Handle a new DT property power-supply-needs-irq that boards can
use if the power-domains' power-supply node (regulator) needs
interrupts to function. An example is a pmic regulator on i2c.
Since suspend/resume doesn't work in these cases currently, switch
to the normal, outer suspend/resume callbacks.
Usually such a regulator already times out during suspend_noirq:
[ 41.024193] buck4: failed to disable: -ETIMEDOUT
Initially systemd suspend problems had been discussed at
https://lore.kernel.org/linux-arm-kernel/[email protected]/
which led to discussing the pmic that contains the regulators which
serve as power-domain power-supplies:
https://lore.kernel.org/linux-pm/[email protected]/T/
So this enables boards to use power-supply-needs-irq and fix systemd
suspend and resume.
Possibly one can find more changes this property should configure. They
can be added later when testing them.
Signed-off-by: Martin Kepplinger <[email protected]>
---
drivers/base/power/domain.c | 14 ++++++++++++++
1 file changed, 14 insertions(+)
diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c
index 3e86772d5fac..c8fecba2c191 100644
--- a/drivers/base/power/domain.c
+++ b/drivers/base/power/domain.c
@@ -2298,6 +2298,16 @@ static bool genpd_present(const struct generic_pm_domain *genpd)
return ret;
}
+static void of_genpd_get_power_supply_irq(struct generic_pm_domain *pd)
+{
+ if (of_property_read_bool(pd->dev.of_node, "power-supply-needs-irq")) {
+ pd->domain.ops.suspend = genpd_suspend_noirq;
+ pd->domain.ops.resume = genpd_resume_noirq;
+ pd->domain.ops.suspend_noirq = NULL;
+ pd->domain.ops.resume_noirq = NULL;
+ }
+}
+
/**
* of_genpd_add_provider_simple() - Register a simple PM domain provider
* @np: Device node pointer associated with the PM domain provider.
@@ -2343,6 +2353,8 @@ int of_genpd_add_provider_simple(struct device_node *np,
genpd->provider = &np->fwnode;
genpd->has_provider = true;
+ of_genpd_get_power_supply_irq(genpd);
+
return 0;
}
EXPORT_SYMBOL_GPL(of_genpd_add_provider_simple);
@@ -2394,6 +2406,8 @@ int of_genpd_add_provider_onecell(struct device_node *np,
genpd->provider = &np->fwnode;
genpd->has_provider = true;
+
+ of_genpd_get_power_supply_irq(genpd);
}
ret = genpd_add_provider(np, data->xlate, data);
--
2.30.2
Add the power-supply-needs-irq board description property for power domains.
Signed-off-by: Martin Kepplinger <[email protected]>
---
.../devicetree/bindings/power/power-domain.yaml | 10 ++++++++++
1 file changed, 10 insertions(+)
diff --git a/Documentation/devicetree/bindings/power/power-domain.yaml b/Documentation/devicetree/bindings/power/power-domain.yaml
index 889091b9814f..e82c2f7ccb97 100644
--- a/Documentation/devicetree/bindings/power/power-domain.yaml
+++ b/Documentation/devicetree/bindings/power/power-domain.yaml
@@ -70,6 +70,16 @@ properties:
by the given provider should be subdomains of the domain specified
by this binding.
+ power-supply: true
+
+ power-supply-needs-irq:
+ type: boolean
+ description:
+ A power-supply can link for example to a regulator controlled via
+ i2c or otherwise needing interrupts enabled to be able to enable and
+ disable. This property makes various callbacks usually run in the
+ noirq phase, being run when interrupts are available.
+
required:
- "#power-domain-cells"
--
2.30.2
On 11/07/2022 11:45, Martin Kepplinger wrote:
> Add the power-supply-needs-irq board description property for power domains.
Where is a board description here? I think you just meant
"power-supply-needs-irq property"?
>
> Signed-off-by: Martin Kepplinger <[email protected]>
> ---
> .../devicetree/bindings/power/power-domain.yaml | 10 ++++++++++
> 1 file changed, 10 insertions(+)
>
> diff --git a/Documentation/devicetree/bindings/power/power-domain.yaml b/Documentation/devicetree/bindings/power/power-domain.yaml
> index 889091b9814f..e82c2f7ccb97 100644
> --- a/Documentation/devicetree/bindings/power/power-domain.yaml
> +++ b/Documentation/devicetree/bindings/power/power-domain.yaml
> @@ -70,6 +70,16 @@ properties:
> by the given provider should be subdomains of the domain specified
> by this binding.
