Re: [PATCH block/for-3.4/core] cfq: fix cfqg ref handling whenBLK_CGROUP && !CFQ_GROUP_IOSCHED

[Tejun Heo]

On Thu, Mar 15, 2012 at 01:01:00PM -0400, Vivek Goyal wrote:
> On Thu, Mar 15, 2012 at 09:50:57AM -0700, Tejun Heo wrote:
> 
> > > [..]
> > > > @@ -3533,7 +3551,7 @@ static int cfq_init_queue(struct request
> > > >  
> > > >  	spin_lock_irq(q->queue_lock);
> > > >  	cfq_link_cfqq_cfqg(&cfqd->oom_cfqq, cfqd->root_group);
> > > > -	blkg_put(cfqg_to_blkg(cfqd->root_group));
> > > > +	cfqg_put(cfqd->root_group);
> > > 
> > > This seems to be a spurious cfqg_put()? Which reference we are putting
> > > down here?
> > 
> > The extra ref from cfq_link_cfqq_cfqg() for oom_cfqq; otherwise, we
> > need an extra cfq_put() in cfq_exit_queue().  I thought I wrote
> > comment about that somewhere.  Hmmm.... apparently not.  The thing is
> > that oom_cfqq doesn't go through proper cfqq destruction and thus
> > never puts the extra ref to root cfqg.
> 
> Ok. Is cfq_exit_queue() a better place to put down this reference
> explicitly with a comment. Even if you keep it here, atleast a comment
> is required. It is not obvious at all (atleast to me).

Ah... the comment actually already is there.

	/*
	 * Our fallback cfqq if cfq_find_alloc_queue() runs into OOM issues.
	 * Grab a permanent reference to it, so that the normal code flow
	 * will not attempt to free it.  oom_cfqq is linked to root_group
	 * but shouldn't hold a reference as it'll never be unlinked.  Lose
	 * the reference from linking right away.
	 */

-- 
tejun
--
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/