Add support for two bdevs in blkdev_copy_offload

Signed-off-by: Ameer Hamza <[email protected]>

Author: ixhamza

block/blk-core.c

@@ -854,18 +854,18 @@ void submit_bio_noacct(struct bio *bio)
 		if (!bdev_is_zoned(bio->bi_bdev))
 			goto not_supported;
 		break;
+	case REQ_OP_COPY_SRC:
+	case REQ_OP_COPY_DST:
+		if (!q->limits.max_copy_sectors)
+			goto not_supported;
+		break;
 	case REQ_OP_DRV_IN:
 	case REQ_OP_DRV_OUT:
 		/*
 		 * Driver private operations are only used with passthrough
 		 * requests.
 		 */
 		fallthrough;
-	case REQ_OP_COPY_SRC:
-	case REQ_OP_COPY_DST:
-		if (!q->limits.max_copy_sectors)
-			goto not_supported;
-		break;
 	default:
 		goto not_supported;
 	}

block/blk-lib.c

@@ -10,22 +10,6 @@
 
 #include "blk.h"
 
-/* Keeps track of all outstanding copy IO */
-struct blkdev_copy_io {
-	atomic_t refcount;
-	ssize_t copied;
-	int status;
-	struct task_struct *waiter;
-	void (*endio)(void *private, int status, ssize_t copied);
-	void *private;
-};
-
-/* Keeps track of single outstanding copy offload IO */
-struct blkdev_copy_offload_io {
-	struct blkdev_copy_io *cio;
-	loff_t offset;
-};
-
 static sector_t bio_discard_limit(struct block_device *bdev, sector_t sector)
 {
 	unsigned int discard_granularity = bdev_discard_granularity(bdev);
@@ -180,14 +164,17 @@ static void blkdev_copy_offload_src_endio(struct bio *bio)
 			cio->status = blk_status_to_errno(bio->bi_status);
 	}
 	bio_put(bio);
+	if (offload_io->dst_bio)
+		bio_put(offload_io->dst_bio);
+
 	kfree(offload_io);
 
 	if (atomic_dec_and_test(&cio->refcount))
 		blkdev_copy_endio(cio);
 }
 
 /*
- * @bdev:	block device
+ * @bdev:	source block device
  * @pos_in:	source offset
  * @pos_out:	destination offset
  * @len:	length in bytes to be copied
@@ -196,6 +183,7 @@ static void blkdev_copy_offload_src_endio(struct bio *bio)
  * @private:	endio function will be called with this private data,
  *		for synchronous operation this should be NULL
  * @gfp_mask:	memory allocation flags (for bio_alloc)
+ * @bdev_out:	destination block device
  *
  * For synchronous operation returns the length of bytes copied or error
  * For asynchronous operation returns -EIOCBQUEUED or error
@@ -216,20 +204,38 @@ static void blkdev_copy_offload_src_endio(struct bio *bio)
 ssize_t blkdev_copy_offload(struct block_device *bdev, loff_t pos_in,
 			    loff_t pos_out, size_t len,
 			    void (*endio)(void *, int, ssize_t),
-			    void *private, gfp_t gfp)
+			    void *private, gfp_t gfp, struct block_device *bdev_out)
 {
 	struct blkdev_copy_io *cio;
 	struct blkdev_copy_offload_io *offload_io;
 	struct bio *src_bio, *dst_bio;
 	size_t rem, chunk;
-	size_t max_copy_bytes = bdev_max_copy_sectors(bdev) << SECTOR_SHIFT;
 	ssize_t ret;
 	struct blk_plug plug;
+	int is_mq = 0;
+	size_t max_copy_bytes = min(bdev_max_copy_sectors(bdev) << SECTOR_SHIFT,
+	    bdev_max_copy_sectors(bdev_out) << SECTOR_SHIFT);
 
 	if (!max_copy_bytes)
 		return -EOPNOTSUPP;
 
-	ret = blkdev_copy_sanity_check(bdev, pos_in, bdev, pos_out, len);
+	if (queue_is_mq(bdev->bd_queue)) {
+		if (bdev->bd_queue->mq_ops != bdev_out->bd_queue->mq_ops)
+			return -EOPNOTSUPP;
+		is_mq = 1;
+	} else if (!bdev->bd_disk->fops->submit_bio ||
+	    bdev->bd_disk->fops->submit_bio != bdev_out->bd_disk->fops->submit_bio) {
+			return -EOPNOTSUPP;
+	}
+
+	/*
+	 * Cross device copy only supported for zvols
+	 */
+	if (strncmp(bdev_out->bd_disk->disk_name, "zd", 2))
+		return -EOPNOTSUPP;
+
+	ret = blkdev_copy_sanity_check(bdev, pos_in, bdev_out, pos_out, len);
+
 	if (ret)
 		return ret;
 
