RFR: 7902982: jcstress: Add samples for some mutex algorithms [v3]

[Michael Mirwaldt]

On Tue, 29 Jun 2021 13:46:44 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

>> What about this?
>>         taken1 = true; // critical section starts
>>         r.r1 = taken2;
>>         taken1 = false; // critical section ends
>
> Maybe we should recast the whole thing into atomicity test. This would alleviate any questions about how `taken` fields are actually working. I.e.:
> 
> 
> int v;
> 
> @Actor 
> public void actor1(II_Result r) {
>   { // critical section (broken)
>     r.r1 = ++v;
>   }
> }
> 
> @Actor 
> public void actor2(II_Result r) {
>   { // critical section (broken)
>      r.r2 = ++v;
>   }
> }
> 
> 
> ...then check that `(1, 1)` case is not possible in case of properly working lock.

That's an interesting idea. If I get you right, you use here the fact that incrementing v is not an atomic operation. I guess it works because if both actors enter the critical section at the same time, they will both read 0 for v and increment it to 1 for their results.

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

PR: https://git.openjdk.java.net/jcstress/pull/85