Re: [PATCH 4/7] phy: qcom-qmp-combo: Introduce orientation switching

[Johan Hovold <johan@kernel.org>]

On Mon, Apr 24, 2023 at 08:40:07PM -0700, Bjorn Andersson wrote:
> The data lanes of the QMP PHY is swapped in order to handle changing
> orientation of the USB Type-C cable. Register a typec_switch device to
> allow a TCPM to configure the orientation.
> 
> The newly introduced orientation variable is adjusted based on the
> request, and the initialized components are brought down and up again.
> To keep track of what parts needs to be cycled new variables to keep
> track of the individual init_count is introduced.
> 
> Both the USB and the DisplayPort altmode signals are properly switched.
> For DisplayPort the controller will after the TCPM having established
> orientation power on the PHY, so this is not done implicitly, but for
> USB the PHY typically is kept initialized across the switch, and must
> therefor then be reinitialized.
> 
> This is based on initial work by Wesley Cheng.
> 
> Link: https://lore.kernel.org/r/20201009082843.28503-3-wcheng@codeaurora.org/
> Signed-off-by: Bjorn Andersson <quic_bjorande@quicinc.com>
> ---
>  drivers/phy/qualcomm/phy-qcom-qmp-combo.c | 92 ++++++++++++++++++++---
>  1 file changed, 83 insertions(+), 9 deletions(-)
> 
> diff --git a/drivers/phy/qualcomm/phy-qcom-qmp-combo.c b/drivers/phy/qualcomm/phy-qcom-qmp-combo.c
> index 6748f31da7a3..5d6d6ef3944b 100644
> --- a/drivers/phy/qualcomm/phy-qcom-qmp-combo.c
> +++ b/drivers/phy/qualcomm/phy-qcom-qmp-combo.c

> @@ -2567,8 +2573,9 @@ static int qmp_combo_dp_exit(struct phy *phy)
>  
>  	mutex_lock(&qmp->phy_mutex);
>  
> -	qmp_combo_com_exit(qmp);
> +	qmp_combo_com_exit(qmp, false);
>  
> +	qmp->dp_init_count--;

Nit: add a newline for symmetry.

>  	mutex_unlock(&qmp->phy_mutex);
>  
>  	return 0;

> @@ -3351,6 +3362,65 @@ static struct phy *qmp_combo_phy_xlate(struct device *dev, struct of_phandle_arg
>  	return ERR_PTR(-EINVAL);
>  }
>  
> +#if IS_ENABLED(CONFIG_TYPEC)
> +static int qmp_combo_typec_switch_set(struct typec_switch_dev *sw,
> +				      enum typec_orientation orientation)
> +{
> +	struct qmp_combo *qmp = typec_switch_get_drvdata(sw);
> +	const struct qmp_phy_cfg *cfg = qmp->cfg;
> +
> +	if (orientation == qmp->orientation || orientation == TYPEC_ORIENTATION_NONE)
> +		return 0;
> +
> +	mutex_lock(&qmp->phy_mutex);
> +	qmp->orientation = orientation;
> +
> +	if (qmp->init_count) {
> +		if (qmp->usb_init_count)
> +			qmp_combo_usb_power_off(qmp->usb_phy);
> +		qmp_combo_com_exit(qmp, true);
> +
> +		qmp_combo_com_init(qmp, true);
> +		if (qmp->usb_init_count)
> +			qmp_combo_usb_power_on(qmp->usb_phy);
> +		if (qmp->dp_init_count)
> +			cfg->dp_aux_init(qmp);
> +	}
> +	mutex_unlock(&qmp->phy_mutex);
> +
> +	return 0;
> +}
> +
> +static void qmp_combo_typec_unregister(void *data)
> +{
> +	struct qmp_combo *qmp = data;
> +
> +	typec_switch_unregister(qmp->sw);
> +}
> +
> +static int qmp_combo_typec_switch_register(struct qmp_combo *qmp)
> +{
> +	struct typec_switch_desc sw_desc = {};
> +	struct device *dev = qmp->dev;
> +
> +	sw_desc.drvdata = qmp;
> +	sw_desc.fwnode = dev->fwnode;
> +	sw_desc.set = qmp_combo_typec_switch_set;

Nit: I'd add a newline here for readability.

> +	qmp->sw = typec_switch_register(dev, &sw_desc);
> +	if (IS_ERR(qmp->sw)) {
> +		dev_err(dev, "Unable to register typec switch: %pe\n", qmp->sw);
> +		return PTR_ERR(qmp->sw);
> +	}
> +
> +	return devm_add_action_or_reset(dev, qmp_combo_typec_unregister, qmp);
> +}
> +#else
> +static int qmp_combo_typec_switch_register(struct qmp_combo *qmp)
> +{
> +	return 0;
> +}
> +#endif

Perhaps you can move the type-c block after qmp_combo_register_clocks()
above to keep the type-c and later added bridge functions together and
better reflect the probe init order (e.g. keeping the dt-functions just
above probe()).

> +
>  static int qmp_combo_probe(struct platform_device *pdev)
>  {
>  	struct qmp_combo *qmp;
> @@ -3385,6 +3455,10 @@ static int qmp_combo_probe(struct platform_device *pdev)
>  	if (ret)
>  		return ret;
>  
> +	ret = qmp_combo_typec_switch_register(qmp);
> +	if (ret)
> +		return ret;
> +
>  	/* Check for legacy binding with child nodes. */
>  	usb_np = of_get_child_by_name(dev->of_node, "usb3-phy");
>  	if (usb_np) {

Looks good otherwise:

Reviewed-by: Johan Hovold <johan+linaro@kernel.org>

Johan