Re: [PATCH v2 4/4] spi: spi-davinci: Add support for SPI_CS_WORD

[David Lechner]

On 09/13/2018 08:44 AM, Geert Uytterhoeven wrote:
> Hi David,
>
> On Thu, Sep 13, 2018 at 2:40 AM David Lechner <david@xxxxxxxxxxxxxx> wrote:
> > This adds support for the SPI_CS_WORD flag to the TI DaVinci SPI
> > driver. This mode can be used as long as we are using the hardware
> > chip select and not a GPIO chip select.
> >
> > Signed-off-by: David Lechner <david@xxxxxxxxxxxxxx>
> > ---
> >   drivers/spi/spi-davinci.c | 11 ++++++++---
> >   1 file changed, 8 insertions(+), 3 deletions(-)
> >
> > diff --git a/drivers/spi/spi-davinci.c b/drivers/spi/spi-davinci.c
> > index d502cf504deb..8f7dcbc53c57 100644
> > --- a/drivers/spi/spi-davinci.c
> > +++ b/drivers/spi/spi-davinci.c
> > @@ -230,7 +230,8 @@ static void davinci_spi_chipselect(struct spi_device *spi, int value)
> >                                  !(spi->mode & SPI_CS_HIGH));
> >          } else {
> >                  if (value == BITBANG_CS_ACTIVE) {
> > -                       spidat1 |= SPIDAT1_CSHOLD_MASK;
> > +                       if (!(spi->mode & SPI_CS_WORD))
> > +                               spidat1 |= SPIDAT1_CSHOLD_MASK;
> >                          spidat1 &= ~(0x1 << chip_sel);
> >                  }
> >          }
> > @@ -440,8 +441,12 @@ static int davinci_spi_setup(struct spi_device *spi)
> >                          return retval;
> >                  }
> >
> > -               if (internal_cs)
> > +               if (internal_cs) {
> >                          set_io_bits(dspi->base + SPIPC0, 1 << spi->chip_select);
> > +               } else if (spi->mode & SPI_CS_WORD) {
> > +                       dev_err(&spi->dev, "SPI_CS_WORD can't be use with GPIO CS\n");
> > +                       return -EINVAL;
>
> Does the SPI core fall back to splitting the transfer in this case?

Hmm... it doesn't look like it.

I suppose it might be best to modify the SPI core to say:

	if ((spi->mode & SPI_CS_WORD) && (!(ctlr->mode_bits & SPI_CS_WORD) ||
					  gpio_is_valid(spi->cs_gpio)) {

instead of:

	if ((spi->mode & SPI_CS_WORD) && !(ctlr->mode_bits & SPI_CS_WORD)) {

Then we could drop the error above.

> > +               }
>
> Gr{oetje,eeting}s,
>
>                          Geert
>
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
>
> In personal conversations with technical people, I call myself a hacker. But
> when I'm talking to journalists I just say "programmer" or something like that.
>                                  -- Linus Torvalds