Re: [PATCH][BUGFIX] cgroups: more safe tasklist locking in cgroup_attach_proc

[Paul Menage <menage-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>]

On Fri, Jul 29, 2011 at 7:28 AM, Ben Blum <bblum-OM76b2Iv3yLQjUSlxSEPGw@public.gmane.org> wrote:
> Fix unstable tasklist locking in cgroup_attach_proc.
>
> From: Ben Blum <bblum-OM76b2Iv3yLQjUSlxSEPGw@public.gmane.org>
>
> According to this thread - https://lkml.org/lkml/2011/7/27/243 - RCU is
> not sufficient to guarantee the tasklist is stable w.r.t. de_thread and
> exit. Taking tasklist_lock for reading, instead of rcu_read_lock,
> ensures proper exclusion.
>
> Signed-off-by: Ben Blum <bblum-OM76b2Iv3yLQjUSlxSEPGw@public.gmane.org>

Acked-by: Paul Menage <menage-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>

Thanks,
Paul

> ---
>  kernel/cgroup.c |    6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff a/kernel/cgroup.c b/kernel/cgroup.c
> --- a/kernel/cgroup.c   2011-07-21 19:17:23.000000000 -0700
> +++ b/kernel/cgroup.c   2011-07-29 06:17:47.000000000 -0700
> @@ -2024,7 +2024,7 @@
>                goto out_free_group_list;
>
>        /* prevent changes to the threadgroup list while we take a snapshot. */
> -       rcu_read_lock();
> +       read_lock(&tasklist_lock);
>        if (!thread_group_leader(leader)) {
>                /*
>                 * a race with de_thread from another thread's exec() may strip
> @@ -2033,7 +2033,7 @@
>                 * throw this task away and try again (from cgroup_procs_write);
>                 * this is "double-double-toil-and-trouble-check locking".
>                 */
> -               rcu_read_unlock();
> +               read_unlock(&tasklist_lock);
>                retval = -EAGAIN;
>                goto out_free_group_list;
>        }
> @@ -2054,7 +2054,7 @@
>        } while_each_thread(leader, tsk);
>        /* remember the number of threads in the array for later. */
>        group_size = i;
> -       rcu_read_unlock();
> +       read_unlock(&tasklist_lock);
>
>        /*
>         * step 1: check that we can legitimately attach to the cgroup.
>