Hello Guiseppe.
Thanks for incorporating the changes.
On Fri, Dec 13, 2019 at 11:28:08AM +0100, Giuseppe Scrivano <gscrivan@xxxxxxxxxx> wrote:
> @@ -202,8 +234,11 @@ int hugetlb_cgroup_charge_cgroup(int idx, unsigned long nr_pages,
> }
> rcu_read_unlock();
>
> - if (!page_counter_try_charge(&h_cg->hugepage[idx], nr_pages, &counter))
> + if (!page_counter_try_charge(&h_cg->hugepage[idx], nr_pages,
> + &counter)) {
> ret = -ENOMEM;
> + hugetlb_event(h_cg, idx, HUGETLB_MAX);
Here should be something like
- hugetlb_event(h_cg, idx, HUGETLB_MAX);
+ hugetlb_event(hugetlb_cgroup_from_counter(counter), idx, HUGETLB_MAX);
in order to have consistent behavior with memcg events (because
page_counter_try_charge may fail higher in the hierarchy than h_cg).
Michal
Attachment:
signature.asc
Description: Digital signature
