RFR 8170870: Fix synchronization of access to PackageEntryTables and ModuleEntryTables

[serguei.spitsyn at oracle.com]

On 12/22/16 03:28, serguei.spitsyn at oracle.com wrote:
> Hi Harold and David,
>
>
> On 12/21/16 15:48, David Holmes wrote:
>> Hi Harold,
>>
>> On 22/12/2016 12:00 AM, harold seigel wrote:
>>> Hi,
>>>
>>> Please review this fix for bug JDK-8170870.  The fix adds a missing
>>> load_ptr_acquire and fixes some asserts.
>>
>> Sorry but I do not understand these changes. The existing asserts 
>> show that the code is executed at a safepoint or with the Module_lock 
>> held so how does this suddenly become lock-free? If you are "simply" 
>> intending this code to be allowed to be called other than at a 
>> safepoint or with the Module_lock held, then it may require a lot 
>> more than a load_ptr_acquire to make it safe!
>
> I'm puzzled too.
>
>
>>
>> The bug synopsis and description do not describe the problem being 
>> addressed - or even explain what the problem is, or the proposed 
>> solution.
>
> I guess, the suggestion was from Markus in one of the comments.
. . .

Please, do not follow my attempt to discuss this confidential comment on 
the open mailing list.
Sorry for my mistake.

Thanks,
Serguei

>
>>
>>> It also removes the
>>> exported_pending_delete array, improving the management of package
>>> qualified export lists.
>>
>> That seems to be a completely unrelated issue which should have its 
>> own bug IMHO.
>
> I agree, separating this into a different bug would simplify things.
>
>
> Thanks,
> Serguei
>
>
>>
>> Thanks,
>> David
>>
>>> Open Webrev:
>>> http://cr.openjdk.java.net/~hseigel/bug_8170870/webrev/index.html
>>>
>>> JBS Bug: https://bugs.openjdk.java.net/browse/JDK-8170870
>>>
>>> The fix was tested with the JCK Lang and vm tests, the JTreg hotspot,
>>> java/io, java/lang, java/util and other JTReg tests, and the NSK
>>> co-located and non-colocated tests.
>>>
>>> Thanks, Harold
>>>
>