RFR: Merge remaining jdk11

Zhengyu Gu zgu at redhat.com
Fri Jun 8 18:49:46 UTC 2018


Okay, upstream actually removed StringTable::rehash() from SP cleanup 
task, just as our patch does.

I guess this just a merging issue then.

Thanks,

-Zhengyu



On 06/08/2018 12:48 PM, Roman Kennke wrote:
> Please push that. We need to carefully check a couple of places (ref-proc, cleanup, etc) anyway. Thanks for looking into it.
> 
> Are we implementing/using the conc cleanup stuff already?
> 
> Thanks, Roman
> 
> Am 8. Juni 2018 18:42:46 MESZ schrieb Zhengyu Gu <zgu at redhat.com>:
>>
>>
>> On 06/08/2018 11:31 AM, Roman Kennke wrote:
>>> Thanks!
>>> I've pushed a minor release build fix after the fact. It was not
>>> committed to the merge changeset when I pushed the main batch. Please
>>> have a look at the concurrentHashMap issue ASAP.
>>
>> Looks like that it still tries to rehash the string table at
>> safepoints,
>> which defeats the purpose of concurrent cleanup ...
>>
>> http://cr.openjdk.java.net/~zgu/shenandoah/stringtable_no_par_cleanup/webrev.00/
>> Seems to fix that. But this looks like an upstream bug (?) Do we know
>> if
>> there is a reported one? I will try to create a reproducer for upstream
>> ...
>>
>> Thanks,
>>
>> -Zhengyu
>>
>>>
>>> Roman
>>>
>>>> Ok!
>>>>
>>>> Thanks,
>>>>
>>>> -Zhengyu
>>>>
>>>> On 06/08/2018 10:36 AM, Roman Kennke wrote:
>>>>> This merges from jdk/jdk up to:
>>>>>
>>>>> changeset:   50805:ec70aad8aebe
>>>>> user:        simonis
>>>>> date:        Fri Jun 08 10:22:26 2018 +0200
>>>>> summary:     8204335: [ppc] Assembler::add_const_optimized
>> incorrect for
>>>>> some inputs
>>>>>
>>>>>
>>>>> There've only been smallish stuff, and a bunch of improvements that
>> I
>>>>> pushed to upstream meanwhile.
>>>>>
>>>>> Testing looks good, except that StringInternCleanup.java appears to
>>>>> trigger a bug in concurrentHashMap when run from fastdebug. I hope
>>>>> Zhengyu can look at that?
>>>>>
>>>>> List of changesets:
>>>>>
>> http://cr.openjdk.java.net/~rkennke/upstream-merge-2018-06-08/outgoing.txt
>>>>>
>>>>>
>>>>> Ok?
>>>>>
>>>>> Roman
>>>>>
>>>
>>>
> 


More information about the shenandoah-dev mailing list