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

[Gustavo Romero]

Hi David,

On 18-08-2016 21:52, David Holmes wrote:
> Hi Gustavo,
> 
> On 19/08/2016 6:48 AM, Gustavo Romero wrote:
>> Hi,
>>
>> On jdk9/hs-comp 37de4195dd18 tip, jdk9/hs-comp/hostpot f932a58df902 tip
>> all jtreg tests found under compiler/rtm/locking (https://goo.gl/VuGX1L)
>> are failing with a message similar like:
>>
>> Exception in thread "main" java.lang.IllegalAccessError: class jdk.test.lib.Utils (in unnamed module @0x5c7fa833) cannot access class jdk.internal.misc.Unsafe (in module java.base) because module
>> java.base does not export jdk.internal.misc to unnamed module @0x5c7fa833
>>
>> Please, find a full log example here: https://goo.gl/1PkJ6x
> 
> I also note this:
> 
> OpenJDK 64-Bit Server VM warning: UseRTMLocking is only available as experimental option on this platform.
> 
> Does the failure mode change if that warning is fixed?

No. On non-experimental platforms, like x64, the tests in question fail
the same way. For instance, that log shows that on x64 with RTM support
tests fail just like on POWER8: https://paste.fedoraproject.org/413588/raw/


>> I understand this failure is related to
>>
>> https://bugs.openjdk.java.net/browse/JDK-8164323
>> https://bugs.openjdk.java.net/browse/JDK-8164324
>>
>> which are duplicates of
>>
>> https://bugs.openjdk.java.net/browse/JDK-8163871
>>
>> that, by its turn, is not publicly available.
>>
>> If it's correct, is this issue understood and expected to be fixed
>> before the rampdown start, on 2016/09/01?
> 
> The above issues pertain to specific non-open tests that were failing in specific ways, so they have been fixed.

Thanks for clarifying that.


> 
> I do not see any such failures of the compiler/rtm/locking tests in our nightly testing, nor any open bugs. If this is not being caused by the "experimental" warning then a new bug will need to be
> filed. If it is caused by the experiment warning a different new bug would need to be filed.

If the platform has no support for RTM it's expected all test runs will
not be treated as a failure, so I guess the nightly build run on a system
without RTM? For example, on x64 without RTM support all tests passed:
https://paste.fedoraproject.org/413597/raw/

Thank you for checking any open bugs that could be related to this issue.

I'll file a bug for that issue if you agree.

Thank you and best regards,
Gustavo