[4/5] phy: tegra: xusb: Add support for power supplies

Message ID	20190125112525.10697-4-thierry.reding@gmail.com
State	Changes Requested
Headers	show Return-Path: <linux-tegra-owner@vger.kernel.org> From: Thierry Reding <thierry.reding@gmail.com> To: Kishon Vijay Abraham I <kishon@ti.com> Cc: Jonathan Hunter <jonathanh@nvidia.com>, JC Kuo <jckuo@nvidia.com>, linux-tegra@vger.kernel.org, linux-kernel@vger.kernel.org Subject: [PATCH 4/5] phy: tegra: xusb: Add support for power supplies Date: Fri, 25 Jan 2019 12:25:24 +0100 Message-Id: <20190125112525.10697-4-thierry.reding@gmail.com> In-Reply-To: <20190125112525.10697-1-thierry.reding@gmail.com> References: <20190125112525.10697-1-thierry.reding@gmail.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Sender: linux-tegra-owner@vger.kernel.org Precedence: bulk
Series	[1/5] dt-bindings: phy: tegra: Add Tegra186 support \| expand [1/5] dt-bindings: phy: tegra: Add Tegra186 support [2/5] phy: tegra: xusb: Skip single function lane programming [3/5] phy: tegra: xusb: Parse dual-role mode property [4/5] phy: tegra: xusb: Add support for power supplies [5/5] phy: tegra: xusb: Add Tegra186 support

Thierry Reding Jan. 25, 2019, 11:25 a.m. UTC

From: Thierry Reding <treding@nvidia.com>

Support enabling various supplies needed to provide power to the PLLs
and logic used to drive the USB, PCI and SATA pads.

Signed-off-by: Thierry Reding <treding@nvidia.com>
---
 drivers/phy/tegra/xusb.c | 34 +++++++++++++++++++++++++++++++++-
 drivers/phy/tegra/xusb.h |  5 +++++
 2 files changed, 38 insertions(+), 1 deletion(-)

JC Kuo Jan. 28, 2019, 7:22 a.m. UTC | #1

Hi Thierry,

I think any non-zero return value of 
regulator_bulk_enable()/devm_regulator_bulk_get() means error.

