Re: [PATCH v4 2/5] iio: adc: ad7949: fix spi messages on non 14-bit controllers

From: Jonathan Cameron
Date: Mon Aug 02 2021 - 06:21:10 EST


On Sun, 01 Aug 2021 19:01:10 -0400
"Liam Beguin" <liambeguin@xxxxxxxxx> wrote:

> On Sat Jul 31, 2021 at 10:52 AM EDT, Jonathan Cameron wrote:
> > On Sat, 31 Jul 2021 15:29:21 +0100
> > Jonathan Cameron <jic23@xxxxxxxxxx> wrote:
> >
> > > On Tue, 27 Jul 2021 19:29:03 -0400
> > > Liam Beguin <liambeguin@xxxxxxxxx> wrote:
> > >
> > > > From: Liam Beguin <lvb@xxxxxxxxxx>
> > > >
> > > > This driver supports devices with 14-bit and 16-bit sample sizes.
> > > > This is not always handled properly by spi controllers and can fail. To
> > > > work around this limitation, pad samples to 16-bit and split the sample
> > > > into two 8-bit messages in the event that only 8-bit messages are
> > > > supported by the controller.
> > > >
> > > > Signed-off-by: Liam Beguin <lvb@xxxxxxxxxx>
> > > > ---
> > > > drivers/iio/adc/ad7949.c | 62 ++++++++++++++++++++++++++++++++++------
> > > > 1 file changed, 54 insertions(+), 8 deletions(-)
> > > >
> > > > diff --git a/drivers/iio/adc/ad7949.c b/drivers/iio/adc/ad7949.c
> > > > index 0b549b8bd7a9..f1702c54c8be 100644
> > > > --- a/drivers/iio/adc/ad7949.c
> > > > +++ b/drivers/iio/adc/ad7949.c
> > > > @@ -67,6 +67,7 @@ static const struct ad7949_adc_spec ad7949_adc_spec[] = {
> > > > * @indio_dev: reference to iio structure
> > > > * @spi: reference to spi structure
> > > > * @resolution: resolution of the chip
> > > > + * @bits_per_word: number of bits per SPI word
> > > > * @cfg: copy of the configuration register
> > > > * @current_channel: current channel in use
> > > > * @buffer: buffer to send / receive data to / from device
> > > > @@ -77,6 +78,7 @@ struct ad7949_adc_chip {
> > > > struct iio_dev *indio_dev;
> > > > struct spi_device *spi;
> > > > u8 resolution;
> > > > + u8 bits_per_word;
> > > > u16 cfg;
> > > > unsigned int current_channel;
> > > > u16 buffer ____cacheline_aligned;
> > > > @@ -86,19 +88,34 @@ static int ad7949_spi_write_cfg(struct ad7949_adc_chip *ad7949_adc, u16 val,
> > > > u16 mask)
> > > > {
> > > > int ret;
> > > > - int bits_per_word = ad7949_adc->resolution;
> > > > - int shift = bits_per_word - AD7949_CFG_REG_SIZE_BITS;
> > >
> > > The define for this was removed in patch 1. I'll fix that up whilst applying by
> > > keeping it until this patch. Please check build passes on intermediate points
> > > during a patch series as otherwise we may break bisectability and that's really
> > > annoying if you are bisecting!
> > >
> > > Jonathan
> > >
> > > > struct spi_message msg;
> > > > struct spi_transfer tx[] = {
> > > > {
> > > > .tx_buf = &ad7949_adc->buffer,
> > > > .len = 2,
> > > > - .bits_per_word = bits_per_word,
> > > > + .bits_per_word = ad7949_adc->bits_per_word,
> > > > },
> > > > };
> > > >
> > > > + ad7949_adc->buffer = 0;
> > > > ad7949_adc->cfg = (val & mask) | (ad7949_adc->cfg & ~mask);
> > > > - ad7949_adc->buffer = ad7949_adc->cfg << shift;
> > > > +
> > > > + switch (ad7949_adc->bits_per_word) {
> > > > + case 16:
> > > > + ad7949_adc->buffer = ad7949_adc->cfg << 2;
> > > > + break;
> > > > + case 14:
> > > > + ad7949_adc->buffer = ad7949_adc->cfg;
> > > > + break;
> > > > + case 8:
> > > > + /* Here, type is big endian as it must be sent in two transfers */
> > > > + ad7949_adc->buffer = (u16)cpu_to_be16(ad7949_adc->cfg << 2);
> >
> > Gah, I wasn't thinking clearly when I suggested this. Sparse warns on
> > the
> > endian conversion
> >
> > One option is to resort to ignoring the fact we know it's aligned and
> > use the put_unaligned_be16() and get_unaligned_be16 calls which sparse
> > seems to be
> > happy with. Alternative would be to just have a be16 buffer after the
> > existing
> > one in the iio_priv structure. Then you will have to change the various
> > users
> > of iio_priv()->buffer to point to the new value if we are doing 8 bit
> > transfers.
> >
> > Whilst more invasive, this second option is the one I'd suggest.
>
> Understood, I'll go with your suggestion.
>
> Out of curiosity, other that being more explicit, is there another
> we'd rather not use {get,put}_unaligned_be16()?

