Re: [PATCH v4 2/2] media: i2c: add Himax HM1246 image sensor driver

[Sakari Ailus <sakari.ailus@linux.intel.com>]

Hi Matthias,

On Mon, Nov 03, 2025 at 05:19:47PM +0100, Matthias Fend wrote:
> > > > > > +static int hm1246_parse_fwnode(struct hm1246 *hm1246)
> > > > > > +{
> > > > > > +    struct fwnode_handle *endpoint;
> > > > > > +    struct v4l2_fwnode_endpoint bus_cfg = {
> > > > > > +        .bus_type = V4L2_MBUS_PARALLEL,
> > > > > > +    };
> > > > > > +    int ret;
> > > > > > +
> > > > > > +    endpoint =
> > > > > > fwnode_graph_get_endpoint_by_id(dev_fwnode(hm1246- >dev), 0,
> > > > > > +                           0,
> > > > > > +                           FWNODE_GRAPH_ENDPOINT_NEXT);
> > > > > > +    if (!endpoint)
> > > > > > +        return dev_err_probe(hm1246->dev, -EINVAL,
> > > > > > +                     "missing endpoint node\n");
> > > > > > +
> > > > > > +    ret = v4l2_fwnode_endpoint_parse(endpoint, &bus_cfg);
> > > > > 
> > > > > What about validating the link frequencies? You can use
> > > > > v4l2_link_freq_to_bitmap(), too.
> > > > 
> > > > I was under the impression that for sensors with a parallel interface,
> > > > no frequency information is provided in the device tree (because there's
> > > > no need for it). Since there are no frequency entries, they can't be
> > > > verified.
> > > > 
> > > > Am I wrong, or did you perhaps mean something else?
> > 
> > The current documentation
> > <URL:https://hverkuil.home.xs4all.nl/spec/driver-api/camera-sensor.html>
> > doesn't distinguish CSI-2 and parallel interfaces in this respect. It's a
> > good idea to ensure a safe frequency is used as the driver works the same
> > way in all cases, whether or not using one is mandatory.
> 
> If I understand correctly, this means that in the bindings, the port
> property 'link-frequencies' should be marked as 'required', and the port in
> the example node should be extended with the line 'link-frequencies = /bits/
> 64 <42174000>;'.
> Then, during probe, it can be checked with v4l2_link_freq_to_bitmap()
> whether the link frequency entered in the device tree is supported (this
> also requires switching to v4l2_fwnode_endpoint_alloc_parse).
> 
> Does this describe the desired change?

Yes, it does.

-- 
Regards,

Sakari Ailus