kthread_exit(9) unexpectedness

[John Baldwin]

On Tuesday 02 December 2008 08:45:25 pm Lawrence Stewart wrote:
> > might even be better to add a dedicated condvar to 'struct thread' in 8.x 
> > that is used for the wakeup and do the wakeup on that rather than the 
thread 
> > pointer to be honest.
> > 
> 
> What are the pros/cons of using mtx_sleep/wakeup vs cv_wait/cv_broadcast?

Forces you to use explicit wait channels as opposed to some of the problems we 
have now with 3-4 places sleeping on proc pointers for example.

-- 
John Baldwin