diff --git a/include/linux/gfp.h b/include/linux/gfp.h
index f78d1e89593f..194da88f128f 100644
--- a/include/linux/gfp.h
+++ b/include/linux/gfp.h
@@ -44,6 +44,7 @@ struct vm_area_struct;
 #else
 #define ___GFP_NOLOCKDEP	0
 #endif
+#define ___GFP_CMA		0x1000000u
 /* If the above are modified, __GFP_BITS_SHIFT may need updating */
 
 /*
@@ -57,6 +58,7 @@ struct vm_area_struct;
 #define __GFP_HIGHMEM	((__force gfp_t)___GFP_HIGHMEM)
 #define __GFP_DMA32	((__force gfp_t)___GFP_DMA32)
 #define __GFP_MOVABLE	((__force gfp_t)___GFP_MOVABLE)  /* ZONE_MOVABLE allowed */
+#define __GFP_CMA	((__force gfp_t)___GFP_CMA)
 #define GFP_ZONEMASK	(__GFP_DMA|__GFP_HIGHMEM|__GFP_DMA32|__GFP_MOVABLE)
 
 /**
@@ -217,8 +219,13 @@ struct vm_area_struct;
 #define __GFP_NOLOCKDEP ((__force gfp_t)___GFP_NOLOCKDEP)
 
 /* Room for N __GFP_FOO bits */
-#define __GFP_BITS_SHIFT (23 + IS_ENABLED(CONFIG_LOCKDEP))
+#define __GFP_BITS_SHIFT (25)
+#ifdef CONFIG_LOCKDEP
 #define __GFP_BITS_MASK ((__force gfp_t)((1 << __GFP_BITS_SHIFT) - 1))
+#else
+#define __GFP_BITS_MASK (((__force gfp_t)((1 << __GFP_BITS_SHIFT) - 1)) & \
+				~0x800000u)
+#endif
 
 /**
  * DOC: Useful GFP flag combinations
diff --git a/include/linux/highmem.h b/include/linux/highmem.h
index 0690679832d4..e80602ee359b 100644
--- a/include/linux/highmem.h
+++ b/include/linux/highmem.h
@@ -181,7 +181,12 @@ static inline struct page *
 alloc_zeroed_user_highpage_movable(struct vm_area_struct *vma,
 					unsigned long vaddr)
 {
+#ifndef CONFIG_CMA
 	return __alloc_zeroed_user_highpage(__GFP_MOVABLE, vma, vaddr);
+#else
+	return __alloc_zeroed_user_highpage(__GFP_MOVABLE|__GFP_CMA, vma,
+						vaddr);
+#endif
 }
 
 static inline void clear_highpage(struct page *page)
diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
index e8fbc2c74093..769bab28dc1c 100644
--- a/include/linux/mmzone.h
+++ b/include/linux/mmzone.h
@@ -387,6 +387,10 @@ struct zone {
 	struct pglist_data	*zone_pgdat;
 	struct per_cpu_pageset __percpu *pageset;
 
+#ifdef CONFIG_CMA
+	bool			cma_alloc;
+#endif
+
 #ifndef CONFIG_SPARSEMEM
 	/*
 	 * Flags for a pageblock_nr_pages block. See pageblock-flags.h.
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index f680a6532cc3..ee41b849d276 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -2579,14 +2579,30 @@ __rmqueue(struct zone *zone, unsigned int order, int migratetype,
 
 retry:
 	page = __rmqueue_smallest(zone, order, migratetype);
-	if (unlikely(!page)) {
-		if (migratetype == MIGRATE_MOVABLE)
-			page = __rmqueue_cma_fallback(zone, order);
 
