On 27/06/2022 13:49, Viorel Suman wrote: > On 22-06-24 12:25:44, Krzysztof Kozlowski wrote: >> On 16/06/2022 18:42, Viorel Suman wrote: >>> From: Abel Vesa <abel.vesa@xxxxxxx> >>> >>> In order to replace the fsl,scu txt file from bindings/arm/freescale, >>> we need to split it between the right subsystems. This patch adds the >>> fsl,scu.yaml in the firmware bindings folder. This one is only for >>> the main SCU node. The old txt file will be removed only after all >>> the child nodes have been properly switch to yaml. >>> >>> Signed-off-by: Abel Vesa <abel.vesa@xxxxxxx> >>> Signed-off-by: Viorel Suman <viorel.suman@xxxxxxx> >>> --- >>> .../devicetree/bindings/firmware/fsl,scu.yaml | 170 ++++++++++++++++++ >>> 1 file changed, 170 insertions(+) >>> create mode 100644 Documentation/devicetree/bindings/firmware/fsl,scu.yaml >>> >>> diff --git a/Documentation/devicetree/bindings/firmware/fsl,scu.yaml b/Documentation/devicetree/bindings/firmware/fsl,scu.yaml > > [...] > >>> +properties: >>> + $nodename: >>> + const: 'scu' >> >> Why enforcing node name? Second point is that node names should be >> generic, so I wonder what "SCU" exactly means and whether it is generic? >> > > It stands for "System Control Unit" - looks generic to me. Nope, it's specific, just like other scu - Snoop Control Unit. What's more, reusing the same acronym leads to confusions. If it was generic, then I expect it to be present in several other places, but it's not (except the other SCU). Generic name is for example "system-controller". > I guess a reason to enforce it - need to check with Abel - might be > the need to group multiple SCU implementations under a common known name. Device bindings do not enforce the names, unless it's really needed, and I doubt there is a need here. Just drop it and rename nodes in DTS to something generic. Best regards, Krzysztof