The goal of the whole series is to have logging around linknode fixup. It's a slow operation, and we have two heuristics to make it faster. Unfortunately we have no idea about how well these heuristics actually help. This series of diffs aims to fix this problem. After this series is landed and logging is enabled, I hope that we can find out if we really these heuristics at all.
Let's log more data to make it easier to analyze where and how often does
linknode adjusing happen.