Re: [PATCH v3 2/2] mm, memcg: Decouple e{low,min} state mutations from protection checks

[Naresh Kamboju]

On Fri, 22 May 2020 at 17:15, Chris Down <chris@xxxxxxxxxxxxxx> wrote:
>
> Naresh Kamboju writes:
> >This patch is causing oom-killer while running mkfs -t ext4 on i386 kernel
> >running on x86_64 machine version linux-next 5.7.0-rc6-next-20200521.
>
> I think I see what's wrong here -- if we bail out early, memory.e{min,low}
> might be uninitialised.
>
> Does this patch fix it, by any chance?
>
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index d3b23c57bed4..aa902b9cbb79 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -5046,6 +5046,9 @@ mem_cgroup_css_alloc(struct cgroup_subsys_state *parent_css)
>                         memory_cgrp_subsys.broken_hierarchy = true;
>         }
>
> +       memcg->memory.emin = 0;
> +       memcg->memory.elow = 0;
> +
>         /* The following stuff does not apply to the root */
>         if (!parent) {
>   #ifdef CONFIG_MEMCG_KMEM

This patch did not fix the problem.
The test log showing oom-killer link,
https://lkft.validation.linaro.org/scheduler/job/1445128#L1194

Build images:
https://builds.tuxbuild.com/YqW37GVsCHf8ndxOT2z1iQ/

- Naresh