Ok, that makes sense. Can you explain why you think it's better to return EBUSY rather than blocking on the mutex in -destroy, instead of relying on the user checking the return code?
> On 17 Mar 2016, at 08:05, Paul Irofti <[email protected]> wrote: > > On Wed, Mar 16, 2016 at 10:16:45PM +0000, Kári Tristan Helgason wrote: >> This is very similar to the patch I had come up with for this issue. >> >> What is the reason for not using the local variable b in the trylock >> in barrier_destroy()? > > I want that assignment to happen once we are holding the lock and we're > sure *barrier is valid. Otherwise another thread could clear it between > the copy and the mutex lock.
