diff --git a/include/linux/shmem_fs.h b/include/linux/shmem_fs.h
index 17e9969db499..51b1dcfb5022 100644
--- a/include/linux/shmem_fs.h
+++ b/include/linux/shmem_fs.h
@@ -85,6 +85,7 @@ extern bool shmem_huge_enabled(struct vm_area_struct *vma);
 extern unsigned long shmem_swap_usage(struct vm_area_struct *vma);
 extern unsigned long shmem_partial_swap_usage(struct address_space *mapping,
 						pgoff_t start, pgoff_t end);
+extern void shmem_mark_page_lazyfree(struct page *page);
 
 /* Flag allocation requirements to shmem_getpage */
 enum sgp_type {
diff --git a/include/linux/swap.h b/include/linux/swap.h
index e68ae91643f3..8ed2f6e51ae7 100644
--- a/include/linux/swap.h
+++ b/include/linux/swap.h
@@ -363,6 +363,7 @@ extern void rotate_reclaimable_page(struct page *page);
 extern void deactivate_file_page(struct page *page);
 extern void deactivate_page(struct page *page);
 extern void mark_page_lazyfree(struct page *page);
+extern void mark_page_lazyfree_movetail(struct page *page);
 extern void swap_setup(void);
 
 extern void __lru_cache_add_inactive_or_unevictable(struct page *page,
diff --git a/mm/shmem.c b/mm/shmem.c
index d8f45f30b656..a4b5012b1267 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -4284,3 +4284,9 @@ struct page *shmem_read_mapping_page_gfp(struct address_space *mapping,
 #endif
 }
 EXPORT_SYMBOL_GPL(shmem_read_mapping_page_gfp);
+
+void shmem_mark_page_lazyfree(struct page *page)
+{
+	mark_page_lazyfree_movetail(page);
+}
+EXPORT_SYMBOL_GPL(shmem_mark_page_lazyfree);
diff --git a/mm/swap.c b/mm/swap.c
index b6c5e44e49f0..171213a64202 100644
--- a/mm/swap.c
+++ b/mm/swap.c
@@ -65,6 +65,7 @@ struct lru_pvecs {
 	struct pagevec lru_deactivate_file;
 	struct pagevec lru_deactivate;
 	struct pagevec lru_lazyfree;
+	struct pagevec lru_lazyfree_movetail;
 #ifdef CONFIG_SMP
 	struct pagevec activate_page;
 #endif
@@ -630,6 +631,21 @@ static void lru_lazyfree_fn(struct page *page, struct lruvec *lruvec,
 	}
 }
 
+static void lru_lazyfree_movetail_fn(struct page *page, struct lruvec *lruvec,
+			    void *arg)
+{
+	if (PageLRU(page) && !PageUnevictable(page) && PageSwapBacked(page) &&
+		!PageSwapCache(page)) {
+		bool active = PageActive(page);
+
+		del_page_from_lru_list(page, lruvec,
+				       LRU_INACTIVE_ANON + active);
+		ClearPageActive(page);
+		ClearPageReferenced(page);
+		add_page_to_lru_list_tail(page, lruvec, LRU_INACTIVE_FILE);
+	}
+}
+
 /*
  * Drain pages out of the cpu's pagevecs.
  * Either "cpu" is the current CPU, and preemption has already been
@@ -665,6 +681,10 @@ void lru_add_drain_cpu(int cpu)
 	if (pagevec_count(pvec))
 		pagevec_lru_move_fn(pvec, lru_lazyfree_fn, NULL);
 
+	pvec = &per_cpu(lru_pvecs.lru_lazyfree_movetail, cpu);
+	if (pagevec_count(pvec))
+		pagevec_lru_move_fn(pvec, lru_lazyfree_movetail_fn, NULL);
+
 	activate_page_drain(cpu);
 	invalidate_bh_lrus_cpu(cpu);
 }
@@ -742,6 +762,29 @@ void mark_page_lazyfree(struct page *page)
 	}
 }
 
+/**
+ * mark_page_lazyfree_movetail - make a swapbacked page lazyfree
+ * @page: page to deactivate
+ *
+ * mark_page_lazyfree_movetail() moves @page to the tail of inactive file list.
+ * This is done to accelerate the reclaim of @page.
+ */
+void mark_page_lazyfree_movetail(struct page *page)
+{
+	if (PageLRU(page) && !PageUnevictable(page) && PageSwapBacked(page) &&
+		!PageSwapCache(page)) {
+		struct pagevec *pvec;
+
+		local_lock(&lru_pvecs.lock);
+		pvec = this_cpu_ptr(&lru_pvecs.lru_lazyfree_movetail);
+		get_page(page);
+		if (pagevec_add_and_need_flush(pvec, page))
+			pagevec_lru_move_fn(pvec,
+					lru_lazyfree_movetail_fn, NULL);
+		local_unlock(&lru_pvecs.lock);
+	}
+}
+
 void lru_add_drain(void)
 {
 	local_lock(&lru_pvecs.lock);
@@ -854,6 +897,7 @@ inline void __lru_add_drain_all(bool force_all_cpus)
 		    pagevec_count(&per_cpu(lru_pvecs.lru_deactivate_file, cpu)) ||
 		    pagevec_count(&per_cpu(lru_pvecs.lru_deactivate, cpu)) ||
 		    pagevec_count(&per_cpu(lru_pvecs.lru_lazyfree, cpu)) ||
+		    pagevec_count(&per_cpu(lru_pvecs.lru_lazyfree_movetail, cpu)) ||
 		    need_activate_page_drain(cpu) ||
 		    has_bh_in_lru(cpu, NULL)) {
 			INIT_WORK(work, lru_add_drain_per_cpu);