Re: [RFC 6/7] arm64: dts: qcom: x1e80100: Enable cpufreq

From: Sibi Sankar
Date: Mon Feb 12 2024 - 04:29:20 EST

On 1/18/24 20:55, Sudeep Holla wrote:
(Generic note: It is middle of merge window and I have seen multiple
series posted by you. Since I am mainly looking for bug fixes only ATM,
I may miss to look at few. You may have to ping or repost after the merge
window, just responding to this for now as it caught my attention)


On Wed, Jan 17, 2024 at 11:04:57PM +0530, Sibi Sankar wrote:
Enable cpufreq on X1E80100 SoCs through the SCMI perf protocol node.

Signed-off-by: Sibi Sankar <quic_sibis@xxxxxxxxxxx>
arch/arm64/boot/dts/qcom/x1e80100.dtsi | 27 ++++++++++++++++++++++++++
1 file changed, 27 insertions(+)

diff --git a/arch/arm64/boot/dts/qcom/x1e80100.dtsi b/arch/arm64/boot/dts/qcom/x1e80100.dtsi
index afdbd27f8346..6856a206f7fc 100644
--- a/arch/arm64/boot/dts/qcom/x1e80100.dtsi
+++ b/arch/arm64/boot/dts/qcom/x1e80100.dtsi
@@ -62,6 +62,7 @@ CPU0: cpu@0 {
compatible = "qcom,oryon";
reg = <0x0 0x0>;
enable-method = "psci";
+ clocks = <&scmi_dvfs 0>;

I would use genpd bindings Ulf added recently. The reason I ask is I remember
one of the Qcom platform had both clocks and qcom,freq-domain and each one
served different purpose with latter one being used for cpufreq. So will
that be an issue here ?

The cpufreq-hw node that Qualcomm used had a opp-table associated with
it to vote for various buses which in turn required both clock and freq-
domain. However the memory buses voting is done by the vendor protocol 0x80 on X1E and hence won't be an issue here.