Re: [PATCH 4/7] locking/rwsem: threshold limited spinning for active readers

[Davidlohr Bueso]

On Mon, 2014-08-04 at 22:30 -0700, Davidlohr Bueso wrote:
> On Mon, 2014-08-04 at 21:54 -0700, Davidlohr Bueso wrote:
> > >  #ifdef CONFIG_RWSEM_SPIN_ON_OWNER
> > > +/*
> > > + * The owner field is set to RWSEM_READ_OWNED if the last owner(s) are
> > > + * readers. It is not reset until a writer takes over and set it to its
> > > + * task structure pointer or NULL when it frees the lock. So a value
> > > + * of RWSEM_READ_OWNED doesn't mean it currently has active readers.
> > > + */
> > > +#define RWSEM_READ_OWNED	((struct task_struct *)-1)
> > 
> > Looks rather weird...
> 
> Instead of populating owner when taking the reader lock, why not just
> leave it NULL. Then, we can differentiate between the owner being NULL
> either because it is taken by reader(s) or simply because it is not
> taken. So something like this:
> 
> static inline bool rwsem_owner_is_reader(struct rw_semaphore *sem)
> {
> 	return !sem->owner && rwsem_is_locked(sem));
> }

Although we could race between both checks, specially when going into
slowpaths, too bad. Probably a read barrier before calling is_locked
too. Anyway, you see, things start getting funky.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/