RFR 8258945: test/jdk/tools/jlink/JLinkReproducibleTest.java fails randomly when cds is disabled

[Langer, Christoph]

Hi hedongbo,

I went again through this topic and I don't see any other fix at the moment than to exclude this test in 11u. It would need JEP 341 to be backported which is probably out of scope.

I'll file an exclusion.

Best regards
Christoph

> -----Original Message-----
> From: hedongbo <hedongbo at huawei.com>
> Sent: Dienstag, 19. Januar 2021 08:04
> To: Langer, Christoph <christoph.langer at sap.com>
> Cc: jdk-updates-dev at openjdk.java.net
> Subject: RE: RFR 8258945: test/jdk/tools/jlink/JLinkReproducibleTest.java
> fails randomly when cds is disabled
> 
> Thanks Christoph!
> 
> Glad to hear the news and look forward to you can post a PR ASAP.
> 
> 
> Thanks,
> dongbohe
> 
> -----Original Message-----
> From: Langer, Christoph [mailto:christoph.langer at sap.com]
> Sent: Friday, January 15, 2021 1:15 AM
> To: hedongbo <hedongbo at huawei.com>; jdk-updates-
> dev at openjdk.java.net
> Subject: RE: RFR 8258945: test/jdk/tools/jlink/JLinkReproducibleTest.java
> fails randomly when cds is disabled
> 
> Hi dongbohe,
> 
> we also see this issue in our nightly testing and are interested in fixing this.
> 
> However, backporting https://bugs.openjdk.java.net/browse/JDK-8202951
> doesn't seem appropriate for backporting to 11u to me. Don't know if other
> people would have a different opinion - if so, please chime in.
> 
> We should find a different solution, maybe we can exclude the test from
> running when CDS is disabled by some jtreg directive?
> 
> Thanks
> Christoph
> 
> > -----Original Message-----
> > From: jdk-updates-dev <jdk-updates-dev-retn at openjdk.java.net> On
> > Behalf Of hedongbo
> > Sent: Donnerstag, 14. Januar 2021 02:54
> > To: jdk-updates-dev at openjdk.java.net
> > Subject: RFR 8258945: test/jdk/tools/jlink/JLinkReproducibleTest.java
> > fails randomly when cds is disabled
> >
> > Hi,
> > Can someone help me look at this problem?
> >
> > BugID: http://bugs.openjdk.java.net/browse/JDK-8258945
> >
> > As analyzed in JDK-8258945, this is caused by the fact that the object
> > hashcode of enum may be different in different JVMs. It is not a good
> > idea to modify the calculation of the hashcode of the
> > ModuleDescriptor, and it is also difficult.  Therefore, I think we can
> > backport JDK-
> > 8202951<https://bugs.openjdk.java.net/browse/JDK-8202951> to jdk11 to
> workarounds.
> >
> > Any suggestions? If backport is needed, I will start this work.
> >
> >
> >
> > Thanks,
> > dongbohe