On 09/01/2026 15:19, Ivan Vecera wrote:

On 1/9/26 10:55 AM, Krzysztof Kozlowski wrote:

quoted

On Thu, Jan 08, 2026 at 07:23:09PM +0100, Ivan Vecera wrote:

quoted

Add helper functions to the DPLL core to retrieve a DPLL pin's firmware
node handle based on the "dpll-pins" and "dpll-pin-names" properties.

* `fwnode_dpll_pin_node_get()`: matches the given name against the
   "dpll-pin-names" property to find the correct index, then retrieves
   the reference from "dpll-pins".
* `device_dpll_pin_node_get()`: a wrapper around the fwnode helper for
   convenience when using a `struct device`.

These helpers simplify the process for consumer drivers (such as Ethernet
controllers or PHYs) to look up their associated DPLL pins defined in
the DT or ACPI, which can then be passed to the DPLL subsystem to acquire
the pin object.

Signed-off-by: Ivan Vecera <ivecera@redhat.com>
---
  drivers/dpll/dpll_core.c | 20 ++++++++++++++++++++
  include/linux/dpll.h     | 15 +++++++++++++++
  2 files changed, 35 insertions(+)

I don't see cells defined in your binding. Neither updated property.c.

And if the cells are not required? I mean that dpll-names only specifies
array of phandles without parameters...

e.g.
dpll-pin-names = "abc", "def";
dpll-pins = <&dpll_pin_abc>, <&dpll_pin_def>;

Should '#dpll-pin-cells' be defined as constantly equal to 0?

I don't understand how can you guarantee for every possible future
device to have always cells=0. If that's the case then indeed you do not
need cells, but this needs explanation. You are designing now entire
ABI, so you must design it fully, not just "works for me now".

Lack of complete DTS - nothing here, nothing in the changelog - is IMO
reason to NAK this patchset completely, also for reason me guessing the
entire design instead of seeing the big picture.

Best regards,
Krzysztof