2018-02-22 17:44 GMT+01:00 David Lechner <[email protected]>: > On 02/22/2018 05:34 AM, Philipp Zabel wrote: >> >> On Tue, 2018-02-20 at 10:40 -0600, David Lechner wrote: >> [...] >>>> >>>> In your case the platform code that adds the lookup may be identical to >>>> the code that registers the struct reset_controller_dev, but that >>>> doesn't have to be the case. I'm not sure how that is supposed to work >>>> for the phy framework (I see no platform code adding phy lookups, only >>>> drivers). >>>> >>> In our use case, we would be adding the lookup in the driver rather than >>> in the platform code, which is why I am suggesting doing it like the phy >>> framework. >> >> >> Shouldn't it be the job of the platform code to describe the connections >> between reset controller and peripheral module reset >> inputs? > > > > I guess that depends on who you ask. There are many clock driver that > register clkdev lookups in drivers/clk/, so that is what we have done > with the clock driver we are working on. The clock device is also the > reset controller, so it makes sense to me to register the reset lookup > in the same place that we are registering the clkdev lookup. > > We have a platform_device_id for each possible configuration, so that > it works very much like the device tree compatible string. You register > a platform device in the board file with the proper name and the driver > takes care of the reset because it knows which connections there are > based on the device name.
I sent another attempt. Since the in-progress psc driver has all the clocks in the driver code, it makes sense to do the same for resets. Bart
