| Message ID | 20181109220044.24843-4-matheus.bernardino@usp.br (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | staging:iio:ad2s90: Add dt support and move out of staging | expand |
On Fri, 9 Nov 2018 20:00:41 -0200 Matheus Tavares <matheus.bernardino@usp.br> wrote: > This patch adds a max frequency check at the beginning of ad2s90_probe > function so that when it is set to a value above 0.83Mhz, dev_err is > called with an appropriate message and -EINVAL is returned. > > The defined limit is 0.83Mhz instead of 2Mhz, which is the chip's max > frequency as specified in the datasheet, because, as also specified in > the datasheet, a 600ns delay is expected between the application of a > logic LO to CS and the application of SCLK. Since the delay is not > implemented in the spi code, to satisfy it, SCLK's period should be at > most 2 * 600ns, so the max frequency should be 1 / (2 * 6e-7), which > gives roughly 830000Hz. > > Signed-off-by: Matheus Tavares <matheus.bernardino@usp.br> > Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com> > --- > Alex's S-o-B was added because he gave the code suggestion for this > patch. If he gave the actual code then he should be the credited author git commit --amend --author=... If it was just a suggestion, then an informal tag such as Suggested-by is usually used. Signed-off-by has formal legal meaning to do with the developer certificate of origin, it's not valid as a way of crediting someone with a contribution to the patch. > > drivers/staging/iio/resolver/ad2s90.c | 12 ++++++++++++ > 1 file changed, 12 insertions(+) > > diff --git a/drivers/staging/iio/resolver/ad2s90.c b/drivers/staging/iio/resolver/ad2s90.c > index 95c118c48400..949ff55ac6b0 100644 > --- a/drivers/staging/iio/resolver/ad2s90.c > +++ b/drivers/staging/iio/resolver/ad2s90.c > @@ -19,6 +19,12 @@ > #include <linux/iio/iio.h> > #include <linux/iio/sysfs.h> > > +/* > + * Although chip's max frequency is 2Mhz, it needs 600ns between CS and the > + * first falling edge of SCLK, so frequency should be at most 1 / (2 * 6e-7) > + */ > +#define AD2S90_MAX_SPI_FREQ_HZ 830000 > + > struct ad2s90_state { > struct mutex lock; > struct spi_device *sdev; > @@ -78,6 +84,12 @@ static int ad2s90_probe(struct spi_device *spi) > struct iio_dev *indio_dev; > struct ad2s90_state *st; > > + if (spi->max_speed_hz > AD2S90_MAX_SPI_FREQ_HZ) { > + dev_err(&spi->dev, "SPI CLK, %d Hz exceeds %d Hz\n", > + spi->max_speed_hz, AD2S90_MAX_SPI_FREQ_HZ); > + return -EINVAL; > + } Now this is interesting as a follow up to the previous and may actually answer the question I raised. Let's see what Mark and Rob come back with. If possible I'd like us to resolve this once and for all with a thread to point people back to! > + > indio_dev = devm_iio_device_alloc(&spi->dev, sizeof(*st)); > if (!indio_dev) > return -ENOMEM;
diff --git a/drivers/staging/iio/resolver/ad2s90.c b/drivers/staging/iio/resolver/ad2s90.c index 95c118c48400..949ff55ac6b0 100644 --- a/drivers/staging/iio/resolver/ad2s90.c +++ b/drivers/staging/iio/resolver/ad2s90.c @@ -19,6 +19,12 @@ #include <linux/iio/iio.h> #include <linux/iio/sysfs.h> +/* + * Although chip's max frequency is 2Mhz, it needs 600ns between CS and the + * first falling edge of SCLK, so frequency should be at most 1 / (2 * 6e-7) + */ +#define AD2S90_MAX_SPI_FREQ_HZ 830000 + struct ad2s90_state { struct mutex lock; struct spi_device *sdev; @@ -78,6 +84,12 @@ static int ad2s90_probe(struct spi_device *spi) struct iio_dev *indio_dev; struct ad2s90_state *st; + if (spi->max_speed_hz > AD2S90_MAX_SPI_FREQ_HZ) { + dev_err(&spi->dev, "SPI CLK, %d Hz exceeds %d Hz\n", + spi->max_speed_hz, AD2S90_MAX_SPI_FREQ_HZ); + return -EINVAL; + } + indio_dev = devm_iio_device_alloc(&spi->dev, sizeof(*st)); if (!indio_dev) return -ENOMEM;