Re: [PATCH v1] Bluetooth: hci_qca: Enable clocks required for BT SOC

[bgodavar]

Hi Matthias,

sorry missed you mail.
On 2019-12-12 23:13, Matthias Kaehlcke wrote:
> On Thu, Nov 14, 2019 at 01:44:30PM +0530, Balakrishna Godavarthi wrote:
> > Instead of relying on other subsytem to turn ON clocks
> > required for BT SoC to operate, voting them from the driver.
> >
> > Signed-off-by: Balakrishna Godavarthi <bgodavar@xxxxxxxxxxxxxx>
> > ---
> >  drivers/bluetooth/hci_qca.c | 31 +++++++++++++++++++++++++++++--
> >  1 file changed, 29 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/bluetooth/hci_qca.c b/drivers/bluetooth/hci_qca.c
> > index f10bdf8e1fc5..dc95e378574b 100644
> > --- a/drivers/bluetooth/hci_qca.c
> > +++ b/drivers/bluetooth/hci_qca.c
> > @@ -164,6 +164,7 @@ struct qca_serdev {
> >  };
> >
> >  static int qca_regulator_enable(struct qca_serdev *qcadev);
> > +static int qca_power_on(struct qca_serdev *qcadev);
> >  static void qca_regulator_disable(struct qca_serdev *qcadev);
> >  static void qca_power_shutdown(struct hci_uart *hu);
> >  static int qca_power_off(struct hci_dev *hdev);
> > @@ -528,7 +529,7 @@ static int qca_open(struct hci_uart *hu)
> >  		} else {
> >  			hu->init_speed = qcadev->init_speed;
> >  			hu->oper_speed = qcadev->oper_speed;
> > -			ret = qca_regulator_enable(qcadev);
> > +			ret = qca_power_on(qcadev);
> >  			if (ret) {
> >  				destroy_workqueue(qca->workqueue);
> >  				kfree_skb(qca->rx_skb);
> > @@ -1214,7 +1215,7 @@ static int qca_wcn3990_init(struct hci_uart *hu)
> >  	qcadev = serdev_device_get_drvdata(hu->serdev);
> >  	if (!qcadev->bt_power->vregs_on) {
> >  		serdev_device_close(hu->serdev);
> > -		ret = qca_regulator_enable(qcadev);
> > +		ret = qca_power_on(qcadev);
> >  		if (ret)
> >  			return ret;
> >
> > @@ -1408,6 +1409,9 @@ static void qca_power_shutdown(struct hci_uart 
> > *hu)
> >  	host_set_baudrate(hu, 2400);
> >  	qca_send_power_pulse(hu, false);
> >  	qca_regulator_disable(qcadev);
> > +
> > +	if (qcadev->susclk)
> > +		clk_disable_unprepare(qcadev->susclk);
> >  }
> >
> >  static int qca_power_off(struct hci_dev *hdev)
> > @@ -1423,6 +1427,20 @@ static int qca_power_off(struct hci_dev *hdev)
> >  	return 0;
> >  }
> >
> > +static int qca_power_on(struct qca_serdev *qcadev)
> > +{
> > +	int err;
> > +
> > +	if (qcadev->susclk) {
> > +		err = clk_prepare_enable(qcadev->susclk);
> > +		if (err)
> > +			return err;
> > +	}
> > +
> > +	qca_regulator_enable(qcadev);
> > +	return 0;
> > +}
> > +
> >  static int qca_regulator_enable(struct qca_serdev *qcadev)
> >  {
> >  	struct qca_power *power = qcadev->bt_power;
> > @@ -1523,6 +1541,15 @@ static int qca_serdev_probe(struct 
> > serdev_device *serdev)
> >
> >  		qcadev->bt_power->vregs_on = false;
> >
> > +		if (qcadev->btsoc_type == QCA_WCN3990 ||
> > +		    qcadev->btsoc_type == QCA_WCN3991) {
> > +			qcadev->susclk = devm_clk_get(&serdev->dev, NULL);
> > +			if (IS_ERR(qcadev->susclk)) {
> > +				dev_err(&serdev->dev, "failed to acquire clk\n");
> > +				return PTR_ERR(qcadev->susclk);
> > +			}
>
> This will break existing users. Use devm_clk_get_optional() and at most
> raise a warning if the clock doesn't exist.
>
> It would also be nice to add the clock to the affected devices in the 
> tree
> if possible:
[Bala]: Sure we will use devm_clk_get_optional() in the next patch.

We will check the effected areas and update the necessary as i see some 
projects are not existing.

> arch/arm64/boot/dts/qcom/msm8998-clamshell.dtsi:
> compatible = "qcom,wcn3990-bt";
> arch/arm64/boot/dts/qcom/msm8998-mtp.dtsi:              compatible =
> "qcom,wcn3990-bt";
> arch/arm64/boot/dts/qcom/qcs404-evb.dtsi:               compatible =
> "qcom,wcn3990-bt";
> arch/arm64/boot/dts/qcom/sdm845-cheza.dtsi:             compatible =
> "qcom,wcn3990-bt";
> arch/arm64/boot/dts/qcom/sdm845-db845c.dts:             compatible =
> "qcom,wcn3990-bt";
> arch/arm64/boot/dts/qcom/sdm850-lenovo-yoga-c630.dts:
> compatible = "qcom,wcn3990-bt";