Re: [PATCH] arm64: dts: renesas: r8a779g3: Update thermal trip points on V4H Sparrow Hawk

[Geert Uytterhoeven <geert@linux-m68k.org>]

Hi Marek,

On Wed, 25 Jun 2025 at 12:03, Marek Vasut
<marek.vasut+renesas@mailbox.org> wrote:
> Since the Sparrow Hawk has a smaller PCB than the White Hawk, it tends
> to generate more heat. To prevent potential damage to the board, adjust
> the temperature trip points.
>
> Add four "passive" trip points which increasingly throttle the CPU to
> prevent overheating. The first trip point at 68°C disables the 1.8 GHz
> and 1.7 GHz modes and limits the CPU to 1.5 GHz frequency. The second
> trip point at 72°C disables the 1.5 GHz mode and limits the CPU to 1.0
> GHz frequency. The third trip point at 76°C uses thermal-idle to start
> inserting idle cycles into the CPU instruction stream to cool the CPU
> cores down. The fourth and last trip point at 80°C disables the 1.0 GHz
> mode and limits the CPU to 500 MHz frequency.
>
> In case the SoC heats up further, in case either of the thermal sensors
> readings passes the 100°C, a thermal shutdown is triggered to prevent
> any damage to the hardware.
>
> Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>

Thanks for your patch!

> --- a/arch/arm64/boot/dts/renesas/r8a779g3-sparrow-hawk.dts
> +++ b/arch/arm64/boot/dts/renesas/r8a779g3-sparrow-hawk.dts
> @@ -38,6 +38,7 @@
>
>  /dts-v1/;
>  #include <dt-bindings/gpio/gpio.h>
> +#include <dt-bindings/thermal/thermal.h>
>
>  #include "r8a779g3.dtsi"
>
> @@ -797,3 +798,139 @@ &rwdt {
>  &scif_clk {    /* X12 */
>         clock-frequency = <24000000>;
>  };
> +
> +/* thermal-idle cooling for all SoC cores */
> +&a76_0 {

Please keep nodes sorted (alphabetically by label).

> +       #cooling-cells = <2>;

This is only present for the first CPU core, and map{0,1,3} refer
only to a76_0, because all four CPU cores are driven by a single clock
(Z0), right?

> +
> +       a76_0_thermal_idle: thermal-idle {
> +               #cooling-cells = <2>;
> +               duration-us = <10000>;
> +               exit-latency-us = <500>;
> +       };
> +};

> +/* THS sensor in SoC near CA76 cores does more progressive cooling. */
> +&sensor_thermal_ca76 {
> +       critical-action = "shutdown";
> +
> +       cooling-maps {
> +               /*
> +                * The cooling-device minimum and maximum parameters inversely
> +                * match opp-table-0 {} node entries in r8a779g0.dtsi, in other
> +                * words, 0 refers to 1.8 GHz OPP and 4 refers to 500 MHz OPP.
> +                * This is because they refer to cooling levels, where maximum
> +                * cooling level happens at 500 MHz OPP, when the CPU core is
> +                * running slowly and therefore generates least heat.

That applies to cooling-device = <&a76_[0-3] ...>...

> +                */
> +               map0 {
> +                       /* At 68C, inhibit 1.7 GHz and 1.8 GHz modes */
> +                       trip = <&sensor3_passive_low>;
> +                       cooling-device = <&a76_0 2 4>;
> +                       contribution = <128>;
> +               };
> +
> +               map1 {
> +                       /* At 72C, inhibit 1.5 GHz mode */
> +                       trip = <&sensor3_passive_mid>;
> +                       cooling-device = <&a76_0 3 4>;
> +                       contribution = <256>;
> +               };
> +
> +               map2 {
> +                       /* At 76C, start injecting idle states */
> +                       trip = <&sensor3_passive_hi>;
> +                       cooling-device = <&a76_0_thermal_idle 0 80>,
> +                                        <&a76_1_thermal_idle 0 80>,
> +                                        <&a76_2_thermal_idle 0 80>,
> +                                        <&a76_3_thermal_idle 0 80>;

... but what do "0 80" refer to? I couldn't find in the thermal-idle
bindings what exactly are the minimum and maximum cooling states here.

> +                       contribution = <512>;
> +               };

The rest LGTM, so with the sort order fixed, and the thermal-idle
states clarified:
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds