Re: [PATCH v2 02/12] media: v4l: fwnode: Support ACPI's _PLD for v4l2_fwnode_device_parse

[Ricardo Ribalda]

Hi Sakari

Thanks for your review!

On Mon, 30 Jun 2025 at 09:06, Sakari Ailus <sakari.ailus@xxxxxxxxxxxxxxx> wrote:
>
> Hi Ricardo,
>
> Thanks for the update.
>
> On 6/5/25 20:52, Ricardo Ribalda wrote:
> > Currently v4l2_fwnode_device_parse() obtains the orientation and
> > rotation via fwnode properties.
> >
> > Extend the function to support as well ACPI devices with _PLD info.
> >
> > We give a higher priority to fwnode, because it might contain quirks
> > injected via swnodes.
> >
> > Signed-off-by: Ricardo Ribalda <ribalda@xxxxxxxxxxxx>
> > ---
> >   drivers/media/v4l2-core/v4l2-fwnode.c | 85 ++++++++++++++++++++++++++++++++---
> >   1 file changed, 79 insertions(+), 6 deletions(-)
> >
> > diff --git a/drivers/media/v4l2-core/v4l2-fwnode.c b/drivers/media/v4l2-core/v4l2-fwnode.c
> > index cb153ce42c45d69600a3ec4e59a5584d7e791a2a..379290ab3cfde74c8f663d61837a9a95011b5ae0 100644
> > --- a/drivers/media/v4l2-core/v4l2-fwnode.c
> > +++ b/drivers/media/v4l2-core/v4l2-fwnode.c
> > @@ -15,6 +15,7 @@
> >    * Author: Guennadi Liakhovetski <g.liakhovetski@xxxxxx>
> >    */
> >   #include <linux/acpi.h>
> > +#include <acpi/acpi_bus.h>
> >   #include <linux/kernel.h>
> >   #include <linux/mm.h>
> >   #include <linux/module.h>
> > @@ -807,16 +808,65 @@ int v4l2_fwnode_connector_add_link(struct fwnode_handle *fwnode,
> >   }
> >   EXPORT_SYMBOL_GPL(v4l2_fwnode_connector_add_link);
> >
> > -int v4l2_fwnode_device_parse(struct device *dev,
> > -                          struct v4l2_fwnode_device_properties *props)
> > +static int v4l2_fwnode_device_parse_acpi(struct device *dev,
> > +                                      struct v4l2_fwnode_device_properties *props)
> > +{
> > +     struct acpi_pld_info *pld;
> > +     int ret = 0;
> > +
> > +     if (!is_acpi_device_node(dev_fwnode(dev)))
> > +             return 0;
> > +
> > +     if (!acpi_get_physical_device_location(ACPI_HANDLE(dev), &pld)) {
> > +             dev_dbg(dev, "acpi _PLD call failed\n");
>
> I'd do:
>
> acpi_handle_debug(ACPI_HANDLE(dev), "cannot obtain _PLD\n");
ack
>
> > +             return 0;
> > +     }
> > +
> > +     if (props->orientation != V4L2_FWNODE_PROPERTY_UNSET) {
> > +             switch (pld->panel) {
> > +             case ACPI_PLD_PANEL_FRONT:
> > +                     props->orientation = V4L2_FWNODE_ORIENTATION_FRONT;
> > +                     break;
> > +             case ACPI_PLD_PANEL_BACK:
> > +                     props->orientation = V4L2_FWNODE_ORIENTATION_BACK;
> > +                     break;
> > +             case ACPI_PLD_PANEL_TOP:
> > +             case ACPI_PLD_PANEL_LEFT:
> > +             case ACPI_PLD_PANEL_RIGHT:
> > +             case ACPI_PLD_PANEL_UNKNOWN:
> > +                     props->orientation = V4L2_FWNODE_ORIENTATION_EXTERNAL;
> > +                     break;
> > +             default:
> > +                     dev_dbg(dev, "Unknown _PLD panel val %d\n", pld->panel);
>
> Similarly:
>
> acpi_handle_debug(ACPI_HANDLE(dev), "invalid panel %u in _PLD\n",
>                   pld->panel);
>
> > +                     ret = -EINVAL;
>
> Should this be an error or should we simply ignore it here (and maybe
> use acpi_handle_warn())?

v4l2_fwnode_device_parse_of() returns -EINVAL for a similar situation,
so I think it is better to be consistent and return -EINVAL here.
But I agree that acpi_handle_warn() better suits here than _dbg.

