On Wed, Feb 6, 2013 at 2:35 AM, Grant Likely <grant.lik...@secretlab.ca> wrote:
> On Tue, 5 Feb 2013 15:09:42 -0800, Girish K S <girishks2...@gmail.com> wrote:
>> The 64xx spi driver supports partial polling mode.
>> Only the last chunk of the transfer length is transferred
>> or recieved in polling mode.
>>
>> Some SoC's that adopt this controller might not have have dma
>> interface. This patch adds support for complete polling mode
>> and gives flexibity for the user to select poll/dma mode.
>>
>> Signed-off-by: Girish K S <ks.g...@samsung.com>
>> ---
>> drivers/spi/spi-s3c64xx.c | 65
>> +++++++++++++++++++++------------------------
>> 1 file changed, 30 insertions(+), 35 deletions(-)
>>
>> diff --git a/drivers/spi/spi-s3c64xx.c b/drivers/spi/spi-s3c64xx.c
>> index b770f88..90770bd 100644
>> --- a/drivers/spi/spi-s3c64xx.c
>> +++ b/drivers/spi/spi-s3c64xx.c
>> @@ -345,19 +348,7 @@ static void enable_datapath(struct
>> s3c64xx_spi_driver_data *sdd,
>>
>> chcfg = readl(regs + S3C64XX_SPI_CH_CFG);
>> chcfg &= ~S3C64XX_SPI_CH_TXCH_ON;
>> -
>> - if (dma_mode) {
>> chcfg &= ~S3C64XX_SPI_CH_RXCH_ON;
>> - } else {
>> - /* Always shift in data in FIFO, even if xfer is Tx only,
>> - * this helps setting PCKT_CNT value for generating clocks
>> - * as exactly needed.
>> - */
>> - chcfg |= S3C64XX_SPI_CH_RXCH_ON;
>> - writel(((xfer->len * 8 / sdd->cur_bpw) & 0xffff)
>> - | S3C64XX_SPI_PACKET_CNT_EN,
>> - regs + S3C64XX_SPI_PACKET_CNT);
>> - }
>
> The removes a block of code, but leaves the modification of chcfg where
> it is without fixing the indentation. I could also use some help
Yes sorry missed that indentation.
> understanding why this particular else block is moved down in the
> function.
The else block is related only to polling mode, so moved it inside the
rx block.
Clock generation and RX is enabled only when there is a valid Rx
buffer to receive the data from FIFO.
>
>> if (xfer->tx_buf != NULL) {
>> sdd->state |= TXBUSY;
>> @@ -385,6 +376,10 @@ static void enable_datapath(struct
>> s3c64xx_spi_driver_data *sdd,
>>
>> if (xfer->rx_buf != NULL) {
>> sdd->state |= RXBUSY;
>> + chcfg |= S3C64XX_SPI_CH_RXCH_ON;
>> + writel(((xfer->len * 8 / sdd->cur_bpw) & 0xffff)
>> + | S3C64XX_SPI_PACKET_CNT_EN,
>> + regs + S3C64XX_SPI_PACKET_CNT);
>>
>> if (sdd->port_conf->high_speed && sdd->cur_speed >= 30000000UL
>> && !(sdd->cur_mode & SPI_CPHA))
>> @@ -392,10 +387,6 @@ static void enable_datapath(struct
>> s3c64xx_spi_driver_data *sdd,
>>
>> if (dma_mode) {
>> modecfg |= S3C64XX_SPI_MODE_RXDMA_ON;
>> - chcfg |= S3C64XX_SPI_CH_RXCH_ON;
>> - writel(((xfer->len * 8 / sdd->cur_bpw) & 0xffff)
>> - | S3C64XX_SPI_PACKET_CNT_EN,
>> - regs + S3C64XX_SPI_PACKET_CNT);
>> prepare_dma(&sdd->rx_dma, xfer->len, xfer->rx_dma);
>> }
>> }
>> @@ -421,6 +412,9 @@ static inline void enable_cs(struct
>> s3c64xx_spi_driver_data *sdd,
>>
>> cs = spi->controller_data;
>> gpio_set_value(cs->line, spi->mode & SPI_CS_HIGH ? 1 : 0);
>> +
>> + /* Start the signals */
>> + writel(0, sdd->regs + S3C64XX_SPI_SLAVE_SEL);
>> }
>>
>> static int wait_for_xfer(struct s3c64xx_spi_driver_data *sdd,
>> @@ -480,16 +474,19 @@ static int wait_for_xfer(struct
>> s3c64xx_spi_driver_data *sdd,
>>
>> switch (sdd->cur_bpw) {
>> case 32:
>> - ioread32_rep(regs + S3C64XX_SPI_RX_DATA,
>> - xfer->rx_buf, xfer->len / 4);
>> + for (val = 0; val < (xfer->len / 4); val++)
>> + *((u32 *)xfer->rx_buf + val) =
>> + ioread32(regs + S3C64XX_SPI_RX_DATA);
>> break;
>> case 16:
>> - ioread16_rep(regs + S3C64XX_SPI_RX_DATA,
>> - xfer->rx_buf, xfer->len / 2);
>> + for (val = 0; val < (xfer->len / 2); val++)
>> + *((u16 *)xfer->rx_buf + val) =
>> + ioread16(regs + S3C64XX_SPI_RX_DATA);
>> break;
>> default:
>> - ioread8_rep(regs + S3C64XX_SPI_RX_DATA,
>> - xfer->rx_buf, xfer->len);
>> + for (val = 0; val < xfer->len; val++)
>> + *((u8 *)xfer->rx_buf + val) =
>> + ioread8(regs + S3C64XX_SPI_RX_DATA);
>
> What are all of the above lines changed? That seems to have nothing to
> do with making the driver be albe to do polling transfers. Nor is it
> described in the patch description.
As per the original implementation, the polling path is taken only for
the last chunk of tarnsfer size. The last chunk is always
<= all other other previous transfers. so readx_rep can read the FIFO
data successfully for the given transfer size.
But in polling mode, readx_rep would read junk if the transfer size is
more than the threshold limit.
e.g
dd if=/dev/mtd0 of=temp.bin bs=512 count=1
In this case transfer size to receive is 512. The threshold for FIFO
is 128 bytes (half of FIFO depth, i didnt modify this as it is in DMA
transfer also)
Now in the above receive case with repx_read, i always received first
128 valid bytes and all the rest as junk. With the above modification
I just make sure i read out all the data whatever is received 1 by 1.
>
> I think this patch needs some more work. It certainly need to be
> described better, and it appears that some of the changes really should be
> in a separate patch.
sure I will give a clear description of what is done.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
