Re: [PATCH tip/core/rcu 09/15] rcu: Increasing rcu_barrier() concurrency

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

On Fri, Jun 15, 2012 at 04:31:51PM -0700, Josh Triplett wrote:
> On Fri, Jun 15, 2012 at 02:06:04PM -0700, Paul E. McKenney wrote:
> > From: "Paul E. McKenney" <paul.mckenney@linaro.org>
> > 
> > The traditional rcu_barrier() implementation has serialized all requests,
> > regardless of RCU flavor, and also does not coalesce concurrent requests.
> > In the past, this has been good and sufficient.
> > 
> > However, systems are getting larger and use of rcu_barrier() has been
> > increasing.  This commit therefore introduces a counter-based scheme
> > that allows _rcu_barrier() calls for the same flavor of RCU to take
> > advantage of each others' work.
> > 
> > Signed-off-by: Paul E. McKenney <paul.mckenney@linaro.org>
> > Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
> > ---
> >  kernel/rcutree.c |   27 ++++++++++++++++++++++++++-
> >  kernel/rcutree.h |    2 ++
> >  2 files changed, 28 insertions(+), 1 deletions(-)
> > 
> > diff --git a/kernel/rcutree.c b/kernel/rcutree.c
> > index 93358d4..7c299d3 100644
> > --- a/kernel/rcutree.c
> > +++ b/kernel/rcutree.c
> > @@ -2291,13 +2291,32 @@ static void _rcu_barrier(struct rcu_state *rsp)
> >  	unsigned long flags;
> >  	struct rcu_data *rdp;
> >  	struct rcu_data rd;
> > +	unsigned long snap = ACCESS_ONCE(rsp->n_barrier_done);
> > +	unsigned long snap_done;
> >  
> >  	init_rcu_head_on_stack(&rd.barrier_head);
> >  
> >  	/* Take mutex to serialize concurrent rcu_barrier() requests. */
> >  	mutex_lock(&rsp->barrier_mutex);
> >  
> > -	smp_mb();  /* Prevent any prior operations from leaking in. */
> > +	/*
> > +	 * Ensure tht all prior references, including to ->n_barrier_done,
> > +	 * are ordered before the _rcu_barrier() machinery.
> > +	 */
> > +	smp_mb();  /* See above block comment. */
> 
> If checkpatch complains about the lack of a comment to the right of a
> barrier even when the barrier has a comment directly above it, that
> seems like a bug in checkpatch that needs fixing, to prevent developers
> from having to add noise like "See above block comment.". :)

;-)

> Also: what type of barriers do mutex_lock and mutex_unlock imply?  I
> assume they imply some weaker barrier than smp_mb, but I'd still assume
> they imply *some* barrier.

mutex_lock() prevents code from leaving the critical section, but is
not guaranteed to prevent code from entering the critical section.

> > +	/* Recheck ->n_barrier_done to see if others did our work for us. */
> > +	snap_done = ACCESS_ONCE(rsp->n_barrier_done);
> > +	if (ULONG_CMP_GE(snap_done, ((snap + 1) & ~0x1) + 2)) {
> 
> This calculation seems sufficiently clever that it merits an explanatory
> comment.

I will see what I can come up with.

> > +		smp_mb();
> > +		mutex_unlock(&rsp->barrier_mutex);
> > +		return;
> > +	}
> > +
> > +	/* Increment ->n_barrier_done to avoid duplicate work. */
> > +	ACCESS_ONCE(rsp->n_barrier_done)++;
> 
> Interesting dissonance here: the use of ACCESS_ONCE with ++ implies
> exactly two accesses, rather than exactly one.  What makes it safe to
> not use atomic_inc here, but not safe to drop the ACCESS_ONCE?
> Potential use of a cached value read earlier in the function?

Or, worse yet, the compiler speculating the increment and then backing
it out if the early-exit path is taken.

							Thanx, Paul