[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20180907202429.GA97913@dennisz-mbp.dhcp.thefacebook.com>
Date: Fri, 7 Sep 2018 16:24:30 -0400
From: Dennis Zhou <dennisszhou@...il.com>
To: Tejun Heo <tj@...nel.org>
Cc: Jens Axboe <axboe@...nel.dk>, Johannes Weiner <hannes@...xchg.org>,
Josef Bacik <josef@...icpanda.com>, kernel-team@...com,
linux-block@...r.kernel.org, cgroups@...r.kernel.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH 09/12] blkcg: remove additional reference to the css
On Fri, Sep 07, 2018 at 10:54:46AM -0700, Tejun Heo wrote:
> Hello,
>
> On Thu, Sep 06, 2018 at 05:10:42PM -0400, Dennis Zhou wrote:
> > +struct cgroup_subsys_state *cgroup_e_css(struct cgroup *cgrp,
> > + struct cgroup_subsys *ss)
> > +{
> > + struct cgroup_subsys_state *css;
> > +
> > + rcu_read_lock();
> > +
> > + do {
> > + css = cgroup_css(cgrp, ss);
> > +
> > + if (css)
> > + goto out_unlock;
> > + cgrp = cgroup_parent(cgrp);
> > + } while (cgrp);
> > +
> > + css = init_css_set.subsys[ss->id];
> > +out_unlock:
> > + rcu_read_unlock();
>
> Nothing protects @css here tho. It can be released before the caller
> is done with it. The caller must ensure that it's holding rcu read
> lock to protect the lookup and the subsequent uses. cgroup_css()
> already checks for rcu locking, so if you just drop
> rcu_read_lock/unlock(), everything should work fine.
>
Ah yes, that's my bad. I've removed it and the unnecessary goto now.
Thanks,
Dennis
Powered by blists - more mailing lists