } else { + writel(SSP_CTRL1_DMA_ENABLE, &ssp_regs->hw_ssp_ctrl1_set); + return mxs_spi_xfer_dma(mxs_slave, data, len, write, flags); + } }