Thanks,

JC

On 1/25/19 7:25 PM, Thierry Reding wrote:
> From: Thierry Reding <treding@nvidia.com>
>
> Support enabling various supplies needed to provide power to the PLLs
> and logic used to drive the USB, PCI and SATA pads.
>
> Signed-off-by: Thierry Reding <treding@nvidia.com>
> ---
>   drivers/phy/tegra/xusb.c | 34 +++++++++++++++++++++++++++++++++-
>   drivers/phy/tegra/xusb.h |  5 +++++
>   2 files changed, 38 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/phy/tegra/xusb.c b/drivers/phy/tegra/xusb.c
> index 57a2d08ef6da..e510629f4f1c 100644
> --- a/drivers/phy/tegra/xusb.c
> +++ b/drivers/phy/tegra/xusb.c
> @@ -864,6 +864,7 @@ static int tegra_xusb_padctl_probe(struct platform_device *pdev)
>   	struct tegra_xusb_padctl *padctl;
>   	const struct of_device_id *match;
>   	struct resource *res;
> +	unsigned int i;
>   	int err;
>   
>   	/* for backwards compatibility with old device trees */
> @@ -901,14 +902,38 @@ static int tegra_xusb_padctl_probe(struct platform_device *pdev)
>   		goto remove;
>   	}
>   
> +	padctl->supplies = devm_kcalloc(&pdev->dev, padctl->soc->num_supplies,
> +					sizeof(*padctl->supplies), GFP_KERNEL);
> +	if (!padctl->supplies) {
> +		err = -ENOMEM;
> +		goto remove;
> +	}
> +
> +	for (i = 0; i < padctl->soc->num_supplies; i++)
> +		padctl->supplies[i].supply = padctl->soc->supply_names[i];
> +
> +	err = devm_regulator_bulk_get(&pdev->dev, padctl->soc->num_supplies,
> +				      padctl->supplies);
> +	if (err < 0) {
> +		dev_err(&pdev->dev, "failed to get regulators: %d\n", err);
> +		goto remove;
> +	}
> +
>   	err = reset_control_deassert(padctl->rst);
>   	if (err < 0)
>   		goto remove;
>   
> +	err = regulator_bulk_enable(padctl->soc->num_supplies,
> +				    padctl->supplies);
> +	if (err < 0) {
> +		dev_err(&pdev->dev, "failed to enable supplies: %d\n", err);
> +		goto reset;
> +	}
> +
>   	err = tegra_xusb_setup_pads(padctl);
>   	if (err < 0) {
>   		dev_err(&pdev->dev, "failed to setup pads: %d\n", err);
> -		goto reset;
> +		goto power_down;
>   	}
>   
>   	err = tegra_xusb_setup_ports(padctl);
> @@ -921,6 +946,8 @@ static int tegra_xusb_padctl_probe(struct platform_device *pdev)
>   
>   remove_pads:
>   	tegra_xusb_remove_pads(padctl);
> +power_down:
> +	regulator_bulk_disable(padctl->soc->num_supplies, padctl->supplies);
>   reset:
>   	reset_control_assert(padctl->rst);
>   remove:
> @@ -936,6 +963,11 @@ static int tegra_xusb_padctl_remove(struct platform_device *pdev)
>   	tegra_xusb_remove_ports(padctl);
>   	tegra_xusb_remove_pads(padctl);
>   
> +	err = regulator_bulk_disable(padctl->soc->num_supplies,
> +				     padctl->supplies);
> +	if (err < 0)
> +		dev_err(&pdev->dev, "failed to disable supplies: %d\n", err);
> +
>   	err = reset_control_assert(padctl->rst);
>   	if (err < 0)
>   		dev_err(&pdev->dev, "failed to assert reset: %d\n", err);
> diff --git a/drivers/phy/tegra/xusb.h b/drivers/phy/tegra/xusb.h
> index bb60fc09c752..5d5d22f6cb41 100644
> --- a/drivers/phy/tegra/xusb.h
> +++ b/drivers/phy/tegra/xusb.h
> @@ -370,6 +370,9 @@ struct tegra_xusb_padctl_soc {
>   	} ports;
>   
>   	const struct tegra_xusb_padctl_ops *ops;
> +
> +	const char * const *supply_names;
> +	unsigned int num_supplies;
>   };
>   
>   struct tegra_xusb_padctl {
> @@ -393,6 +396,8 @@ struct tegra_xusb_padctl {
>   	unsigned int enable;
>   
>   	struct clk *clk;
> +
> +	struct regulator_bulk_data *supplies;
>   };
>   
>   static inline void padctl_writel(struct tegra_xusb_padctl *padctl, u32 value,

Thierry Reding Jan. 28, 2019, 8 a.m. UTC | #2

On Mon, Jan 28, 2019 at 03:22:09PM +0800, jckuo wrote:
> Hi Thierry,
> 
> I think any non-zero return value of
> regulator_bulk_enable()/devm_regulator_bulk_get() means error.
> 
> Thanks,
> 
> JC

Theoretically I think only regulator_bulk_enable() could return a
positive value, but even so it never will in practice because all of the
regulator_enable() (see _regulator_enable()) calls will only ever return
negative error codes.

I can change this and resend if you have strong concerns about this
possibly missing legitimate error cases.

