On 29-07-24, 22:30, Ulf Hansson wrote: > In regards to the clocks, I assume this is handled correctly too, as > the clocks are per device clocks that don't belong to the genpd. It would be if the clk node is present in the device's node. I was talking about a clock node in the genpd's table earlier. If that is ever the case, we will end up programming the wrong clk here. > That's right, but why do we want to call dev_pm_opp_set_opp() for the > multiple PM domain case then? It makes the behaviour inconsistent. To have a common path for all required OPP device types, irrespective of the fact that the required OPP device is a genpd or not. And we are talking about a required OPP of a separate device here, it must be set via this call only, technically speaking. Genpd makes it a little complex, and I agree to that. But I strongly feel this code needs to be generic and not genpd specific. The OPP core should have as less genpd specific code as possible. It must handle all device types with a single code path. -- viresh