>
> + power-supply: true
This is a new property not described in the commit msg.
> +
> + power-supply-needs-irq:
> + type: boolean
> + description:
> + A power-supply can link for example to a regulator controlled via
> + i2c or otherwise needing interrupts enabled to be able to enable and
> + disable.
Not really a property of power domain. How the regulator supply works is
entirely up to regulator. Otherwise such property should appear for
every device.
> This property makes various callbacks usually run in the
> + noirq phase, being run when interrupts are available.
Last sentence does not fit - you embed Linux implementation into DT
bindings. noirq phase is Linux specific.
> +
> required:
> - "#power-domain-cells"
>
Best regards,
Krzysztof
Am Montag, dem 11.07.2022 um 12:38 +0200 schrieb Krzysztof Kozlowski:
> On 11/07/2022 11:45, Martin Kepplinger wrote:
> > Add the power-supply-needs-irq board description property for power
> > domains.
>
> Where is a board description here? I think you just meant
> "power-supply-needs-irq property"?
> >
> > Signed-off-by: Martin Kepplinger <[email protected]>
> > ---
> > .../devicetree/bindings/power/power-domain.yaml | 10
> > ++++++++++
> > 1 file changed, 10 insertions(+)
> >
> > diff --git a/Documentation/devicetree/bindings/power/power-
> > domain.yaml b/Documentation/devicetree/bindings/power/power-
> > domain.yaml
> > index 889091b9814f..e82c2f7ccb97 100644
> > --- a/Documentation/devicetree/bindings/power/power-domain.yaml
> > +++ b/Documentation/devicetree/bindings/power/power-domain.yaml
> > @@ -70,6 +70,16 @@ properties:
> > by the given provider should be subdomains of the domain
> > specified
> > by this binding.
> >
> > + power-supply: true
>
> This is a new property not described in the commit msg.
true, I think it's missing and could be added as a separate patch.
>
> > +
> > + power-supply-needs-irq:
> > + type: boolean
> > + description:
> > + A power-supply can link for example to a regulator
> > controlled via
> > + i2c or otherwise needing interrupts enabled to be able to
> > enable and
> > + disable.
>
> Not really a property of power domain. How the regulator supply works
> is
> entirely up to regulator. Otherwise such property should appear for
> every device.
you're right. The power-domain driver could read the power-supply
regulator node directly. Still, I think then a new regulator property
is needed instead, or is it?
>
> > This property makes various callbacks usually run in the
> > + noirq phase, being run when interrupts are available.
>
> Last sentence does not fit - you embed Linux implementation into DT
> bindings. noirq phase is Linux specific.
oh I keep making this mistake. thanks for the fast review!
>
> > +
> > required:
> > - "#power-domain-cells"
> >
>
>
> Best regards,
> Krzysztof
On 11/07/2022 15:17, Martin Kepplinger wrote:
> Am Montag, dem 11.07.2022 um 12:38 +0200 schrieb Krzysztof Kozlowski:
>> On 11/07/2022 11:45, Martin Kepplinger wrote:
>>> Add the power-supply-needs-irq board description property for power
>>> domains.
>>
>> Where is a board description here? I think you just meant
>> "power-supply-needs-irq property"?
>>>
>>> Signed-off-by: Martin Kepplinger <[email protected]>
>>> ---
>>> .../devicetree/bindings/power/power-domain.yaml | 10
>>> ++++++++++
>>> 1 file changed, 10 insertions(+)
>>>
>>> diff --git a/Documentation/devicetree/bindings/power/power-
>>> domain.yaml b/Documentation/devicetree/bindings/power/power-
>>> domain.yaml
>>> index 889091b9814f..e82c2f7ccb97 100644
>>> --- a/Documentation/devicetree/bindings/power/power-domain.yaml
>>> +++ b/Documentation/devicetree/bindings/power/power-domain.yaml
>>> @@ -70,6 +70,16 @@ properties:
>>> by the given provider should be subdomains of the domain
>>> specified
>>> by this binding.
