On Tue 07-04-20 08:36:24, Michal Hocko wrote:
> On Mon 06-04-20 23:48:54, Yafang Shao wrote:
> > When I run my memcg testcase which creates lots of memcgs, I found
> > there're unexpected out of memory logs while there're still enough
> > available free memory. The error log is,
> > mkdir: cannot create directory 'foo.65533': Cannot allocate memory
> >
> > The reason is when we try to create more than MEM_CGROUP_ID_MAX memcgs, an
> > -ENOMEM errno will be set by mem_cgroup_css_alloc(), but the right errno
> > should be -EBUSY "Device or resource busy". That is same with
> > memcg_alloc_cache_id().
>
> I do not see EBUSY being listed as expected return value for mkdir(2)
> which is the primary way to create a cgroup.
>
> > As the errno really misled me, we should make it right. After this patch,
> > the error log will be,
> > mkdir: cannot create directory 'foo.65533': Device or resource busy
>
> I do see ENOMEM being slightly confusing but if we really need to fix
> this then ENOSPC sounds like a better fit to me.
Btw. I have just checked 73f576c04b94 ("mm: memcontrol: fix cgroup
creation failure after many small jobs") and it explicitly talks about
ENOSPC being returned prior to this patch.
--
Michal Hocko
SUSE Labs
