Re: [PATCH v3 06/15] iio: sx9310: Fixes various memory handling
From: Jonathan Cameron
Date: Sat Aug 01 2020 - 10:52:46 EST
On Fri, 31 Jul 2020 22:24:47 +0300
Andy Shevchenko <andy.shevchenko@xxxxxxxxx> wrote:
> On Fri, Jul 31, 2020 at 7:49 PM Daniel Campello <campello@xxxxxxxxxxxx> wrote:
> >
> > Makes use __aligned(8) to ensure that the timestamp is correctly aligned
> > when we call io_push_to_buffers_with_timestamp().
> > Also makes use of sizeof() for regmap_bulk_read instead of static value.
>
> Reviewed-by: Andy Shevchenko <andy.shevchenko@xxxxxxxxx>
>
> > Signed-off-by: Daniel Campello <campello@xxxxxxxxxxxx>
> > ---
> >
> > Changes in v3:
> > - Changed buffer to struct type to align timestamp memory properly.
> >
> > Changes in v2:
> > - Fixed commit message from "iio: sx9310: Align memory"
> >
> > drivers/iio/proximity/sx9310.c | 13 ++++++++-----
> > 1 file changed, 8 insertions(+), 5 deletions(-)
> >
> > diff --git a/drivers/iio/proximity/sx9310.c b/drivers/iio/proximity/sx9310.c
> > index 2ed062d01634bc..c46584b4817b4a 100644
> > --- a/drivers/iio/proximity/sx9310.c
> > +++ b/drivers/iio/proximity/sx9310.c
> > @@ -132,8 +132,11 @@ struct sx9310_data {
> > */
> > bool prox_stat[SX9310_NUM_CHANNELS];
> > bool trigger_enabled;
> > - __be16 buffer[SX9310_NUM_CHANNELS +
> > - 4]; /* 64-bit data + 64-bit timestamp */
> > + /* Ensure correct alignment of timestamp when present. */
> > + struct {
> > + __be16 channels[SX9310_NUM_CHANNELS];
> > + s64 ts __aligned(8);
> > + } buffer;
> > /* Remember enabled channels and sample rate during suspend. */
> > unsigned int suspend_ctrl0;
> > struct completion completion;
> > @@ -346,7 +349,7 @@ static int sx9310_read_prox_data(struct sx9310_data *data,
> > if (ret < 0)
> > return ret;
> >
> > - return regmap_bulk_read(data->regmap, chan->address, val, 2);
> > + return regmap_bulk_read(data->regmap, chan->address, val, sizeof(*val));
> > }
> >
> > /*
> > @@ -697,10 +700,10 @@ static irqreturn_t sx9310_trigger_handler(int irq, void *private)
> > if (ret < 0)
> > goto out;
> >
> > - data->buffer[i++] = val;
> > + data->buffer.channels[i++] = val;
> > }
> >
> > - iio_push_to_buffers_with_timestamp(indio_dev, data->buffer,
> > + iio_push_to_buffers_with_timestamp(indio_dev, data->buffer.channels,
Whilst it's the same address, it makes more sense to push &data->buffer
Technically buffer.channels isn't large enough to meant the requirements
of iio_push_to_buffers_with_timestamp.
Otherwise, looks good.
Thanks,
Jonathan
> > pf->timestamp);
> >
> > out:
> > --
> > 2.28.0.163.g6104cc2f0b6-goog
> >
>
>
> --
> With Best Regards,
> Andy Shevchenko