977027_en-US

[#1]

ls1043a - should thermal_zone5 be active in the Linux BSP?

I work on with the reference board for ls1043 processor ls1043ardb and having an issue with the thermal subsystem. The error I see is:

 [ 116.704475] thermal thermal_zone5: critical temperature reached (104 C), shutting down

It seems a bit sporadic but when it comes it comes quite soon after boot. It does not come all the time. When the system is stable the temperature from thermal_zone5 is always 0.

The dts file fsl-ls1043a.dtsi have thermal_zone0 - thernal_zone5 active. (fsl-ls1043a.dtsi\freescale\dts\boot\arm64\arch - qoriq-components/linux - Linux Tree for QorIQ suppo... )

The question is if ls1043 should have thermal_zone5 active? Looking in to the "QorIQ LS1043A Reference Manual, Rev. 5, 04/2019" chapter 35.1.1 "Local temperature sensor placement" I can read a table saying that the ls1043 have Temperature sensor ID 0-4 while ID 5-15 is marked as reserved. Is it correct that hte dtsi file enables thermal_zone5 and is it available in ls1043?

Thanks,
/Peter

Re: ls1043a - should thermal_zone5 be active in the Linux BSP?

Hello,

We are seeing a similar issue on an LS1043A platform running Linux 4.19.68.

The system occasionally reports:

thermal thermal_zone5: critical temperature reached (104 C), shutting down

One observation is that thermal zones 0-4 generally track each other closely, while thermal_zone5 often reports significantly different values and behaves differently from the other zones.

Prior to shutdown, the thermal zones report values similar to:

thermal_zone0: 75000

thermal_zone1: 76000

thermal_zone2: 76000

thermal_zone3: 75000

thermal_zone4: 74000

thermal_zone5: 0

In a previous reply, NXP mentioned that thermal_zone5 values were indeterminate and that a fix would be provided in a future LSDK release.

Could you please confirm whether this issue was ever fixed, and if so, which LSDK/kernel release contains the fix?

Thanks.

Re: ls1043a - should thermal_zone5 be active in the Linux BSP?

I am also facing the similar issue on the latest LSDK 20.12, Kernel 5.4.47. 

[ 2115.927267] thermal thermal_zone1: critical temperature reached (85 C), shutting down
[ 2116.951246] thermal thermal_zone1: critical temperature reached (85 C), shutting down
[ 2117.975285] thermal thermal_zone1: critical temperature reached (85 C), shutting down

Could you please let me know if any workaround is available to fix this?

What is the root cause of this issue?

Re: ls1043a - should thermal_zone5 be active in the Linux BSP?

We acknowledge the issue, currently the values reported by thermal zone 5 are indeterminate and can cause such issue at any random point We will be providing an official fix in our upcoming LSDK releases.

For now you can remove thermal zone 5 from your dtsi  and see if it helps