Thierry

> On 1/25/19 7:25 PM, Thierry Reding wrote:
> > From: Thierry Reding <treding@nvidia.com>
> > 
> > Support enabling various supplies needed to provide power to the PLLs
> > and logic used to drive the USB, PCI and SATA pads.
> > 
> > Signed-off-by: Thierry Reding <treding@nvidia.com>
> > ---
> >   drivers/phy/tegra/xusb.c | 34 +++++++++++++++++++++++++++++++++-
> >   drivers/phy/tegra/xusb.h |  5 +++++
> >   2 files changed, 38 insertions(+), 1 deletion(-)
> > 
> > diff --git a/drivers/phy/tegra/xusb.c b/drivers/phy/tegra/xusb.c
> > index 57a2d08ef6da..e510629f4f1c 100644
> > --- a/drivers/phy/tegra/xusb.c
> > +++ b/drivers/phy/tegra/xusb.c
> > @@ -864,6 +864,7 @@ static int tegra_xusb_padctl_probe(struct platform_device *pdev)
> >   	struct tegra_xusb_padctl *padctl;
> >   	const struct of_device_id *match;
> >   	struct resource *res;
> > +	unsigned int i;
> >   	int err;
> >   	/* for backwards compatibility with old device trees */
> > @@ -901,14 +902,38 @@ static int tegra_xusb_padctl_probe(struct platform_device *pdev)
> >   		goto remove;
> >   	}
> > +	padctl->supplies = devm_kcalloc(&pdev->dev, padctl->soc->num_supplies,
> > +					sizeof(*padctl->supplies), GFP_KERNEL);
> > +	if (!padctl->supplies) {
> > +		err = -ENOMEM;
> > +		goto remove;
> > +	}
> > +
> > +	for (i = 0; i < padctl->soc->num_supplies; i++)
> > +		padctl->supplies[i].supply = padctl->soc->supply_names[i];
> > +
> > +	err = devm_regulator_bulk_get(&pdev->dev, padctl->soc->num_supplies,
> > +				      padctl->supplies);
> > +	if (err < 0) {
> > +		dev_err(&pdev->dev, "failed to get regulators: %d\n", err);
> > +		goto remove;
> > +	}
> > +
> >   	err = reset_control_deassert(padctl->rst);
> >   	if (err < 0)
> >   		goto remove;
> > +	err = regulator_bulk_enable(padctl->soc->num_supplies,
> > +				    padctl->supplies);
> > +	if (err < 0) {
> > +		dev_err(&pdev->dev, "failed to enable supplies: %d\n", err);
> > +		goto reset;
> > +	}
> > +
> >   	err = tegra_xusb_setup_pads(padctl);
> >   	if (err < 0) {
> >   		dev_err(&pdev->dev, "failed to setup pads: %d\n", err);
> > -		goto reset;
> > +		goto power_down;
> >   	}
> >   	err = tegra_xusb_setup_ports(padctl);
> > @@ -921,6 +946,8 @@ static int tegra_xusb_padctl_probe(struct platform_device *pdev)
> >   remove_pads:
> >   	tegra_xusb_remove_pads(padctl);
> > +power_down:
> > +	regulator_bulk_disable(padctl->soc->num_supplies, padctl->supplies);
> >   reset:
> >   	reset_control_assert(padctl->rst);
> >   remove:
> > @@ -936,6 +963,11 @@ static int tegra_xusb_padctl_remove(struct platform_device *pdev)
> >   	tegra_xusb_remove_ports(padctl);
> >   	tegra_xusb_remove_pads(padctl);
> > +	err = regulator_bulk_disable(padctl->soc->num_supplies,
> > +				     padctl->supplies);
> > +	if (err < 0)
> > +		dev_err(&pdev->dev, "failed to disable supplies: %d\n", err);
> > +
> >   	err = reset_control_assert(padctl->rst);
> >   	if (err < 0)
> >   		dev_err(&pdev->dev, "failed to assert reset: %d\n", err);
> > diff --git a/drivers/phy/tegra/xusb.h b/drivers/phy/tegra/xusb.h
> > index bb60fc09c752..5d5d22f6cb41 100644
> > --- a/drivers/phy/tegra/xusb.h
> > +++ b/drivers/phy/tegra/xusb.h
> > @@ -370,6 +370,9 @@ struct tegra_xusb_padctl_soc {
> >   	} ports;
> >   	const struct tegra_xusb_padctl_ops *ops;
> > +
> > +	const char * const *supply_names;
> > +	unsigned int num_supplies;
> >   };
> >   struct tegra_xusb_padctl {
> > @@ -393,6 +396,8 @@ struct tegra_xusb_padctl {
> >   	unsigned int enable;
> >   	struct clk *clk;
> > +
> > +	struct regulator_bulk_data *supplies;
> >   };
> >   static inline void padctl_writel(struct tegra_xusb_padctl *padctl, u32 value,

JC Kuo Jan. 29, 2019, 6:50 a.m. UTC | #3

Thanks Thierry.

