Re: [PATCH] thermal: samsung: acpm-tmu: Fix spurious CPU throttling
From: Christian Loehle
Date: Thu Nov 19 2026 - 08:24:41 EST
On 11/18/26 14:01, Tudor Ambarus wrote:
>
>
> On 11/18/26 3:28 PM, Tudor Ambarus wrote:
>> Would something along the lines work?
>
> Thinking more about it, caching the window instead of the raw
> thresholds is simpler and closer to your v1, as inten derives from
> low/high anyway. Maybe this:
>
> diff --git a/drivers/thermal/samsung/acpm-tmu.c b/drivers/thermal/samsung/acpm-tmu.c
> index 7c30401f8fc7..8422a7c970e7 100644
> --- a/drivers/thermal/samsung/acpm-tmu.c
> +++ b/drivers/thermal/samsung/acpm-tmu.c
> @@ -39,6 +39,10 @@
>
> #define GS101_REG_INTPEND(i) ((i) * 0x50 + 0xf8)
>
> +#define ACPM_TMU_NUM_THRESHOLDS 8
> +/* Threshold 7 throttles the CPUs in hardware, regardless of the INTEN mask. */
> +#define ACPM_TMU_HW_TRIP_THRESHOLD 7
> +
> enum {
> P0_INTPEND,
> P1_INTPEND,
> @@ -70,6 +74,9 @@ struct acpm_tmu_sensor {
> struct acpm_tmu_priv *priv;
> struct mutex lock; /* protects sensor state */
> enum thermal_device_mode mode;
> + /* Last programmed window, protected by the thermal zone lock */
> + int low;
> + int high;
> bool enabled;
> };
>
> @@ -228,7 +235,8 @@ static int acpm_tmu_get_temp(struct thermal_zone_device *tz, int *temp)
> }
>
> static int acpm_tmu_update_thresholds(struct acpm_tmu_sensor *sensor,
> - u8 thresholds[2], u8 inten)
> + u8 thresholds[ACPM_TMU_NUM_THRESHOLDS],
> + u8 inten)
> {
> struct acpm_tmu_priv *priv = sensor->priv;
> struct acpm_handle *handle = priv->handle;
> @@ -249,7 +257,7 @@ static int acpm_tmu_update_thresholds(struct acpm_tmu_sensor *sensor,
> }
>
> ret = ops->set_threshold(handle, mbox_chan_id, acpm_sensor_id,
> - thresholds, 2);
> + thresholds, ACPM_TMU_NUM_THRESHOLDS);
> if (!ret) {
> ret = ops->set_interrupt_enable(handle, mbox_chan_id,
> acpm_sensor_id, inten);
> @@ -267,12 +275,22 @@ static int acpm_tmu_update_thresholds(struct acpm_tmu_sensor *sensor,
> return ret;
> }
>
> -static int acpm_tmu_set_trips(struct thermal_zone_device *tz, int low, int high)
> +static int acpm_tmu_get_crit_temp(struct thermal_trip *trip, void *arg)
> {
> - struct acpm_tmu_sensor *sensor = thermal_zone_device_priv(tz);
> - struct acpm_tmu_priv *priv = sensor->priv;
> - struct device *dev = priv->dev;
> - u8 thresholds[2] = {};
> + int *temp = arg;
> +
> + if (trip->type == THERMAL_TRIP_CRITICAL &&
> + trip->temperature != THERMAL_TEMP_INVALID)
> + *temp = min(*temp, trip->temperature);
> +
> + return 0;
> +}
> +
> +static int acpm_tmu_program_thresholds(struct acpm_tmu_sensor *sensor,
> + int low, int high, int crit)
> +{
> + struct device *dev = sensor->priv->dev;
> + u8 thresholds[ACPM_TMU_NUM_THRESHOLDS] = {};
> u8 inten = 0;
> int ret;
>
> @@ -288,17 +306,61 @@ static int acpm_tmu_set_trips(struct thermal_zone_device *tz, int low, int high)
> inten |= BIT(1);
> }
>
> + /*
> + * The firmware programs all thresholds at once. Set the hardware trip
> + * to the critical trip temperature, so that it never fires below the
> + * point where the thermal core shuts the system down.
> + */
> + thresholds[ACPM_TMU_HW_TRIP_THRESHOLD] =
> + clamp_val(crit / MILLIDEGREE_PER_DEGREE, 0, 255);
> +
> ret = pm_runtime_resume_and_get(dev);
> if (ret)
> return ret;
>
> ret = acpm_tmu_update_thresholds(sensor, thresholds, inten);
> + if (!ret) {
> + sensor->low = low;
> + sensor->high = high;
> + }
>
> pm_runtime_put_autosuspend(dev);
>
> return ret;
> }
>
> +static int acpm_tmu_set_trips(struct thermal_zone_device *tz, int low, int high)
> +{
> + struct acpm_tmu_sensor *sensor = thermal_zone_device_priv(tz);
> + int crit = INT_MAX;
> +
> + /*
> + * Use the lowest critical trip temperature. If the zone has no
> + * critical trip, crit stays INT_MAX and is clamped to the highest
> + * threshold value.
> + */
> + for_each_thermal_trip(tz, acpm_tmu_get_crit_temp, &crit);
> +
> + return acpm_tmu_program_thresholds(sensor, low, high, crit);
> +}
> +
> +static int acpm_tmu_set_trip_temp(struct thermal_zone_device *tz,
> + const struct thermal_trip *trip, int temp)
> +{
> + struct acpm_tmu_sensor *sensor = thermal_zone_device_priv(tz);
> +
> + if (trip->type != THERMAL_TRIP_CRITICAL)
> + return 0;
> +
> + /*
> + * The thermal core skips .set_trips() when the low/high window does
> + * not change, so update the hardware trip threshold here, keeping the
> + * last programmed window.
> + */
> + return acpm_tmu_program_thresholds(sensor, sensor->low, sensor->high,
> + temp);
> +}
> +
> static int acpm_tmu_change_mode(struct thermal_zone_device *tz,
> enum thermal_device_mode mode)
> {
> @@ -329,6 +391,7 @@ static const struct thermal_zone_device_ops acpm_tmu_sensor_ops = {
> .change_mode = acpm_tmu_change_mode,
> .get_temp = acpm_tmu_get_temp,
> .set_trips = acpm_tmu_set_trips,
> + .set_trip_temp = acpm_tmu_set_trip_temp,
> };
>
> static bool acpm_tmu_clear_pending_irq(struct acpm_tmu_sensor *sensor)
> @@ -479,6 +542,8 @@ static int acpm_tmu_probe(struct platform_device *pdev)
> mutex_init(&sensor->lock);
> sensor->group = &data->sensor_groups[i];
> sensor->priv = priv;
> + sensor->low = -INT_MAX;
> + sensor->high = INT_MAX;
>
> tzd = devm_thermal_of_zone_register(dev, i, sensor,
> &acpm_tmu_sensor_ops);
>
Yup something along those lines would work with some minor fixes if we wanna
handle program failures and restore/undo partials.
The alternative is of course to just hardcode slot 7, downstream uses 115C
as a default value (supplied by DT trip point 'hot'). Or do that as a fix and
then follow-up with a proper wire-up? What do you prefer?