Re: [PATCH WIP v4 1/9] media: qcom: camss: csiphy: Introduce PHY configuration

[Bryan O'Donoghue <bryan.odonoghue@linaro.org>]

On 01/03/2026 00:51, David Heidelberg via B4 Relay wrote:
> From: David Heidelberg <david@ixit.cz>
> 
> Read PHY configuration from the device-tree bus-type and save it into
> the csiphy structure for later use.
> 
> For C-PHY, skip clock line configuration, as there is none.
> 
> Signed-off-by: David Heidelberg <david@ixit.cz>
> ---
>   drivers/media/platform/qcom/camss/camss-csiphy.h |  2 ++
>   drivers/media/platform/qcom/camss/camss.c        | 18 +++++++++++-------
>   2 files changed, 13 insertions(+), 7 deletions(-)
> 
> diff --git a/drivers/media/platform/qcom/camss/camss-csiphy.h b/drivers/media/platform/qcom/camss/camss-csiphy.h
> index 2d5054819df7f..d198171700e73 100644
> --- a/drivers/media/platform/qcom/camss/camss-csiphy.h
> +++ b/drivers/media/platform/qcom/camss/camss-csiphy.h
> @@ -28,11 +28,13 @@ struct csiphy_lane {
>   
>   /**
>    * struct csiphy_lanes_cfg - CSIPHY lanes configuration
> + * @phy_cfg:  interface selection (C-PHY or D-PHY)
>    * @num_data: number of data lanes
>    * @data:     data lanes configuration
>    * @clk:      clock lane configuration (only for D-PHY)
>    */
>   struct csiphy_lanes_cfg {
> +	enum v4l2_mbus_type phy_cfg;
>   	int num_data;
>   	struct csiphy_lane *data;
>   	struct csiphy_lane clk;
> diff --git a/drivers/media/platform/qcom/camss/camss.c b/drivers/media/platform/qcom/camss/camss.c
> index 00b87fd9afbd8..ea0c8cf3cd806 100644
> --- a/drivers/media/platform/qcom/camss/camss.c
> +++ b/drivers/media/platform/qcom/camss/camss.c
> @@ -4411,11 +4411,11 @@ static int camss_parse_endpoint_node(struct device *dev,
>   	if (ret)
>   		return ret;
>   
> -	/*
> -	 * Most SoCs support both D-PHY and C-PHY standards, but currently only
> -	 * D-PHY is supported in the driver.
> -	 */
> -	if (vep.bus_type != V4L2_MBUS_CSI2_DPHY) {
> +	switch (vep.bus_type) {
> +	case V4L2_MBUS_CSI2_CPHY:
> +	case V4L2_MBUS_CSI2_DPHY:
> +		break;
> +	default:
>   		dev_err(dev, "Unsupported bus type %d\n", vep.bus_type);
>   		return -EINVAL;
>   	}
> @@ -4423,9 +4423,13 @@ static int camss_parse_endpoint_node(struct device *dev,
>   	csd->interface.csiphy_id = vep.base.port;
>   
>   	mipi_csi2 = &vep.bus.mipi_csi2;
> -	lncfg->clk.pos = mipi_csi2->clock_lane;
> -	lncfg->clk.pol = mipi_csi2->lane_polarities[0];
>   	lncfg->num_data = mipi_csi2->num_data_lanes;
> +	lncfg->phy_cfg = vep.bus_type;
> +
> +	if (lncfg->phy_cfg != V4L2_MBUS_CSI2_CPHY) {
> +		lncfg->clk.pos = mipi_csi2->clock_lane;
> +		lncfg->clk.pol = mipi_csi2->lane_polarities[0];
> +	}
>   
>   	lncfg->data = devm_kcalloc(dev,
>   				   lncfg->num_data, sizeof(*lncfg->data),
> 

This patch should come last in the series - i.e. it should be invalid to 
select CPHY mode until all of the other code has advanced to accommodate.

I guess you end up depending on this variable a little later on.

I think this patch looks good with the one caveat that I'd like 
bisectability here.

So perhaps take the switch statement and have that part be a last patch 
at the end formally enabling CPHY mode.

---
bod