[PATCH v4 3/3] mm/hugetlb: improve locking in dissolve_free_huge_pages()

[Gerald Schaefer <gerald.schaefer@de.ibm.com>]

For every pfn aligned to minimum_order, dissolve_free_huge_pages() will
call dissolve_free_huge_page() which takes the hugetlb spinlock, even if
the page is not huge at all or a hugepage that is in-use.

Improve this by doing the PageHuge() and page_count() checks already in
dissolve_free_huge_pages() before calling dissolve_free_huge_page(). In
dissolve_free_huge_page(), when holding the spinlock, those checks need
to be revalidated.

Signed-off-by: Gerald Schaefer <gerald.schaefer@de.ibm.com>
---
 mm/hugetlb.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)

diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index 91ae1f5..770d83e 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -1476,14 +1476,20 @@ static int dissolve_free_huge_page(struct page *page)
 int dissolve_free_huge_pages(unsigned long start_pfn, unsigned long end_pfn)
 {
 	unsigned long pfn;
+	struct page *page;
 	int rc = 0;
 
 	if (!hugepages_supported())
 		return rc;
 
-	for (pfn = start_pfn; pfn < end_pfn; pfn += 1 << minimum_order)
-		if (rc = dissolve_free_huge_page(pfn_to_page(pfn)))
-			break;
+	for (pfn = start_pfn; pfn < end_pfn; pfn += 1 << minimum_order) {
+		page = pfn_to_page(pfn);
+		if (PageHuge(page) && !page_count(page)) {
+			rc = dissolve_free_huge_page(page);
+			if (rc)
+				break;
+		}
+	}
 
 	return rc;
 }
-- 
2.8.4