On Wed, 2018-02-07 at 11:48 +0530, Viresh Kumar wrote:

On 07-02-18, 14:16, Sean Wang wrote:

quoted

On Wed, 2018-02-07 at 09:03 +0530, Viresh Kumar wrote:

quoted

On 06-02-18, 17:52, sean.wang-NuS5LvNUpcJWk0Htik3J/w@public.gmane.org wrote:

quoted

 	cpus {
 		#address-cells = <2>;
 		#size-cells = <0>;

@@ -26,6 +70,10 @@
 			device_type = "cpu";
 			compatible = "arm,cortex-a53", "arm,armv8";
 			reg = <0x0 0x0>;
+			clocks = <&infracfg CLK_INFRA_MUX1_SEL>,
+				 <&apmixedsys CLK_APMIXED_MAIN_CORE_EN>;
+			clock-names = "cpu", "intermediate";
+			operating-points-v2 = <&cpu_opp_table>;
 			enable-method = "psci";
 			clock-frequency = <1300000000>;
 		};

@@ -34,6 +82,7 @@
 			device_type = "cpu";
 			compatible = "arm,cortex-a53", "arm,armv8";
 			reg = <0x0 0x1>;
+			operating-points-v2 = <&cpu_opp_table>;
 			enable-method = "psci";
 			clock-frequency = <1300000000>;
 		};

Sorry for not picking this earlier, but you should probably add the same clock
related properties for both cpu nodes here. Things will break if CPU1 is used by
the cpufreq core to bring the cpufreq policy online.

This can happen if cpufreq driver is a module, CPU0 is hotplugged out and then
the cpufreq driver is inserted.

mt7622 cpu0 does not support hotplug. do I still need to add same clock
related properties for both cpu nodes here?

Normally we should always add these properties to all the CPUs, as that's the
real scenario hardware configuration wise.

Agree on, I will add these missing clock properties also into the cpu1
node to reflect the hardware actually should have. 

That also is devicetree wants us to do to describe the device more
closely.

But I am not sure if something else will break if you don't provide clocks in
CPU1.

@Rob @Mark: What do you suggest ?

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html