On Friday 21 November 2014 23:05:28 Jassi Brar wrote: > On 21 November 2014 22:39, Arnd Bergmann <arnd@xxxxxxxx> wrote: > > On Friday 21 November 2014 22:19:43 Jassi Brar wrote: > >> > > >> > This seems weird: rather than have one clock controller node per > >> > clock source, why not just have one node with #clock-cells = >? > >> > > >> > Your controller seems perfectly capable of this, and you even > >> > put the numbers in the label. It would be much more straightforward > >> > to do > >> > > >> > clocks = <&clk_alw 0 1>; > >> > > >> > than > >> > > >> > clocks = <&clk_alw_0_1>; > >> > > >> > here. Possibly even > >> > > >> > clocks = <&clk MB86S70_CRG11_ALW 0 1>; > >> > > >> > with #clock-cells = > > >> > > >> Is there a way to compact the dtsi and yet populate only clocks that > >> are actually ever used (instead of 6x16x9)? > > > > I believe the clock driver just wouldn't create the clocks until > > a device driver requests them. > > > Seems like the chicken-egg problem. Let me check. What I mean is that you should call of_clk_add_provider once for each of the six clock controllers, but not call clk_register_gate or clk_register for each possible output until it gets requested. Arnd -- 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
