diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c
index 51f2547c20070d7f248a431604081d66bcf7380f..21b348408478d03118742db9440896c366be9cc9 100644
--- a/drivers/md/raid1.c
+++ b/drivers/md/raid1.c
@@ -1331,6 +1331,7 @@ static void raid1_write_request(struct mddev *mddev, struct bio *bio,
 	struct raid1_plug_cb *plug = NULL;
 	int first_clone;
 	int max_sectors;
+	bool write_behind = false;
 
 	if (mddev_is_clustered(mddev) &&
 	     md_cluster_ops->area_resyncing(mddev, WRITE,
@@ -1383,6 +1384,15 @@ static void raid1_write_request(struct mddev *mddev, struct bio *bio,
 	max_sectors = r1_bio->sectors;
 	for (i = 0;  i < disks; i++) {
 		struct md_rdev *rdev = rcu_dereference(conf->mirrors[i].rdev);
+
+		/*
+		 * The write-behind io is only attempted on drives marked as
+		 * write-mostly, which means we could allocate write behind
+		 * bio later.
+		 */
+		if (rdev && test_bit(WriteMostly, &rdev->flags))
+			write_behind = true;
+
 		if (rdev && unlikely(test_bit(Blocked, &rdev->flags))) {
 			atomic_inc(&rdev->nr_pending);
 			blocked_rdev = rdev;
@@ -1456,6 +1466,15 @@ static void raid1_write_request(struct mddev *mddev, struct bio *bio,
 		goto retry_write;
 	}
 
+	/*
+	 * When using a bitmap, we may call alloc_behind_master_bio below.
+	 * alloc_behind_master_bio allocates a copy of the data payload a page
+	 * at a time and thus needs a new bio that can fit the whole payload
+	 * this bio in page sized chunks.
+	 */
+	if (write_behind && bitmap)
+		max_sectors = min_t(int, max_sectors,
+				    BIO_MAX_VECS * (PAGE_SIZE >> 9));
 	if (max_sectors < bio_sectors(bio)) {
 		struct bio *split = bio_split(bio, max_sectors,
 					      GFP_NOIO, &conf->bio_split);