On Fri 15-01-21 20:49:40, Muchun Song wrote:
There is a race condition between __free_huge_page() and dissolve_free_huge_page().
CPU0: CPU1:
// page_count(page) == 1 put_page(page) __free_huge_page(page) dissolve_free_huge_page(page) spin_lock(&hugetlb_lock) // PageHuge(page) && !page_count(page) update_and_free_page(page) // page is freed to the buddy spin_unlock(&hugetlb_lock) spin_lock(&hugetlb_lock) clear_page_huge_active(page) enqueue_huge_page(page) // It is wrong, the page is already freed spin_unlock(&hugetlb_lock)
The race windows is between put_page() and dissolve_free_huge_page().
We should make sure that the page is already on the free list when it is dissolved.
As a result __free_huge_page would corrupt page(s) already in the buddy allocator.
Fixes: c8721bbbdd36 ("mm: memory-hotplug: enable memory hotplug to handle hugepage") Signed-off-by: Muchun Song songmuchun@bytedance.com Reviewed-by: Mike Kravetz mike.kravetz@oracle.com Cc: stable@vger.kernel.org
Acked-by: Michal Hocko mhocko@suse.com
Thanks!
mm/hugetlb.c | 39 +++++++++++++++++++++++++++++++++++++++ 1 file changed, 39 insertions(+)
diff --git a/mm/hugetlb.c b/mm/hugetlb.c index 4741d60f8955..b99fe4a2b435 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c @@ -79,6 +79,21 @@ DEFINE_SPINLOCK(hugetlb_lock); static int num_fault_mutexes; struct mutex *hugetlb_fault_mutex_table ____cacheline_aligned_in_smp; +static inline bool PageHugeFreed(struct page *head) +{
- return page_private(head + 4) == -1UL;
+}
+static inline void SetPageHugeFreed(struct page *head) +{
- set_page_private(head + 4, -1UL);
+}
+static inline void ClearPageHugeFreed(struct page *head) +{
- set_page_private(head + 4, 0);
+}
/* Forward declaration */ static int hugetlb_acct_memory(struct hstate *h, long delta); @@ -1028,6 +1043,7 @@ static void enqueue_huge_page(struct hstate *h, struct page *page) list_move(&page->lru, &h->hugepage_freelists[nid]); h->free_huge_pages++; h->free_huge_pages_node[nid]++;
- SetPageHugeFreed(page);
} static struct page *dequeue_huge_page_node_exact(struct hstate *h, int nid) @@ -1044,6 +1060,7 @@ static struct page *dequeue_huge_page_node_exact(struct hstate *h, int nid) list_move(&page->lru, &h->hugepage_activelist); set_page_refcounted(page);
ClearPageHugeFreed(page);@@ -1504,6 +1521,7 @@ static void prep_new_huge_page(struct hstate *h, struct page *page, int nid) spin_lock(&hugetlb_lock); h->nr_huge_pages++; h->nr_huge_pages_node[nid]++;
- ClearPageHugeFreed(page); spin_unlock(&hugetlb_lock);
} @@ -1754,6 +1772,7 @@ int dissolve_free_huge_page(struct page *page) { int rc = -EBUSY; +retry: /* Not to disrupt normal path by vainly holding hugetlb_lock */ if (!PageHuge(page)) return 0; @@ -1770,6 +1789,26 @@ int dissolve_free_huge_page(struct page *page) int nid = page_to_nid(head); if (h->free_huge_pages - h->resv_huge_pages == 0) goto out;
/** We should make sure that the page is already on the free list* when it is dissolved.*/if (unlikely(!PageHugeFreed(head))) {spin_unlock(&hugetlb_lock);cond_resched();/** Theoretically, we should return -EBUSY when we* encounter this race. In fact, we have a chance* to successfully dissolve the page if we do a* retry. Because the race window is quite small.* If we seize this opportunity, it is an optimization* for increasing the success rate of dissolving page.*/goto retry;}- /*
- Move PageHWPoison flag from head page to the raw error page,
- which makes any subpages rather than the error page reusable.
-- 2.11.0