I just opened a recording where more than 20% of the GC were delayed due to GC Locker (JNI Critical I guess). In terms of time spent in GC, one of the GC locker paused GCs ended up being three times as long as the normal GCs. The rules should at least not have resulted in 0 with the text "The GC Locker did not cause any garbage collections.". Also, that text rubs me the wrong way since the GC isn't "caused" by the GC locker, it is initiated after the call to release the critical array from JNI, due to a GC not being able to be run whilst it was held. I think it should be reworded as part of fixing this bug.