On 26/04/2022 15:52, Hawkins, Nick wrote: > Apologies, I did miss the comment about the double spacing around the label and the label not being necessary. I will not make this mistake again. I became focused about the comment of mapping an entire register space which indirectly lead me on to the path which I am now having the gxp-timer have the gxp-wdt as a child. To be specific the feedback I was speaking of above was about the gxp-timer which is here: https://lore.kernel.org/all/704ffa56-4bae-fc33-fddf-3e3dd8be0db9@xxxxxxxxxx/ That is the children must be defined for a simple-mfd device. This was comment for this v5, not for previous patches. In this v5, you have a child of timer, so it has to be defined in timer schema. This was not a comment whether a child should exist or should not. It was made under the assumption that you want to have a child node. > Hence the plan I have now is to remove the hpe,gxp-wdt.yaml entirely and include it in the hpe,gxp-timer.yaml. I assume that is the correct thing to do? I would follow here the advice from Rob, so since the blocks are mixed significantly (same address space), then let's assume it's actually one device with two functions. In such case Rob pointed out that child node is not necessary. The implementation might differ, depending how the features are mixed-up with each other. It might be one driver having timer and watchdog, or several drivers (usually bound together with a MFD driver which serves as parents and binds to the OF node). Best regards, Krzysztof
