Previously badcache used one single mutex for everything, which was causing performance issues. Use one global rwlock for the whole hashtable and per-bucket mutexes. (cherry picked from commit 47e5f5564c)
47e5f5564c
1296e48e11