On Wed 06-01-21 16:47:38, Muchun Song wrote:
> There is a race between isolate_huge_page() and __free_huge_page().
>
> CPU0: CPU1:
>
> if (PageHuge(page))
> put_page(page)
> __free_huge_page(page)
> spin_lock(&hugetlb_lock)
> update_and_free_page(page)
> set_compound_page_dtor(page,
> NULL_COMPOUND_DTOR)
> spin_unlock(&hugetlb_lock)
> isolate_huge_page(page)
> // trigger BUG_ON
> VM_BUG_ON_PAGE(!PageHead(page), page)
> spin_lock(&hugetlb_lock)
> page_huge_active(page)
> // trigger BUG_ON
> VM_BUG_ON_PAGE(!PageHuge(page), page)
> spin_unlock(&hugetlb_lock)
>
> When we isolate a HugeTLB page on CPU0. Meanwhile, we free it to the
> buddy allocator on CPU1. Then, we can trigger a BUG_ON on CPU0. Because
> it is already freed to the buddy allocator.
>
> Fixes: c8721bbbdd36 ("mm: memory-hotplug: enable memory hotplug to handle hugepage")
> Signed-off-by: Muchun Song <songmuchun@xxxxxxxxxxxxx>
> Reviewed-by: Mike Kravetz <mike.kravetz@xxxxxxxxxx>
Acked-by: Michal Hocko <mhocko@xxxxxxxx>
Thanks!
> ---
> mm/hugetlb.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/mm/hugetlb.c b/mm/hugetlb.c
> index bf02e81e3953..67200dd25b1d 100644
> --- a/mm/hugetlb.c
> +++ b/mm/hugetlb.c
> @@ -5587,9 +5587,9 @@ bool isolate_huge_page(struct page *page, struct list_head *list)
> {
> bool ret = true;
>
> - VM_BUG_ON_PAGE(!PageHead(page), page);
> spin_lock(&hugetlb_lock);
> - if (!page_huge_active(page) || !get_page_unless_zero(page)) {
> + if (!PageHeadHuge(page) || !page_huge_active(page) ||
> + !get_page_unless_zero(page)) {
> ret = false;
> goto unlock;
> }
> --
> 2.11.0
--
Michal Hocko
SUSE Labs
