[PATCH v2] net: phy: Handle both led@0 and led subnode name for single-LED PHYs

[PATCH v2] net: phy: Handle both led@0 and led subnode name for single-LED PHYs

[Marek Vasut]

In case a PHY supports only one LED in total, like ADIN1300, and this LED
is described in DT, it is currently necessary to include unit address in
the LED node name and the address-cells have to be set to 1:

leds {
  #address-cells = <1>;
  ...
  led@0 {
    reg = <0>;
    ...
  };
};

For a single LED PHY, this should not be necessary and plain 'led' node
without unit should be acceptable as well:

leds {
  ...
  led {
    ...
  };
};

Handle this special case. In case reg property is not present in the leds
node subnode, test whether the leds node contains exactly one subnode, and
if so, assume this is the one single LED with reg property set to 0.

Signed-off-by: Marek Vasut <marex@denx.de>
---
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Andrew Lunn <andrew@lunn.ch>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Heiner Kallweit <hkallweit1@gmail.com>
Cc: Jakub Kicinski <kuba@kernel.org>
Cc: Paolo Abeni <pabeni@redhat.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: netdev@vger.kernel.org
---
V2: Fix up variable rename
---
 drivers/net/phy/phy_device.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)

diff --git a/drivers/net/phy/phy_device.c b/drivers/net/phy/phy_device.c
index 5b34d39d1d52a..e36e65e672e79 100644
--- a/drivers/net/phy/phy_device.c
+++ b/drivers/net/phy/phy_device.c
@@ -3346,6 +3346,7 @@ static int of_phy_led(struct phy_device *phydev,
 	struct led_classdev *cdev;
 	unsigned long modes = 0;
 	struct phy_led *phyled;
+	int led_count;
 	u32 index;
 	int err;
 
@@ -3357,8 +3358,18 @@ static int of_phy_led(struct phy_device *phydev,
 	phyled->phydev = phydev;
 
 	err = of_property_read_u32(led, "reg", &index);
-	if (err)
-		return err;
+	if (err) {
+		led_count = of_get_available_child_count(of_get_parent(led));
+
+		/*
+		 * If there is one PHY LED in total, accept 'led' subnode
+		 * in addition to 'led@0' subnode, and assume reg = <0>;
+		 */
+		if (led_count != 1)
+			return err;
+		index = 0;
+		err = 0;
+	}
 	if (index > U8_MAX)
 		return -EINVAL;
 
-- 
2.45.2

Re: [PATCH v2] net: phy: Handle both led@0 and led subnode name for single-LED PHYs

[Andrew Lunn]

On Mon, Jan 20, 2025 at 10:02:46AM +0100, Marek Vasut wrote:
> In case a PHY supports only one LED in total, like ADIN1300, and this LED
> is described in DT, it is currently necessary to include unit address in
> the LED node name and the address-cells have to be set to 1:
> 
> leds {
>   #address-cells = <1>;
>   ...
>   led@0 {
>     reg = <0>;
>     ...
>   };
> };
> 
> For a single LED PHY, this should not be necessary and plain 'led' node
> without unit should be acceptable as well:
> 
> leds {
>   ...
>   led {
>     ...
>   };
> };

So how do other subsystems handle this? SPI with only a single chip
select line? Standalone LED controllers with a single LED? A PWM with
a single output?

>  drivers/net/phy/phy_device.c | 15 +++++++++++++--

What about the device tree binding? Does it already have the reg
property as optional?

	Andrew

Re: [PATCH v2] net: phy: Handle both led@0 and led subnode name for single-LED PHYs

[Marek Vasut]

On 1/20/25 5:35 PM, Andrew Lunn wrote:
> On Mon, Jan 20, 2025 at 10:02:46AM +0100, Marek Vasut wrote:
>> In case a PHY supports only one LED in total, like ADIN1300, and this LED
>> is described in DT, it is currently necessary to include unit address in
>> the LED node name and the address-cells have to be set to 1:
>>
>> leds {
>>    #address-cells = <1>;
>>    ...
>>    led@0 {
>>      reg = <0>;
>>      ...
>>    };
>> };
>>
>> For a single LED PHY, this should not be necessary and plain 'led' node
>> without unit should be acceptable as well:
>>
>> leds {
>>    ...
>>    led {
>>      ...
>>    };
>> };
> 
> So how do other subsystems handle this? SPI with only a single chip
> select line?

I believe for SPI, it is always mandatory to have 'reg' property.

> Standalone LED controllers with a single LED?

The GPIO LEDs driver does not enumerate the LEDs at all, so there is no 
'reg' property needed in any case, but I don't think this is really 
applicable for the PHY LEDs, which depend on the ordering within each PHY.

> A PWM with a single output?

Reference the controller node itself, similar to how single output clock 
controller works.

>>   drivers/net/phy/phy_device.c | 15 +++++++++++++--
> 
> What about the device tree binding? Does it already have the reg
> property as optional?
Documentation/devicetree/bindings/net/ethernet-phy.yaml does not, but 
that can be changed if desirable . If not desirable, then I can just 
discard this patch ?