Re: [patch 05/15] Generic Mutex Subsystem, mutex-core.patch

[Ingo Molnar <mingo@elte.hu>]

* Oleg Nesterov <oleg@tv-sign.ru> wrote:

> > +__mutex_wakeup_waiter(struct mutex *lock __IP_DECL__)
> > +{
> > +	struct mutex_waiter *waiter;
> > ...
> > +	if (!waiter->woken) {
> > +		waiter->woken = 1;
> > +		wake_up_process(waiter->ti->task);
> > +	}
> 
> Is it optimization? If yes - why? From mutex.h:
> 
> 	- only one task can hold the mutex at a time
> 	- only the owner can unlock the mutex
> 
> So, how can this help?

yes, it's an optimization. I've removed it from the latest queue because 
it didnt trigger all that often, but the optimization is valid: while we 
have a 'waiter in flight', another (fast) task might grab the mutex, and 
might release it - in which case it could attempt to wake up the waiter 
again - which this flag optimizes.

	Ingo