Re: [PATCH 6/8] cgroup: mount cgroupns-root when inside non-init cgroupns

[Sergey Senozhatsky <sergey.senozhatsky-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>]

On (12/31/15 18:58), Serge E. Hallyn wrote:
[..]
> Thanks!  Following trivial patch should fix this:

confirmed, thanks!

Tested-by: Sergey Senozhatsky <sergey.senozhatsky-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>

	-ss

> From fc54592077533ff2ff90ed54b72bf03b4378ca9f Mon Sep 17 00:00:00 2001
> From: Serge Hallyn <serge.hallyn-GeWIH/nMZzLQT0dZR+AlfA@public.gmane.org>
> Date: Thu, 31 Dec 2015 16:55:19 -0800
> Subject: [PATCH 1/1] cgroup_release_agent: grab css_set_lock around
>  cgroup_path()
> 
> Reported-by: Sergey Senozhatsky <sergey.senozhatsky-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> Signed-off-by: Serge Hallyn <serge.hallyn-GeWIH/nMZzLQT0dZR+AlfA@public.gmane.org>
> ---
>  kernel/cgroup.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/kernel/cgroup.c b/kernel/cgroup.c
> index 4812f59..d10f127 100644
> --- a/kernel/cgroup.c
> +++ b/kernel/cgroup.c
> @@ -5769,7 +5769,9 @@ static void cgroup_release_agent(struct work_struct *work)
>  	if (!pathbuf || !agentbuf)
>  		goto out;
>  
> +	spin_lock_bh(&css_set_lock);
>  	path = cgroup_path(cgrp, pathbuf, PATH_MAX);
> +	spin_unlock_bh(&css_set_lock);
>  	if (!path)
>  		goto out;
>  
> -- 
> 2.5.0
>