Whenever one uses 2 thermostats in TPCONTROL, the temperature reported in CHARMM log is that of the FIRST THERMOSTAT (rather than the temperature of the whole system which one would expect). AVER> and FLUC> values are of the same thing.

Also, degrees of freedom for the subsystems are calculated in a strange way, as described in this thread

So far I'm not sure if the latter is a bug, but the former certainly is.