Re: [PATCH] fix spurious EBUSY on memory cgroup removal

[Balbir Singh <balbir-23VcF4HTsmIX0ybBhKVfKdBPR1lH4CV8@public.gmane.org>]

YAMAMOTO Takashi wrote:
> hi,
> 
> the following patch is to fix spurious EBUSY on cgroup removal.
> 
> YAMAMOTO Takashi
> 
> 
> call mm_free_cgroup earlier.
> otherwise a reference due to lazy mm switching can prevent cgroup removal.
> 
> Signed-off-by: YAMAMOTO Takashi <yamamoto-jCdQPDEk3idL9jVzuh4AOg@public.gmane.org>
> ---
> 
> --- linux-2.6.24-rc8-mm1/kernel/fork.c.BACKUP	2008-01-23 14:43:29.000000000 +0900
> +++ linux-2.6.24-rc8-mm1/kernel/fork.c	2008-01-31 17:26:31.000000000 +0900
> @@ -393,7 +393,6 @@ void __mmdrop(struct mm_struct *mm)
>  {
>  	BUG_ON(mm == &init_mm);
>  	mm_free_pgd(mm);
> -	mm_free_cgroup(mm);
>  	destroy_context(mm);
>  	free_mm(mm);
>  }
> @@ -415,6 +414,7 @@ void mmput(struct mm_struct *mm)
>  			spin_unlock(&mmlist_lock);
>  		}
>  		put_swap_token(mm);
> +		mm_free_cgroup(mm);
>  		mmdrop(mm);
>  	}
>  }

The difference I see with the change is that mmput() will now call
mm_free_cgroup() when mm->mm_users drop to 0 and __mmdrop() will call it when
mm->mm_count drops to 0.

I think this change makes sense

Acked-by: Balbir Singh <balbir-23VcF4HTsmIX0ybBhKVfKdBPR1lH4CV8@public.gmane.org>

-- 
	Warm Regards,
	Balbir Singh
	Linux Technology Center
	IBM, ISTL