On Wed, Apr 23, 2014 at 04:50:39PM +0200, Krzysztof Kozlowski wrote: > This patch adds support for mode REGULATOR_MODE_STANDBY (and NORMAL) to > LDO regulators by implementing the set_mode() and get_mode() operations. > However the necessary regulator constraints (valid modes) are not parsed > by of_regulator_match() so the driver adds them manually to the > regulator init_data.
No, that's not the idea here. The reason that the modes need to be explicitly enabled is that there's an element of board design in determining if a given mode can satisfy the required current demand for the board with sufficient quality (usually the lower power modes have both a lower maximum current and poorer regulation accuracy especially as the current rises). Doing it unconditionally isn't in general reliable. The reason that the modes aren't supported by DT is that defining a binding is hard - it's not clear what exactly a "mode" means since it's basically a Linux internal thing. We probably need to explicitly add definitions of the modes to the bindings for individual devices unfortunately (ie, saying "mode X maps to Y in the datasheet", possibly using the datasheet modes in the binding for ease of use and having that translation in the driver). Ideally we'd be able to have the automatic mode setting working for devices but in practice nobody wants to publish the numbers and working out how much the board needs can also be hard so that isn't really practical.
signature.asc
Description: Digital signature
