[PATCH v5 3/3] riscv: dts: spacemit: Add thermal sensor for K1 SoC

Shuwei Wu shuwei.wu at mailbox.org
Mon May 18 19:41:27 PDT 2026


Hi Yixun,

On Tue May 19, 2026 at 8:26 AM CST, Yixun Lan wrote:
> Hi Shuwei,
>
> On 15:15 Mon 27 Apr     , Shuwei Wu wrote:
>> Include the Thermal Sensor node in the SpacemiT K1 dtsi
>> with definitions for registers, clocks, and interrupts.
>> Additionally, configure thermal zones for the soc, package, gpu, and
>> clusters to enable temperature monitoring via the thermal framework.
>> 
>> Tested-by: Vincent Legoll <legoll at online.fr> # OrangePi-RV2
>> Tested-by: Gong Shuai <gsh517025 at gmail.com>
>> Signed-off-by: Shuwei Wu <shuwei.wu at mailbox.org>
>> 
>> ---
>> Changes in v2:
>> - Update compatible to "spacemit,k1-tsensor"
>> ---
>>  arch/riscv/boot/dts/spacemit/k1.dtsi | 101 +++++++++++++++++++++++++++++++++++
>>  1 file changed, 101 insertions(+)
>> 
>> diff --git a/arch/riscv/boot/dts/spacemit/k1.dtsi b/arch/riscv/boot/dts/spacemit/k1.dtsi
>> index 529ec68e9c23..e9952204224e 100644
>> --- a/arch/riscv/boot/dts/spacemit/k1.dtsi
>> +++ b/arch/riscv/boot/dts/spacemit/k1.dtsi
>> @@ -339,6 +339,96 @@ osc_32k: clock-32k {
>>  		};
>>  	};
>>  
>> +	thermal-zones {
>> +		soc-thermal {
>> +			polling-delay-passive = <0>;
>> +			polling-delay = <0>;
>> +			thermal-sensors = <&thermal 0>;
>> +
>> +			trips {
>> +				soc-crit {
>> +					temperature = <115000>;
>> +					hysteresis = <0>;
>> +					type = "critical";
>> +				};
>> +			};
>> +		};
>> +
>> +		package-thermal {
>> +			polling-delay-passive = <0>;
>> +			polling-delay = <0>;
>> +			thermal-sensors = <&thermal 1>;
>> +
>> +			trips {
>> +				package-crit {
>> +					temperature = <115000>;
>> +					hysteresis = <0>;
>> +					type = "critical";
>> +				};
>> +			};
>> +		};
>> +
>> +		gpu-thermal {
>> +			polling-delay-passive = <100>;
>> +			polling-delay = <0>;
>> +			thermal-sensors = <&thermal 2>;
>> +
>> +			trips {
>> +				gpu-alert {
>> +					temperature = <85000>;
>> +					hysteresis = <2000>;
>> +					type = "passive";
>> +				};
>> +
>> +				gpu-crit {
>> +					temperature = <115000>;
>> +					hysteresis = <0>;
>> +					type = "critical";
>> +				};
>> +			};
>> +		};
>> +
>> +		cluster0-thermal {
>> +			polling-delay-passive = <100>;
>> +			polling-delay = <0>;
>> +			thermal-sensors = <&thermal 3>;
>> +
>> +			trips {
>> +				cluster0-alert {
>> +					temperature = <85000>;
>> +					hysteresis = <2000>;
>> +					type = "passive";
>> +				};
>> +
>> +				cluster0-crit {
>> +					temperature = <115000>;
>> +					hysteresis = <0>;
>> +					type = "critical";
>> +				};
>> +			};
>> +		};
>> +
>> +		cluster1-thermal {
>> +			polling-delay-passive = <100>;
>> +			polling-delay = <0>;
>> +			thermal-sensors = <&thermal 4>;
>> +
>> +			trips {
>> +				cluster1-alert {
>> +					temperature = <85000>;
>> +					hysteresis = <2000>;
>> +					type = "passive";
>> +				};
>> +
>> +				cluster1-crit {
>> +					temperature = <115000>;
>> +					hysteresis = <0>;
>> +					type = "critical";
>> +				};
>> +			};
>> +		};
>> +	};
>> +
>>  	soc {
>>  		compatible = "simple-bus";
>>  		interrupt-parent = <&plic>;
>> @@ -494,6 +584,17 @@ syscon_apbc: system-controller at d4015000 {
>>  			#reset-cells = <1>;
>>  		};
>>  
>> +		thermal: thermal at d4018000 {
>> +			compatible = "spacemit,k1-tsensor";
>> +			reg = <0x0 0xd4018000 0x0 0x100>;
>> +			clocks = <&syscon_apbc CLK_TSEN>,
>> +				 <&syscon_apbc CLK_TSEN_BUS>;
>> +			clock-names = "core", "bus";
>> +			interrupts = <61>;
>> +			resets = <&syscon_apbc RESET_TSEN>;
>> +			#thermal-sensor-cells = <1>;
>> +		};
> Ok, so if I understand correctly the thermal is a SoC feature, so with
> above it will be enabled by default for all boards, but for the convention
> we usually disable it in dtsi file, and enable it at board dts level
>
> Please convince me doing above is better? as I'm not sure if there is
> cases that user want it disabled (but could possible)..

I had two reasons for keeping the thermal node enabled in k1.dtsi,
rather than requiring each board to opt in:

1. Other SoCs follow the same pattern: the RISC-V JH7110 and Allwinner D1s
both define their thermal sensor nodes and thermal-zones in the SoC dtsi
without status = "disabled".

2. Thermal protection is safety-critical, and the silicon trip points
don't vary between boards. Leaving it disabled by default risks boards
booting without thermal shutdown.

-- 
Best regards,
Shuwei Wu



More information about the linux-riscv mailing list