Allow for the PHY to be put into a powered down state when possible.
Add the required suspend and resume callbacks, which will determine
what resources can be turned off depending on the cable status.
Signed-off-by: Wesley Cheng <[email protected]>
---
Depends-on: phy: qcom-snps: Add SNPS USB PHY driver for QCOM based SOCs
(https://patchwork.kernel.org/patch/11486171/)
---
drivers/phy/qualcomm/phy-qcom-snps-femto-v2.c | 93 +++++++++++++++++++++++++++
1 file changed, 93 insertions(+)
diff --git a/drivers/phy/qualcomm/phy-qcom-snps-femto-v2.c b/drivers/phy/qualcomm/phy-qcom-snps-femto-v2.c
index 20442a3..f48d877 100644
--- a/drivers/phy/qualcomm/phy-qcom-snps-femto-v2.c
+++ b/drivers/phy/qualcomm/phy-qcom-snps-femto-v2.c
@@ -76,7 +76,9 @@
* @iface_clk: phy interface clock
* @phy_reset: phy reset control
* @vregs: regulator supplies bulk data
+ * @suspended: PHY is in the suspended state
* @phy_initialized: if PHY has been initialized correctly
+ * @mode: contains the current mode the PHY is in
*/
struct qcom_snps_hsphy {
struct phy *phy;
@@ -87,7 +89,9 @@ struct qcom_snps_hsphy {
struct reset_control *phy_reset;
struct regulator_bulk_data vregs[SNPS_HS_NUM_VREGS];
+ bool suspended;
bool phy_initialized;
+ enum phy_mode mode;
};
static inline void qcom_snps_hsphy_write_mask(void __iomem *base, u32 offset,
@@ -104,6 +108,77 @@ static inline void qcom_snps_hsphy_write_mask(void __iomem *base, u32 offset,
readl_relaxed(base + offset);
}
+static int qcom_snps_hsphy_suspend(struct qcom_snps_hsphy *hsphy)
+{
+ if (hsphy->suspended)
+ return 0;
+
+ dev_dbg(&hsphy->phy->dev, "Suspend QCOM SNPS PHY, mode = %d \n", hsphy->mode);
+
+ if (hsphy->mode == PHY_MODE_USB_HOST) {
+ /* Enable auto-resume to meet remote wakeup timing */
+ qcom_snps_hsphy_write_mask(hsphy->base, USB2_PHY_USB_PHY_HS_PHY_CTRL2,
+ USB2_AUTO_RESUME, USB2_AUTO_RESUME);
+ usleep_range(500, 1000);
+ qcom_snps_hsphy_write_mask(hsphy->base, USB2_PHY_USB_PHY_HS_PHY_CTRL2,
+ 0, USB2_AUTO_RESUME);
+ }
+
+ clk_disable_unprepare(hsphy->cfg_ahb_clk);
+ hsphy->suspended = true;
+
+ return 0;
+}
+
+static int qcom_snps_hsphy_resume(struct qcom_snps_hsphy *hsphy)
+{
+ int ret = 0;
+
+ if (!hsphy->suspended)
+ return 0;
+
+ dev_dbg(&hsphy->phy->dev, "Resume QCOM SNPS PHY, mode = %d \n", hsphy->mode);
+
+ ret = clk_prepare_enable(hsphy->cfg_ahb_clk);
+ if (ret) {
+ dev_err(&hsphy->phy->dev, "failed to enable cfg ahb clock, %d\n", ret);
+ return ret;
+ }
+
+ hsphy->suspended = false;
+ return 0;
+}
+
+static int __maybe_unused qcom_snps_hsphy_runtime_suspend(struct device *dev)
+{
+ struct qcom_snps_hsphy *hsphy = dev_get_drvdata(dev);
+
+ if (!hsphy->phy_initialized)
+ return 0;
+
+ qcom_snps_hsphy_suspend(hsphy);
+ return 0;
+}
+
+static int __maybe_unused qcom_snps_hsphy_runtime_resume(struct device *dev)
+{
+ struct qcom_snps_hsphy *hsphy = dev_get_drvdata(dev);
+
+ if (!hsphy->phy_initialized)
+ return 0;
+
+ qcom_snps_hsphy_resume(hsphy);
+ return 0;
+}
+
+static int qcom_snps_hsphy_set_mode(struct phy *phy, enum phy_mode mode, int submode)
+{