We know it is aligned (as u16) so on a big endian platform that happens
to not handle unaligned accesses we will now be doing work which wouldn't
be needed if there was a get_aligned_be16() that was more relaxed about
types than cpu_to_be16() etc.

So basically it looks odd and we will will be hiding that we are smashing
data of potentially different ordering into the same structure field.

>
> > Note that there will be no need to add an __cacheline_aligned marking to
> > this
> > new element because it will be in a cachline that is only used for DMA
> > simply being
> > after the other buffer element which is force to start on a new
> > cacheline.
>
> Noted, Thanks for taking the time to explaining this.
>
> Liam
>
> >
> > Jonathan
> >
> > > > + break;
> > > > + default:
> > > > + dev_err(&ad7949_adc->indio_dev->dev, "unsupported BPW\n");
> > > > + return -EINVAL;
> > > > + }
> > > > +
> > > > spi_message_init_with_transfers(&msg, tx, 1);
> > > > ret = spi_sync(ad7949_adc->spi, &msg);
> > > >
> > > > @@ -115,14 +132,12 @@ static int ad7949_spi_read_channel(struct ad7949_adc_chip *ad7949_adc, int *val,
> > > > {
> > > > int ret;
> > > > int i;
> > > > - int bits_per_word = ad7949_adc->resolution;
> > > > - int mask = GENMASK(ad7949_adc->resolution - 1, 0);
> > > > struct spi_message msg;
> > > > struct spi_transfer tx[] = {
> > > > {
> > > > .rx_buf = &ad7949_adc->buffer,
> > > > .len = 2,
> > > > - .bits_per_word = bits_per_word,
> > > > + .bits_per_word = ad7949_adc->bits_per_word,
> > > > },
> > > > };
> > > >
> > > > @@ -157,7 +172,25 @@ static int ad7949_spi_read_channel(struct ad7949_adc_chip *ad7949_adc, int *val,
> > > >
> > > > ad7949_adc->current_channel = channel;
> > > >
> > > > - *val = ad7949_adc->buffer & mask;
> > > > + switch (ad7949_adc->bits_per_word) {
> > > > + case 16:
> > > > + *val = ad7949_adc->buffer;
> > > > + /* Shift-out padding bits */
> > > > + *val >>= 16 - ad7949_adc->resolution;
> > > > + break;
> > > > + case 14:
> > > > + *val = ad7949_adc->buffer & GENMASK(13, 0);
> > > > + break;
> > > > + case 8:
> > > > + /* Here, type is big endian as data was sent in two transfers */
> > > > + *val = be16_to_cpu(ad7949_adc->buffer);
> > > > + /* Shift-out padding bits */
> > > > + *val >>= 16 - ad7949_adc->resolution;
> > > > + break;
> > > > + default:
> > > > + dev_err(&ad7949_adc->indio_dev->dev, "unsupported BPW\n");
> > > > + return -EINVAL;
> > > > + }
> > > >
> > > > return 0;
> > > > }
> > > > @@ -265,6 +298,7 @@ static int ad7949_spi_init(struct ad7949_adc_chip *ad7949_adc)
> > > >
> > > > static int ad7949_spi_probe(struct spi_device *spi)
> > > > {
> > > > + u32 spi_ctrl_mask = spi->controller->bits_per_word_mask;
> > > > struct device *dev = &spi->dev;
> > > > const struct ad7949_adc_spec *spec;
> > > > struct ad7949_adc_chip *ad7949_adc;
> > > > @@ -291,6 +325,18 @@ static int ad7949_spi_probe(struct spi_device *spi)
> > > > indio_dev->num_channels = spec->num_channels;
> > > > ad7949_adc->resolution = spec->resolution;
> > > >
> > > > + /* Set SPI bits per word */
> > > > + if (spi_ctrl_mask & SPI_BPW_MASK(ad7949_adc->resolution)) {
> > > > + ad7949_adc->bits_per_word = ad7949_adc->resolution;
> > > > + } else if (spi_ctrl_mask == SPI_BPW_MASK(16)) {
> > > > + ad7949_adc->bits_per_word = 16;
> > > > + } else if (spi_ctrl_mask == SPI_BPW_MASK(8)) {
> > > > + ad7949_adc->bits_per_word = 8;
> > > > + } else {
> > > > + dev_err(dev, "unable to find common BPW with spi controller\n");
> > > > + return -EINVAL;
> > > > + }
> > > > +
> > > > ad7949_adc->vref = devm_regulator_get(dev, "vref");
> > > > if (IS_ERR(ad7949_adc->vref)) {
> > > > dev_err(dev, "fail to request regulator\n");
> > >
>