RFR(S): 8072016: Array copy may cause infinite cycle of deoptimization/compilation

Roland Westrelin roland.westrelin at oracle.com
Mon Feb 9 14:18:15 UTC 2015


http://cr.openjdk.java.net/~roland/8072016/webrev.00/

When an allocation is tightly coupled and one of the test that guards an array copy fail, we record the trap at the bci of the allocation but on next compilation we don’t test for a trap at the bci of the allocation so we can see cycles of deoptimizations/recompilations.

Roland.


More information about the hotspot-compiler-dev mailing list