Can it really take over 10k additional lines to fix a pointer issue bug like this, where the problem is that the code uses memory that it has freed already?
I wished they had put the tests in a commit on their own so it would be immediately clear how large the patch itself is.