-		if (!page && __rmqueue_fallback(zone, order, migratetype,
-						alloc_flags))
-			goto retry;
-	}
+	if (unlikely(!page) && __rmqueue_fallback(zone, order, migratetype,
+						  alloc_flags))
+		goto retry;
+
+	trace_mm_page_alloc_zone_locked(page, order, migratetype);
+	return page;
+}
+
+static struct page *__rmqueue_cma(struct zone *zone, unsigned int order,
+					int migratetype)
+{
+	struct page *page = 0;
+
+retry:
+#ifdef CONFIG_CMA
+	if (migratetype == MIGRATE_MOVABLE && !zone->cma_alloc)
+		page = __rmqueue_cma_fallback(zone, order);
+	else
+#endif
+		page = __rmqueue_smallest(zone, order, migratetype);
+
+	if (unlikely(!page) && __rmqueue_fallback(zone, order, migratetype))
+		goto retry;
 
 	trace_mm_page_alloc_zone_locked(page, order, migratetype);
 	return page;
@@ -2599,14 +2615,19 @@ __rmqueue(struct zone *zone, unsigned int order, int migratetype,
  */
 static int rmqueue_bulk(struct zone *zone, unsigned int order,
 			unsigned long count, struct list_head *list,
-			int migratetype, unsigned int alloc_flags)
+			int migratetype, unsigned int alloc_flags, int cma)
 {
 	int i, alloced = 0;
 
 	spin_lock(&zone->lock);
 	for (i = 0; i < count; ++i) {
-		struct page *page = __rmqueue(zone, order, migratetype,
-								alloc_flags);
+		struct page *page;
+
+		if (cma)
+			page = __rmqueue_cma(zone, order, migratetype);
+		else
+			page = __rmqueue(zone, order, migratetype, alloc_flags);
+
 		if (unlikely(page == NULL))
 			break;
 
@@ -3064,7 +3085,7 @@ static inline void zone_statistics(struct zone *preferred_zone, struct zone *z)
 static struct page *__rmqueue_pcplist(struct zone *zone, int migratetype,
 			unsigned int alloc_flags,
 			struct per_cpu_pages *pcp,
-			struct list_head *list)
+			struct list_head *list, gfp_t gfp_flags)
 {
 	struct page *page;
 
@@ -3072,7 +3093,8 @@ static struct page *__rmqueue_pcplist(struct zone *zone, int migratetype,
 		if (list_empty(list)) {
 			pcp->count += rmqueue_bulk(zone, 0,
 					pcp->batch, list,
-					migratetype, alloc_flags);
+					migratetype, alloc_flags,
+					gfp_flags & __GFP_CMA);
 			if (unlikely(list_empty(list)))
 				return NULL;
 		}
@@ -3099,7 +3121,8 @@ static struct page *rmqueue_pcplist(struct zone *preferred_zone,
 	local_irq_save(flags);
 	pcp = &this_cpu_ptr(zone->pageset)->pcp;
 	list = &pcp->lists[migratetype];
-	page = __rmqueue_pcplist(zone,  migratetype, alloc_flags, pcp, list);
+	page = __rmqueue_pcplist(zone,  migratetype, alloc_flags, pcp, list,
+				 gfp_flags);
 	if (page) {
 		__count_zid_vm_events(PGALLOC, page_zonenum(page), 1 << order);
 		zone_statistics(preferred_zone, zone);
@@ -3140,8 +3163,13 @@ struct page *rmqueue(struct zone *preferred_zone,
 			if (page)
 				trace_mm_page_alloc_zone_locked(page, order, migratetype);
 		}
-		if (!page)
-			page = __rmqueue(zone, order, migratetype, alloc_flags);
+		if (!page) {
+			if (gfp_flags & __GFP_CMA)
+				page = __rmqueue_cma(zone, order, migratetype);
+			else
+				page = __rmqueue(zone, order, migratetype,
+						 alloc_flags);
+		}
 	} while (page && check_new_pages(page, order));
 	spin_unlock(&zone->lock);
 	if (!page)
@@ -8159,6 +8187,7 @@ int alloc_contig_range(unsigned long start, unsigned long end,
 	if (ret)
 		return ret;
 
+	cc.zone->cma_alloc = 1;
 	/*
 	 * In case of -EBUSY, we'd like to know which page causes problem.
 	 * So, just fall through. test_pages_isolated() has a tracepoint
@@ -8241,6 +8270,7 @@ int alloc_contig_range(unsigned long start, unsigned long end,
 done:
 	undo_isolate_page_range(pfn_max_align_down(start),
 				pfn_max_align_up(end), migratetype);
+	cc.zone->cma_alloc = 0;
 	return ret;
 }