+ struct qcom_snps_hsphy *hsphy = phy_get_drvdata(phy);
+
+ hsphy->mode = mode;
+ return 0;
+}
+
static int qcom_snps_hsphy_init(struct phy *phy)
{
struct qcom_snps_hsphy *hsphy = phy_get_drvdata(phy);
@@ -171,6 +246,7 @@ static int qcom_snps_hsphy_init(struct phy *phy)
UTMI_PHY_CMN_CTRL_OVERRIDE_EN, 0);
hsphy->phy_initialized = true;
+ hsphy->suspended = false;
return 0;
@@ -197,6 +273,7 @@ static int qcom_snps_hsphy_exit(struct phy *phy)
static const struct phy_ops qcom_snps_hsphy_gen_ops = {
.init = qcom_snps_hsphy_init,
.exit = qcom_snps_hsphy_exit,
+ .set_mode = qcom_snps_hsphy_set_mode,
.owner = THIS_MODULE,
};
@@ -208,6 +285,11 @@ static int qcom_snps_hsphy_exit(struct phy *phy)
};
MODULE_DEVICE_TABLE(of, qcom_snps_hsphy_of_match_table);
+static const struct dev_pm_ops qcom_snps_hsphy_pm_ops = {
+ SET_RUNTIME_PM_OPS(qcom_snps_hsphy_runtime_suspend,
+ qcom_snps_hsphy_runtime_resume, NULL)
+};
+
static int qcom_snps_hsphy_probe(struct platform_device *pdev)
{
struct device *dev = &pdev->dev;
@@ -251,6 +333,14 @@ static int qcom_snps_hsphy_probe(struct platform_device *pdev)
return ret;
}
+ pm_runtime_set_active(dev);
+ pm_runtime_enable(dev);
+ /*
+ * Prevent runtime pm from being ON by default. Users can enable
+ * it using power/control in sysfs.
+ */
+ pm_runtime_forbid(dev);
+
generic_phy = devm_phy_create(dev, NULL, &qcom_snps_hsphy_gen_ops);
if (IS_ERR(generic_phy)) {
ret = PTR_ERR(generic_phy);
@@ -265,6 +355,8 @@ static int qcom_snps_hsphy_probe(struct platform_device *pdev)
phy_provider = devm_of_phy_provider_register(dev, of_phy_simple_xlate);
if (!IS_ERR(phy_provider))
dev_dbg(dev, "Registered Qcom-SNPS HS phy\n");
+ else
+ pm_runtime_disable(dev);
return PTR_ERR_OR_ZERO(phy_provider);
}
@@ -273,6 +365,7 @@ static int qcom_snps_hsphy_probe(struct platform_device *pdev)
.probe = qcom_snps_hsphy_probe,
.driver = {
.name = "qcom-snps-hs-femto-v2-phy",
+ .pm = &qcom_snps_hsphy_pm_ops,
.of_match_table = qcom_snps_hsphy_of_match_table,
},
};
--
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project
On 23-04-20, 10:26, Wesley Cheng wrote:
> +static int qcom_snps_hsphy_suspend(struct qcom_snps_hsphy *hsphy)
> +{
> + if (hsphy->suspended)
> + return 0;
> +
> + dev_dbg(&hsphy->phy->dev, "Suspend QCOM SNPS PHY, mode = %d \n", hsphy->mode);
> +
> + if (hsphy->mode == PHY_MODE_USB_HOST) {
> + /* Enable auto-resume to meet remote wakeup timing */
> + qcom_snps_hsphy_write_mask(hsphy->base, USB2_PHY_USB_PHY_HS_PHY_CTRL2,
> + USB2_AUTO_RESUME, USB2_AUTO_RESUME);
> + usleep_range(500, 1000);
> + qcom_snps_hsphy_write_mask(hsphy->base, USB2_PHY_USB_PHY_HS_PHY_CTRL2,
> + 0, USB2_AUTO_RESUME);
Kernel has a coding guideline where we try to "stick" to 80 char limit
and is sometimes okay like debug logs. Above is not okay. Please fix it
and run ./scripts/checkpatch.pl --strict on your patch and fix all
errors. Warning and checks at your discretion using common sense. When
in doubt do ask :)
> + }
> +
> + clk_disable_unprepare(hsphy->cfg_ahb_clk);
> + hsphy->suspended = true;
why do you need to track this?
