Document the 'open-drain' property that allows configuring the PWM
controller outputs in open drain versus totem pole.

Signed-off-by: Florian Fainelli <florian.fainelli@broadcom.com>
---
 Documentation/devicetree/bindings/pwm/brcm,bcm7038-pwm.yaml | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/Documentation/devicetree/bindings/pwm/brcm,bcm7038-pwm.yaml b/Documentation/devicetree/bindings/pwm/brcm,bcm7038-pwm.yaml
index 119de3d7f9dd..12851c43a612 100644
--- a/Documentation/devicetree/bindings/pwm/brcm,bcm7038-pwm.yaml
+++ b/Documentation/devicetree/bindings/pwm/brcm,bcm7038-pwm.yaml
@@ -25,6 +25,12 @@ properties:
   clocks:
     maxItems: 1
 
+  open-drain:
+    type: boolean
+    description:
+      Configure the outputs with open-drain structure, if omitted totem pole
+      structure is used.
+
 required:
   - compatible
   - reg
-- 
2.34.1

On Fri, Oct 11, 2024 at 07:56:02PM -0700, Florian Fainelli wrote:
> Document the 'open-drain' property that allows configuring the PWM
> controller outputs in open drain versus totem pole.
> 
> Signed-off-by: Florian Fainelli <florian.fainelli@broadcom.com>
> ---
>  Documentation/devicetree/bindings/pwm/brcm,bcm7038-pwm.yaml | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/Documentation/devicetree/bindings/pwm/brcm,bcm7038-pwm.yaml b/Documentation/devicetree/bindings/pwm/brcm,bcm7038-pwm.yaml
> index 119de3d7f9dd..12851c43a612 100644
> --- a/Documentation/devicetree/bindings/pwm/brcm,bcm7038-pwm.yaml
> +++ b/Documentation/devicetree/bindings/pwm/brcm,bcm7038-pwm.yaml
> @@ -25,6 +25,12 @@ properties:
>    clocks:
>      maxItems: 1
>  
> +  open-drain:
> +    type: boolean
> +    description:
> +      Configure the outputs with open-drain structure, if omitted totem pole
> +      structure is used.
> +

There's one other PWM with 'open-drain' so move the definition to 
pwm.yaml.

Alternatively, 'drive-open-drain' is a much more commonly used 
variation.

Another thing to consider is for any PWM controller with more than 
1 output, you might want this to be per output and therefore should be 
a flag in the cells.

Rob

On 10/15/24 09:32, Rob Herring wrote:
> On Fri, Oct 11, 2024 at 07:56:02PM -0700, Florian Fainelli wrote:
>> Document the 'open-drain' property that allows configuring the PWM
>> controller outputs in open drain versus totem pole.
>>
>> Signed-off-by: Florian Fainelli <florian.fainelli@broadcom.com>
>> ---
>>   Documentation/devicetree/bindings/pwm/brcm,bcm7038-pwm.yaml | 6 ++++++
>>   1 file changed, 6 insertions(+)
>>
>> diff --git a/Documentation/devicetree/bindings/pwm/brcm,bcm7038-pwm.yaml b/Documentation/devicetree/bindings/pwm/brcm,bcm7038-pwm.yaml
>> index 119de3d7f9dd..12851c43a612 100644
>> --- a/Documentation/devicetree/bindings/pwm/brcm,bcm7038-pwm.yaml
>> +++ b/Documentation/devicetree/bindings/pwm/brcm,bcm7038-pwm.yaml
>> @@ -25,6 +25,12 @@ properties:
>>     clocks:
>>       maxItems: 1
>>   
>> +  open-drain:
>> +    type: boolean
>> +    description:
>> +      Configure the outputs with open-drain structure, if omitted totem pole
>> +      structure is used.
>> +
> 
> There's one other PWM with 'open-drain' so move the definition to
> pwm.yaml.

Ah indeed, there is one now, great.

> 
> Alternatively, 'drive-open-drain' is a much more commonly used
> variation.

That name works just as well.

> 
> Another thing to consider is for any PWM controller with more than
> 1 output, you might want this to be per output and therefore should be
> a flag in the cells.

Yes, that is a good point, this controller has two channels, so it seems 
like increasing the #pwm-cells might be the way to go.

Thanks!
-- 
Florian

Hello,

On Tue, Oct 15, 2024 at 10:07:10AM -0700, Florian Fainelli wrote:
> On 10/15/24 09:32, Rob Herring wrote:
> > Another thing to consider is for any PWM controller with more than
> > 1 output, you might want this to be per output and therefore should be
> > a flag in the cells.
> 
> Yes, that is a good point, this controller has two channels, so it seems
> like increasing the #pwm-cells might be the way to go.

