On Tue, Sep 17, 2013 at 10:14:01AM -0600, Stephen Warren wrote: > On 09/17/2013 01:59 AM, Sebastian Hesselbarth wrote: > > On 09/16/2013 08:37 PM, Stephen Warren wrote: > ... > >> Perhaps if clock-frequency is specified, the driver should refuse to > >> provide anything else. If clock-frequency isn't specified, the driver > >> shouldn't touch the HW when it initializes, but should honor any > >> requests that come in from other drivers? That would maintain what I > >> feel is clock-frequency's connection to being a fixed clock. > > > > For the clk-si5351 programmable clock driver in mainline, it already > > uses "clock-frequency" for initial clock setup but allows to set it > > later on. IMHO that is ok, because from a initial point-of-view, an > > initial frequency is fixed. As soon as the driver takes over, the user > > is free to do whatever he wants and should not be limited by DT. > > > > But if we vote against that approach, we should probably also modifiy > > clk-si5351 accordingly. > > I suppose that approach isn't unreasonable. So, if there's precedent for > it, this driver may as well follow it. All right, this sounds the most flexible to me and requires the least changes as well :). I think all I have to change for v2 in regards of this, is to rename the property to 'clock-frequency'. I'll do that and send out v2 later today. Sören -- 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