IllegalAccessError: class jdk.test.lib.Utils cannot access class jdk.internal.misc.Unsafe

[Dmitry Samersoff]

Gustavo,

Thank you for detailed report.

> Should I open a bug for it?

Yes. Please open CR to be able to track down the issue.

-Dmitry

On 2016-08-25 02:04, Gustavo Romero wrote:
> Hi Dmitry,
> 
> On 19-08-2016 09:51, Dmitry Samersoff wrote:
>> 1. Make sure you have fresh jdk9 workspace
>> 2. Make sure you have recent jtreg
> 
> 1. Checked. It's purged and I'm using only the forest tips;
> 2. I'm using the most recent found on cloudbees ( > b03 ) [1].
> 
> I'm testing on jdk9/hs-comp repo (instead of jdk9/jdk9) because I'm
> assuming that most fresh changesets on hotpot will be there. Am I wrong?
> (Did you mean jdk9/jdk9 by "jdk9 workspace"?)
> 
> Anyway, the same problem occurs on jdk9/jdk9:
> 
> Error log for jdk9/jdk9:
> https://paste.fedoraproject.org/413618/raw/
> 
> Error log for jdk9/hs-comp:
> https://paste.fedoraproject.org/413619/raw/
> 
> 
>>
>> Please check:
>>
>> https://bugs.openjdk.java.net/browse/JDK-8154258
>> https://bugs.openjdk.java.net/browse/JDK-8155957
> 
> Thank you for pointing out references on that issue. As these patches are
> already on jdk9/jdk9 I'm assuming that they do not fix the issue directly.
> 
> I'm not much familiar with the transition from class path to the module
> system. I  was able to workaround the issue using the following patch
> (which obviously is not the right way to fix it - but I had to use it since
> I'm primarily focused on the RTM tests internal functionality):
> 
> diff -r 84edc0512b0e src/java.base/share/classes/module-info.java
> --- a/src/java.base/share/classes/module-info.java      Thu Aug 11 16:22:09 2016 -0700
> +++ b/src/java.base/share/classes/module-info.java      Wed Aug 24 18:58:19 2016 -0400
> @@ -152,6 +152,8 @@
>          jdk.dynalink,
>          jdk.jartool,
>          jdk.jlink;
> +    exports jdk.internal.misc;
> +/*
>      exports jdk.internal.misc to
>          java.corba,
>          java.desktop,
> @@ -170,6 +172,7 @@
>          jdk.scripting.nashorn,
>          jdk.unsupported,
>          jdk.vm.ci;
> +*/
>      exports jdk.internal.perf to
>          java.desktop,
>          java.management,
> 
> Now they run fine, for instance:
> https://paste.fedoraproject.org/413621/raw/
> 
> (It fails but it's not the run, but the test itself. The remaining tests
> also run fine now)
> 
> Should I open a bug for it?
> 
> 
> Thanks a lot!
> 
> Best regards,
> Gustavo
> 
> [1] https://adopt-openjdk.ci.cloudbees.com/job/jtreg/lastSuccessfulBuild/artifact/
> 


-- 
Dmitry Samersoff
Oracle Java development team, Saint Petersburg, Russia
* I would love to change the world, but they won't give me the sources.