Reviewed-by: JC Kuo <jckuo@nvidia.com>

On 1/28/19 4:00 PM, Thierry Reding wrote:
> On Mon, Jan 28, 2019 at 03:22:09PM +0800, jckuo wrote:
>> Hi Thierry,
>>
>> I think any non-zero return value of
>> regulator_bulk_enable()/devm_regulator_bulk_get() means error.
>>
>> Thanks,
>>
>> JC
> Theoretically I think only regulator_bulk_enable() could return a
> positive value, but even so it never will in practice because all of the
> regulator_enable() (see _regulator_enable()) calls will only ever return
> negative error codes.
>
> I can change this and resend if you have strong concerns about this
> possibly missing legitimate error cases.
>
> Thierry
>
>> On 1/25/19 7:25 PM, Thierry Reding wrote:
>>> From: Thierry Reding <treding@nvidia.com>
>>>
>>> Support enabling various supplies needed to provide power to the PLLs
>>> and logic used to drive the USB, PCI and SATA pads.
>>>
>>> Signed-off-by: Thierry Reding <treding@nvidia.com>
>>> ---
>>>    drivers/phy/tegra/xusb.c | 34 +++++++++++++++++++++++++++++++++-
>>>    drivers/phy/tegra/xusb.h |  5 +++++
>>>    2 files changed, 38 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/phy/tegra/xusb.c b/drivers/phy/tegra/xusb.c
>>> index 57a2d08ef6da..e510629f4f1c 100644
>>> --- a/drivers/phy/tegra/xusb.c
>>> +++ b/drivers/phy/tegra/xusb.c
>>> @@ -864,6 +864,7 @@ static int tegra_xusb_padctl_probe(struct platform_device *pdev)
>>>    	struct tegra_xusb_padctl *padctl;
>>>    	const struct of_device_id *match;
>>>    	struct resource *res;
>>> +	unsigned int i;
>>>    	int err;
>>>    	/* for backwards compatibility with old device trees */
>>> @@ -901,14 +902,38 @@ static int tegra_xusb_padctl_probe(struct platform_device *pdev)
>>>    		goto remove;
>>>    	}
>>> +	padctl->supplies = devm_kcalloc(&pdev->dev, padctl->soc->num_supplies,
>>> +					sizeof(*padctl->supplies), GFP_KERNEL);
>>> +	if (!padctl->supplies) {
>>> +		err = -ENOMEM;
>>> +		goto remove;
>>> +	}
>>> +
>>> +	for (i = 0; i < padctl->soc->num_supplies; i++)
>>> +		padctl->supplies[i].supply = padctl->soc->supply_names[i];
>>> +
>>> +	err = devm_regulator_bulk_get(&pdev->dev, padctl->soc->num_supplies,
>>> +				      padctl->supplies);
>>> +	if (err < 0) {
>>> +		dev_err(&pdev->dev, "failed to get regulators: %d\n", err);
>>> +		goto remove;
>>> +	}
>>> +
>>>    	err = reset_control_deassert(padctl->rst);
>>>    	if (err < 0)
>>>    		goto remove;
>>> +	err = regulator_bulk_enable(padctl->soc->num_supplies,
>>> +				    padctl->supplies);
>>> +	if (err < 0) {
>>> +		dev_err(&pdev->dev, "failed to enable supplies: %d\n", err);
>>> +		goto reset;
>>> +	}
>>> +
>>>    	err = tegra_xusb_setup_pads(padctl);
>>>    	if (err < 0) {
>>>    		dev_err(&pdev->dev, "failed to setup pads: %d\n", err);
>>> -		goto reset;
>>> +		goto power_down;
>>>    	}
>>>    	err = tegra_xusb_setup_ports(padctl);
>>> @@ -921,6 +946,8 @@ static int tegra_xusb_padctl_probe(struct platform_device *pdev)
>>>    remove_pads:
>>>    	tegra_xusb_remove_pads(padctl);
>>> +power_down:
>>> +	regulator_bulk_disable(padctl->soc->num_supplies, padctl->supplies);
>>>    reset:
>>>    	reset_control_assert(padctl->rst);
>>>    remove:
>>> @@ -936,6 +963,11 @@ static int tegra_xusb_padctl_remove(struct platform_device *pdev)
>>>    	tegra_xusb_remove_ports(padctl);
>>>    	tegra_xusb_remove_pads(padctl);
>>> +	err = regulator_bulk_disable(padctl->soc->num_supplies,
>>> +				     padctl->supplies);
>>> +	if (err < 0)
>>> +		dev_err(&pdev->dev, "failed to disable supplies: %d\n", err);
>>> +
>>>    	err = reset_control_assert(padctl->rst);
>>>    	if (err < 0)
>>>    		dev_err(&pdev->dev, "failed to assert reset: %d\n", err);
>>> diff --git a/drivers/phy/tegra/xusb.h b/drivers/phy/tegra/xusb.h
>>> index bb60fc09c752..5d5d22f6cb41 100644
>>> --- a/drivers/phy/tegra/xusb.h
>>> +++ b/drivers/phy/tegra/xusb.h
>>> @@ -370,6 +370,9 @@ struct tegra_xusb_padctl_soc {
>>>    	} ports;
>>>    	const struct tegra_xusb_padctl_ops *ops;
>>> +
>>> +	const char * const *supply_names;
>>> +	unsigned int num_supplies;
>>>    };
>>>    struct tegra_xusb_padctl {
>>> @@ -393,6 +396,8 @@ struct tegra_xusb_padctl {
>>>    	unsigned int enable;
>>>    	struct clk *clk;
>>> +
>>> +	struct regulator_bulk_data *supplies;
>>>    };
>>>    static inline void padctl_writel(struct tegra_xusb_padctl *padctl, u32 value,

