With JDK-6898462, OOME when scalar replaced objects are reallocated during deoptimization causes all interpreter frames for the deoptimized compiled frame to be popped. One suggestion to improve the behaviour in this case is to reserve cold, never-used memory to recharge a thread's TLAB in some TBD gated manner, to allow catch clauses a better chance to complete.
- relates to
-
JDK-8227309 Handle OOM when rematerializing objects less surprising
-
- Open
-
-
JDK-6898462 The escape analysis with G1 cause crash assertion src/share/vm/runtime/vframeArray.cpp:94
-
- Resolved
-