The option G1TraceReclaimDeadHumongousObjectsAtYoungGC prints some information about live and dead humongous objects during early reclaim of humongous objects (JDK-8048180).
However, it only does so when there is at least one humongous object candidate. This is problematic since this makes getting further information about humongous objects somewhat unreliable.
The problem is that early in the GC phase, we check whether there are any candidates, and if not, skip the actual humongous reclaim phase (because there is nothing that can be reclaimed).
It would be preferable that if G1TraceReclaimDeadHumongousObjectsAtYoungGC is set, do not skip the humongous reclaim phase even if it cannot reclaim anything, but at least print the information.
Also, add humongous object size to the output.
This would help analyzing for JDK-8048179.