There are some mistakes in the thermal trip points currently used on
X1E80100. Several of the critical trip points are too high, so the hardware
will trigger an emergency reset before the kernel has the chance to
shutdown cleanly. We're also missing GPU cooling in the device tree, which
prevents running certain GPU-intensive applications without reaching the
critical shutdown temperatures. The CPU on the other hand throttles itself
automatically, so we can just drop the passive trip points there.
This series addresses these problems. It's quite similar to Neil's thermal
zone rework for SM8650 [1], which provides some extra background on the
mechanisms behind the throttling.
[1]: https://lore.kernel.org/linux-arm-msm/20250203-topic-sm8650-thermal-cpu-idle-v4-0-65e35f307301@linaro.org/T/
Signed-off-by: Stephan Gerhold <stephan.gerhold@linaro.org>
---
Stephan Gerhold (4):
arm64: dts: qcom: x1e80100: Fix video thermal zone
arm64: dts: qcom: x1e80100: Apply consistent critical thermal shutdown
arm64: dts: qcom: x1e80100: Add GPU cooling
arm64: dts: qcom: x1e80100: Drop unused passive thermal trip points for CPU
arch/arm64/boot/dts/qcom/x1e80100.dtsi | 679 ++++++++-------------------------
1 file changed, 160 insertions(+), 519 deletions(-)
---
base-commit: c177fed7617d6306541305e93e575c0c01600ff0
change-id: 20250218-x1e80100-thermal-fixes-774d08de4e3c
Best regards,
--
Stephan Gerhold <stephan.gerhold@linaro.org>