Re: [PATCH v2] introduce atomic_pointer to fix a race condition in cancelable mcs spinlocks

["Paul E. McKenney" <paulmck@linux.vnet.ibm.com>]

On Mon, Jun 02, 2014 at 03:44:48PM -0700, Eric Dumazet wrote:
> On Mon, 2014-06-02 at 15:08 -0700, Paul E. McKenney wrote:
> 
> > diff --git a/kernel/rcu/srcu.c b/kernel/rcu/srcu.c
> > index c639556f3fa0..c0120279dead 100644
> > --- a/kernel/rcu/srcu.c
> > +++ b/kernel/rcu/srcu.c
> > @@ -295,12 +295,15 @@ EXPORT_SYMBOL_GPL(cleanup_srcu_struct);
> >  int __srcu_read_lock(struct srcu_struct *sp)
> >  {
> >  	int idx;
> > +	unsigned long *lp;
> >  
> >  	idx = ACCESS_ONCE(sp->completed) & 0x1;
> >  	preempt_disable();
> > -	ACCESS_ONCE(this_cpu_ptr(sp->per_cpu_ref)->c[idx]) += 1;
> > +	lp = this_cpu_ptr(&sp->per_cpu_ref->c[idx]);
> > +	ACCESS_ONCE(*lp) = *lp + 1;
> >  	smp_mb(); /* B */  /* Avoid leaking the critical section. */
> > -	ACCESS_ONCE(this_cpu_ptr(sp->per_cpu_ref)->seq[idx]) += 1;
> > +	lp = this_cpu_ptr(&sp->per_cpu_ref->seq[idx]);
> > +	ACCESS_ONCE(*lp) = *lp + 1;
> >  	preempt_enable();
> >  	return idx;
> >  
> 
> This probably could use the following 
> 
> diff --git a/kernel/rcu/srcu.c b/kernel/rcu/srcu.c
> index c639556f3fa0..3a97eb6f9076 100644
> --- a/kernel/rcu/srcu.c
> +++ b/kernel/rcu/srcu.c
> @@ -298,9 +298,9 @@ int __srcu_read_lock(struct srcu_struct *sp)
> 
>  	idx = ACCESS_ONCE(sp->completed) & 0x1;
>  	preempt_disable();
> -	ACCESS_ONCE(this_cpu_ptr(sp->per_cpu_ref)->c[idx]) += 1;
> +	this_cpu_inc(sp->per_cpu_ref->c[idx]);
>  	smp_mb(); /* B */  /* Avoid leaking the critical section. */
> -	ACCESS_ONCE(this_cpu_ptr(sp->per_cpu_ref)->seq[idx]) += 1;
> +	this_cpu_inc(sp->per_cpu_ref->seq[idx]);
>  	preempt_enable();
>  	return idx;
>  }

Good point!

But given that I already have preemption disabled and given that
__srcu_read_lock() is not to be used by irq handlers, I should be able to
use __this_cpu_inc(), correct?  Just to avoid unnecessary irq disabling
on non-x86 platforms...

Seems to pass a quick build, so trying a bit heavier testing.

							Thanx, Paul