[PATCH v1] hugetlb, userfaultfd: Fix reservation restore on userfaultfd error

From: Mina Almasry
Date: Tue Nov 16 2021 - 18:58:00 EST

Currently in the is_continue case in hugetlb_mcopy_atomic_pte(), if we
bail out using "goto out_release_unlock;" in the cases where idx >=
size, or !huge_pte_none(), the code will detect that new_pagecache_page
== false, and so call restore_reserve_on_error().
In this case I see restore_reserve_on_error() delete the reservation,
and the following call to remove_inode_hugepages() will increment
h->resv_hugepages causing a 100% reproducible leak.

We should treat the is_continue case similar to adding a page into the
pagecache and set new_pagecache_page to true, to indicate that there is
no reservation to restore on the error path, and we need not call

Cc: Wei Xu <weixugc@xxxxxxxxxx>

Fixes: c7b1850dfb41 ("hugetlb: don't pass page cache pages to restore_reserve_on_error")
Signed-off-by: Mina Almasry <almasrymina@xxxxxxxxxx>
Reported-by: James Houghton <jthoughton@xxxxxxxxxx>
mm/hugetlb.c | 8 ++++++++
1 file changed, 8 insertions(+)

diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index e09159c957e3..25a7a3d84607 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -5741,6 +5741,14 @@ int hugetlb_mcopy_atomic_pte(struct mm_struct *dst_mm,
page = find_lock_page(mapping, idx);
if (!page)
goto out;
+ /*
+ * Set new_pagecache_page to true, as we've added a page to the
+ * pagecache, but userfaultfd hasn't set up a mapping for this
+ * page yet. If we bail out before setting up the mapping, we
+ * want to indicate to restore_reserve_on_error() that we've
+ * added the page to the page cache.
+ */
+ new_pagecache_page = true;
} else if (!*pagep) {
/* If a page already exists, then it's UFFDIO_COPY for
* a non-missing case. Return -EEXIST.