[PATCH v5 08/14] mm/gup: do not allow zero page for pinned pages
From: Pavel Tatashin <pasha.tatashin@soleen.com>
Date: 2021-01-19 04:49:13
Also in:
linux-doc, linux-kselftest, lkml
Subsystem:
memory management, memory management - gup (get user pages), the rest · Maintainers:
Andrew Morton, David Hildenbrand, Linus Torvalds
Zero page should not be used for long term pinned pages. Once pages are pinned their physical addresses cannot changed until they are unpinned. Guarantee to always return real pages when they are pinned by adding FOLL_WRITE. Signed-off-by: Pavel Tatashin <pasha.tatashin@soleen.com> --- mm/gup.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-)
diff --git a/mm/gup.c b/mm/gup.c
index 857b273e32ac..9a817652f501 100644
--- a/mm/gup.c
+++ b/mm/gup.c@@ -1668,8 +1668,16 @@ static long __gup_longterm_locked(struct mm_struct *mm, unsigned long flags = 0; long rc; - if (gup_flags & FOLL_LONGTERM) + if (gup_flags & FOLL_LONGTERM) { + /* + * We are long term pinning pages and their PA's should not + * change until unpinned. Without FOLL_WRITE we might get zero + * page which we do not want. Force creating normal + * pages by adding FOLL_WRITE. + */ + gup_flags |= FOLL_WRITE; flags = memalloc_pin_save(); + } rc = __get_user_pages_locked(mm, start, nr_pages, pages, vmas, NULL, gup_flags);
--
2.25.1