Re: [PATCH] memcg: Fix memcg_kmem_bypass() for remote memcg charging

[Michal Hocko <mhocko@kernel.org>]

On Wed 13-05-20 15:28:28, Li Zefan wrote:
> While trying to use remote memcg charging in an out-of-tree kernel module
> I found it's not working, because the current thread is a workqueue thread.
> 
> Signed-off-by: Zefan Li <lizefan@huawei.com>
> ---
> 
> No need to queue this for v5.7 as currently no upstream users of this memcg
> feature suffer from this bug.
> 
> ---
>  mm/memcontrol.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index a3b97f1..db836fc 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -2802,6 +2802,8 @@ static void memcg_schedule_kmem_cache_create(struct mem_cgroup *memcg,
>  
>  static inline bool memcg_kmem_bypass(void)
>  {
> +	if (unlikely(current->active_memcg))
> +		return false;

I am confused. Why the check below is insufficient? It checks for both mm
and PF_KTHREAD?

>  	if (in_interrupt() || !current->mm || (current->flags & PF_KTHREAD))
>  		return true;
>  	return false;
> -- 
> 2.7.4

-- 
Michal Hocko
SUSE Labs