[CRIU] Fix cpuset restore-in-root bug

[Pavel Emelyanov]

On 10/08/2014 07:05 PM, Tycho Andersen wrote:
> On Wed, Oct 08, 2014 at 06:46:37PM +0400, Pavel Emelyanov wrote:
>>
>> Yes, you're right. I meant the prepare_fds() one -- it closes proc service
>> fd and tty one. I think the cgyard can be closed even earlier, but this place
>> is at least about fds :)
> 
> Yes, this works for me:

Great! Patch applied. Thanks for the report and help with the fix :)

> 
>>From a6ef413cbcf6cbbf8b5182f7f4caf8f51256faf5 Mon Sep 17 00:00:00 2001
> From: Tycho Andersen <tycho.andersen at canonical.com>
> Date: Tue, 7 Oct 2014 17:25:26 -0500
> Subject: [PATCH] restore: don't race when closing cg yard
> 
> TASK_HELPERs are created with CLONE_FILES, so if we always close the cg yard
> here, it will close it for the other helpers and cause problems. Instead, we
> close it much later, in code only called by alive tasks, to ensure that there
> is no conflict.
> 
> Signed-off-by: Tycho Andersen <tycho.andersen at canonical.com>
> ---
>  cr-restore.c | 8 --------
>  files.c      | 7 +++++++
>  2 files changed, 7 insertions(+), 8 deletions(-)
> 
> diff --git a/cr-restore.c b/cr-restore.c
> index fd35bef..c8cb428 100644
> --- a/cr-restore.c
> +++ b/cr-restore.c
> @@ -1461,14 +1461,6 @@ static int restore_task_with_children(void *_arg)
>  	if (create_children_and_session())
>  		goto err_fini_mnt;
>  
> -	/*
> -	 * This must be done after forking to allow child
> -	 * to get the cgroup fd so it can move into the
> -	 * correct /tasks file if it is in a different cgroup
> -	 * set than its parent
> -	 */
> -	close_service_fd(CGROUP_YARD);
> -
>  	if (restore_task_mnt_ns(current))
>  		goto err_fini_mnt;
>  
> diff --git a/files.c b/files.c
> index b975cd4..ce83736 100644
> --- a/files.c
> +++ b/files.c
> @@ -949,6 +949,13 @@ int prepare_fds(struct pstree_item *me)
>  
>  	pr_info("Opening fdinfo-s\n");
>  
> +	/*
> +	 * This must be done after forking to allow child
> +	 * to get the cgroup fd so it can move into the
> +	 * correct /tasks file if it is in a different cgroup
> +	 * set than its parent
> +	 */
> +	close_service_fd(CGROUP_YARD);
>  	close_pid_proc(); /* flush any proc cached fds we may have */
>  
>  	if (rsti(me)->fdt) {
>