Hi,

On Tue, Sep 01, 2026 at 07:04:28PM +0200, Maurizio Casciano wrote:
> Thank you for the review.
> 
> I agree that firmware-node references should retain priority over the
> legacy name-based mechanism. The "supplied-from" property should only be
> used as a fallback when no "power-supplies" references are present,
> including the case where the consumer has no firmware node.
> 
> For v2 I will preserve the existing order as follows:
> 
> 1. honor an explicitly pre-populated supplied_from list;
> 2. resolve "power-supplies" firmware-node references when present;
> 3. fall back to the name-based "supplied-from" property only when no
>    references were specified.
> 
> Errors while resolving specified references, including -EPROBE_DEFER,
> will still be propagated rather than triggering the fallback.
> 
> I will wait for Sebastian's opinion before posting v2.

Ack.

Also FWIW the power-supply core has been migrated to be fully fwnode
based, so the CONFIG_OF is not really correct anymore anyways. Let's
have a single implementation for power_supply_check_supplies().

Greetings,

-- Sebastian