The bug affects all ICM20602 sensors from invensense, but it is extreamly random in its nature.
it simply corrupts the temperature reading on the sensor in the FIFO data. this is very easy to detect, and a soft reset of the fifo throws out the bad data and doesn’t continue showing issues.
the issue was more about how we reacted to the data, than the issue itself. we over reacted by making this an “internal error” when it should smply have reset the data… however, at the time, more effort was needed to ensure there was no risk to fligh, and there is no risk to flight.