Quoting Taniya Das (2018-07-12 11:05:44)
[..]

+                       compatible = "qcom,kryo385";
+                       reg = <0x0 0x600>;
+                       enable-method = "psci";
+                       next-level-cache = <&L2_600>;
+                       qcom,freq-domain = <&freq_domain_table1>;
+                       L2_600: l2-cache {
+                               compatible = "cache";
+                               next-level-cache = <&L3_0>;
+                       };
+               };
+
+               CPU7: cpu@700 {
+                       device_type = "cpu";
+                       compatible = "qcom,kryo385";
+                       reg = <0x0 0x700>;
+                       enable-method = "psci";
+                       next-level-cache = <&L2_700>;
+                       qcom,freq-domain = <&freq_domain_table1>;
+                       L2_700: l2-cache {
+                               compatible = "cache";
+                               next-level-cache = <&L3_0>;
+                       };
+               };
+       };
+
+       qcom,cpufreq-hw {
+               compatible = "qcom,cpufreq-hw";
+               #address-cells = <2>;
+               #size-cells = <2>;
+               ranges;
+               freq_domain_table0: freq_table0 {
+                       reg = <0 0x17d43000 0 0x1400>;
+               };
+
+               freq_domain_table1: freq_table1 {
+                       reg = <0 0x17d45800 0 0x1400>;
+               };

It seems that we need to map the CPUs in the cpus node to the frequency
domains in the cpufreq-hw node. Wouldn't that be better served via a
#foo-cells and <&phandle foo-cell> property in the CPU node? It's
annoying that the cpufreq-hw node doesn't have a reg property, when it
really should have one that goes over the whole register space (or is
split across the frequency domains so that there are two reg properties
here).