So the idea is something like:

diff --git a/include/dt-bindings/pwm/pwm.h b/include/dt-bindings/pwm/pwm.h
index ab9a077e3c7d..d18b006a7399 100644
--- a/include/dt-bindings/pwm/pwm.h
+++ b/include/dt-bindings/pwm/pwm.h
@@ -11,5 +11,6 @@
 #define _DT_BINDINGS_PWM_PWM_H
 
 #define PWM_POLARITY_INVERTED			(1 << 0)
+#define PWM_OUTPUT_OPEN_DRAIN			(1 << 1)
 
 #endif

and then add support for that to the core and drivers? There is some
intersection with pinctrl (depending on hardware). I wonder if
abstracting this somehow using the typical pinctrl properties would be a
saner option??

Best regards
Uwe

On 10/29/24 03:44, Uwe Kleine-König wrote:
> Hello,
> 
> On Tue, Oct 15, 2024 at 10:07:10AM -0700, Florian Fainelli wrote:
>> On 10/15/24 09:32, Rob Herring wrote:
>>> Another thing to consider is for any PWM controller with more than
>>> 1 output, you might want this to be per output and therefore should be
>>> a flag in the cells.
>>
>> Yes, that is a good point, this controller has two channels, so it seems
>> like increasing the #pwm-cells might be the way to go.
> 
> So the idea is something like:
> 
> diff --git a/include/dt-bindings/pwm/pwm.h b/include/dt-bindings/pwm/pwm.h
> index ab9a077e3c7d..d18b006a7399 100644
> --- a/include/dt-bindings/pwm/pwm.h
> +++ b/include/dt-bindings/pwm/pwm.h
> @@ -11,5 +11,6 @@
>   #define _DT_BINDINGS_PWM_PWM_H
>   
>   #define PWM_POLARITY_INVERTED			(1 << 0)
> +#define PWM_OUTPUT_OPEN_DRAIN			(1 << 1)
>   
>   #endif
> 
> and then add support for that to the core and drivers? There is some
> intersection with pinctrl (depending on hardware). I wonder if
> abstracting this somehow using the typical pinctrl properties would be a
> saner option??

But what if the pin is not managed by a pinctrl provider? I have started 
going the route of implementing the PWM_OUTPUT_OPEN_DRAIN bit as an 
additional specifier in the #pwm-cells, but I am not sure to what extent 
this should be allowed to be changed at runtime.
-- 
Florian

Hello Florian,

[adding Linus and linux-gpio to Cc:]

On Tue, Oct 29, 2024 at 09:03:57AM -0700, Florian Fainelli wrote:
> On 10/29/24 03:44, Uwe Kleine-König wrote:
> > On Tue, Oct 15, 2024 at 10:07:10AM -0700, Florian Fainelli wrote:
> > > On 10/15/24 09:32, Rob Herring wrote:
> > > > Another thing to consider is for any PWM controller with more than
> > > > 1 output, you might want this to be per output and therefore should be
> > > > a flag in the cells.
> > > 
> > > Yes, that is a good point, this controller has two channels, so it seems
> > > like increasing the #pwm-cells might be the way to go.
> > 
> > So the idea is something like:
> > 
> > diff --git a/include/dt-bindings/pwm/pwm.h b/include/dt-bindings/pwm/pwm.h
> > index ab9a077e3c7d..d18b006a7399 100644
> > --- a/include/dt-bindings/pwm/pwm.h
> > +++ b/include/dt-bindings/pwm/pwm.h
> > @@ -11,5 +11,6 @@
> >   #define _DT_BINDINGS_PWM_PWM_H
> >   #define PWM_POLARITY_INVERTED			(1 << 0)
> > +#define PWM_OUTPUT_OPEN_DRAIN			(1 << 1)
> >   #endif
> > 
> > and then add support for that to the core and drivers? There is some
> > intersection with pinctrl (depending on hardware). I wonder if
> > abstracting this somehow using the typical pinctrl properties would be a
> > saner option??
> 
> But what if the pin is not managed by a pinctrl provider?

Then create one? If that's the PWM itself that is the pinctrl device it
would look as follows:

	pwm@f0408000 {
		compatible = "brcm,bcm7038-pwm";
		pinctrl-0 = <&pwm_pins>;
		reg = <0xf0408000 0x28>;
		#pwm-cells = <2>;
		#pinctrl-cells = <0>;
		clocks = <&upg_fixed>;
		
		pinctrl {
			pwm_pins: pwm-pins {
				pins = "A", "B";
				drive-open-drain;
			};
		};
	};

Maybe this is difficult if there is a pinctrl that configures the output
as "PWM" and then there is that additional register in the PWM IP to
make this pin open drain? One could just use

	pinctrl-0 = <&pwm_pins>, <&system_pinctrl_pwm>;

then. Not entirely sure this is overengineered, but the dt
representation would be nice (IMHO). Thoughts?

> I have started
> going the route of implementing the PWM_OUTPUT_OPEN_DRAIN bit as an
> additional specifier in the #pwm-cells, but I am not sure to what extent
> this should be allowed to be changed at runtime.

I would not expect that the open-drainness needs to change at runtime.

Best regards
Uwe