Marc, can you please test these 3 patches on top of my original series? Andy, patches 1 and 2 fix problems that already exist upstream. They were not introduced by my series. Patch 3 fixes an issue with my series where the name of the thermal zone was too long. You can choose to fold it into the patch titled "arm64: dts: msm8998: thermal: Add temperature sensors near major peripherals" or add it as a new patch with a Fixes tag. On Wed, Apr 24, 2019 at 8:00 PM Marc Gonzalez wrote: > > On 23/04/2019 15:10, Marc Gonzalez wrote: > > > Sure. I can give you a Tested-by once everything works on the MEDIABOX I have. > > > > You mention the 4 thermal_zone*/temp failures. > > > > What about the cluster0-mhm-thermal and cluster1-mhm-thermal failures? > > I have synced up with today's linux-next. Thanks for the report. Can you add the attached couple of patches and retest? > # dmesg | grep thermal > Failed to build cluster0-mhm-thermal zone -22 > Failed to build cluster1-mhm-thermal zone -22 This one turned out to be a silly one, The name of the thermal zone is restricted to under 20 characters. Fixed that with patch 3. > # for F in $(find /sys/class/thermal/thermal_zone?/ /sys/class/thermal/thermal_zone??/ -type f); do echo $F; cat $F; done > /sys/class/thermal/thermal_zone0/uevent > /sys/class/thermal/thermal_zone0/trip_point_0_hyst > 2000 > /sys/class/thermal/thermal_zone0/mode > enabled > /sys/class/thermal/thermal_zone0/power/runtime_active_time > 0 > /sys/class/thermal/thermal_zone0/power/runtime_status > unsupported > /sys/class/thermal/thermal_zone8/type > gpu-thermal-bottom > /sys/class/thermal/thermal_zone8/offset > 0 > /sys/class/thermal/thermal_zone8/slope > 1 > /sys/class/thermal/thermal_zone8/trip_point_0_type > hot > /sys/class/thermal/thermal_zone8/k_po > 0 > /sys/class/thermal/thermal_zone8/integral_cutoff > 0 > /sys/class/thermal/thermal_zone8/k_i > 0 > /sys/class/thermal/thermal_zone8/k_pu > 0 > /sys/class/thermal/thermal_zone8/temp > cat: read error: Invalid argument Patch 2 should fix this and thermal_zone9 below. > /sys/class/thermal/thermal_zone8/trip_point_0_temp > 90000 > /sys/class/thermal/thermal_zone9/uevent > /sys/class/thermal/thermal_zone9/trip_point_0_hyst > 2000 > /sys/class/thermal/thermal_zone9/passive > 0 > /sys/class/thermal/thermal_zone9/mode > disabled > /sys/class/thermal/thermal_zone9/power/runtime_active_time > 0 > /sys/class/thermal/thermal_zone9/power/runtime_status > unsupported > /sys/class/thermal/thermal_zone9/power/autosuspend_delay_ms > cat: read error: Input/output error > /sys/class/thermal/thermal_zone9/power/runtime_suspended_time > 0 > /sys/class/thermal/thermal_zone9/power/control > auto > /sys/class/thermal/thermal_zone9/available_policies > step_wise > /sys/class/thermal/thermal_zone9/policy > step_wise > /sys/class/thermal/thermal_zone9/k_d > 0 > /sys/class/thermal/thermal_zone9/sustainable_power > 0 > /sys/class/thermal/thermal_zone9/type > gpu-thermal-top > /sys/class/thermal/thermal_zone9/offset > 0 > /sys/class/thermal/thermal_zone9/slope > /sys/class/thermal/thermal_zone18/available_policies > step_wise > /sys/class/thermal/thermal_zone18/policy > step_wise > /sys/class/thermal/thermal_zone18/k_d > 0 > /sys/class/thermal/thermal_zone18/sustainable_power > 0 > /sys/class/thermal/thermal_zone18/type > battery-thermal This look wrong. TSENS shouldn't have board-specific thermal zones. > /sys/class/thermal/thermal_zone19/k_d > 0 > /sys/class/thermal/thermal_zone19/sustainable_power > 0 > /sys/class/thermal/thermal_zone19/type > skin-thermal And this too. See patch 1 to fix this existing issue (not introduced by my series).