| Message ID | 20240920-ad7606_add_iio_backend_support-v2-10-0e78782ae7d0@baylibre.com |
|---|---|
| State | New |
| Headers | show |
| Series | Add iio backend compatibility for ad7606 | expand |
On Fri, 20 Sep 2024 17:33:30 +0000 Guillaume Stols <gstols@baylibre.com> wrote: > Since the pwm was introduced before backend, there was an example use > whit triggered buffers. However, using it may be dangerous, because if > the PWM goes too fast, a new conversion can be triggered before the > transmission is over, whit the associated mess in the buffers. > Until a solution is found to mitigate this risk, for instante CRC > support, the PWM will be disabled. > > Signed-off-by: Guillaume Stols <gstols@baylibre.com> Whilst I can see the logic introducing it in two steps, maybe shout a bit more about that in the original patch as otherwise some foolish unnamed maintainer might pick up part of this series leaving things in a bad state for a kernel release. The earlier patch 'recommends' it isn't used which is a bit week. Say support is going to be removed later. > --- > drivers/iio/adc/ad7606.c | 8 ++++++-- > 1 file changed, 6 insertions(+), 2 deletions(-) > > diff --git a/drivers/iio/adc/ad7606.c b/drivers/iio/adc/ad7606.c > index f710445bdc22..0c12ca237ee9 100644 > --- a/drivers/iio/adc/ad7606.c > +++ b/drivers/iio/adc/ad7606.c > @@ -599,7 +599,6 @@ static int ad7606_buffer_postenable(struct iio_dev *indio_dev) > struct ad7606_state *st = iio_priv(indio_dev); > > gpiod_set_value(st->gpio_convst, 1); > - ad7606_pwm_set_swing(st); > > return 0; > } > @@ -609,7 +608,6 @@ static int ad7606_buffer_predisable(struct iio_dev *indio_dev) > struct ad7606_state *st = iio_priv(indio_dev); > > gpiod_set_value(st->gpio_convst, 0); > - ad7606_pwm_set_low(st); > > return 0; > } > @@ -838,6 +836,12 @@ int ad7606_probe(struct device *dev, int irq, void __iomem *base_address, > indio_dev->setup_ops = &ad7606_pwm_buffer_ops; > } else { > init_completion(&st->completion); > + > + /* Reserve the PWM use only for backend (force gpio_convst definition) */ > + if (!st->gpio_convst) > + return dev_err_probe(dev, -EINVAL, > + "No backend, connect convst to a GPIO"); > + > ret = devm_request_threaded_irq(dev, irq, > NULL, > &ad7606_interrupt, >
diff --git a/drivers/iio/adc/ad7606.c b/drivers/iio/adc/ad7606.c index f710445bdc22..0c12ca237ee9 100644 --- a/drivers/iio/adc/ad7606.c +++ b/drivers/iio/adc/ad7606.c @@ -599,7 +599,6 @@ static int ad7606_buffer_postenable(struct iio_dev *indio_dev) struct ad7606_state *st = iio_priv(indio_dev); gpiod_set_value(st->gpio_convst, 1); - ad7606_pwm_set_swing(st); return 0; } @@ -609,7 +608,6 @@ static int ad7606_buffer_predisable(struct iio_dev *indio_dev) struct ad7606_state *st = iio_priv(indio_dev); gpiod_set_value(st->gpio_convst, 0); - ad7606_pwm_set_low(st); return 0; } @@ -838,6 +836,12 @@ int ad7606_probe(struct device *dev, int irq, void __iomem *base_address, indio_dev->setup_ops = &ad7606_pwm_buffer_ops; } else { init_completion(&st->completion); + + /* Reserve the PWM use only for backend (force gpio_convst definition) */ + if (!st->gpio_convst) + return dev_err_probe(dev, -EINVAL, + "No backend, connect convst to a GPIO"); + ret = devm_request_threaded_irq(dev, irq, NULL, &ad7606_interrupt,
Since the pwm was introduced before backend, there was an example use whit triggered buffers. However, using it may be dangerous, because if the PWM goes too fast, a new conversion can be triggered before the transmission is over, whit the associated mess in the buffers. Until a solution is found to mitigate this risk, for instante CRC support, the PWM will be disabled. Signed-off-by: Guillaume Stols <gstols@baylibre.com> --- drivers/iio/adc/ad7606.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)