Re: [PATCH net-next 1/4] dt-bindings: net: document st,phy-wol property
From: Gatien CHEVALLIER
Date: Mon Jul 21 2025 - 12:01:04 EST
Hello Andrew,
On 7/21/25 15:18, Andrew Lunn wrote:
On Mon, Jul 21, 2025 at 02:10:48PM +0200, Gatien CHEVALLIER wrote:
Hello Krzysztof,
On 7/21/25 13:30, Krzysztof Kozlowski wrote:
On 21/07/2025 13:14, Gatien Chevallier wrote:
The "st,phy-wol" property can be set to use the wakeup capability of
the PHY instead of the MAC.
And why would that be property of a SoC or board? Word "can" suggests
you are documenting something which exists, but this does not exist.
Can you elaborate a bit more on the "not existing" part please?
For the WoL from PHY to be supported, the PHY line that is raised
(On nPME pin for this case) when receiving a wake up event has to be
wired to a wakeup event input of the Extended interrupt and event
controller(EXTI), and that's implementation dependent.
How does this differ from normal interrupts from the PHY? Isn't the
presence of an interrupt in DT sufficient to indicate the PHY can wake
the system?
Andrew
Here's an extract from the Microchip datasheet for the LAN8742A PHY:
"In addition to the main interrupts described in this section, an nPME
pin is provided exclusively for WoL specific interrupts."
I'm not an expert of the different PHYs, but I guess there may be a
distinction needed between some "main" PHY interrupts and the WoL one
at least for deep low-power use cases.
Because this line is wired to a peripheral managed by our
TEE, the ultimate goal here would be to declare the OP-TEE node as
an interrupt provider and to forward the interrupt to the kernel using
the asynchronous notification mechanism. Then, reference the OP-TEE
node in the "interrupts-extended" property in the PHY node so that it
can be acked by the PHY driver. As of now, this patch set at least allow
to wakeup from a deep low power mode using the WoL capability of the
PHY.
Regarding this property, somewhat similar to "mediatek,mac-wol",
I need to position a flag at the mac driver level. I thought I'd go
using the same approach.
Best regards,
Gatien