rbd: don't use snapc->seq that way

In what appears to be an artifact of a different way of encoding
whether an rbd image maps a snapshot, __rbd_refresh_header() has
code that arranges to update the seq value in an rbd image's
snapshot context to point to the first entry in its snapshot
array if that's where it was pointing initially.

We now use rbd_dev->snap_id to record the snapshot id--using the
special value CEPH_NOSNAP to indicate the rbd_dev is not mapping a
snapshot at all.

There is therefore no need to check for this case, nor to update the
seq value, in __rbd_refresh_header().  Just preserve the seq value
that rbd_read_header() provides (which, at the moment, is nothing).

Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

drivers/block/rbd.c

@@ -1718,8 +1718,6 @@ static int __rbd_refresh_header(struct rbd_device *rbd_dev)
 {
 	int ret;
 	struct rbd_image_header h;
-	u64 snap_seq;
-	int follow_seq = 0;
 
 	ret = rbd_read_header(rbd_dev, &h);
 	if (ret < 0)
@@ -1735,13 +1733,6 @@ static int __rbd_refresh_header(struct rbd_device *rbd_dev)
 		set_capacity(rbd_dev->disk, size);
 	}
 
-	snap_seq = rbd_dev->header.snapc->seq;
-	if (rbd_dev->header.total_snaps &&
-	    rbd_dev->header.snapc->snaps[0] == snap_seq)
-		/* pointing at the head, will need to follow that
-		   if head moves */
-		follow_seq = 1;
-
 	/* rbd_dev->header.object_prefix shouldn't change */
 	kfree(rbd_dev->header.snap_sizes);
 	kfree(rbd_dev->header.snap_names);
@@ -1759,11 +1750,6 @@ static int __rbd_refresh_header(struct rbd_device *rbd_dev)
 	WARN_ON(strcmp(rbd_dev->header.object_prefix, h.object_prefix));
 	kfree(h.object_prefix);
 
-	if (follow_seq)
-		rbd_dev->header.snapc->seq = rbd_dev->header.snapc->snaps[0];
-	else
-		rbd_dev->header.snapc->seq = snap_seq;
-
 	ret = __rbd_init_snaps_header(rbd_dev);
 
 	up_write(&rbd_dev->header_rwsem);