Re: [PATCH] taskstats: fix data-race

[Marco Elver <elver@google.com>]

On Wed, 9 Oct 2019 at 13:31, Christian Brauner
<christian.brauner@ubuntu.com> wrote:
>
> When assiging and testing taskstats in taskstats_exit() there's a race
> when writing and reading sig->stats when a thread-group with more than
> one thread exits:
>
> cpu0:
> thread catches fatal signal and whole thread-group gets taken down
>  do_exit()
>  do_group_exit()
>  taskstats_exit()
>  taskstats_tgid_alloc()
> The tasks reads sig->stats without holding sighand lock seeing garbage.
>
> cpu1:
> task calls exit_group()
>  do_exit()
>  do_group_exit()
>  taskstats_exit()
>  taskstats_tgid_alloc()
> The task takes sighand lock and assigns new stats to sig->stats.
>
> Fix this by using smp_load_acquire() and smp_store_release().
>
> Reported-by: syzbot+c5d03165a1bd1dead0c1@syzkaller.appspotmail.com
> Fixes: 34ec12349c8a ("taskstats: cleanup ->signal->stats allocation")
> Cc: stable@vger.kernel.org
> Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>
> Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
> ---
> /* v1 */
> Link: https://lore.kernel.org/r/20191005112806.13960-1-christian.brauner@ubuntu.com
>
> /* v2 */
> Link: https://lore.kernel.org/r/20191006235216.7483-1-christian.brauner@ubuntu.com
> - Dmitry Vyukov <dvyukov@google.com>, Marco Elver <elver@google.com>:
>   - fix the original double-checked locking using memory barriers
>
> /* v3 */
> Link: https://lore.kernel.org/r/20191007110117.1096-1-christian.brauner@ubuntu.com/
> - Andrea Parri <parri.andrea@gmail.com>:
>   - document memory barriers to make checkpatch happy
>
> /* v4 */
> - Andrea Parri <parri.andrea@gmail.com>:
>   - use smp_load_acquire(), not READ_ONCE()
>   - update commit message

Acked-by: Marco Elver <elver@google.com>

Note that this now looks almost like what I suggested, except the
return at the end of the function is accessing sig->stats again. In
this case, it seems it's fine assuming sig->stats cannot be written
elsewhere. Just wanted to point it out to make sure it's considered.

Thanks!

> ---
>  kernel/taskstats.c | 24 +++++++++++++++---------
>  1 file changed, 15 insertions(+), 9 deletions(-)
>
> diff --git a/kernel/taskstats.c b/kernel/taskstats.c
> index 13a0f2e6ebc2..e6b45315607a 100644
> --- a/kernel/taskstats.c
> +++ b/kernel/taskstats.c
> @@ -554,24 +554,30 @@ static int taskstats_user_cmd(struct sk_buff *skb, struct genl_info *info)
>  static struct taskstats *taskstats_tgid_alloc(struct task_struct *tsk)
>  {
>         struct signal_struct *sig = tsk->signal;
> -       struct taskstats *stats;
> +       struct taskstats *stats_new, *stats;
>
> -       if (sig->stats || thread_group_empty(tsk))
> -               goto ret;
> +       /* Pairs with smp_store_release() below. */
> +       stats = smp_load_acquire(sig->stats);
> +       if (stats || thread_group_empty(tsk))
> +               return stats;
>
>         /* No problem if kmem_cache_zalloc() fails */
> -       stats = kmem_cache_zalloc(taskstats_cache, GFP_KERNEL);
> +       stats_new = kmem_cache_zalloc(taskstats_cache, GFP_KERNEL);
>
>         spin_lock_irq(&tsk->sighand->siglock);
>         if (!sig->stats) {
> -               sig->stats = stats;
> -               stats = NULL;
> +               /*
> +                * Pairs with smp_store_release() above and order the
> +                * kmem_cache_zalloc().
> +                */
> +               smp_store_release(&sig->stats, stats_new);
> +               stats_new = NULL;
>         }
>         spin_unlock_irq(&tsk->sighand->siglock);
>
> -       if (stats)
> -               kmem_cache_free(taskstats_cache, stats);
> -ret:
> +       if (stats_new)
> +               kmem_cache_free(taskstats_cache, stats_new);
> +
>         return sig->stats;
>  }
>
> --
> 2.23.0
>