Re: [PATCH 04/10] ARM: dts: enable clock support for BCM5301X

[Stephen Boyd <sboyd@codeaurora.org>]

On 10/02, Jon Mason wrote:
> Replace current device tree dummy clocks with real clock support for
> Broadcom Northstar SoCs.
> 
> Signed-off-by: Jon Mason <jonmason@broadcom.com>
> ---

I'd rather not take any dts changes through clk tree.

>  arch/arm/boot/dts/bcm5301x.dtsi | 67 ++++++++++++++++++++++++++++++++++++-----
>  1 file changed, 60 insertions(+), 7 deletions(-)
> 
> diff --git a/arch/arm/boot/dts/bcm5301x.dtsi b/arch/arm/boot/dts/bcm5301x.dtsi
> index 6f50f67..f717859 100644
> --- a/arch/arm/boot/dts/bcm5301x.dtsi
> +++ b/arch/arm/boot/dts/bcm5301x.dtsi
> @@ -55,14 +56,14 @@
>  			compatible = "arm,cortex-a9-global-timer";
>  			reg = <0x0200 0x100>;
>  			interrupts = <GIC_PPI 11 IRQ_TYPE_LEVEL_HIGH>;
> -			clocks = <&clk_periph>;
> +			clocks = <&periph_clk>;
>  		};
>  
>  		local-timer@0600 {
>  			compatible = "arm,cortex-a9-twd-timer";
>  			reg = <0x0600 0x100>;
>  			interrupts = <GIC_PPI 13 IRQ_TYPE_LEVEL_HIGH>;
> -			clocks = <&clk_periph>;
> +			clocks = <&periph_clk>;
>  		};
>  
>  		gic: interrupt-controller@1000 {
> @@ -94,14 +95,66 @@
>  
>  	clocks {

I'd expect this to only contain nodes that don't have a reg
property. Clock providers that have a reg property would go into
some soc node or bus. Perhaps that's the chipcommonA node, or
axi?

>  		#address-cells = <1>;
> -		#size-cells = <0>;
> +		#size-cells = <1>;
> +		ranges;
>  
> -		/* As long as we do not have a real clock driver us this
> -		 * fixed clock */
> -		clk_periph: periph {
> +		osc: oscillator {
> +			#clock-cells = <0>;
>  			compatible = "fixed-clock";
> +			clock-frequency = <25000000>;
> +		};
> +
> +		lcpll0: lcpll0@1800c100 {
> +			#clock-cells = <1>;
> +			compatible = "brcm,nsp-lcpll0";
> +			reg = <0x1800c100 0x14>;
> +			clocks = <&osc>;
> +			clock-output-names = "lcpll0", "pcie_phy", "sdio",
> +					     "ddr_phy";
> +		};
> +
> +		genpll: genpll@1800c140 {
> +			#clock-cells = <1>;
> +			compatible = "brcm,nsp-genpll";
> +			reg = <0x1800c140 0x24>;
> +			clocks = <&osc>;
> +			clock-output-names = "genpll", "phy", "ethernetclk",
> +					     "usbclk", "iprocfast", "sata1",
> +					     "sata2";
> +		};
> +
> +		iprocmed: iprocmed {
> +			#clock-cells = <0>;
> +			compatible = "fixed-factor-clock";
> +			clocks = <&genpll BCM_NSP_GENPLL_IPROCFAST_CLK>;
> +			clock-div = <2>;
> +			clock-mult = <1>;
> +			clock-output-names = "iprocmed";
> +		};
> +
> +		iprocslow: iprocslow {
> +			#clock-cells = <0>;
> +			compatible = "fixed-factor-clock";
> +			clocks = <&genpll BCM_NSP_GENPLL_IPROCFAST_CLK>;
> +			clock-div = <4>;
> +			clock-mult = <1>;
> +			clock-output-names = "iprocslow";
> +		};
> +
> +
> +		a9pll: arm_clk@19000000 {
> +			#clock-cells = <0>;
> +			compatible = "brcm,nsp-armpll";
> +			clocks = <&osc>;
> +			reg = <0x19000000 0x1000>;
> +		};
> +
> +		periph_clk: periph_clk {
>  			#clock-cells = <0>;
> -			clock-frequency = <400000000>;
> +			compatible = "fixed-factor-clock";
> +			clocks = <&a9pll>;
> +			clock-div = <2>;
> +			clock-mult = <1>;
>  		};
>  	};
>  

We're trying to move away from putting individual clocks into DT
like this. Is there some sort of clock controller that's at
0x1800c000, but we're just not exposing all the clocks in there?
See this thread for more information on why we'd like to avoid
this sort of design:

http://lkml.kernel.org/r/<20150416192014.19585.9663@quantum>

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project