On Sat, 14 Sep 2024 11:21:34 +0200
Guillaume Stols <gstols@xxxxxxxxxxxx> wrote:
> On 8/17/24 17:33, Jonathan Cameron wrote:
> > On Thu, 15 Aug 2024 12:12:01 +0000
> > Guillaume Stols <gstols@xxxxxxxxxxxx> wrote:
> >
> >> On the parallel version, the current implementation is only compatible
> >> with id tables and won't work with fx_nodes. So in this commit, the goal
> >> is to switch to use get_device_match_data, in order to simplify the
> >> logic of retrieving chip data.
> >>
> >> Also, chip info is moved in the .h file so to be accessible to all the
> >> driver files that can set a pointer to the corresponding chip as the
> >> driver data.
> > This means each driver gets their own copy.
> >
> > Better to use an extern in the header and keep the actual data
> > in the core module.
>
> ack.
>
> Given your previous comment about introducing
> platform_device_get_match_data, I guess I should instead do it directly
> in the driver's probe, like its done in axp20x_adc.c ? Somehting like that:
>
> if (!dev_fwnode(&pdev->dev)) {
> const struct platform_device_id *id;
>
> id = platform_get_device_id(pdev);
> chip_info = (const struct ad7606_chip_info *)id->driver_data;
> } else {
> struct device *dev = &pdev->dev;
> chip_info = device_get_match_data(dev);
> }
Yes, something along those lines makes sense.
If there are enough instances of this we can have a standard
definition for this similar to the i2c / spi ones that defaults
to device_get_match_data() if available, and falls back to the old
way if not.
If you want to add that great, if not it can be a separate
bit of work for another day.
Jonathan
