Randomizing IGVN worklist access might help to trigger some bugs more often:
- Node* n = _worklist.pop();
+ // Randomize access to worklist
+ int idx = (os::random() % _worklist.size());
+ Node* n = _worklist.remove(idx);
I gave above code a quick try and executed tier1 - tier5: No new failures but it may still pay off to introduce this for stress testing.
- relates to
-
JDK-8253748 StressIGV tests fail with release VMs
- Resolved
-
JDK-8253766 compiler/debug/TestGenerateStressSeed.java fails with NoSuchFileException
- Closed
-
JDK-8253767 compiler/debug/TestStressIGVN.java fails with RuntimeException: got the same IGVN trace for different seeds: expected to not equal
- Closed
-
JDK-8254810 gtest for os::random() fails on windows x64
- Closed
-
JDK-8325482 Test that distinct seeds produce distinct traces for compiler stress flags
- Open
-
JDK-8253765 C2: Control randomization in StressLCM and StressGCM
- Resolved
-
JDK-8272703 StressSeed should be set via FLAG_SET_ERGO
- Resolved
-
CODETOOLS-7902768 jcstress should support -XX:+StressIGVN
- Closed
-
CODETOOLS-7903041 jcstress: Provide explicit StressSeed when available
- Closed
-
JDK-8314836 4 compiler/debug tests ignore tested VM flags
- Open
-
JDK-8311279 TestStressIGVNAndCCP.java failed with different IGVN traces for the same seed
- Resolved