>>>
>>> + power-supply: true
>>
>> This is a new property not described in the commit msg.
>
> true, I think it's missing and could be added as a separate patch.
>
>>
>>> +
>>> + power-supply-needs-irq:
>>> + type: boolean
>>> + description:
>>> + A power-supply can link for example to a regulator
>>> controlled via
>>> + i2c or otherwise needing interrupts enabled to be able to
>>> enable and
>>> + disable.
>>
>> Not really a property of power domain. How the regulator supply works
>> is
>> entirely up to regulator. Otherwise such property should appear for
>> every device.
>
> you're right. The power-domain driver could read the power-supply
> regulator node directly. Still, I think then a new regulator property
> is needed instead, or is it?
In case of regulator, I am not so sure it needs a dedicated property of
DT. If it is I2C regulator - the parent node is I2C bus and regulator
device is some child of I2C controller (could be via a MFD device), so
no need for dedicated property.
If it uses interrupts, then:
1. The presence of interrupts is already known - "interrupts" property.
2. The actual use of interrupts is DT independent and only driver knows it.
Best regards,
Krzysztof
Am Dienstag, dem 12.07.2022 um 09:01 +0200 schrieb Krzysztof Kozlowski:
> On 11/07/2022 15:17, Martin Kepplinger wrote:
> > Am Montag, dem 11.07.2022 um 12:38 +0200 schrieb Krzysztof
> > Kozlowski:
> > > On 11/07/2022 11:45, Martin Kepplinger wrote:
> > > > Add the power-supply-needs-irq board description property for
> > > > power
> > > > domains.
> > >
> > > Where is a board description here? I think you just meant
> > > "power-supply-needs-irq property"?
> > > >
> > > > Signed-off-by: Martin Kepplinger <[email protected]>
> > > > ---
> > > > .../devicetree/bindings/power/power-domain.yaml | 10
> > > > ++++++++++
> > > > 1 file changed, 10 insertions(+)
> > > >
> > > > diff --git a/Documentation/devicetree/bindings/power/power-
> > > > domain.yaml b/Documentation/devicetree/bindings/power/power-
> > > > domain.yaml
> > > > index 889091b9814f..e82c2f7ccb97 100644
> > > > --- a/Documentation/devicetree/bindings/power/power-domain.yaml
> > > > +++ b/Documentation/devicetree/bindings/power/power-domain.yaml
> > > > @@ -70,6 +70,16 @@ properties:
> > > > by the given provider should be subdomains of the domain
> > > > specified
> > > > by this binding.
> > > >
> > > > + power-supply: true
> > >
> > > This is a new property not described in the commit msg.
> >
> > true, I think it's missing and could be added as a separate patch.
> >
> > >
> > > > +
> > > > + power-supply-needs-irq:
> > > > + type: boolean
> > > > + description:
> > > > + A power-supply can link for example to a regulator
> > > > controlled via
> > > > + i2c or otherwise needing interrupts enabled to be able
> > > > to
> > > > enable and
> > > > + disable.
> > >
> > > Not really a property of power domain. How the regulator supply
> > > works
> > > is
> > > entirely up to regulator. Otherwise such property should appear
> > > for
> > > every device.
> >
> > you're right. The power-domain driver could read the power-supply
> > regulator node directly. Still, I think then a new regulator
> > property
> > is needed instead, or is it?
>
> In case of regulator, I am not so sure it needs a dedicated property
> of
> DT. If it is I2C regulator - the parent node is I2C bus and regulator
> device is some child of I2C controller (could be via a MFD device),
> so
> no need for dedicated property.
>
> If it uses interrupts, then:
> 1. The presence of interrupts is already known - "interrupts"
> property.
> 2. The actual use of interrupts is DT independent and only driver
> knows it.
thanks for this great suggestion! for the imx8mq devices this (1.)
indeed is the case for exactly the 3 regulators I manually describe
here. v2 of this patch looks very elegant (and I guess I could have
removed the DT people from the email, I forgot). here it is:
https://lore.kernel.org/linux-arm-kernel/[email protected]/T/#u
>
> Best regards,
> Krzysztof