Re: [PATCH v4] locking/mutex: Prevent lock starvation when spinning is disabled

[Peter Zijlstra]

Please trim your emails..

On Fri, Aug 19, 2016 at 12:57:06PM -0400, Waiman Long wrote:

> >+static inline bool __mutex_trylock_pending(struct mutex *lock)
> >+{
> >+	return atomic_read(&lock->count)>= 0&&
> >+	       atomic_xchg_acquire(&lock->count, -1) == 1;
> >+}
> >+
> 
> Maybe you can make a more general __mutex_trylock function that is used in
> all three trylock attempts in the slowpath. For example,
> 
> static inline bool __mutex_trylock(struct mutex *lock, bool waiter)
> {
>     if (waiter) {
>         return atomic_read(&lock->count) >= 0 &&
>                atomic_xchg_acquire(&lock->count, -1) == 1;
>     } else {
>         return !need_yield_to_waiter(lock) &&
>                !mutex_is_locked(lock) &&
>                ((atomic_xchg_acquire(&lock->count, 0) == 1);
>     }
> }

That seems more messy to me..