> +
> + return 0;
> +}
> +
> +static int qcom_snps_hsphy_resume(struct qcom_snps_hsphy *hsphy)
> +{
> + int ret = 0;
superfluous init..
> static int qcom_snps_hsphy_probe(struct platform_device *pdev)
> {
> struct device *dev = &pdev->dev;
> @@ -251,6 +333,14 @@ static int qcom_snps_hsphy_probe(struct platform_device *pdev)
> return ret;
> }
>
> + pm_runtime_set_active(dev);
> + pm_runtime_enable(dev);
would it not make sense to enable this after pjy in initialized?
--
~Vinod
On 4/27/2020 9:59 AM, Vinod Koul wrote:
> On 23-04-20, 10:26, Wesley Cheng wrote:
>
>> +static int qcom_snps_hsphy_suspend(struct qcom_snps_hsphy *hsphy)
>> +{
>> + if (hsphy->suspended)
>> + return 0;
>> +
>> + dev_dbg(&hsphy->phy->dev, "Suspend QCOM SNPS PHY, mode = %d \n", hsphy->mode);
>> +
>> + if (hsphy->mode == PHY_MODE_USB_HOST) {
>> + /* Enable auto-resume to meet remote wakeup timing */
>> + qcom_snps_hsphy_write_mask(hsphy->base, USB2_PHY_USB_PHY_HS_PHY_CTRL2,
>> + USB2_AUTO_RESUME, USB2_AUTO_RESUME);
>> + usleep_range(500, 1000);
>> + qcom_snps_hsphy_write_mask(hsphy->base, USB2_PHY_USB_PHY_HS_PHY_CTRL2,
>> + 0, USB2_AUTO_RESUME);
>
> Kernel has a coding guideline where we try to "stick" to 80 char limit
> and is sometimes okay like debug logs. Above is not okay. Please fix it
> and run ./scripts/checkpatch.pl --strict on your patch and fix all
> errors. Warning and checks at your discretion using common sense. When
> in doubt do ask :)
>
Hi Vinod,
Thanks for the input. I do run the checkpatch script before sending
patches, and addressing the errors. However, seems this was tagged as a
warning instead. Will keep in mind to address as many warnings as I can
as well.
>> + }
>> +
>> + clk_disable_unprepare(hsphy->cfg_ahb_clk);
>> + hsphy->suspended = true;
>
> why do you need to track this?
>
More for debug purposes in case the RPM state becomes out of sync with
the expected PHY state. We've seen some situations during PM
suspend/resume testing where our RPM routines aren't executed, as PM
will disable RPM briefly. It would be nice to be able to catch these
situations after collecting our crash dumps.
>> +
>> + return 0;
>> +}
>> +
>> +static int qcom_snps_hsphy_resume(struct qcom_snps_hsphy *hsphy)
>> +{
>> + int ret = 0;
>
> superfluous init..
>
Agreed.
>> static int qcom_snps_hsphy_probe(struct platform_device *pdev)
>> {
>> struct device *dev = &pdev->dev;
>> @@ -251,6 +333,14 @@ static int qcom_snps_hsphy_probe(struct platform_device *pdev)
>> return ret;
>> }
>>
>> + pm_runtime_set_active(dev);
>> + pm_runtime_enable(dev);
>
> would it not make sense to enable this after pjy in initialized?
>
Not sure we want to put this in the phy_init() callback, as we can't
guarantee how the driver registering the PHY will use it. It'll put the
requirement of having to call phy_exit() for every phy_init() sequence,
in order to avoid unbalanced disable_depth warnings from the RPM driver.
--
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project