On Tue, Jan 29, 2019 at 11:46:52PM +0100, Niklas Cassel wrote: > Adding Mark Brown on CC. It really helps if you ask a specific question when doing something like this rather than just have a big quoted mail - it makes it much easier to find what's relevant rather than trying to find things, especially when they're buried behind several layers of quoting. > On Tue, Jan 29, 2019 at 10:58:47PM +0100, Jorge Ramirez wrote: > > On 1/26/19 00:29, Bjorn Andersson wrote: > > the question is, should this property contain only hardware achievable > > values? or should drivers only request hardware achievable values? the > > way the constrains are implemented it has to be one of the two (I think > > the former would be more intuitive - ie if the dts > > regulator-min-microvolt is a valid value) Drivers should not be coded with a specific regulator or board in mind and should just request whatever they need. This will then be matched with whatever the board is actually able to deliver. Similarly there is no requirement that machine constraints be written with specific reference to what the physical regulator on the board is able to do, for example the constraints will come from electrical engineering restrictions like the specifications of the parts connected to the regulator rather than from what the regulator can actually do so people should feel free to just write down the actual physical constraint and let the regulator API ensure that the constraint is met.
Attachment:
signature.asc
Description: PGP signature
