Hi,
We are running Linux on Beaglebone, and are seeing an issue we would like some input on.
We are reading the CPU temperature from /sys/class/hwmon/hwmon0/device/temp1_input, whose temperature source is bandgap_ctrl in the AM335x as far as I know. This usually works, but sometimes and on some boards the value suddenly reads 127 degrees celcius. Whenever this happens, any further reads will give a value of 127 degrees until the board is rebooted.
The boards seem to work just fine, and getting a correct temperature is not very important to us. However, we do need to know if a reading of 127 can be a symptom of something bad, or if getting into the stuck state can be the cause of other issues down the line. So:
1) What can cause the temp sensor to suddenly return 127 (which is a slightly suspicious value)?
2) Would you expect that using single-sample mode instead of continuous mode (which the Linux kernel uses) would help avoid the sensor value becoming “stuck” forever?
Thanks for any insight,
Oyvind
