Add the  "st,phy-wol" to indicate the MAC to use the wakeup capability
of the PHY instead of the MAC.

Signed-off-by: Gatien Chevallier <gatien.chevallier@foss.st.com>
---
 Documentation/devicetree/bindings/net/stm32-dwmac.yaml | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/Documentation/devicetree/bindings/net/stm32-dwmac.yaml b/Documentation/devicetree/bindings/net/stm32-dwmac.yaml
index 987254900d0da7aab81237f20b1540ad8a17bd21..985bd4c320b3e07fd1cd0aa398d6cce0b55a4e4d 100644
--- a/Documentation/devicetree/bindings/net/stm32-dwmac.yaml
+++ b/Documentation/devicetree/bindings/net/stm32-dwmac.yaml
@@ -121,6 +121,12 @@ properties:
     minItems: 1
     maxItems: 2
 
+  st,phy-wol:
+    description:
+      set this property to use the wakeup capability from the PHY, if supported, instead of the
+      MAC.
+    type: boolean
+
 required:
   - compatible
   - clocks

-- 
2.35.3

On Wed, Sep 17, 2025 at 05:36:36PM +0200, Gatien Chevallier wrote:
> Add the  "st,phy-wol" to indicate the MAC to use the wakeup capability
> of the PHY instead of the MAC.

Why is this ST specific? PHYs being wakeup capable or not is independent 
of ST. If you want to or can use wakeup from the PHY, shouldn't that be 
a property in the PHY?

Seems to me you would want to define what all components are wakeup 
capable and then let the kernel decide which component to use. I'd think 
the kernel would prefer the PHY as that's closest to the wire and 
probably lowest power.

That's my 2 cents spending all of 5 minutes thinking about it. I'll 
defer to Russell and Andrew...

Rob

On 9/22/25 19:05, Rob Herring wrote:
> On Wed, Sep 17, 2025 at 05:36:36PM +0200, Gatien Chevallier wrote:
>> Add the  "st,phy-wol" to indicate the MAC to use the wakeup capability
>> of the PHY instead of the MAC.
> 
> Why is this ST specific? PHYs being wakeup capable or not is independent
> of ST. If you want to or can use wakeup from the PHY, shouldn't that be
> a property in the PHY?
> 
> Seems to me you would want to define what all components are wakeup
> capable and then let the kernel decide which component to use. I'd think
> the kernel would prefer the PHY as that's closest to the wire and
> probably lowest power.
> 
> That's my 2 cents spending all of 5 minutes thinking about it. I'll
> defer to Russell and Andrew...
> 
> Rob

Hi Rob,

I think we're all aligned on this now. The thing about this property
being added is really related to how the STMMAC driver was interacting
with the PHY drivers and how the PHY drivers communicate their WoL
capability. It has been done previously for the "mediatek,mac-wol"
property.

The complicated thing to handle is: Is the PHY capable of doing WoL
AND is the wiring actually present on the platform on the correct
pin.

I'll send a V3 taking into account the different discussions we had
with Andrew and Russell.

Best regards,
Gatien