We might want to see what call sites allocate objects that are never freed, once we have #6 . Once the bump allocation indexes are randomised (#5), should we just consistently sample calls that are at a few fixed offsets from the beginning of a chunk? Or should we store additional random numbers in the span metadata?
We might want to see what call sites allocate objects that are never freed, once we have #6 . Once the bump allocation indexes are randomised (#5), should we just consistently sample calls that are at a few fixed offsets from the beginning of a chunk? Or should we store additional random numbers in the span metadata?