2017-03-13 09:41:47

by Yoshihiro Shimoda

Subject: [PATCH v2] phy: rcar-gen3-usb2: fix implementation for runtime PM

This patch fixes an issue that this driver doesn't take care of the runtime
PM. This code assumed that devm_phy_create() called pm_runtime_enable(dev),
but it misunderstood the dev_phy_create()'s specification.
This driver should call its own pm_runtime_pm() before dev_phy_create().

Fixes: f3b5a8d9b50d ("phy: rcar-gen3-usb2: Add R-Car Gen3 USB2 PHY driver")
Signed-off-by: Yoshihiro Shimoda <[email protected]>
---
Changes from v1:
- revise the commit log (s/the own/its own/)
- simplify return code in probe().

drivers/phy/phy-rcar-gen3-usb2.c | 31 ++++++++++++++++++++++++-------
1 file changed, 24 insertions(+), 7 deletions(-)

diff --git a/drivers/phy/phy-rcar-gen3-usb2.c b/drivers/phy/phy-rcar-gen3-usb2.c
index afb4d04..54c3429 100644
--- a/drivers/phy/phy-rcar-gen3-usb2.c
+++ b/drivers/phy/phy-rcar-gen3-usb2.c
@@ -20,6 +20,7 @@
#include <linux/of_address.h>
#include <linux/phy/phy.h>
#include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
#include <linux/regulator/consumer.h>
#include <linux/workqueue.h>

@@ -395,7 +396,7 @@ static int rcar_gen3_phy_usb2_probe(struct platform_device *pdev)
struct rcar_gen3_chan *channel;
struct phy_provider *provider;
struct resource *res;
- int irq;
+ int irq, ret = 0;

if (!dev->of_node) {
dev_err(dev, "This driver needs device tree\n");
@@ -434,17 +435,24 @@ static int rcar_gen3_phy_usb2_probe(struct platform_device *pdev)
}
}

- /* devm_phy_create() will call pm_runtime_enable(dev); */
+ /*
+ * devm_phy_create() will call pm_runtime_enable(&phy->dev);
+ * And then, phy-core will manage runtime pm for this device.
+ */
+ pm_runtime_enable(dev);
channel->phy = devm_phy_create(dev, NULL, &rcar_gen3_phy_usb2_ops);
if (IS_ERR(channel->phy)) {
dev_err(dev, "Failed to create USB2 PHY\n");
- return PTR_ERR(channel->phy);
+ ret = PTR_ERR(channel->phy);
+ goto error;
}

channel->vbus = devm_regulator_get_optional(dev, "vbus");
if (IS_ERR(channel->vbus)) {
- if (PTR_ERR(channel->vbus) == -EPROBE_DEFER)
- return PTR_ERR(channel->vbus);
+ if (PTR_ERR(channel->vbus) == -EPROBE_DEFER) {
+ ret = PTR_ERR(channel->vbus);
+ goto error;
+ }
channel->vbus = NULL;
}

@@ -454,15 +462,22 @@ static int rcar_gen3_phy_usb2_probe(struct platform_device *pdev)
provider = devm_of_phy_provider_register(dev, of_phy_simple_xlate);
if (IS_ERR(provider)) {
dev_err(dev, "Failed to register PHY provider\n");
+ ret = PTR_ERR(provider);
+ goto error;
} else if (channel->has_otg) {
int ret;

ret = device_create_file(dev, &dev_attr_role);
if (ret < 0)
- return ret;
+ goto error;
}

- return PTR_ERR_OR_ZERO(provider);
+ return 0;
+
+error:
+ pm_runtime_disable(dev);
+
+ return ret;
}

static int rcar_gen3_phy_usb2_remove(struct platform_device *pdev)
@@ -472,6 +487,8 @@ static int rcar_gen3_phy_usb2_remove(struct platform_device *pdev)
if (channel->has_otg)
device_remove_file(&pdev->dev, &dev_attr_role);

+ pm_runtime_disable(&pdev->dev);
+
return 0;
};

--
1.9.1

2017-03-13 11:57:15

by Sergei Shtylyov

Subject: Re: [PATCH v2] phy: rcar-gen3-usb2: fix implementation for runtime PM

On 03/13/2017 12:39 PM, Yoshihiro Shimoda wrote:

> This patch fixes an issue that this driver doesn't take care of the runtime
> PM. This code assumed that devm_phy_create() called pm_runtime_enable(dev),
> but it misunderstood the dev_phy_create()'s specification.
> This driver should call its own pm_runtime_pm() before dev_phy_create().

You probably meant pm_runtime_enable()? Sorry for not noticing the 1st time...

> Fixes: f3b5a8d9b50d ("phy: rcar-gen3-usb2: Add R-Car Gen3 USB2 PHY driver")
> Signed-off-by: Yoshihiro Shimoda <[email protected]>
[...]

MBR, Sergei

2017-03-13 23:39:39

by Yoshihiro Shimoda

Subject: RE: [PATCH v2] phy: rcar-gen3-usb2: fix implementation for runtime PM

Hi!

> From: Sergei Shtylyov
> Sent: Monday, March 13, 2017 8:57 PM
>
> On 03/13/2017 12:39 PM, Yoshihiro Shimoda wrote:
>
> > This patch fixes an issue that this driver doesn't take care of the runtime
> > PM. This code assumed that devm_phy_create() called pm_runtime_enable(dev),
> > but it misunderstood the dev_phy_create()'s specification.
> > This driver should call its own pm_runtime_pm() before dev_phy_create().
>
> You probably meant pm_runtime_enable()? Sorry for not noticing the 1st time...

Yes, I meant pm_runtime_enable(). Thank you for the review again and again.

Best regards,
Yoshihiro Shimoda