On 27/09/2024 11:35, Markus Schneider-Pargmann wrote:

quoted

quoted

It's not obvious for me. Maybe you can elaborate a bit.

The hardware has a different set of wakeup sources depending on the
power mode it is in and I would like to describe these different sets of
wakeup sources in the devicetree as for me this is a hardware property,
not a driver thing.

I stated the argument to which you did not respond: it will not matter
for the device whether this is wakeup-source = S2R or wakeup-source =
TI-Partial-IO or whatever.

Each device is or is not wakeup source.

And just because your device has some registers or some configuration
does not mean this property is suitable for DT.

I came up with a different (better) way to model this in the devicetree.
This group of units that are powered in Partial-IO are all powered by
just one regulator that is always on. I can simply describe this in the
devicetree by defining the regulator and consumer relationship:

Defining the regulator as described in the board schematic:

  vddshv_canuart: regulator-7 {
         /* TPS22965DSGT */
         compatible = "regulator-fixed";
         regulator-name = "vddshv_canuart";
         regulator-min-microvolt = <3300000>;
         regulator-max-microvolt = <3300000>;
         vin-supply = <&vcc_3v3_main>;
         regulator-always-on;
         regulator-boot-on;
  };

Adding vio-supply to mcan and uarts, note, this binding does not exist
yet:

  &mcu_mcan0 {
         vio-supply = <&vddshv_canuart>;
  };

  &mcu_mcan1 {
         vio-supply = <&vddshv_canuart>;
  };

  &mcu_uart0 {
         vio-supply = <&vddshv_canuart>;
  };

  &wkup_uart0 {
         vio-supply = <&vddshv_canuart>;
  };

I am happy that problem is solved, but it really, really puzzles me how
above fits wakeup-mode-problem at all. This is so different that I doubt
you came up with proper hardware description.

Best regards,
Krzysztof