On Thu, Jan 08, 2015 at 01:07:02PM -0800, Mike Stump wrote:
> On Jan 8, 2015, at 11:29 AM, Jakub Jelinek <[email protected]> wrote:
> > I disagree. Busy waiting of this kind is not appropriate for the test
> > suite,
>
> What busy waiting, there is none in the last version of the patch?
It was still while (... != i - 1), wasn't it?
While pthread_barrier_wait/futex would typically only wake you up when
needed.
> > tsan can't intercept the calls that you do through dlsym, because you
> > explicitly bypass tsan in that case.
>
> Ah, yes, right, I had pthread_barrier_wait on the brain, sorry. The direct
> use of it would be problematic. The dlopen use of it, is safe.
>
> So, that removes the objection I had to his patch. Jakub, since he has a
> complete solution to the problem submitted with all the test cases fixed, I
> think it should go in.
>
> Any objections to approving it now?
LGTM.
Jakub
