diff --git a/block/blk-mq.c b/block/blk-mq.c
index b038ec680e84..81409162e662 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -1959,10 +1959,14 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio)
 	data.cmd_flags = bio->bi_opf;
 	rq = blk_mq_get_request(q, bio, &data);
 	if (unlikely(!rq)) {
+		blk_qc_t ret = BLK_QC_T_NONE;
+
 		rq_qos_cleanup(q, bio);
-		if (bio->bi_opf & REQ_NOWAIT)
+		if (bio->bi_opf & REQ_NOWAIT_RET)
+			ret = BLK_QC_T_EAGAIN;
+		else if (bio->bi_opf & REQ_NOWAIT)
 			bio_wouldblock_error(bio);
-		return BLK_QC_T_NONE;
+		return ret;
 	}
 
 	trace_block_getrq(q, bio, bio->bi_opf);
diff --git a/fs/block_dev.c b/fs/block_dev.c
index f00b569a9f89..6a09f483f15c 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -344,15 +344,24 @@ __blkdev_direct_IO(struct kiocb *iocb, struct iov_iter *iter, int nr_pages)
 	struct bio *bio;
 	bool is_poll = (iocb->ki_flags & IOCB_HIPRI) != 0;
 	bool is_read = (iov_iter_rw(iter) == READ), is_sync;
+	bool nowait = (iocb->ki_flags & IOCB_NOWAIT) != 0;
 	loff_t pos = iocb->ki_pos;
 	blk_qc_t qc = BLK_QC_T_NONE;
+	gfp_t gfp;
 	int ret = 0;
 
 	if ((pos | iov_iter_alignment(iter)) &
 	    (bdev_logical_block_size(bdev) - 1))
 		return -EINVAL;
 
-	bio = bio_alloc_bioset(GFP_KERNEL, nr_pages, &blkdev_dio_pool);
+	if (nowait)
+		gfp = GFP_NOWAIT;
+	else
+		gfp = GFP_KERNEL;
+
+	bio = bio_alloc_bioset(gfp, nr_pages, &blkdev_dio_pool);
+	if (!bio)
+		return -EAGAIN;
 
 	dio = container_of(bio, struct blkdev_dio, bio);
 	dio->is_sync = is_sync = is_sync_kiocb(iocb);
@@ -397,6 +406,8 @@ __blkdev_direct_IO(struct kiocb *iocb, struct iov_iter *iter, int nr_pages)
 			bio->bi_opf = dio_bio_write_op(iocb);
 			task_io_account_write(bio->bi_iter.bi_size);
 		}
+		if (nowait)
+			bio->bi_opf |= (REQ_NOWAIT | REQ_NOWAIT_RET);
 
 		dio->size += bio->bi_iter.bi_size;
 		pos += bio->bi_iter.bi_size;
@@ -411,6 +422,10 @@ __blkdev_direct_IO(struct kiocb *iocb, struct iov_iter *iter, int nr_pages)
 			}
 
 			qc = submit_bio(bio);
+			if (qc == BLK_QC_T_EAGAIN) {
+				ret = -EAGAIN;
+				goto err;
+			}
 
 			if (polled)
 				WRITE_ONCE(iocb->ki_cookie, qc);
@@ -431,7 +446,12 @@ __blkdev_direct_IO(struct kiocb *iocb, struct iov_iter *iter, int nr_pages)
 			atomic_inc(&dio->ref);
 		}
 
-		submit_bio(bio);
+		qc = submit_bio(bio);
+		if (qc == BLK_QC_T_EAGAIN) {
+			ret = -EAGAIN;
+			goto err;
+		}
+
 		bio = bio_alloc(GFP_KERNEL, nr_pages);
 	}
 
@@ -452,6 +472,7 @@ __blkdev_direct_IO(struct kiocb *iocb, struct iov_iter *iter, int nr_pages)
 	}
 	__set_current_state(TASK_RUNNING);
 
+out:
 	if (!ret)
 		ret = blk_status_to_errno(dio->bio.bi_status);
 	if (likely(!ret))
@@ -459,6 +480,10 @@ __blkdev_direct_IO(struct kiocb *iocb, struct iov_iter *iter, int nr_pages)
 
 	bio_put(&dio->bio);
 	return ret;
+err:
+	if (!is_poll)
+		blk_finish_plug(&plug);
+	goto out;
 }
 
 static ssize_t
diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
index feff3fe4467e..3d89a3c04977 100644
--- a/include/linux/blk_types.h
+++ b/include/linux/blk_types.h
@@ -311,6 +311,7 @@ enum req_flag_bits {
 	__REQ_RAHEAD,		/* read ahead, can fail anytime */
 	__REQ_BACKGROUND,	/* background IO */
 	__REQ_NOWAIT,           /* Don't wait if request will block */
+	__REQ_NOWAIT_RET,	/* Return would-block error inline */
 	/*
 	 * When a shared kthread needs to issue a bio for a cgroup, doing
 	 * so synchronously can lead to priority inversions as the kthread
@@ -345,6 +346,7 @@ enum req_flag_bits {
 #define REQ_RAHEAD		(1ULL << __REQ_RAHEAD)
 #define REQ_BACKGROUND		(1ULL << __REQ_BACKGROUND)
 #define REQ_NOWAIT		(1ULL << __REQ_NOWAIT)
+#define REQ_NOWAIT_RET		(1ULL << __REQ_NOWAIT_RET)
 #define REQ_CGROUP_PUNT		(1ULL << __REQ_CGROUP_PUNT)
 
 #define REQ_NOUNMAP		(1ULL << __REQ_NOUNMAP)
@@ -418,6 +420,7 @@ static inline int op_stat_group(unsigned int op)
 
 typedef unsigned int blk_qc_t;
 #define BLK_QC_T_NONE		-1U
+#define BLK_QC_T_EAGAIN		-2U
 #define BLK_QC_T_SHIFT		16
 #define BLK_QC_T_INTERNAL	(1U << 31)