[LTP] [PATCH v7 3/6] Refactor mqns_03 using new LTP API

[Petr Vorel]

> Hi!
> > > I do not think that we need atomicity here, the cleanup code does not
> > > run concurently at all as the cleanup in the parent is triggered after
> > > the child did exit. I suppose that instead we need to set the mq_freed
> > > to be volatile because it's shared memory which may change at any
> > > change, so we need to tell that to the compiler.
> > That's fine, but I followed suggestions in the reviews. I think that 
> > having 3 people reviewing the same patch doesn't help the development 
> > process. Now I'm not sure who I should follow :-)

> It's actually the other way around, the more people look at the code the
> better, at least that way we have potential to catch more problems
> earlier. And if the reviewers disagree, let them fight for the right
> answer.

+1

Kind regards,
Petr

> I think that in this case this all can be actually simplified and we can
> get rid of the mq_freed flag as I tried to outline below.