RFR(S) : 8160673 : Jittester: investigate bytecode generation hangs during bytecode tests generation

Igor Ignatyev igor.ignatyev at oracle.com
Tue Jun 26 23:21:12 UTC 2018


http://cr.openjdk.java.net/~iignatyev//8160673/webrev.00/index.html
> 32 lines changed: 6 ins; 12 del; 14 mod; 

Hi all,

could you please review this small fix for jit-tester library?

ByteCodeVisitor loaded generated classes, so ContextDependedClassWriter::getCommonSuperClass would be able to find common supper class using regular Class methods. in some cases, bytrecode for classes isn't yet generated (and hence not loaded) when getCommonSuperClass needs to get information about them (e.g. when generating bytecode for Class1 which is using instances of Class2 and a super class of Class2), which resulted in infinity loop inside asm library.

the fix changes getCommonSuperClass method to use information about classes from jit-tester maintained list of all generated and used types.

JBS: https://bugs.openjdk.java.net/browse/JDK-8160673
webrev: http://cr.openjdk.java.net/~iignatyev//8160673/webrev.00/index.html
testing: generated 1k tests w/ jit-tester

Thanks,
-- Igor


More information about the hotspot-compiler-dev mailing list