>
> > +                     goto done;
> > +             }
> > +     }
> > +
> > +     if (props->rotation != V4L2_FWNODE_PROPERTY_UNSET) {
> > +             switch (pld->rotation) {
> > +             case 0 ... 7:
> > +                     props->rotation = pld->rotation * 45;
> > +                     break;
> > +             default:
> > +                     dev_dbg(dev, "Unknown _PLD rotation val %d\n", pld->panel);
>
> acpi_handle_debug(ACPI_HANDLE(dev), "invalid rotation %u in _PLD\n",
>                   pld->rotation);
>
> > +                     ret = -EINVAL;
> > +                     goto done;
> > +             }
> > +     }
> > +
> > +done:
> > +     ACPI_FREE(pld);
> > +     return ret;
> > +}
> > +
> > +static int v4l2_fwnode_device_parse_dt(struct device *dev,
>
> I'd call this v4l2_fwnode_device_parse_of() as we're parsing OF nodes
> and properties here.
ack
>
> > +                                    struct v4l2_fwnode_device_properties *props)
> >   {
> >       struct fwnode_handle *fwnode = dev_fwnode(dev);
> >       u32 val;
> >       int ret;
> >
> > -     memset(props, 0, sizeof(*props));
> > -
> > -     props->orientation = V4L2_FWNODE_PROPERTY_UNSET;
> >       ret = fwnode_property_read_u32(fwnode, "orientation", &val);
> >       if (!ret) {
> >               switch (val) {
> > @@ -833,7 +883,6 @@ int v4l2_fwnode_device_parse(struct device *dev,
> >               dev_dbg(dev, "device orientation: %u\n", val);
> >       }
> >
> > -     props->rotation = V4L2_FWNODE_PROPERTY_UNSET;
> >       ret = fwnode_property_read_u32(fwnode, "rotation", &val);
> >       if (!ret) {
> >               if (val >= 360) {
> > @@ -847,6 +896,30 @@ int v4l2_fwnode_device_parse(struct device *dev,
> >
> >       return 0;
> >   }
> > +
> > +int v4l2_fwnode_device_parse(struct device *dev,
> > +                          struct v4l2_fwnode_device_properties *props)
> > +{
> > +     int ret;
> > +
> > +     memset(props, 0, sizeof(*props));
> > +
> > +     props->orientation = V4L2_FWNODE_PROPERTY_UNSET;
> > +     props->rotation = V4L2_FWNODE_PROPERTY_UNSET;
> > +
> > +     /* Start by looking into swnodes and dt. */
> > +     ret =  v4l2_fwnode_device_parse_dt(dev, props);
> > +     if (ret)
> > +             return ret;
> > +
> > +     /* Orientation and rotation found!, we are ready. */
> > +     if (props->orientation != V4L2_FWNODE_PROPERTY_UNSET &&
> > +         props->rotation != V4L2_FWNODE_PROPERTY_UNSET)
> > +             return 0;
>
> I think you can remove this check without affecting the functionality.
I want to avoid calling an acpi method unless it is strictly
necessary. The check is not that ugly... if it is ok with you i'd
rather keep it.

>
> > +
> > +     /* Let's check the acpi table. */
> > +     return v4l2_fwnode_device_parse_acpi(dev, props);
> > +}
> >   EXPORT_SYMBOL_GPL(v4l2_fwnode_device_parse);
> >
> >   /*
> >
>
> --
> Regards,
>
> Sakari Ailus

Thanks!

-- 
Ricardo Ribalda