Re: [PATCH] cgroup: list_for_each cleanup

[Paul Menage]

Hi Kosaki,

On Wed, Jun 4, 2008 at 1:36 AM, KOSAKI Motohiro
<kosaki.motohiro@xxxxxxxxxxxxxx> wrote:
> @@ -3072,16 +3070,19 @@ void __css_put(struct cgroup_subsys_stat
>  */
>  static void cgroup_release_agent(struct work_struct *work)
>  {
> +       struct cgroup *cgrp;
> +       struct cgroup *saved_cgrp;
> +
>        BUG_ON(work != &release_agent_work);
>        mutex_lock(&cgroup_mutex);
>        spin_lock(&release_list_lock);
> -       while (!list_empty(&release_list)) {
> +
> +       list_for_each_entry_safe(cgrp, saved_cgrp, &release_list,
> +                                release_list) {
>                char *argv[3], *envp[3];
>                int i;
>                char *pathbuf;
> -               struct cgroup *cgrp = list_entry(release_list.next,
> -                                                   struct cgroup,
> -                                                   release_list);
> +
>                list_del_init(&cgrp->release_list);
>                spin_unlock(&release_list_lock);
>                pathbuf = kmalloc(PAGE_SIZE, GFP_KERNEL);
>

I don't think that list_for_each_entry_safe() is still safe when we've
dropped release_list_lock. saved_cgroup could end up pointing to a
cgroup that's freed while we're forking the usermode helper.

Paul
--
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/