[RFC][PATCH] introduce task cgroup (#task restrictioon for prevent fork bomb by cgroup)

Paul Menage menage at google.com
Thu Jun 5 15:04:12 PDT 2008


Hi Kosaki,

The basic idea of a task-limiting subsystem is good, thanks.

On Wed, Jun 4, 2008 at 9:43 PM, KOSAKI Motohiro
<kosaki.motohiro at jp.fujitsu.com> wrote:
> --- a/kernel/cgroup.c
> +++ b/kernel/cgroup.c
> @@ -2719,13 +2719,27 @@ static struct file_operations proc_cgrou
>  * At the point that cgroup_fork() is called, 'current' is the parent
>  * task, and the passed argument 'child' points to the child task.
>  */
> -void cgroup_fork(struct task_struct *child)
> +int cgroup_fork(struct task_struct *child)
>  {
> +       int i;
> +       int ret;
> +
> +       for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) {
> +               struct cgroup_subsys *ss = subsys[i];
> +               if (ss->can_fork) {
> +                       ret = ss->can_fork(ss, child);
> +                       if (ret)
> +                               return ret;
> +               }
> +       }
> +
>        task_lock(current);
>        child->cgroups = current->cgroups;
>        get_css_set(child->cgroups);
>        task_unlock(current);
>        INIT_LIST_HEAD(&child->cg_list);
> +
> +       return 0;
>  }

I don't think this is the right way to handle this check. This isn't a
generic control groups callback, it's one that specific for a
particular subsystem. So the right way to handle it is to call
task_cgroup_can_fork() from the same place that the RLIM_NPROC limit
is checked.

If it later turned out that multiple cgroup subsystems wanted to be
able to prevent forking, then it might make sense to have a generic
cgroup callback, but for just one subsystem it's cleaner to call
directly.

> +
> +static int task_cgroup_populate(struct cgroup_subsys *ss,
> +                               struct cgroup *cgrp)
> +{
> +       if (task_cgroup_subsys.disabled)
> +               return 0;

I don't think you should need this check - if the subsystem is
disabled, it'll never be mounted in the first place.

Paul


More information about the Containers mailing list