RFR: 8203188: Add JEP-181 support to the Zero interpreter

[David Holmes]

Hi Severin,

On 15/06/2018 7:01 PM, Severin Gehwolf wrote:
> Hi David,
> 
> On Tue, 2018-06-05 at 19:46 +1000, David Holmes wrote:
>> Looks good.
>>
>> I'll push this with the nestmate changes later in the week.
> 
> Any update on this?

A last minute hurdle to overcome. Hopefully early next week now.

David

> Thanks,
> Severin
> 
>> Thanks,
>> David
>>
>> On 5/06/2018 7:40 PM, Severin Gehwolf wrote:
>>> Hi David,
>>>
>>> Thanks for the review!
>>>
>>> On Tue, 2018-06-05 at 14:44 +1000, David Holmes wrote:
>>>> Hi Severin,
>>>>
>>>> On 5/06/2018 1:26 AM, Severin Gehwolf wrote:
>>>>> Hi,
>>>>>
>>>>> Could I please get a review of this change adding support for JEP-181 -
>>>>> a.k.a Nestmates - to Zero. This patch depends on David Holmes'
>>>>> Nestmates implementation via JDK-8010319. Thanks to David Holmes and
>>>>> Chris Phillips for their initial reviews prior to this RFR.
>>>>>
>>>>> Bug: https://bugs.openjdk.java.net/browse/JDK-8203188
>>>>> webrev: http://cr.openjdk.java.net/~sgehwolf/webrevs/JDK-8203188/webrev.02/
>>>>
>>>> src/hotspot/cpu/zero/methodHandles_zero.cpp
>>>>
>>>> The change here seems to be an existing bug unrelated to nestmate
>>>> changes.
>>>
>>> Agreed.
>>>
>>>> IT also begs the question as to what happens in the same
>>>> circumstance with a removed static or "special" method? (I thought I had
>>>> a test for that in the nestmates changes ... will need to double-check
>>>> and add it if missing!).
>>>
>>> It might bomb in the same way (NULL dereference). I'm currently looking
>>> at some other potential issues in this area...
>>>
>>>> src/hotspot/share/interpreter/bytecodeInterpreter.cpp
>>>>
>>>> Interpreter changes seem fine - mirroring what is done elsewhere. You
>>>> can delete these incorrect comments:
>>>>
>>>> 2576         // This code isn't produced by javac, but could be produced by
>>>> 2577         // another compliant java compiler.
>>>>
>>>> That code path is taken in more circumstances than the author of that
>>>> comment realized. :)
>>>
>>> Done.
>>>
>>>>> Testing:
>>>>>
>>>>> Zero on Linux-x86_64 with the following test set:
>>>>>
>>>>> test/jdk/java/lang/invoke/AccessControlTest.java
>>>>> test/jdk/java/lang/invoke/FinalVirtualCallFromInterface.java
>>>>> test/jdk/java/lang/invoke/PrivateInterfaceCall.java
>>>>> test/jdk/java/lang/invoke/SpecialInterfaceCall.java
>>>>> test/jdk/java/lang/reflect/Nestmates
>>>>> test/hotspot/jtreg/runtime/SelectionResolution/InvokeInterfaceICCE.java
>>>>> test/hotspot/jtreg/runtime/SelectionResolution/InvokeInterfaceSuccessTest.java
>>>>> test/hotspot/jtreg/runtime/Nestmates
>>>>>
>>>>> I cannot run this through the submit repo since the main Nestmates
>>>>> patch hasn't yet landed in JDK 11. Currently testing a Zero bootcycle-
>>>>> images build on x86_64. Thoughts?
>>>
>>> FWIW, bootcycle-images build passed on linux x86_64 Zero.
>>>
>>>> I can bundle this in with the nestmate changes when I push them later
>>>> this week. Just send me a pointer to the finalized changeset once its
>>>> finalized. I'll run it all through a final step of testing equivalent
>>>> (actually more than) the submit repo.
>>>
>>> OK, thanks!
>>>
>>> Latest webrev:
>>> http://cr.openjdk.java.net/~sgehwolf/webrevs/JDK-8203188/webrev.03/
>>>
>>> Thanks,
>>> Severin
>>>
>>>> Thanks,
>>>> David
>>>>
>>>>> Thanks,
>>>>> Severin
>>>>>