Kishon Vijay Abraham I Feb. 5, 2019, 12:55 p.m. UTC | #4

Hi,

On 28/01/19 1:30 PM, Thierry Reding wrote:
> On Mon, Jan 28, 2019 at 03:22:09PM +0800, jckuo wrote:
>> Hi Thierry,
>>
>> I think any non-zero return value of
>> regulator_bulk_enable()/devm_regulator_bulk_get() means error.
>>
>> Thanks,
>>
>> JC
> 
> Theoretically I think only regulator_bulk_enable() could return a
> positive value, but even so it never will in practice because all of the
> regulator_enable() (see _regulator_enable()) calls will only ever return
> negative error codes.
> 
> I can change this and resend if you have strong concerns about this
> possibly missing legitimate error cases.

All error codes are negative, so this is fine IMO.

Thanks
Kishon
> 
> Thierry
> 
>> On 1/25/19 7:25 PM, Thierry Reding wrote:
>>> From: Thierry Reding <treding@nvidia.com>
>>>
>>> Support enabling various supplies needed to provide power to the PLLs
>>> and logic used to drive the USB, PCI and SATA pads.
>>>
>>> Signed-off-by: Thierry Reding <treding@nvidia.com>
>>> ---
>>>   drivers/phy/tegra/xusb.c | 34 +++++++++++++++++++++++++++++++++-
>>>   drivers/phy/tegra/xusb.h |  5 +++++
>>>   2 files changed, 38 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/phy/tegra/xusb.c b/drivers/phy/tegra/xusb.c
>>> index 57a2d08ef6da..e510629f4f1c 100644
>>> --- a/drivers/phy/tegra/xusb.c
>>> +++ b/drivers/phy/tegra/xusb.c
>>> @@ -864,6 +864,7 @@ static int tegra_xusb_padctl_probe(struct platform_device *pdev)
>>>   	struct tegra_xusb_padctl *padctl;
>>>   	const struct of_device_id *match;
>>>   	struct resource *res;
>>> +	unsigned int i;
>>>   	int err;
>>>   	/* for backwards compatibility with old device trees */
>>> @@ -901,14 +902,38 @@ static int tegra_xusb_padctl_probe(struct platform_device *pdev)
>>>   		goto remove;
>>>   	}
>>> +	padctl->supplies = devm_kcalloc(&pdev->dev, padctl->soc->num_supplies,
>>> +					sizeof(*padctl->supplies), GFP_KERNEL);
>>> +	if (!padctl->supplies) {
>>> +		err = -ENOMEM;
>>> +		goto remove;
>>> +	}
>>> +
>>> +	for (i = 0; i < padctl->soc->num_supplies; i++)
>>> +		padctl->supplies[i].supply = padctl->soc->supply_names[i];
>>> +
>>> +	err = devm_regulator_bulk_get(&pdev->dev, padctl->soc->num_supplies,
>>> +				      padctl->supplies);
>>> +	if (err < 0) {
>>> +		dev_err(&pdev->dev, "failed to get regulators: %d\n", err);
>>> +		goto remove;
>>> +	}
>>> +
>>>   	err = reset_control_deassert(padctl->rst);
>>>   	if (err < 0)
>>>   		goto remove;
>>> +	err = regulator_bulk_enable(padctl->soc->num_supplies,
>>> +				    padctl->supplies);
>>> +	if (err < 0) {
>>> +		dev_err(&pdev->dev, "failed to enable supplies: %d\n", err);
>>> +		goto reset;
>>> +	}
>>> +
