Also, I think you can get rid of 'ti,pruss-gp-mux-sel'. Can't it just be an arg cell in 'ti,prus' entries? Rob
+1 for using cells instead of a separate property. FYI, we will have a similar issue with the PRUSSEVTSEL signal for the interrupt controller on the AM18XX. I am still of the opinion (described in more detail at [1]) that using a cell for this makes for both better device tree bindings and easier driver implementation. So I am interested to see what the resolution is here. [1]: https://patchwork.kernel.org/project/linux-arm-kernel/patch/20190708035243.12170-5-s-a...@ti.com/
