Am Freitag, den 22.07.2016, 21:25 +0200 schrieb Arnd Bergmann: > On Thursday, July 21, 2016 7:16:02 PM CEST Paul Gortmaker wrote: > > > > > ...we probably don't need module.h here or any of the other > > > > MODULE_<blah> tags/macros either. Use the builtin for the > > > > register and all should be good. > > > > > > I wonder if we should have separate symbols for each reset driver > > > instead. Those could be allowed to be tristate, and could also > > > be enabled for compile testing on other architectures. > > > > I'm not quite clear on what you mean by the above. Maybe if you have a > > pointer to an example -- say like the work you mention below -- then > > I'll understand what you have in mind. > > I mean instead of having each driver compiled implicitly when the > ARCH_* symbol is enabled, make the drivers user-selectable > and just default to 'y' when the platform is enabled. This would > let us actually use modular reset drivers where appropriate > (which might be nowhere, but at least we wouldn't forbid it > for no reason). I agree with this for the drivers that could be built as a module (which is most of them). I don't think it is useful for the builtin drivers to have user visible Kconfig symbols. > Another annoying thing is that all platforms today need to > select two symbols > > select ARCH_HAS_RESET_CONTROLLER > select RESET_CONTROLLER > > in order to actually build the driver, and the second one is > user-selectable, meaning you can still disable all of them today, > just not individual drivers. RESET_CONTROLLER is default y if ARCH_HAS_RESET_CONTROLLER, so if platforms select ARCH_HAS_RESET_CONTROLLER, that should be enough. regards Philipp -- To unsubscribe from this list: send the line "unsubscribe devicetree" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