@@ -258,25 +264,33 @@ ssize_t blkdev_copy_offload(struct block_device *bdev, loff_t pos_in,
 		 * successful copy length
 		 */
 		offload_io->offset = len - rem;
+		if (bdev_out)
+			offload_io->driver_private = bdev_out->bd_queue->queuedata;
 
 		dst_bio = bio_alloc(bdev, 0, REQ_OP_COPY_DST, gfp);
 		if (!dst_bio)
 			goto err_free_offload_io;
 		dst_bio->bi_iter.bi_size = chunk;
 		dst_bio->bi_iter.bi_sector = pos_out >> SECTOR_SHIFT;
 
-		blk_start_plug(&plug);
-		src_bio = blk_next_bio(dst_bio, bdev, 0, REQ_OP_COPY_SRC, gfp);
+		if (is_mq) {
+			blk_start_plug(&plug);
+			src_bio = blk_next_bio(dst_bio, bdev, 0, REQ_OP_COPY_SRC, gfp);
+		} else {
+			src_bio = bio_alloc(bdev, 0, REQ_OP_COPY_SRC, gfp);
+		}
 		if (!src_bio)
 			goto err_free_dst_bio;
 		src_bio->bi_iter.bi_size = chunk;
 		src_bio->bi_iter.bi_sector = pos_in >> SECTOR_SHIFT;
 		src_bio->bi_end_io = blkdev_copy_offload_src_endio;
 		src_bio->bi_private = offload_io;
+		offload_io->dst_bio = (is_mq) ? NULL : dst_bio;
 
 		atomic_inc(&cio->refcount);
 		submit_bio(src_bio);
-		blk_finish_plug(&plug);
+		if (is_mq)
+			blk_finish_plug(&plug);
 		pos_in += chunk;
 		pos_out += chunk;
 	}
@@ -289,6 +303,8 @@ ssize_t blkdev_copy_offload(struct block_device *bdev, loff_t pos_in,
 	return blkdev_copy_wait_for_completion_io(cio);
 
 err_free_dst_bio:
+	if (is_mq)
+		blk_finish_plug(&plug);
 	bio_put(dst_bio);
 err_free_offload_io:
 	kfree(offload_io);

block/fops.c

@@ -784,12 +784,12 @@ static ssize_t blkdev_copy_file_range(struct file *file_in, loff_t pos_in,
 	struct block_device *out_bdev = I_BDEV(bdev_file_inode(file_out));
 	ssize_t copied = 0;
 
-	if ((in_bdev == out_bdev) && bdev_max_copy_sectors(in_bdev) &&
+	if (bdev_max_copy_sectors(in_bdev) &&
 	    (file_in->f_iocb_flags & IOCB_DIRECT) &&
 	    (file_out->f_iocb_flags & IOCB_DIRECT)) {
 		copied = blkdev_copy_offload(in_bdev, pos_in, pos_out, len,
-					     NULL, NULL, GFP_KERNEL);
-		if (copied < 0)
+					     NULL, NULL, GFP_KERNEL, out_bdev);
+		if (copied != -EOPNOTSUPP && copied < 0)
 			copied = 0;
 	} else {
 		copied = splice_copy_file_range(file_in, pos_in + copied,

include/linux/blkdev.h

@@ -409,6 +409,24 @@ struct queue_limits {
 	struct blk_integrity	integrity;
 };
 
+/* Keeps track of all outstanding copy IO */
+struct blkdev_copy_io {
+	atomic_t refcount;
+	ssize_t copied;
+	int status;
+	struct task_struct *waiter;
+	void (*endio)(void *private, int status, ssize_t copied);
+	void *private;
+};
+
+/* Keeps track of single outstanding copy offload IO */
+struct blkdev_copy_offload_io {
+	struct bio *dst_bio;
+	void   *driver_private;
+	struct blkdev_copy_io *cio;
+	loff_t offset;
+};
+
 typedef int (*report_zones_cb)(struct blk_zone *zone, unsigned int idx,
 			       void *data);
 
@@ -1111,7 +1129,8 @@ int blkdev_issue_secure_erase(struct block_device *bdev, sector_t sector,
 ssize_t blkdev_copy_offload(struct block_device *bdev, loff_t pos_in,
 			    loff_t pos_out, size_t len,
 			    void (*endio)(void *, int, ssize_t),
-			    void *private, gfp_t gfp_mask);
+			    void *private, gfp_t gfp_mask,
+			    struct block_device *bdev_out);
 
 #define BLKDEV_ZERO_NOUNMAP	(1 << 0)  /* do not free blocks */
 #define BLKDEV_ZERO_NOFALLBACK	(1 << 1)  /* don't write explicit zeroes */
@@ -1307,8 +1326,8 @@ static inline unsigned int bdev_discard_granularity(struct block_device *bdev)
 	return bdev_get_queue(bdev)->limits.discard_granularity;
 }
 
-/* maximum copy offload length, this is set to 128MB based on current testing */
-#define BLK_COPY_MAX_BYTES		(1 << 27)
+/* Tested till 2GB with zvol testing */
+#define BLK_COPY_MAX_BYTES		(1U << 31)
 
 static inline unsigned int bdev_max_copy_sectors(struct block_device *bdev)
 {