Hi Karol,
Thanks for the patch.
Please refer to the comments below.
On 04/12/2012 08:45 PM, Karol Lewandowski wrote:
> Includes v4l2/dt helper function (hdmi_of_get_i2c_subdev() that probably
> should be implemented in v4l2 core itself.
>
> Signed-off-by: Karol Lewandowski
> Cc: Tomasz Stanislawski
> ---
> drivers/media/video/s5p-tv/hdmi_drv.c | 68
> -
> 1 files changed, 67 insertions(+), 1 deletions(-)
>
> diff --git a/drivers/media/video/s5p-tv/hdmi_drv.c
> b/drivers/media/video/s5p-tv/hdmi_drv.c
> index fff3cab..a402e8f 100644
> --- a/drivers/media/video/s5p-tv/hdmi_drv.c
> +++ b/drivers/media/video/s5p-tv/hdmi_drv.c
> @@ -29,6 +29,8 @@
> #include
> #include
> #include
> +#include
> +#include
>
> #include
> #include
> @@ -712,6 +714,56 @@ static int hdmi_enum_dv_presets(struct v4l2_subdev *sd,
> return v4l_fill_dv_preset_info(hdmi_conf[preset->index].preset, preset);
> }
>
> +#ifdef CONFIG_OF
> +/* Heavily based[1] on v4l2_i2c_new_subdev_board()
> + *
> + * [1] Copy-pasted, that is
> + */
> +struct v4l2_subdev *hdmi_of_get_i2c_subdev(struct v4l2_device *v4l2_dev,
> + struct device_node *np, const char *propname)
> +{
> + struct v4l2_subdev *sd = NULL;
> + struct i2c_client *client;
> + struct device_node *cnp;
> +
> + BUG_ON(!v4l2_dev);
Crashing a kernel is not a good idea in general. Especially when it is possible
to handle a situation. I think that "if (WARN_ON(..)) return NULL" combo
is a preferred way for handling such an error.
> +
> + cnp = of_parse_phandle(np, propname, 0);
> + if (!cnp) {
> + dev_err(v4l2_dev->dev, "Can't find subdev %s\n", propname);
> + goto err;
> + }
> +
> + client = of_find_i2c_device_by_node(cnp);
> + if (!client) {
> + dev_err(v4l2_dev->dev, "subdev %s doesn't reference correct
> node\n",
> + propname);
> + goto err;
> + }
> +
Maybe you should combine the check below with the check above.
> + if (client == NULL || client->driver == NULL)
> + goto err;
> +
> + /* Lock the module so we can safely get the v4l2_subdev pointer */
> + if (!try_module_get(client->driver->driver.owner))
> + goto err;
> + sd = i2c_get_clientdata(client);
> +
> + /* Register with the v4l2_device which increases the module's
> +use count as well. */
> + if (v4l2_device_register_subdev(v4l2_dev, sd)) {
> + printk(KERN_ERR "%s: failed to register subdev\n", __func__);
> + sd = NULL;
> + }
> + /* Decrease the module use count to match the first try_module_get. */
> + module_put(client->driver->driver.owner);
This label 'err' is not only used for error cleanups. Maybe you should rename it
to 'cleanup' or 'done'.
> +err:
> + of_node_put(cnp);
> +
> + return sd;
> +}
> +#endif
I think that you should strongly consider adding the function above to
the V4L2 core to drivers/media/video/videobuf2-core.c file and post
the patch to linux-media list.
> +
> static const struct v4l2_subdev_core_ops hdmi_sd_core_ops = {
> .s_power = hdmi_s_power,
> };
> @@ -875,6 +927,12 @@ static struct v4l2_subdev *hdmi_get_subdev(
> struct i2c_adapter *adapter;
> struct device *dev = hdmi_dev->dev;
>
Now I see why hdmi_get_subdev is called with NULL arguments in the
previous patch :).
However I think that you should introduce two explicit paths in probe
function for handling non-DT and DT cases.
> +#ifdef CONFIG_OF
> + if (dev->of_node)
> + return hdmi_of_get_i2c_subdev(&hdmi_dev->v4l2_dev,
> +dev->of_node, propname);
> +#endif
> +
> if (!bdinfo) {
> dev_err(dev, "%s info is missing in platform data\n",
> propname);
> @@ -913,7 +971,7 @@ static int __devinit hdmi_probe(struct platform_device
> *pdev)
>
> dev_dbg(dev, "probe start\n");
>
> - if (!pdata) {
> + if (!pdata && !dev->of_node) {
> dev_err(dev, "platform data is missing\n");
> ret = -ENODEV;
> goto fail;
> @@ -1037,6 +1095,13 @@ static int __devexit hdmi_remove(struct
> platform_device *pdev)
> return 0;
> }
>
> +#ifdef CONFIG_OF
> +static struct of_device_id hdmi_dt_match[] = {
> + { .compatible = "samsung,s5pv210-hdmi" },
> + { },
> +};
> +#endif
> +
> static struct platform_driver hdmi_driver __refdata = {
> .probe = hdmi_probe,
> .remove = __devexit_p(hdmi_remove),
> @@ -1045,6 +1110,7 @@ static struct platform_driver hdmi_driver __refdata = {
> .name = "s5p-hdmi",
> .owner = THIS_MODULE,
> .pm = &hdmi_pm_ops,
> + .of_match_table = of_match_ptr(hdmi_dt_match),
> }
> };
>
Regards,
Tomasz Stanislawski
--
To unsubscribe from this list: send the line "unsubscribe linux-samsung-soc" in
the body of a message to major
