On Wed, 14 Dec 2022 10:54:32 +0100 Krzysztof Kozlowski <krzysztof.kozlowski@xxxxxxxxxx> wrote: > On 13/12/2022 18:15, Han Xu wrote: > > Add new compatible string for the NXP FXLS8967AF accelerometer sensor. > > > > Signed-off-by: Han Xu <han.xu@xxxxxxx> > > > > --- > > changes in v3 > > - Start commit message in capital > > - Describe all these chips are compatible > > --- > > .../devicetree/bindings/iio/accel/nxp,fxls8962af.yaml | 4 ++++ > > 1 file changed, 4 insertions(+) > > > > diff --git a/Documentation/devicetree/bindings/iio/accel/nxp,fxls8962af.yaml b/Documentation/devicetree/bindings/iio/accel/nxp,fxls8962af.yaml > > index 65ce8ea14b52..8f07ade21abb 100644 > > --- a/Documentation/devicetree/bindings/iio/accel/nxp,fxls8962af.yaml > > +++ b/Documentation/devicetree/bindings/iio/accel/nxp,fxls8962af.yaml > > @@ -14,12 +14,16 @@ description: | > > SPI and I2C interface. > > https://www.nxp.com/docs/en/data-sheet/FXLS8962AF.pdf > > https://www.nxp.com/docs/en/data-sheet/FXLS8964AF.pdf > > + https://www.nxp.com/docs/en/data-sheet/FXLS8967AF.pdf > > > > properties: > > compatible: > > + description: > > + These chips are compatible with each other, just have different IDs. > > As pointed in other mail, the chips are not compatible, so drop the comment. The difference is an ID register. Given we have had a bunch of cases of board manufacturers swapping parts that aren't always compatible we have an old pattern that we are fixing of rejecting unmatched who am I registers. This driver should be relaxed to just print a message when the value doesn't match - that was the compromise we reached with still pointing out possible compatibility problems, whilst assuming the dts is correct. Even better would be to make it a little more clever so it doesn't bother moaning about part changes if the driver knows the new ID is compatible. Once that's done the comment would reflect how we treat it in the driver (which shouldn't matter to the binding anyway). I'm not entirely sure why this driver has handling to allow for different channel sets, but assume there are more parts to be upstreamed where that flexibility will be useful. Jonathan > > Best regards, > Krzysztof >
