RFR: 7902989: jcstress: Add samples for high-level race conditions [v7]

[Aleksey Shipilev]

On Fri, 2 Jul 2021 14:57:11 GMT, Vladimir Sitnikov <vsitnikov at openjdk.org> wrote:

>> Can we resolve that?
>
> Frankly speaking, I think `1: v=200, 1: t1=250, 2: v=200, 2: t2=100, 2: v=t2, 1: v=t1` is too complicated.
> 
> Do you think `200` is important for the test case?
> Would it be better to use `0` starting point?
> What if only `+` was used?
> 
> For instance: "actor 1 uses 10", and "actor2 uses 20". It would be easier than "actor1 uses 100 and actor2 uses 50" because 1-100; 2-50 is harder than 1-10; 2-20;
> 
> suppose the actors only add values to the accumulator, then
> 
> Boring results would be:
> `10, 30, 30: actor1: 0+10; actor2: 10+20`
> `30, 20, 30: actor2: 0+20; actor1: 20+10`
> 
> interesting results would be
> `10, 20, 10; actor1: 0+10; actor2: 0+20; arbiter observes actor1 result`
> `10, 20, 20; actor1: 0+10; actor2: 0+20; arbiter observes actor2 result`

Let's just move on. I prefer this description -- https://github.com/openjdk/jcstress/pull/88#discussion_r663076870 -- and that's it.

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

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