[15u] MonitorInfo bugs: JDK-8249192 and JDK-8251118

[Harold Seigel]

Hi Aleksey,

The fix for JDK-825118, in JDK-16, was to remove the HandleMark from 
biasedLocking.cpp that was added as part of the JDK-16 fix for 
JDK-8249192.  But, my original 15u backport for JDK-8249192 did not 
contain the new HandleMark in biasedLocking.cpp.  So, there was no 
reason to backport 825118 because there was no HandleMark to remove.

I plan to push these changes into the OpenJDK-15u repo in the next few 
days.  Sorry for all the confusion.

Harold

On 11/30/2020 10:02 AM, Harold Seigel wrote:
> Hi Aleksey,
>
> The 15u backports for 8249192 and 8251118 were pushed (by me) to the 
> wrong repo but will be included in the JDK 15.0.2 release.
>
> Harold
>
> On 11/30/2020 8:35 AM, Robert McKenna wrote:
>> W.r.t. 8251118 - I don't see a record of a 15u push after you 
>> requested fix approval.
>>
>>     -Rob
>>
>> On 30/11/2020 13:33, Robert McKenna wrote:
>>> Thomas, I don't see any obvious reason why this was not pushed to 
>>> the OpenJDK repo. Can you rectify this as soon as you get the chance.
>>>
>>> Aleksey, apologies for missing this, and thanks for the heads up. In 
>>> future please include the responsible engineer for the bug in the 
>>> mail (cc'ing the alias and me)
>>>
>>>     -Rob
>>>
>>> On 25/11/2020 08:34, Aleksey Shipilev wrote:
>>>> Rob,
>>>>
>>>> On 11/18/20 2:17 PM, Aleksey Shipilev wrote:
>>>>> On 11/11/20 11:00 AM, Aleksey Shipilev wrote:
>>>>>> Hi again,
>>>>>>
>>>>>> I asked this question in the relevant RFR threads, but they 
>>>>>> apparently went unnoticed. Therefore,
>>>>>> let me ask this in a new thread.
>>>>>>
>>>>>> This is about the following issues:
>>>>>>      8249192: MonitorInfo stores raw oops across safepoints
>>>>>>      8251118: BiasedLocking::preserve_marks should not have a 
>>>>>> HandleMar
>>>>>>
>>>>>> At the time 15.0.1 was open for pushes, it had been stated that 
>>>>>> both issues were pushed to some CPU
>>>>>> repo. I was under impression it was to appear in 15.0.1 release, 
>>>>>> but it did not. So, there are no
>>>>>> changes in current 15u:
>>>>>> https://hg.openjdk.java.net/jdk-updates/jdk15u/log?rev=8249192
>>>>>> https://hg.openjdk.java.net/jdk-updates/jdk15u/log?rev=8251118
>>>>>>
>>>>>> There is still the "Resolved" backport in 15.0.2 for one of the 
>>>>>> issues:
>>>>>>       https://bugs.openjdk.java.net/browse/JDK-8251485
>>>>>>
>>>>>> ...and there is no resolved backport for another.
>>>>>>
>>>>>> So, is there a private 15.0.2 CPU repo within Oracle? Are we 
>>>>>> expecting both issues to come with
>>>>>> 15.0.2 CPU push? Even in this case, can we push the change to 
>>>>>> open 15.0.2, so we would be able to
>>>>>> test it in open repositories before next CPU hits?
>>>>>
>>>>> Friendly reminder.
>>>>
>>>> Another friendly reminder.
>>>>
>>>
>>