Re: [PATCH v2] memcg: Fix memcg_kmem_bypass() for remote memcg charging

[Shakeel Butt <shakeelb@google.com>]

On Wed, May 13, 2020 at 4:47 AM Zefan Li <lizefan@huawei.com> 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.
>
> As we will probably encounter this issue in the future as the users of
> memalloc_use_memcg() grow, it's better we fix it now.
>
> Signed-off-by: Zefan Li <lizefan@huawei.com>
> ---
>
> v2: add a comment as sugguested by Michal. and add changelog to explain why
> upstream kernel needs this fix.
>
> ---
>
>  mm/memcontrol.c | 3 +++
>  1 file changed, 3 insertions(+)
>
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index a3b97f1..43a12ed 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -2802,6 +2802,9 @@ static void memcg_schedule_kmem_cache_create(struct mem_cgroup *memcg,
>
>  static inline bool memcg_kmem_bypass(void)
>  {
> +       /* Allow remote memcg charging in kthread contexts. */
> +       if (unlikely(current->active_memcg))
> +               return false;

What about __GFP_ACCOUNT allocations in the interrupt context?

e.g.

memalloc_use_memcg(memcg);
--->interrupt
--->alloc_page(GFP_KERNEL_ACCOUNT) in interrupt context.
alloc_page(GFP_KERNEL_ACCOUNT);
memalloc_unuse_memcg();


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