The young_index_in_cset is used to calculate per-region survivor rates in young generation regions.

This value is precalculated for all regions that need per-region survival rate calculation at the start of GC, and then used to index some per-thread table.

For some unknown reason it uses the value "-1" for "no survival rate group", which means that for every reference to be copied the code needs to increment it.

While this is no very significant issue, it is some useless calculation in one of the hottest loops in the GC.

Use "0" as "no survival rate group" indicate instead to avoid this additional overhead.

This is afaik a day-one issue.