Hi, the JDK-8198418[1] improvements to lambda bootstrapping meant initialization changed around to allow the possibility of a bootstrap race, which made it possible to cause a class loading deadlock when different threads try to initialize classes like SimpleMethodHandle and SpeciesData at the same time.
Making sure the common ancestor, BoundMethodHandle, is initialized using the same means before going into the synchronized block in LambdaForm:createFormsFor seems to be enough to ensure this race can always be resolved peacefully: http://cr.openjdk.java.net/~redestad/8198492/jdk.00/ Testing: ~5000 runs of the affected tests. /Claes [1] https://bugs.openjdk.java.net/browse/JDK-8198418
