Le Mon, May 09, 2022 at 02:17:27PM +0200, Andrew Lunn a écrit :

On Mon, May 09, 2022 at 07:48:54AM +0000, Corentin Labbe wrote:

quoted

Add entries for the 2 new phy-supply and phy-io-supply.

Signed-off-by: Corentin Labbe <clabbe@baylibre.com>
---
 .../devicetree/bindings/net/ethernet-phy.yaml          | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/Documentation/devicetree/bindings/net/ethernet-phy.yaml b/Documentation/devicetree/bindings/net/ethernet-phy.yaml
index ed1415a4381f..2a6b45ddf010 100644
--- a/Documentation/devicetree/bindings/net/ethernet-phy.yaml
+++ b/Documentation/devicetree/bindings/net/ethernet-phy.yaml

@@ -153,6 +153,16 @@ properties:
       used. The absence of this property indicates the muxers
       should be configured so that the external PHY is used.
 
+  phy-supply:
+    description:
+      Phandle to a regulator that provides power to the PHY. This
+      regulator will be managed during the PHY power on/off sequence.
+
+  phy-io-supply:
+    description:
+      Phandle to a regulator that provides power to the PHY. This
+      regulator will be managed during the PHY power on/off sequence.

If you need two differently named regulators, you need to make it clear
how they differ. My _guess_ would be, you only need the io variant in
order to talk to the PHY registers. However, to talk to a link
partner, you need the other one enabled as well. Which means handling
that regulator probably should be in the PHY driver, so it is enabled
only when the interface is configured up.

If I enable only the IO one, stmmac fail to reset, so both are needed to be up.
I tried also to keep the "phy" one handled by stmmac (by removing patch 2), this lead to the PHY to not be found by MDIO scan.
Proably because stmmac enable the "phy" before the "phy-io".

For the difference between the 2, according to my basic read (I am bad a it) of the shematic
https://linux-sunxi.org/images/5/50/OrangePi_3_Schematics_v1.5.pdf
phy-io(ephy-vdd25) seems to (at least) power MDIO bus.