Re: [PATCH v4 5/6] clone3: allow spawning processes into cgroups

[Christian Brauner]

On Mon, Jan 20, 2020 at 04:39:30PM +0100, Oleg Nesterov wrote:
> On 01/17, Christian Brauner wrote:
> >
> > +static int cgroup_css_set_fork(struct task_struct *parent,
> > +			       struct kernel_clone_args *kargs)
> ...
> > +	kargs->cset = find_css_set(cset, dst_cgrp);
> > +	if (!kargs->cset) {
> > +		ret = -ENOMEM;
> > +		goto err;
> > +	}
> > +
> > +	if (cgroup_is_dead(dst_cgrp)) {
> > +		ret = -ENODEV;
> > +		goto err;
>                 ^^^^^^^^
> 
> this looks wrong... don't we need put_css_set(kargs->cset) before "goto err" ?

Yeah, but we should rather do:

diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c
index 4d36255ef25f..482055d1e64a 100644
--- a/kernel/cgroup/cgroup.c
+++ b/kernel/cgroup/cgroup.c
@@ -5994,6 +5994,8 @@ static int cgroup_css_set_fork(struct kernel_clone_args *kargs)
        if (dst_cgrp)
                cgroup_put(dst_cgrp);
        put_css_set(cset);
+       if (kargs->cset)
+               put_css_set(kargs->cset);
        return ret;
 }

Christian