>>>   	err = tegra_xusb_setup_pads(padctl);
>>>   	if (err < 0) {
>>>   		dev_err(&pdev->dev, "failed to setup pads: %d\n", err);
>>> -		goto reset;
>>> +		goto power_down;
>>>   	}
>>>   	err = tegra_xusb_setup_ports(padctl);
>>> @@ -921,6 +946,8 @@ static int tegra_xusb_padctl_probe(struct platform_device *pdev)
>>>   remove_pads:
>>>   	tegra_xusb_remove_pads(padctl);
>>> +power_down:
>>> +	regulator_bulk_disable(padctl->soc->num_supplies, padctl->supplies);
>>>   reset:
>>>   	reset_control_assert(padctl->rst);
>>>   remove:
>>> @@ -936,6 +963,11 @@ static int tegra_xusb_padctl_remove(struct platform_device *pdev)
>>>   	tegra_xusb_remove_ports(padctl);
>>>   	tegra_xusb_remove_pads(padctl);
>>> +	err = regulator_bulk_disable(padctl->soc->num_supplies,
>>> +				     padctl->supplies);
>>> +	if (err < 0)
>>> +		dev_err(&pdev->dev, "failed to disable supplies: %d\n", err);
>>> +
>>>   	err = reset_control_assert(padctl->rst);
>>>   	if (err < 0)
>>>   		dev_err(&pdev->dev, "failed to assert reset: %d\n", err);
>>> diff --git a/drivers/phy/tegra/xusb.h b/drivers/phy/tegra/xusb.h
>>> index bb60fc09c752..5d5d22f6cb41 100644
>>> --- a/drivers/phy/tegra/xusb.h
>>> +++ b/drivers/phy/tegra/xusb.h
>>> @@ -370,6 +370,9 @@ struct tegra_xusb_padctl_soc {
>>>   	} ports;
>>>   	const struct tegra_xusb_padctl_ops *ops;
>>> +
>>> +	const char * const *supply_names;
>>> +	unsigned int num_supplies;
>>>   };
>>>   struct tegra_xusb_padctl {
>>> @@ -393,6 +396,8 @@ struct tegra_xusb_padctl {
>>>   	unsigned int enable;
>>>   	struct clk *clk;
>>> +
>>> +	struct regulator_bulk_data *supplies;
>>>   };
>>>   static inline void padctl_writel(struct tegra_xusb_padctl *padctl, u32 value,

Thierry Reding Feb. 7, 2019, 11:11 a.m. UTC | #5

On Tue, Feb 05, 2019 at 06:25:04PM +0530, Kishon Vijay Abraham I wrote:
> Hi,
> 
> On 28/01/19 1:30 PM, Thierry Reding wrote:
> > On Mon, Jan 28, 2019 at 03:22:09PM +0800, jckuo wrote:
> >> Hi Thierry,
> >>
> >> I think any non-zero return value of
> >> regulator_bulk_enable()/devm_regulator_bulk_get() means error.
> >>
> >> Thanks,
> >>
> >> JC
> > 
> > Theoretically I think only regulator_bulk_enable() could return a
> > positive value, but even so it never will in practice because all of the
> > regulator_enable() (see _regulator_enable()) calls will only ever return
> > negative error codes.
> > 
> > I can change this and resend if you have strong concerns about this
> > possibly missing legitimate error cases.
> 
> All error codes are negative, so this is fine IMO.

Hi Kishon,

did you have any other comments on this series? Anything keeping this
from going into v5.1?

Thierry

[4/5] phy: tegra: xusb: Add support for power supplies

Commit Message

Comments

Patch