Re: arm64: dts: zx: support cpu-freq for zx296718

[Shawn Guo]

+ Viresh, the author of the bindings.

On Thu, Dec 01, 2016 at 08:08:55PM +0800, Baoyou Xie wrote:
> This patch adds the CPU clock phandle in CPU's node
> and uses operating-points-v2 to register operating points.
> 
> So it can be used by cpufreq-dt driver.
> 
> Signed-off-by: Baoyou Xie <baoyou.xie@xxxxxxxxxx>

Just out of curiosity, why is there no '[PATCH]' prefix in the subject?

> ---
>  arch/arm64/boot/dts/zte/zx296718.dtsi | 37 +++++++++++++++++++++++++++++++++++
>  1 file changed, 37 insertions(+)
> 
> diff --git a/arch/arm64/boot/dts/zte/zx296718.dtsi b/arch/arm64/boot/dts/zte/zx296718.dtsi
> index 7a1aed7..16f7d5e 100644
> --- a/arch/arm64/boot/dts/zte/zx296718.dtsi
> +++ b/arch/arm64/boot/dts/zte/zx296718.dtsi
> @@ -44,6 +44,7 @@
>  #include <dt-bindings/input/input.h>
>  #include <dt-bindings/interrupt-controller/arm-gic.h>
>  #include <dt-bindings/gpio/gpio.h>
> +#include <dt-bindings/clock/zx296718-clock.h>
>  
>  / {
>  	compatible = "zte,zx296718";
> @@ -81,6 +82,8 @@
>  			compatible = "arm,cortex-a53","arm,armv8";
>  			reg = <0x0 0x0>;
>  			enable-method = "psci";
> +			clocks = <&topcrm A53_GATE>;
> +			operating-points-v2 = <&cluster0_opp>;
>  		};
>  
>  		cpu1: cpu@1 {
> @@ -88,6 +91,7 @@
>  			compatible = "arm,cortex-a53","arm,armv8";
>  			reg = <0x0 0x1>;
>  			enable-method = "psci";
> +			operating-points-v2 = <&cluster0_opp>;
>  		};
>  
>  		cpu2: cpu@2 {
> @@ -95,6 +99,7 @@
>  			compatible = "arm,cortex-a53","arm,armv8";
>  			reg = <0x0 0x2>;
>  			enable-method = "psci";
> +			operating-points-v2 = <&cluster0_opp>;
>  		};
>  
>  		cpu3: cpu@3 {
> @@ -102,6 +107,38 @@
>  			compatible = "arm,cortex-a53","arm,armv8";
>  			reg = <0x0 0x3>;
>  			enable-method = "psci";
> +			operating-points-v2 = <&cluster0_opp>;
> +		};
> +	};
> +
> +	cluster0_opp: opp_table0 {

I know this is how examples in the bindings doc written, but it's
recommended to use hyphen rather than underscore in node name.  That
said, the following naming form is better.

	cluster0_opp: opp-table0

> +		compatible = "operating-points-v2";
> +		opp-shared;
> +
> +		opp@500000000 {
> +			opp-hz = /bits/ 64 <500000000>;
> +			opp-microvolt = <857000>;
> +			clock-latency-ns = <500000>;
> +		};

We prefer to have a newline between nodes.

> +		opp@648000000 {
> +			opp-hz = /bits/ 64 <648000000>;
> +			opp-microvolt = <857000>;
> +			clock-latency-ns = <500000>;
> +		};
> +		opp@800000000 {
> +			opp-hz = /bits/ 64 <800000000>;
> +			opp-microvolt = <882000>;
> +			clock-latency-ns = <500000>;
> +		};
> +		opp@1000000000 {
> +			opp-hz = /bits/ 64 <1000000000>;
> +			opp-microvolt = <892000>;
> +			clock-latency-ns = <500000>;
> +		};
> +		opp@1188000000 {
> +			opp-hz = /bits/ 64 <1188000000>;
> +			opp-microvolt = <1009000>;

So we have 5 setpoints with different frequency-voltage pair.  I have
seen 'clocks' specified in cpu0 node and understand how frequency
scaling works.  But what about voltage scaling?  There is no
'cpu-supply' defined, and how does voltage scale among these
opp-microvolt settings?

Another related question: if we do not support voltage scaling for now,
what's the actually voltage when system is up running?  Is that voltage
safe for cpu to run at all those 5 frequencies?

Shawn

> +			clock-latency-ns = <500000>;
>  		};
>  	};
>  
> -- 
> 2.7.4
>