On Mon 08 Jun 23:44 PDT 2020, Amit Kucheria wrote:
> sm8250 has 24 thermal sensors split across two tsens controllers. Add
> the thermal zones to expose them and wireup the cpus to throttle on
> crossing passive temperature thresholds.
>
> Update the comment in the drivers to list the SoCs it supports.
>
> Signed-off-by: Amit Kucheria <[email protected]>
> ---
> arch/arm64/boot/dts/qcom/sm8250.dtsi | 766 +++++++++++++++++++++++++++
> drivers/thermal/qcom/tsens-v2.c | 2 +-
> 2 files changed, 767 insertions(+), 1 deletion(-)
>
> diff --git a/arch/arm64/boot/dts/qcom/sm8250.dtsi
> b/arch/arm64/boot/dts/qcom/sm8250.dtsi
> index deaa8415c7b72..5cd18cd8a675b 100644
> --- a/arch/arm64/boot/dts/qcom/sm8250.dtsi
> +++ b/arch/arm64/boot/dts/qcom/sm8250.dtsi
> @@ -8,6 +8,7 @@
> #include <dt-bindings/clock/qcom,rpmh.h>
> #include <dt-bindings/power/qcom-rpmpd.h>
> #include <dt-bindings/soc/qcom,rpmh-rsc.h>
> +#include <dt-bindings/thermal/thermal.h>
>
> / {
> interrupt-parent = <&intc>;
> @@ -86,6 +87,7 @@ CPU0: cpu@0 {
> enable-method = "psci";
> next-level-cache = <&L2_0>;
> qcom,freq-domain = <&cpufreq_hw 0>;
> + #cooling-cells = <2>;
This doesn't apply to linux-next.
The problem seems to be that, as pointed out when I submitted that
patch, the previously anonymous "cpufreq hardware" is now replaced by
the "EPSS" hardware block.
So we need a new driver (or update the existing one) to support this new
hardware block.
Presumably though, without this there's not much cooling anyways - which
is sad, as your patch looks good.
> L2_0: l2-cache {
> compatible = "cache";
> next-level-cache = <&L3_0>;
[..]
> diff --git a/drivers/thermal/qcom/tsens-v2.c b/drivers/thermal/qcom/tsens-v2.c
> index b293ed32174b5..58cac8f2a358c 100644
> --- a/drivers/thermal/qcom/tsens-v2.c
> +++ b/drivers/thermal/qcom/tsens-v2.c
> @@ -26,7 +26,7 @@
> #define TM_TRDY_OFF 0x00e4
> #define TM_WDOG_LOG_OFF 0x013c
>
> -/* v2.x: 8996, 8998, sdm845 */
> +/* v2.x: 8996, 8998, sc7180, sdm845, sm8150, sm8250 */
Even though it's trivial, can you please send this through the tsens
tree instead, so we don't end up having unnecessary merge conflicts.
Regards,
Bjorn
>
> static struct tsens_features tsens_v2_feat = {
> .ver_major = VER_2_X,
> --
> 2.25.1
>
