Re: [PATCH v4 1/2] rcu: uninline rcu_lock_acquire() and rcu_lock_release()

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

On Tue, Jul 01, 2014 at 06:40:02PM +0200, Oleg Nesterov wrote:
> On 07/01, Peter Zijlstra wrote:
> >
> > On Mon, Jun 30, 2014 at 06:18:49PM +0200, Oleg Nesterov wrote:
> > > +static inline void __rcu_lock_acquire(struct lockdep_map *map, unsigned long ip)
> > >  {
> > > +	lock_acquire(map, 0, 0, 2, 0, NULL, ip);
> > >  }
> >
> > > +extern void rcu_lock_acquire(void);
> > > +extern void rcu_lock_release(void);
> > > +extern void rcu_lock_acquire_bh(void);
> > > +extern void rcu_lock_release_bh(void);
> > > +extern void rcu_lock_acquire_sched(void);
> > > +extern void rcu_lock_release_sched(void);
> >
> > > diff --git a/include/linux/srcu.h b/include/linux/srcu.h
> > > index a2783cb..5c06289 100644
> > > --- a/include/linux/srcu.h
> > > +++ b/include/linux/srcu.h
> > > @@ -219,7 +219,7 @@ static inline int srcu_read_lock(struct srcu_struct *sp) __acquires(sp)
> > >  {
> > >  	int retval = __srcu_read_lock(sp);
> > >
> > > -	rcu_lock_acquire(&(sp)->dep_map);
> > > +	__rcu_lock_acquire(&(sp)->dep_map, _THIS_IP_);
> > >  	return retval;
> > >  }
> >
> > Would an srcu_lock_acquire() not make sense here?
> >
> > In any case, not wrong per se, just a consistency thing that stood out.
> 
> Yes, I looked at this too...
> 
> But probably it would be better to just add __rcu_lock_acquire() into
> __srcu_read_lock(), and kill that inline in srcu.h ?

Makes sense to me!

							Thanx, Paul