On Tue, 20 Oct 2015, Boqun Feng wrote:

>> @@ -93,7 +94,7 @@ static __always_inline void queued_spin_unlock(struct qspinlock *lock)
>>  	/*
>>  	 * smp_mb__before_atomic() in order to guarantee release semantics
>>  	 */
>> -	smp_mb__before_atomic_dec();
>> +	smp_mb__before_atomic();
>>  	atomic_sub(_Q_LOCKED_VAL, &lock->val);
>
>Just be curious, you don't use atomic_sub_release() here on purpose?

atomic_sub() does not imply barriers, so there's no relaxed variants; that's
only for _return() (and such) to the caller.

Thanks,
Davidlohr