RFR: 7902989: jcstress: Add samples for high-level race conditions [v5]

[Michael Mirwaldt]

On Thu, 1 Jul 2021 15:27:26 GMT, Vladimir Sitnikov <vsitnikov at openjdk.org> wrote:

>> I think the synchronization with volatile is just too weak. It keeps the order of the statements, yes, that's right,
>> However, it doesn't force orders with the desired result without the effect of the read-modify-write sequence. 
>> We can remove volatile if you like.
>
>> the order of the statements
> 
> The order of statements is "program order" which always holds no matter what ("final field update" is one of the exceptions, yet updating a final field is really weird)
> 
>> it doesn't force orders
> 
> Please use jls-17 when you reason on the acceptable by Java memory model values.
> See https://shipilev.net/blog/2014/jmm-pragmatics/ ,
> 
>> In order to test if an execution is plausible under JMM, you need to see if it passes all the requirements. Note, however, that you can quickly branch-and-bound the set of considered executions based on the failure of a particular test....

Can we resolve that now, @vlsi?

-------------

PR: https://git.openjdk.java.net/jcstress/pull/88