[15u] RFR (S) 8249192: MonitorInfo stores raw oops across safepoints

[Aleksey Shipilev]

On 9/14/20 9:51 AM, David Holmes wrote:
> Hi Aleksey,
> 
> On 14/09/2020 5:29 pm, Aleksey Shipilev wrote:
>> On 9/9/20 9:10 AM, Aleksey Shipilev wrote:
>>> Original bug:
>>>      https://bugs.openjdk.java.net/browse/JDK-8249192
>>>      https://hg.openjdk.java.net/jdk/jdk/rev/61f9028f360d
>>>
>>> 15u variant:
>>>      https://cr.openjdk.java.net/~shade/8249192/webrev.15u.01/
>>>
>>> It differs from jdk/jdk variant by a conflict in jvmtiEnvBase.cpp,
>>> where removed "ResourceMark rm"
>>> does not have "current_thread" argument yet. It was added later with
>>> JDK-8247729:
>>>      https://hg.openjdk.java.net/jdk/jdk/rev/f8a9be0f9e1a#l2.82
>>>
>>> Testing: tier{1,2} with and without Shenandoah [which seems to find
>>> some of these bugs]
>>
>> Um.
>>
>> I see there is a "backport" to 15.0.2 already:
>>     https://bugs.openjdk.java.net/browse/JDK-8251485
>>
>> ...yet there is no changeset at all:
>>     https://hg.openjdk.java.net/jdk-updates/jdk15u/log?rev=8251485
>>
>> What's up with that? Rob, Thomas?
> 
> I think an oversight on our (Oracle) end. It went into the 15u CPU repo
> instead of the open repo.

Ah. Well, I can push my versions of 8249192 (once reviewed here) and 8251118 to jdk-updates/jdk15u 
then. But it would be nice if someone from Oracle would compare the 8249192 changeset above with 
what is pushed to 15u CPU repo.

-- 
Thanks,
-Aleksey