Re: [PATCH 13/13] media: imx355: Support 2 lane readout.

[Dave Stevenson]

Hi Richard

On Wed, 13 May 2026 at 19:32, Richard Acayan <mailingradian@xxxxxxxxx> wrote:
>
> On Wed, May 06, 2026 at 07:23:51PM +0100, Dave Stevenson wrote:
> > The sensor supports 2 or 4 lane readout, but the driver only allowed
> > for 4 lanes. Add 2 lane support.
> >
> > The clock tree was set to use single PLL mode to feed both IOP (MIPI)
> > and IVT (Pixel array).
> > 2 lane mode supports a MIPI link frequency of up to 445MHz (890Mbit/s)
> > cf 360MHz (720Mbit/s) for 4lane, but that requires switching to dual
> > PLL mode as the rates can't be achieved with simple divisors.
> > The LLP values are extended for each mode to account for the increased
> > time per line over the MIPI link.
> >
> > Signed-off-by: Dave Stevenson <dave.stevenson@xxxxxxxxxxxxxxx>
> > ---
> >  drivers/media/i2c/imx355.c | 135 ++++++++++++++++++++++++++++-----------------
> >  1 file changed, 84 insertions(+), 51 deletions(-)
> >
> > diff --git a/drivers/media/i2c/imx355.c b/drivers/media/i2c/imx355.c
> > index d0e0e81d1e7c..fa1d1c86d5b0 100644
> > --- a/drivers/media/i2c/imx355.c
> > +++ b/drivers/media/i2c/imx355.c
> (snip)
> > @@ -1192,13 +1220,18 @@ static struct imx355_hwcfg *imx355_get_hwcfg(struct device *dev)
> >       if (!cfg)
> >               goto out_err;
> >
> > -     if (bus_cfg.bus.mipi_csi2.num_data_lanes != IMX355_DATA_LANES)
> > +     if (bus_cfg.bus.mipi_csi2.num_data_lanes != 2 &&
> > +         bus_cfg.bus.mipi_csi2.num_data_lanes != 4)
> >               goto out_err;
>
> Now we know that the sensor supports a configuration with 2 lanes.
> Previously, there was a question whether there was a configuration with
> a different number of data lanes (from
> https://lore.kernel.org/linux-arm-msm/3c51f9fe-9c5f-47dd-a971-5b2a9e416230@xxxxxxxxxx/ ):
>
>         Does IMX355 sensor hardware support any other number of lanes?
>
>         1) If no, then it makes no practical sense to check for data lanes number
>         given by firmware, there can be any stored value, but it's known that the
>         number of sensor data lanes is 4.
>
>         2) If yes, then please return to the sensor dt bindings, and reflect any
>         other options, right now it is set strictly to 4 lanes only. In case if
>         'data-lanes' property is omitted, you can use 4 lanes as a default number
>         of data lanes, and this information shall be documented in the dt bindings.
>
> We should adjust the dt-bindings to accept a devicetree with 2 lanes.

Yes, I guess so.

The datasheet states:
"CSI-2 serial data output (selection of 4lane/2lane)"
So 2 or 4 lanes are the only supported options, and there is no
support in the sensor for lane reordering.

I don't think I've ever encountered a sensor that supports 3 lane
readout, and 1 lane is generally too limiting on data rate to be
overly useful.

  Dave